SPENCER: Michael, welcome.

MICHAEL: Hi, Spencer, thanks for having me back.

SPENCER: So how does your approach to figuring out what charities are best to give differ from approaches that others take?

MICHAEL: So I try to take happiness seriously. I've been long-term interested in this utilitarian idea that we should try and create as much happiness as possible. I was looking at what was going on in the world of charities, and I thought, "Why is no one trying to measure the impact of charities on people's happiness?" And so that's what I set out to do.

SPENCER: Now people might think, "Isn't that what effective altruist charities try to do? They maximize total happiness, or isn't that what QALYs try to do?"

MICHAEL: That's what you would think. So what kind of got me going on this was that I'm a philosopher by background, but because I'm interested in this happiness-increasing project, I've also taken quite a bit of an interest in the social science research on happiness, where you're using self-reported measures like 0 to 10, how satisfied are you with your life, that sort of stuff. I started this six years ago while doing a PhD, and I was looking at what was going on in the effective altruist world. I was at Oxford, and back then, there was kind of a buzz about effective altruism. I thought, "Where's the happiness in this? QALYs and DALYs are fine, but surely we can do better. Why aren't we measuring what really matters?"

SPENCER: So how is it different from QALYs or DALYs?

MICHAEL: So the QALYs and DALYs — quality-adjusted life years and disability-adjusted life years — I expect your listeners know much about those, but in case they've momentarily forgotten, are measures of health that have been around for about 20 to 30 years. They haven't been with us since the dawn of time, but they've been going a while, and they've now become quite well-established. The difference between them isn't enormously important. But the main thing they share — in contrast to happiness measures — is that they ask people, typically, who don't have experience of the health conditions how bad they think those health conditions are. For example, they might be asked, "How would you rate this on a scale from zero to one?" Or, "If you had to choose between this many years being healthy, or this many years with the condition, how many would it be for you to feel indifferent between them?" So what's going on is that they are using people's imagination about how much they think something will affect them. Daniel Kahneman would have called this decision utility. What the subjective wellbeing stuff gets you — also in Kahneman's terminology — would be experience utility. It's not about how bad you think something will be, but instead, how you are actually feeling at the moment. How does it affect your life? You're trying to get at people's experience as they live, rather than their hot takes or guesses about how bad they think things will be. As you will know, Spencer, human psychology is a complicated beast, and people are not always very good at imagining how they will feel or how others will act. It's kind of marrying those up and realizing there's a difference.

SPENCER: So at a high level, if we say, "Okay, we want to give money to the charities that do the most good." Then you could say, "Okay, well, what do we mean by doing the most good?" Well, let's suppose we settled on "doing the most good is something like improving people's lives the most." Then it's like, "Okay, well, what does that mean to improve people's lives the most?" QALYs and DALYs are basically saying, "Well, how much do people say they want to avoid certain bad things?" You're saying, "No, let's actually look at people who have those bad things happen and see how much less satisfied they are with their lives compared to people that don't." Is that right?

MICHAEL: That's right. I'll give you a couple of examples. One thing that got me to really recognize, "Oh, there might be quite a difference here," was something I came across when I was doing my master's at the LSE, studying behavioral and wellbeing science. It was looking at time trade-offs for QALYs. If you ask people how much of your life would you trade-off to not have some difficulty walking, it was about 15%. If you ask people how much of your life would you trade off to not have moderate anxiety or depression, it was also about 15%. So people thought these things are as bad as each other, whereas you don't have to think that hard to realize, "Well, moderate anxiety or depression is actually going to make me much, much more miserable than having some mobility issues." Going back to Daniel Kahneman, this would be a failure of affective forecasting, affective emotions, and particularly focusing illusions. We're not that good at putting ourselves in other people's shoes. When you imagine what it's like to be, I don't know, Taylor Swift, you're thinking, "Oh, it must be so great to have all those people cheering you up there on that stage." So Taylor Swift must be having such a great life. But there's probably more to it. When Taylor Swift is having her coffee in the morning, her coffee still probably tastes much the same as yours. People overestimate, on the basis of these focusing illusions, how much difference there is between lives. I'll just mention this other bit, which is that when we do these prospective imaginations, we don't account for what we're going to get used to and what we're not, which is called immune neglect or duration neglect. There are a few kinks in how we imagine our lives compared to how they are that really motivate thinking, "Okay, let's look at the data and see how happy people really are."

SPENCER: And actually, people are pretty famously bad at forecasting how future events are going to affect them. There are studies where they ask, "How bad do you think it would be to have XYZ happen?" Then they look at how bad it actually is to have XYZ happen. Usually, it's much less bad than people estimate because they take however bad they'll feel right away when that thing happens and extend it forward, rather than taking into account the hedonic treadmill and the fact that they'll get used to it significantly.

MICHAEL: Yeah. There are some experiments done by Dan Gilbert and Timothy Wilson. They ask people questions like, "How bad would you feel if your football team loses?" Then they ask people the week after the match has happened, and people feel much less bad than they say they would feel. That's simply because, going back to Daniel Kahneman again — I'm a bit of a broken record — he's got a great line, which is, "Nothing matters as much as you think it does while you're thinking about it." What happens with affective forecasting is that people get the direction right; if you get a promotion, you will feel happier, and if you get dumped, you will feel less happy. But what we get wrong is the duration and the intensity. Basically, things aren't as bad for as long as we expect, and they're not normally as bad as we expect. We think the present is going to be more like the future, and actually, we get used to things. That works the other way around, too. Things aren't as good as we expect, or for as long as we expect. What motivated the kind of work we do is thinking, "Okay, if this is true, then we should probably really focus on the things that are really bad — the kinds of things people don't get used to, like mental illness and pain, where there's internal suffering." We're not very good at imagining what those experiences are like unless we've lived through them — and even then, we tend to forget afterward. So we thought: let's look at those cases and see, "How bad are they really, rather than how bad do we imagine they are?" That was part of the original motivation behind the direction we took. Not just, "Okay, we're measuring happiness — but so what?" but also the idea that we're predictably bad in our predictions about which things will be good or bad for our happiness. And that's going to give us different priorities.

SPENCER: One way I see your approach really differing from other approaches is that things you don't get used to will tend to have much stronger weight in your approach. You mentioned depression. As people have depression longer, they don't necessarily feel better just because they've had depression longer.

MICHAEL: Yeah, that's right. To motivate where you're going to find things that are really important to focus on for charities, this is going to generalize to policies. You've got to think, "What's actually going to affect people's lives as they live moment by moment?" rather than just painting a quick picture in your head.

SPENCER: On that point, it raises a question of how you actually ask people how happy they are. For example, if you ask, "All things considered, how satisfied are you with your life?" — that's a common life satisfaction question. Or you might use what's called Cantril's ladder: "Look at the rungs of this ladder from one to ten, where ten is the best possible life for you and one is the worst possible life. Where would you place yourself?" That might actually weigh things somewhat differently than if you ask, "Okay, at this moment, how happy are you?" or "Today, how happy were you?"

MICHAEL: Yeah. So we've actually been discussing this internally at the Happier Lives Institute. It's a conversation we have basically every six months. When people measure happiness or wellbeing, what they're really measuring is typically these life satisfaction questions. If you're looking at measuring wellbeing, the basic important distinction is between these evaluative measures of wellbeing — how good you think your life is going — and experiential measures of wellbeing, which to the man on the street is what you would call happiness. That's how you feel moment by moment. There's a lot more experience in life satisfaction where you just have these quick measures, whereas it's much harder and costs more money to collect data on people's happiness moment by moment. There is actually not a lot of that stuff. So in practice, we have to rely on life satisfaction measures. But we also look at things like mental health scores — the kind of questions your doctor might ask you if you came in feeling a bit low — to help collect or use interpret data. Sadly, we're not in a world where ideal data exists for these sorts of assessments.

SPENCER: But what would the ideal data from your point of view be?

MICHAEL: Yeah, so I think this isn't a controversial view, but if you're trying to measure experienced wellbeing, aka happiness, then the kind of thing you'd want would be called the experience sampling method. You'd ping someone on their smartphone. It used to be done on pagers a few years ago. You would ask, "How happy do you feel right now? Or how do you feel from very good to very bad?" Then there are no problems with recall or how happy do you think you should feel? That's really capturing it in the moment. It has the highest ecological validity. As you can guess, you have to collect that kind of information from people, so often it's not what's used.

SPENCER: Some people might argue that may not actually be what truly matters the most, that it might be better to ask about life satisfaction, because people don't just care about their moment-to-moment happiness. If someone's imagining having trouble walking, they might think about the effects of that on their day-to-day mood or happiness. But they also might think, "Well, I'm not going to achieve this goal that I have," and that might matter to them independently of that.

MICHAEL: So there is a philosophical choice here. Do you think what matters is life evaluations, or do you think what matters is experiences? Talking to my colleagues who work on wellbeing, I occasionally complain, sometimes politely and sometimes less politely, that I think they're a bit philosophically naive in why they focus on life satisfaction rather than happiness. I've got a paper forthcoming in economics and philosophy, where I say, "Okay, if we really took life satisfaction seriously, what sort of things would that imply?" One oddness of this is, if you think what really matters is how you judge your life, then what are we going to say about animals? If your life goes well, to the extent that you judge it to go well, animals can't judge how their lives are going at all. It seems like the conclusion of the view that says what really matters is how you judge your life is that animals, children, and people with cognitive disabilities just aren't welfare subjects; they don't matter. Nothing can be good or bad for them. If you take your dog and set it on fire, that wasn't bad for the dog, because nothing can be good or bad for the dog, because it can't judge its life as going well or badly. I think that's a very peculiar kind of implication. That's one reason why I see the motivation for life satisfaction, but actually, I think we probably do really want to care about people feeling bad, not just thinking their lives are going badly.

SPENCER: A reasonable pushback on that point could be that it's not that life satisfaction is the fundamental thing we care about and that's all we care about. It's that there's a bundle of values that people care about. Life satisfaction somehow combines these values. People might actually value the wellbeing of themselves, their pets, and their friends, et cetera, which I think they do, but it's not all they value. They value more than that.

MICHAEL: Well, in which case you'd want to ask the question, what are the things that ultimately matter? There have been some studies on this. When you ask people, "What is it you really care about?" People do mostly care about their happiness, how good or bad they feel. Then we get back to the thing we were talking about a moment ago, which is that people do seem to actually care about their experience of life. But how good are we at predicting and forecasting and making good decisions about what is going to make us happier and what isn't? People do care about their experiences as a really big thing, even if not maybe the only thing, but we can still make errors in trying to find it for ourselves and others.

SPENCER: What is the research that suggests that people mostly only care about how good or bad they feel? That surprises me that you would say that.

MICHAEL: Asking people how much they care about happiness, how much they care about health, and then some other things, and asking which of these they think really matter. But yeah, there is this kind of literally asking people how much they care about happiness. People do seem to care about happiness as what they desire. In some sense, you don't really need to conduct an academic study to know that. People do care about how they feel, and they care about it for others. What do parents say? I just want them to be happy. What does the Declaration of the US Constitution say? Life, liberty, and the pursuit of happiness. If you were casting a play for what matters, you'd have happiness as one of the central characters, even if it's not a one-man play.

SPENCER: Right. When we study people's intrinsic values, where we ask people questions or we teach them about what an intrinsic value is. An intrinsic value is something you value for its own sake, not as a means to an end. We teach them about instrumental values — things you value just as an end to get something else. Then we have them submit what they think their intrinsic values are. Then we developed a test from that, which is our Intrinsic Values Test. We find that the statement "I feel happy" is a really common intrinsic value people report, and they often report it as one of their most important intrinsic values. But I'll just point out that there are other ones that are equally often, or sometimes even more often reported. One of them is that "I believe in truth rather than false things." Another is that "I have agency and can make choices for myself." Another is that "I avoid taking actions that significantly harm others." I do think there's a cluster of different things that people tend to value beyond that.

MICHAEL: I guess it depends on how much you want to get into this. The business of philosophy involves thinking about what sort of intuitions we have and then making greater sense of those. There's a question of wellbeing: what makes my life go well for me? And there's a question of morality: what ought I do overall? Some of the things you mentioned could be, "Okay. I think I might say that I value truth intrinsically, but it seems more plausible to me to say that I value truth as a means to an end." I recognize that people may state these things, but that doesn't persuade me that I should change my views on what matters, because you can provide a further explanation as to how this fits together.

SPENCER: It's more convincing if they're willing to sacrifice some of their happiness to get more truth, and if they actually take steps to do that, right?

MICHAEL: I guess, yeah.

SPENCER: One thing I think is kind of interesting — we've got this site called personalitymap.io, where you can look up over a million correlations. I just looked up the correlation between the following two state questions. The first question is: "Right now, at this very moment, how happy rather than unhappy do you feel?" So that's literally asking what you're feeling in the moment. The other one is: "To what extent do you agree with, 'I am satisfied with my life'?" And interestingly, they're really strongly correlated. The empirical correlation is 0.67. It's very interesting because in theory, you'd expect "I am satisfied with my life" would pack in all this other stuff, but it turns out it's very heavily correlated with just how good you feel at this moment.

MICHAEL: Yeah. Part of the explanation there is that there's this kind of discussion in academia that your life satisfaction isn't so bad. When you ask someone how satisfied they are with their life, a big component to that is going to be, "Okay, when you ask me that question, I'm going to ask myself this other question, which is, how happy do I feel?" So that's the plausible psychological mechanism there. But then, there are bits of research looking at the experience data and the life satisfaction data, and it's kind of an open question. These don't say the same thing, even if they often point in the same direction because we don't have lots of really good experience sampling data. It remains an open question as to if you really care about life satisfaction, how far off are you? I can give you an example here, which is that the World Happiness Report uses a measure of life satisfaction. More technically, it's the Cantril's ladder, which is of the same family. You could somewhat meanly say it's the World's Smokingness Report because it's asking people about their life satisfaction rather than how good or bad they feel. But we don't actually have good international measures of happiness, like experiential wellbeing. Something that people often say when the World Happiness Report comes out is that some journalists go to Finland and say, "Oh, these people don't look very happy." On my travels around Africa and Central America, the people do seem happier. They have more positive affect than you seem to encounter in Northern Europe. It wouldn't surprise me if we had a really thorough World Happiness Report that looked at experiential happiness rather than evaluative wellbeing, and that these might come apart in an interesting way. Maybe we would find that Latin America was actually winning the happiness race, even if it's not winning the life satisfaction race.

SPENCER: That can help explain people's observations about, "Why do some countries not seem to smile very much, or they don't seem to show much on their face." You have countries like Finland that do really well, but you kind of look around and think, it's dark a lot of the year, and people don't smile very much. What's going on? Why do they win all the happiness reports? It is kind of funny because I understand why people say it's really hard to measure experiential happiness because you think, "Well, you've got to ping people multiple times a day randomly and ask them how they're feeling at that moment." A cheap proxy of that is, let's say you're doing a survey in some country. You just literally ask them, "Okay, right now, at this very moment, how happy do you feel?" You just do that one time, and it's interesting. It can serve as a proxy for the experience sampling, even though it's not randomly sampling throughout the day.

MICHAEL: Yeah. But I guess you'd want more information over time to see how satisfied you are with your life and how that relates to how happy you are moment by moment in the last week. If you're being kind of a purist about it, which sometimes we should be, the concern is that people's memories can play tricks on us to some extent. This goes back to Daniel Kahneman and remembered utility, contrasted with decision utility and experience utility. You might be missing some of the granularity on people's experience at the time, which they actually don't properly remember.

SPENCER: Okay, so suppose the listener is sold on the idea that we should be giving money to try to maximize the moment-to-moment happiness of other beings, or at least that some of our money should go towards that, even if it's not the only thing we donate to. How do you then approach turning that into a set of recommendations of what to give to?

MICHAEL: Yeah. Okay, well, I started the Happier Lives Institute six years ago, thinking, "Why is no one assessing charities in terms of their people's happiness?" I had been moaning about this online and saying, "Yes, I want you to do this." Eventually, someone said to me, "Michael, what are you going to do about this?" I thought, "What do you mean? I've moaned about it online. What else could I do?" I realized, "Okay, fine, I should do something." At the time, no one had done this before. There had been decades of research into happiness using happiness surveys, and there was the kind of happiness is correlated with this and this, and so on. What hadn't happened is happiness cost-effectiveness, or happiness return on investment, where you look at individual charities, interventions, and policies, and then you say, "Okay, how much bang for your buck do you get?" People thought it was strange when we started, which was somewhat understandable, because it hadn't happened before. But now, actually, quite a lot has happened in this field in the past few years. It has gone from no one having done it to us just having some global comparison, the first-ever global comparison of charities published in The World Happiness Report. There are a few other organizations that do this for charities. There is now some work being done on policies to determine which policies will give you the most wellbeing for your money. Oddly enough, the UK is a sort of world leader in happiness research, not actual happiness, but in efforts to conduct happiness research in terms of happiness cost-effectiveness. Basically, it all seems to be coming out of the UK.

SPENCER: Interesting. Do you have any theories why that is? Is it because of you, or is it just more general?

MICHAEL: I don't know. I'm kind of curious about this. I'm hoping I can find some historians to look into it. It could just go back to Bentham. The English have this sort of utilitarian tradition, Jeremy Bentham, John Stuart Mill, and it has sort of lurked in the shadows. There is also a professor at the London School of Economics called Richard Layard, who, full disclosure, is a member of the Board of Trustees of the Happier Lives Institute, and he has been banging this drum in the UK for about 50 years. He has probably had a big impact in causing it to happen here first. The UK doesn't feel like it's ahead of very much, but it seems to be ahead of the world in happiness research. So, Spencer, the British are coming, and they want you to be happier, and you should be terrified.

SPENCER: It doesn't sound so scary to me [laughs]. Okay, so you were trying to evaluate how effective different charities are from this perspective. What does that research process actually look like? How do you actually make that decision?

MICHAEL: So I tell you about our unit of measurement, which is called the wellbeing life year or WELLBY. If I ask you, "How satisfied are you with life, Spencer, from 0 to 10?" And you say seven, and then you go up by one point for one year, that is one WELLBY, one wellbeing year. So yeah, when we started, that's how we do cost-effectiveness and your wellbeings per dollar. When we began, this was a blank map. We thought, "Okay, where should we start?" So we started with cash transfers to low-income countries, a thousand dollars to families on a couple of dollars a day. We did a big systematic review and meta-analysis of that; that was the first point on the map. Since then, we've looked at other things like deworming and treating mental health with group therapy. We did some work comparing improving lives to saving lives, looking at the anti-malarial bed nets, which have a whole host of other complexities. Now we've looked at things like advocacy to take lead out of cosmetics, acute malnutrition, and iron fortification. If you think of it like a jigsaw board — when we started, there was nothing there. Now we've got a few bits coming together. What we did in the World Happiness Report, which came out two months ago, was we looked at research not just from our own work but also from research other people had done using this same approach.

SPENCER: What is the data that you're basing these conclusions on? Are there actually studies of these interventions where they measured people's happiness before and after, and you can compare the change in a control group versus the group that got the intervention? Or do you have to kind of piece it together from much more auxiliary information?

MICHAEL: Yeah. It's a combination. There are randomized control trials, where one group gets a cash transfer and the other group doesn't. You look at some self-reported measures over time to see how much happier people get. There are also things like monitoring evaluation data from the charities, using different sorts of studies, even if they're not all randomized control trials, to try and establish the causal correlation effect and piece things together to get some estimates.

SPENCER: And so as you go through this space and try to figure out how happiness affects these different interventions, what is the general pattern you're seeing and how it differs from, let's say, if you were prioritizing based on QALYs or DALYs? Are you noticing that the different approaches actually end up recommending different sources of charities?

MICHAEL: Yeah. So this is kind of an apples to oranges comparison. So, okay, I'll give you some problems. Let's say you thought, "Okay, I don't really know how QALYs work, but other people seem to like them, so let's go with them." Well, then these are only limited to health interventions. So if you're looking at the effects of poverty, you probably would think measuring them on health isn't really capturing that, or you're looking at educational outcomes.

SPENCER: Or this presumes an answer shorter, because that's all you can measure.

MICHAEL: So it's like a round peg, square hole, so it doesn't allow you to capture the full spectrum of life. And we've already touched on the differences between people's expectations and their reality. So people often ask me, "Okay, what's your translation grid? Or how do you go from a QALY to a WELLBY?" Part of the motivation for the research is that it's not that simple, because to work it out, you've got to look at how people are doing the QALY or DALY numbers: "How bad is blindness?" or "How bad is cancer?" And then you've got to get some data on people's happiness, as to how much better or worse it makes them as things happen. And so to meld them together, there's no kind of even mapping. It's going to depend upon how far off people's effective forecasts are. And you've got to have the data on the other thing to check. So there isn't a sort of simple answer. What we're doing is thinking, "Well, all this quality-adjusted life year stuff is fine, but we can do better. It gets us closer to what really matters and allows us to have more universal comparisons. So let's just get on with finding out how good and bad things are in terms of WELLBYs."

SPENCER: What about compared to traditional effective altruists' favorite charities? Are there some of those that actually don't look very good from a WELLBY point of view?

MICHAEL: As I said, it's kind of hard to do the comparison. The people who you would have thought were the most relevant here would be the work done by GiveWell, but GiveWell doesn't quite do things in terms of QALYs and DALYs. They have their own system of value where they're looking at lives saved and some sort of trade-off rate for the value of doubled income. What we did earlier on was say, "Okay, let's look at GiveWell's model and then what happens if you do it in terms of wellbeing?" But actually, there aren't very many points of matching. So let's say you wanted to compare cash transfers to treating depression. Treating depression wasn't in the GiveWell schema, so I couldn't tell you what their ratio was compared to ours. Then we looked at deworming interventions compared to those other two, but GiveWell's model of deworming was based on income, not on DALYs. You have deworming, and then later on, your income improves. We looked at the deworming stuff in terms of the wellbeing data. There's one deworming study people rely on, this kind of 20-year follow-up in Kenya, there's KLPS. If you look at the wellbeing improvements, they're non-significant. If you take them at face value, they go down over time. So there doesn't seem to be evidence we can rely on, so then we don't get the GiveWell used to say, which is that deworming is 10 to 20 times better than cash transfers.

SPENCER: I see. So deworming may actually look really bad from a WELLBY perspective, at least given the data we have, even though it's been a top-rated charity by GiveWell for a long time.

MICHAEL: GiveWell has since downgraded their estimation of that one. I'm not a GiveWell scholar, nor do I want to be. There isn't and wasn't a charity ranking based on QALYs, DALYs, and now one based on WELLBYs. You need the ranking, or even better, the cardinal cost-effective numbers on both sides. There just aren't very many places where you have both of them. I guess I just don't worry about it that much and get on with doing the thing, which I think is getting us closer to what really matters.

SPENCER: So tell us more about what you found in the World Happiness Report.

MICHAEL: So as I mentioned, we've been doing this for six years, and we've been focusing on charity interventions that we think are going to be the most effective. We've been looking in the Global South. The World Happiness Report is famous for its inter-country rankings; the Finns are at the top — we were just discussing that before — they've kind of branched out and are trying to get some more interesting happiness research. We thought, "Okay, cool." We're vaguely aware there are other people doing charity work in terms of WELLBYs, and they're based in the UK, but we don't really know what's going on. So we decided to do a global review of the evidence. So that's what we did. We looked at all of the published estimates of charities in terms of wellbeing per dollar. In the UK, it's wellbeing per pound, and then you convert them into wellbeing per dollar. That included about 20 charities, some from us and some from three other organizations: State of Life, Pro Bono Economics, and a group at the LSE. We put them together, did a review of the existing evidence, and our big, important finding was that the difference in cost-effectiveness, in terms of how much happiness charities create per dollar, was enormous. The differences between the best and worst charities were hundreds, even thousands of times when comparing the best charities to some charities working in the UK. I found that it's not the case, which is what people believe, that some charities are about as good as each other; there are really enormous differences. Using the wellbeing approach, we're able to shine a database light on this.

SPENCER: So it turns out, from a wellbeing perspective, it could matter tremendously which of these charities you give to. It's not just a small factor.

MICHAEL: Yeah. Something that effective altruists have said is that the best charity is a hundred times better than the average charity. When we were looking at this, we were like, "Well, what is this actually based on?" Part of what it had been based on is work that Toby Ord had done looking at the cost-effectiveness in DALYs of different health interventions. But this is health interventions focused on international health, specifically in the Global South, and that's not the same as looking at charities. To say the best charities are a hundred times better than the average charities, you actually need to find some average charities to base that on. Who knows if the health interventions are going to be the same as charities? It turns out that there wasn't anything that really looked at a sample of charities from around the world, had some best examples, and then got an average to say, "Okay, here's the difference between the best and the average." What we did is provide the strongest evidence so far, indeed, maybe the only evidence so far, about the difference in cost-effectiveness between the average charity and the top charity. We assessed UK charities and found that the top five were, on average, about 150 times more effective than the typical charity. That's our go-to figure for the difference between the best and the average.

SPENCER: This may be just the mathematician in me speaking, but I wonder how sensible it is to think about the ratio of the averages. It could be that the average charity does no good, and then you're going to get infinity as your ratio. Do you think that thinking in that ratio actually makes sense?

MICHAEL: In some sense, it's a bit silly, but I think we're building on the existing claim that thought it wasn't that silly.

SPENCER: And you're saying that at least through this perspective, you did really good [inaudible].

MICHAEL: Yeah, in some sense, the claim that the best things are going to be loads better than the average things is kind of trivially true if you think the average things don't really do anything. In the stuff we looked at, we didn't have any charities that had negative or zero effects. We did have some with quite small effects, including some household name charities, which maybe I can come to. The stuff that was in our sample were charities that had agreed to have themselves assessed in terms of this new, weird method. So these are probably much better than your truly average charities that are not engaged in the world of research. Our number is probably an underestimate if you're thinking that way.

SPENCER: I don't know if you know this, but I actually published a paper with some colleagues, Lucius Caviola and Stefan Schubert, about how people view the average charity versus the best charity. We found about a 1.5 times difference where people thought that the best charities were about one and a half times better than the average charity, which seems, by your research, just sort of, at least from your perspective of using wellbeing, is vastly off. But that could help explain, in part, why people don't care that much about cause prioritization or charity prioritization, because they don't think there's that much difference.

MICHAEL: Yeah, that's right. So you have to correct me if I got this wrong. But in that paper, that was the difference between charities helping people in poverty. We thought, "Okay, we kind of want to know what the difference is, not just in that particular subset, but across all charities." We did a recent survey with a representative sample of 2,000 Americans. We asked, "What do you think the difference is between the best charity and the average charity, looking at how much happiness they create per dollar?" That wasn't restricted to just poverty charities. You may be interested to know that we found the difference was three. People thought the top charities, asking this kind of general question, were three times better than the average charity. It only goes up by a factor of two if you spread it across all charities, rather than just looking at the global poverty charities, which implies that people think the global poverty charities are only slightly better than domestic charities.

SPENCER: And it seems like there's a lot of evidence against that, given how much easier it is to help people who are in extreme poverty and how much further money goes in some countries compared to others.

MICHAEL: Yeah. So I think this is part of the story. It's sort of understandable. Because you and I have been soaked in this research, there is a plausible belief that there are quite big differences. But unless someone has kind of stuffed this in front of your face, "Well, it doesn't really matter. Do I care about charities? Maybe I care about them a bit. But they all look kind of much of a muchness. I'm not going to go and try to work out the details, and if I did, it probably wouldn't make much difference anyway." So, I'm hoping it's the case, and indeed it is somewhat the case that people who encounter this kind of stronger evidence-based comparison method, comparing charities, where we can say, "Look, there are really big differences, we're hoping that motivates people to take effective giving more seriously. We can really show how big the differences are now that we found that they are big."

SPENCER: Yeah, I wonder if it also, in part, reflects a kind of worldview, where, imagine that you ask people how big are the differences in different products. If you're going to buy a new speaker or a new computer, I think most people would say there's a pretty big difference. You can get quite different results. Some things could be a scam; some things could be really excellent. You would think that that would carry over into thinking about charities, but maybe there's a sense people have that charities are a different kind of class of thing where there aren't big differences between them. I don't know why that would be.

MICHAEL: Yeah, so I think about that in kind of the other way. What we experience in our normal lives as consumers is that you get what you pay for. If you go to a fancy restaurant rather than a basic restaurant, you're going to pay twice as much, but you don't expect the food to be noticeably nicer. That kind of makes sense. There's a market mechanism, so that stuff which is expensive but people don't think is worthwhile, those businesses don't keep existing; the price signal takes care of that. Interestingly, there isn't the same mechanism in charities, because until you have research like ours, you really have no idea what you're paying for. If you're working on the assumption of a £10 burger, you're looking at different restaurants. This place charges £10 for a burger; this place charges £10 for a burger. The sensible, intuitive thing would be to think, "Okay, they're probably going to have the same tastiness," but that's because you have competition. You would extrapolate that and think, "Well, my £10 here to Save the Children does the same as £10 to this other charity." But actually, there's a really big difference because there isn't this kind of market mechanism and competition, and because people don't know what the impacts are.

SPENCER: People might assume, well, if there really were charities that were that much better, people would just give to them, and the information would get out there somehow. In capital markets, if there really was a way to make that much more money doing one thing than another, there would be some greedy capitalist who would do that.

MICHAEL: In trying to explain what we do, one analogy we actually use in the World Happiness Report is, imagine you were buying something and you could make a 20% saving. You'd think, "Ahh that's an amazing deal." You'd go tell your friends about it. But the situation is so weird for charities that I can't even do the percentages in my head, but it's not like 0.2% better. It's like 150 times better. In our normal life, we never encounter a scenario where you can buy one burger for $10, but you can buy exactly as good a burger for one cent. That world doesn't exist.

SPENCER: That would definitely be a scam in the normal world.

MICHAEL: Too good to be true. But that is, in fact, what goes on. Before the wellbeing stuff, it was actually very hard to get an impartial, evidence-based way to compare charities. If you're using income and DALYs and lifesaving, these are all somewhat apples to oranges to pears. Before the kind of work we were doing, it was hard to have a single compelling measure that allows you to see the meaningful differences. With these meaningful differences being very large, we can actually go and say to people, "You wouldn't have expected this, but it's not too good to be true. It really is true. This means you can have a hugely greater impact at no greater cost to yourself."

SPENCER: Even if someone thinks that wellbeing is not the only thing that matters, at least it is something that matters to lots of people. Almost everyone, it matters to some extent to anyone who's altruistic; they want to improve the wellbeing of people, and at least it gives you a common yardstick you can apply to almost any charity whatsoever, unlike DALYs.

MICHAEL: Yeah, that's right. And then, This is, maybe for nerds like you and I, this is kind of interesting, but for the regular person wanting to have an impact, maybe not so much. If you're really thinking that you're a pluralist and you don't just care about people's wellbeing, maybe you care about distributional goods. Or maybe you care about autonomy in addition to wellbeing. What you would want to do if you were trying to be true to your ethics is think, "Okay, what's the autonomy component of these charities? What's the wellbeing component?" And then you'd sort of combine those in some way, but you would still want to account for how miserable people are and how much happier they can be. So it's like the wellbeing stuff is something that anyone that wants to do evidence-based impact should want to include somewhere, even if it's not the whole story of how they think about things.

SPENCER: Now, I imagine a point of contention you might get sometimes is that people might say, "Well, but are these scales really reliable?" If you ask people about how satisfied they are with their life, "What does that really mean?" And additionally, can we really say that improving one point on that scale really means the same thing? If you're doing an intervention in Japan versus in Brazil and you get a one-point improvement, can we really compare those to each other?

MICHAEL: So that does come up quite a bit. We could do a whole podcast on this topic, and in fact, we almost did three years ago where I think this topic came up as well. So, yeah, one response is how bad do you think this is? So I kind of have conversations with people who just think the idea that you can measure happiness is totally baffling. And then, they will look up a restaurant and be like, "Oh, I want to go to that one because it's got five stars, but I don't want to go to that one because it's got three stars." And you might think, "Well, hold on. We use these subjective scales for everything all of the time, even more so in the modern day. And we really think they contain useful information, even if we don't think they're perfect." And, what's the alternative? You're just not going to use this information at all? You're going to use something else that isn't capturing people's experience?

SPENCER: We don't have a device that measures how happy people are in some objective way. People get really attached to these ideas of objective measurement. But I just don't think it's feasible with today's technology.

MICHAEL: No. Sometimes you get people saying, "Could we have an objective measure of happiness?" If you get people to think about it, "Okay, well, let's say you want to look at brain states. How are you going to know which brain states are the happy brain states and which are the unhappy brain states?" You're going to have to ask people; you're going to need some input because no one else knows. There's been a discussion recently about pain scales. Let's have an objective measure of pain. Do you know how they train their objective measure of pain? Through self-report. But as maybe a slightly more conciliatory thing to say. In ordinary language, we use words like good, bad, excellent, and they share a common meaning. If I said that was an excellent meal versus that was an awful meal, you would understand what I said. The raison d'être of language is to be able to communicate. There's really nothing particularly different about using verbal labels like good, bad, ugly compared to using numbers on scales. If you think, "Oh, the person who says I don't trust ratings from other people or I can't use ratings," would just not be able to function in society. They are a totally ingrained, important part of language. The reason we end up with the same numbers is we talk to each other. We say, "Well, that's not good. That's okay." We have this metalinguistic negotiation about what's an appropriate meaning of words.

SPENCER: It does seem reasonable to be concerned that different cultures might use words differently. Do you feel there's a way to address that? Or is the best we can do right now to assume that cultures use them similarly enough that we can translate?

MICHAEL: I think the most obvious test of this is actually with something like the data from the World Happiness Report. Countries that score high on life satisfaction are places like Finland and Denmark; they're about an eight. China's around five and a half, Malaysia is similar. At the bottom of the scale, you have places like Zimbabwe, Iraq, and Afghanistan, which have scores around two or three. That seems like a plausible ordering; there's a shared sense of good and bad lives, and people are using similar reference points for their scales. Conversely, imagine that the Finns said they were one out of ten, and the people in Afghanistan said they were nine out of ten. You'd think, "Oh, maybe there's something weird going on here." But it looks like it's in the right sort of place. Safe, wealthy developed countries are more satisfied, whereas less safe, war-torn, fragile countries are less satisfied. If you're thinking about these things as life satisfaction and other scales as measuring sticks for feelings, the question is, are they all straight and the same length, or are they a bit wonky? It doesn't seem like they're extremely wonky. They might be a bit wonky, but they do allow us to measure something that really matters. So they seem okay. And what are we going to use instead?

SPENCER: What would you like to see happen in terms of improving measurement of WELLBYs?

MICHAEL: What I would like to see happen, there is work I was discussing with some of my colleagues today about trying to use various techniques to accommodate or correct for differences in scale use. One method is vignettes. The person would be asked, "How satisfied are you with your life?" Then you'd give someone descriptions of various other people, such as, "This person has this income and this job, and they have these health conditions. Okay, how satisfied do you think they are?" Then you compare people's own ratings to the ratings they give in vignettes. Using this approach, you can try to see and correct for scale use. There is quite a bit of discussion about this that goes on among wellbeing researchers, how can you check for these sorts of things?

SPENCER: So it sort of anchors people to, "Okay, a five means this kind of thing, and a seven means this kind of thing." Is that the concept behind that?

MICHAEL: Yeah. They are actually called anchoring vignettes. This is one approach that is underway.

SPENCER: So getting really concrete, what are some of the top recommended charities according to your approach?

MICHAEL: Yeah, cool. So I'll give you some of the top ones. The thing that came out at the top in the World Happiness Report was this organization called Pure Earth. We looked at one of their programs, which was doing advocacy to take lead out of cosmetics in Sub-Saharan Africa. I think your listeners probably know that the metal lead is dangerous. That's why we don't put it into our cars — or vehicles, as you call them across the pond. We think that one looks good. Taking lead out of cosmetics affects people's brain development, particularly in children. Another organization is Time Marker, which addresses acute malnutrition for children by providing food supplements in early life, and that has long-term effects. There's no deep mystery about malnutrition, but we found some evidence suggesting the long-term effects are surprisingly durable.

SPENCER: Taking those two as examples, so those are things where lead poisoning or malnutrition could affect you your whole life. So it's kind of cumulative.

MICHAEL: One thing that we were recommending a couple of years ago as our top charities are treatments for depression, organizations like Strong Minds and Friendship Bench. Those are still on the HLI website as our top charities because they are very cost-effective, and there's lots of evidence behind them. There are also a couple of charities we call promising charities, which we think could be more cost-effective, but the evidence base is weaker. Having looked at treating depression in adults, we thought, "Well, what about effects on children? If something is bad for you early in life and it affects you long-term, then maybe that's going to be more cost-effective." That's what got us looking at lead and malnutrition. If your listeners are interested, we write incredibly detailed reports, pretty technical reports, where we explain how we got the evidence and how we do that. These are all on our website.

SPENCER: So we'll put that in the show notes for those who want to check it out.

MICHAEL: Yeah. So then you can kind of see how we come up with the numbers, and some of it's quite amazing what you can find estimates for if you're digging hard enough. So my team does a really good job working that out. And yeah, we do think these long-term effects, these early life effects, can be pretty substantial and therefore cost-effective, and that's why they end up at the top.

SPENCER: What's the mechanism there? Is it people actually feeling a little bit less happy moment to moment because of early childhood lead poisoning or malnutrition? Or is it more about the effects on the choices they're able to make, or the kinds of lives they lead?

MICHAEL: We're looking at mental health scores for these. So if you want the sort of mechanism for malnutrition, you can think of brain development. I would actually have to refer you to the report for the causal story there. I can't quite remember how deep we get into it.

SPENCER: But it affects mental health, basically?

MICHAEL: Yeah. We pick up the effects of mental health, but exactly what the causal pathway is, how malnutrition affects what is happening in your life subsequently, is not a question we've tried to map.

SPENCER: You don't need to answer that. [inaudible]

MICHAEL: And we're interested in what works and what doesn't, whereas the question of why things work and why they don't is a bit less essential than what happens inside the black box. But I think we discussed these in our reports. With lead, there's preferential binding to bits of the brain, and there's an established literature on the effects of lead on IQ. So, yeah, you don't have to have a strong story as to why being poisoned can make your life worse later on.

SPENCER: So some of the recommended charities are kind of that nature, and then you have some that are more direct mental health treatments, like Friendship Bench and Strong Minds. So what's the idea there? What's your view on why they rank really well?

MICHAEL: So, yeah, some people think that we focus on mental health, and that's all we care about. And that's not really true. I mean, what we care about is wellbeing, understood as happiness. There are many things that are good and bad for your happiness. It's not just mental health, but we do think that interventions which directly target mental health, like therapy for depression, which is what Strong Minds and Friendship Bench do, turn out to look really good. But for all of the things we do, we're picking up their internal, subjective psychological impact. Even if you wouldn't categorize treating malnutrition or advocacy for lead as mental health interventions, we are looking at their impacts on people's wellbeing. So that's our common currency.

SPENCER: And my understanding is that depression and anxiety have a shockingly high impact on how happy people are.

MICHAEL: They do. They are amongst the worst of life's common misfortunes. If you're comparing being depressed or anxious to other health conditions or the effects of unemployment, depression and anxiety are right at the top in terms of their reductions in life satisfaction.

SPENCER: Another category of top charities I see in your report are ones that are more about quality of life. So you've got EarthEnable, which I guess is about upgrading people's flooring, and then Give Directly, also on there, where you give money to people in very low-income regions. What do you think the impact is there? Is it that people are just able to use those upgrades to have a better daily quality of life, that life is less unpleasant?

MICHAEL: Yeah. So with EarthEnable, the story is that it provides concrete flooring. This isn't right at the top of what we have in the World Happiness Report, and I would strongly encourage all of your listeners to go and read the chapter. It's chapter eight of the latest World Happiness Report. The story with EarthEnable and its concrete flooring is that people going from dirt to concrete experience quite a lot of effects mediated through health. Because you can clean the floors, people are less ill, and that impacts how they feel about their lives, also serving as a point of pride. We saw that there were really big impacts from concrete flooring, and then we thought, "Okay, why are the impacts so large?" I won't say we've fully got to the bottom of that one, but that has stuck out as a surprise, even if it doesn't turn out to be amongst the most cost-effective things found so far.

SPENCER: Now, a type of charity that stands out to me as missing from this list is ones around malaria, because if you look at GiveWell's top charities, there's a bunch of malaria-related stuff. Do you have an intuition on why those don't make the cut here?

MICHAEL: Yeah. In what we did in the World Happiness Report, we looked at life-improving charities or interventions. The reason we did that is, in part, that's what all of the wellbeing research is on, but also that using the wellbeing approach, you can do apples-to-apples comparisons between different things that improve quality of life. If you're comparing improving lives or improving quality of life to saving lives, then you've got to bring in a whole bunch of moral assumptions about how you trade those things off. That's a point we make: there isn't a direct way of doing that. There are a bunch of different assumptions we can make. Because we were publishing the first collection of results from a new field, we thought this is too complicated to explain. It will just confuse people. If we have all these different estimates of, "Okay, if you put in this parameter, you're going to get this and this." We thought, "Okay, we will leave that out as a question for further work." We've actually found it very hard to get wellbeing researchers to engage in this, or even donors as well. How do you make the trade-off between improving lives and saving lives? It seems to be a question which, understandably, people don't really want to think about: when is a life not worth living? It didn't seem like something we could or should do in that report.

SPENCER: How do you think about saving lives with regard to wellbeing? Do you just not count it because you think that's not what it's trying to measure?

MICHAEL: No. You can count it, but the problem is there are lots of different ways you can count it.

SPENCER: You have to make choices about it.

MICHAEL: Basically, the moral choice you make is going to have a huge impact on how cost-effective one is compared to the other. We wrote this report called The Elephant in the Bed Net three or four years ago, and I thought this was going to be hugely interesting because it had such important consequences, and I basically got zero engagement with it since we wrote it.

SPENCER: We'll see if our listeners are more engaged by it.

MICHAEL: So I could give you an example to try and bring this to life. Let's say you have your zero-to-10 life satisfaction scale, where on the scale is the point that counts as being equivalent to death. Suppose someone says they're a six out of 10 in life satisfaction. And let's assume you think the neutral point — this "equivalent to death" point — is five out of 10. In that case, if you extend this person's life for one year, they get one WELLBY. But you might think: hold on, that's five out of 10. Lots of people report being less satisfied with their lives than that. In many countries, the average life satisfaction is two or three. Are we saying those people are better off dead? What's going on? Now imagine instead that you set the neutral point for life satisfaction to zero out of 10. Extending that same life by a year becomes worth six WELLBYs. You can see that the value of extending a life by one year has just gone up; it's changed by a factor of six purely based on the choice the researcher has made. Whereas, when looking at life-improving charities, it really is apples to apples, where you're looking at evidence and self-reports from people. The comparison isn't so importantly influenced by the moral assumptions that the researcher is making for the comparison.

SPENCER: Does it only hinge on the index point of what you count as being not a life worth living, or does it also depend on other factors, like whether you think you should create people just because they'll be happy or not?

MICHAEL: I gave you the less complicated one, but actually there's a second set of considerations about population ethics and the badness of death, which we talk about in the report. There are lots of different ways you can approach this, which will lead you to have different sorts of conclusions. The range of possible reasonable opinions here is very large. On one view, a life-saving intervention is worth very little, and at other times it's worth many times more. That was why we didn't want to get into that. We thought it was enough to do the first global comparison of charities looking at improving quality of lives with comparable wellbeing data, and we couldn't get into these further considerations.

SPENCER: So groups like the Malaria Consortium or Against Malaria Foundation (AMF) that GiveWell recommends, a lot of the good that those are doing is in saving lives. Is that correct?

MICHAEL: Yeah, that's right. It's sort of straightforward initially to talk about life-saving and life-improving interventions, but often things will do a bit of both, or maybe quite a lot of both. You can disaggregate the life-improving and the life-saving components. Yes. The argument for things like Malaria Consortium and Against Malaria Foundation is with respect to how life-saving they are. At the Happier Lives Institute, where we've argued about this a lot, we don't have a view on what we think the right answer is. We tell people that they can pick their own choice. We even developed a little software tool to allow people to do that, but we've had almost no interest in it. What we've concluded is that donors just seem to want to focus on one or the other, sometimes a bit of both. Maybe you want to make people happier, maybe you want to try and save lives, but I found it almost impossible to get people to engage seriously with the question of what's the appropriate exchange rate for these things. If you're someone that thinks, "I really want to save lives," then the way you can do the numbers is that those will turn out to be probably the most cost-effective thing. We leave it as a choice for donors to make if they want to. In the same way, there are some people that are interested in what happens in the far future or in reducing animal suffering, and you don't get very many detailed conversations about exactly how to trade these things off. People just kind of do one or the other, and they have their intuitions about it. It seems to be a similar thing in this case, that people have intuitions and preferences and don't really want to get stuck into the details of how to do the quantitative trade-offs.

SPENCER: So before we wrap up, what are the things you think people should take away from this conversation? Are there any actions you think people should take based on this?

MICHAEL: Well, I'm hoping that people have listened to this and are thinking that they want to take happiness seriously, take other people's happiness seriously, and realize that there is, in fact, a scientifically sensible and robust way of comparing how much happier we're able to make other people through our charity. I'm proud that at the Happier Lives Institute, we've done some pioneering work starting the field from not even really existing a few years ago to find out what are the most cost-effective ways that we can make other people happier. There are enormous differences in how much better the best charities are than others. There really are enormous differences in how much good you can do between one charity and the next, how much happier these make people. I think that's great news. It means that if we want to make a difference to other people, we can give in a way where we can rely on strong evidence and feel confident that we're really improving what matters most, which is people's happiness. I would encourage your listeners not just to think, "Oh, it'd be a nice idea to make other people happier," but to actually go on and do it. We've identified the best charities that we found so far; you can really make an enormous difference. I encourage you to give to them, to support them. You can also give some money to us so that we can pioneer the next chapter of research into the most cost-effective evidence-based ways to make other people happier and spread the good word.

SPENCER: Michael, thanks for coming on.

MICHAEL: Thank you very much, Spencer.