[music] The Data Skeptic Podcast is a weekly show featuring conversations about skepticism, critical thinking, and data science. [music] Welcome to another episode of the Data Skeptic Podcast. I'm here this week with my guest, Peter Bacchus. Thanks for joining me, Peter. No problem, it's my pleasure. So listeners will know your name from a previous episode where my wife and I discussed this really fun paper you'd written a few years ago called "The Girlfriend Equation." So we'll spend a little time chatting about that, but I also want to get into some of your more formal work. So maybe we can start with your background? Yeah, sure. How far back do you want to go? Whatever you think is relevant. I did my first degree in philosophy, so nothing to do with economics, and then sort of traveled around for about five years working different places. Got interested in politics and less economics, so went back to school and studied. I did my PhD at the University of Warwick, here in the UK, where I think the chapter of the title was essays on the economics of charity or something, so that's been most of my work has been on that, a lot of things have changed recently. I was then at the University of Barcelona for a couple years, and now I'm at the University of Manchester in the economics department here. Very cool. So somewhere along the way, you sat down and scripted this interesting analysis. You started from the point of the Drake equation, which is one of my favorite things to go to for discussion. Could you maybe walk us through your perspective on what motivates you to write this paper? Yeah, it seems like quite an obvious idea, and lots of other people have done it before me, but I was reading a Carl Sagan book and came across it, and I hadn't had a girlfriend in a while, and it just seemed like an obvious application of the Drake equation where you're just applying a series of papers. I was applying a series of increasingly restrictive criteria to a population, so I did it and wrote it as sort of a faux academic paper. It's very tongue-in-cheek. It was only meant to be a joke, which I hope people take it in. Yeah, and just thought it would be sort of a playful thing to do, and people seem to have enjoyed it. Yeah, absolutely. So if you don't mind, I'd like to ask the update. The odds seemed against you at the time of writing. How has your love life panned out since then? Things improved remarkably. I got married about last year. I got married last summer. Congratulations. Thanks very much. Yeah, so there is hope. There is hope. Alright, anyone who is maybe a little downtrodden can note that you can beat these odds. Yes. So let's follow up, too. You had some specific criteria in terms of being an age range, and you were looking for a Londoner and someone college educated. How does your scalar factors figure into the woman you eventually married? She actually fits, I think, just about everyone's criteria in that paper. Not on purpose, but the objective criteria don't seem controversial to me. You want someone that's roughly your age, you want someone who may be roughly with your level of education, somebody nearby is always helpful. So she fortunately fit just about everything. And we get along, and we seem to find each other attractive. So, yeah, it worked pretty well. That's great. So given that you beat the odds, do you consider yourself a lottery winner, or is there maybe a term you might revise in the original analysis? I don't think I would revise anything. I mean, the paper is because it's written as basically a joke. So the odds are based on just selecting a person from random from the population, and that would be the probability. But obviously, we all do things to increase those probabilities. We go to concerts by musicians that we like, hoping to meet somebody else who likes that. We grasp a certain way to fit somebody that fits into our desired subculture, whatever it may be. So the odds are real, I mean, real terms, the odds are much higher, right? Most people end up finding a partner. But if you just look at it statistically, and you are talking about meeting a random person from the population each day, those aren't in fact your odds. So by most people have read it with that sort of idea in mind. I hope so. Yeah, absolutely. So one of the first things I found when I started kind of looking into your background was this great open letter you'd written on post. Well, let's spell it out in case the readers want to check it out. It's post-crasheconomics.com, and I'll put that in the show notes as well. Now, it's a letter written, I think, to economic students, but there's some universal aspects to that and some things that I think, especially the data scientists and the skeptics who listen to my podcast can appreciate. Would you mind summarizing the perspective you shared there? Oh, goodness. So there's a lot in the letter. I should explain that the post-crash society here was a group of students that saw sort of fundamental flaws in economics and were requesting changes in the way economics education is provided. They're great students, they're very engaged and used among the higher achieving students that we have, which is great, but I did find some of their arguments were a bit misguided. So the letter was intended to address that. So the main thing that sort of motivated the post-crash society here, and now there's several other universities and there's been walkouts of various classes at Harvard and other places around the world that reflect this sort of discontent with economics in its current state. But the sort of main issue, the thing that motivated me to write the letter to them was that the emerging narrative seemed to be that economics as some monolithic, you know, discipline, economics failed to predict the financial crisis, therefore economics is fundamentally broken. And we should be appealing to all sorts of alternatives because, quote, unquote, mainstream economics failed us in the most dramatic fashion possible. So the letter was meant to address that idea and some of the assumptions that go into that sort of narrative. I don't know if it got through to any of the students, I hope it did, but that was basically why I wrote the letter. In your opinion, is there truly a failure here or was the field of economics challenged with unprecedented and unpredictable circumstances in the form of the crash? I think there's failures. I don't think that those failures meant all of economics is somehow fatally flawed and we need to just throw everything out and start from scratch. And I think painting blame on economics, whatever that actually means, but painting blame on it sort of misses a lot of the complexity and a lot of the problems that led to the financial crisis that we're still trying to understand ourselves. By us, I mean academics, policymakers, bankers, it's still not 100% clear what went on and the combination of factors that led to something that was as severe as fast and as persistent as the financial crisis started in 2007/08. One of the things you highlight was, and correct me if I'm misquoting you, but there were new financial instruments, often particularly risky ones and ones that perhaps should have been regulated and widely weren't understood is probably one of the causes that went into the crash. Is that controversial or is that pretty universally accepted? Oh goodness, everything that has to do with the financial crisis is controversial at some level. I would say, to my mind, and I'm not a financial economist, I'm not a finance here, I'm a wildly unsuccessful investor. But to my mind, and the books that I've read that were more investigative and were trying to figure out what went on is that there were financial instruments that were in play, there were financial instruments that led to higher degrees of leverage that we had basically ever seen before. And that failure to understand and failure to regulate those certainly exacerbated some of the problems that were already in the system and it certainly exacerbated the severity of the recession. I'm a working data scientist, so I often get involved in predictions of various kinds myself, and one of the challenges my community faces is that they'll come along an event and some algorithm or some regression or process won't have predicted it, or maybe won't have very closely ballparked it. And there's this desire to kind of blame and say, "Oh, the algorithm's broken," or, you know, the method isn't correct. But there's also kind of a question of fairness in that most models, and I'm guessing the same as true in economics, are trained on historical data. So if you have a complete change in your environment and no data that's really relevant to the present, it makes the process of prediction fairly challenging to say the least. Is the lack of historical data a challenge, or is there some responsibility economists need to have to raise their hand and say, "I simply don't know because we're in uncertain times?" I think it would behoove all of us to say, "I don't know a lot more often," and say, "I'm not sure what's going on." I think a little humility in all walks of life, but in academia especially is hugely beneficial to us. But then you're absolutely right, when you talk about using historical data to make predictions about the future, predictions about the future are really hard. What is the same as saying? Prediction is very difficult, especially when it's about the future. So faulting people for failing to predict the future just seems misguided to me, and I talk in that letter, going back to the letter, there is a short discussion. I had a longer discussion in the debate a few months later about sort of the nature of prediction, what do we mean by prediction? And I think that's where a lot of the blame comes in for economists from the public, is that most of the public's interaction with economists and economics is through this idea of prediction. You're watching the news, "Oh, here's the economics and finance report. Joe's going to tell us what's going to happen." Well, we think GDP will grow by 3% next year. And then if that fails to happen, "Oh, Joe, you did a terrible job. You failed to predict what was going on." But is that a failure of prediction? I mean, he doesn't report any sort of standard errors. He's reporting a point estimate. On average, that point estimate might be correct, but we don't get to look at on average going into the future because the time only happens one time. A more nuanced understanding of what we mean by prediction, and I think a more nuanced presentation of prediction within the news media would be great. But that requires us to be able to say, "I don't know exactly what GDP growth is going to be." There's a very good chance it will be between negative 1% and 2%. But that's not overly useful to people, I think, but that requires that sort of humility. And it requires policymakers not to demand point estimates when they're looking at forecasting. Do you think there's a challenge as well in the way science is communicated? Yeah, definitely. With all science, absolutely. The understanding of our average citizen of statistics, I think, is alarmingly low. And so their ability to interpret all sorts of things in the newspaper be it about economics, be it about education policy, all sorts of things is hurt, and that hurts, then sort of a democratic process. Yeah, I think communicating ideas, communicating results is a really complicated thing, as always been. But I think more and more statistics and economics becomes more and more important. And more in everybody's daily life, it would be great to see sort of an increase in education and getting people to sort of a higher standard of understanding basic statistics. And also getting policymakers and economists to do a better job of presenting their work in an accessible way in a useful way. Yeah, absolutely. So I'd like to also touch on some of your work in the charitable areas, which I think are really interesting as well. Oh, good. I've got somebody read that book. Yeah. Would you mind summarizing kind of your interest and findings there? I've done a few different bits and pieces, but so my interest in charitable work goes back to my master's degree or doing work in on the economics of charity because back to my master's degree. And then I wrote a few papers for my PhD and outside of my PhD on different aspects of the charitable sector. I just wanted a really interesting thing to study, especially in economics, where that sort of altruistic behavior or pseudo altruistic behavior is not totally well understood. When you start looking at public goods and the provision of public goods, whether that's by the state, it can also be provided by private money via donations. So some things I'm working on now have to do with the trade-off between, you know, it's a government cut spending for welfare. Do we see an increase in donations to welfare charities? Are people responding to that? The work I've done in the past had to do with tax incentives for charitable giving in the US. It had to do with giving for international development in the UK, and it had to do with sort of changes in the structure of the market for charities in the UK. I think that's a simple way to explain that. I've seen some sort of pragmatic results with respect to charity, like that the average amount of charity in an economy, at least this is what one author was claiming, rarely changes. It's just that different organizations market a little bit better from year to year, so the ebb and flow of where money goes just kind of shifts. Do you think there's an inefficiency there, or can policy directly affect the way people give? I think there's some evidence that policy, in terms of tax incentives, can affect the way that people give, the amounts that people give. I haven't seen any research that looks at how policy would affect, how say, tax policy or tax incentives could affect the way that individuals distribute their donations across different types of charities. I look at tax incentives that were what was the effect of tax incentive for different types of charities, but we didn't look at whether people were substituting from one charity to the other one type of charity to another. But I think it's fairly established evidence that giving has increased over time. Volunteering is a thing that seems very stable, but charitable giving has increased pretty steadily over time, certainly in the UK, and I think to a lesser extent in the US. It seems to me there's a lot, in the same way the financial market was introducing a lot of, I don't want to use the word innovative, but let's say different and interesting types of financial devices. I've seen stuff come up in charities as well, from micro lending and giving directly and more communication on what the success metrics and KPIs for the charities are. Do you think those are shifting the way consumers choose to give? That's a good question. There has been a lot of research on different fundraising strategies and how to sort of get more money out of people. I would love to think that people are responding to the efficiency of charities and some metric of a charity's ability or success in providing whatever it is they're providing. And I think that's happening more in instances where there's a measurable output, which some charities have. Have you heard of impact bonds? I have not. With impact bonds, there's social impact bonds. They're starting to use in the UK. I think they might be using this, but it's a way of trying to line up incentives to get a charity and people giving to a charity to produce results and the government can then fund them. So think of a private sector organization might come along and say, well, we can work with prisoners. We help reform them better than the current system is. The government will say, OK, fine, if you can get the recidivism rate below, I don't know, 50% or whatever it might be, that costs us money when prisoners go back into prison. So if you can get the recidivism rate down, we'll give you the difference, basically. We'll pay you for that. And then private people, private lenders, or private investors come along and they'll buy a bond in this organization. So the organization will be able to make money from the government by doing a better job of reforming prisoners in prisons. But that's something where there's a very measurable output. It's very easy to know what the guy in prison is, not in prison. If you're looking at something like the impact of female empowerment on fertility, on birth rates in India, that's a lot tougher to do. And sometimes trying to measure that leads to more problems than it helps, trying to, you know, if you're measuring, say, a program that's designed to get young girls into school, well, then the incentive is just to get the butts into seats and not worry about actually educating them. And so then the incentives kind of become perverse. But I think there is an effort to try to design mechanisms, which the incentives line up very well for both funders and for the implementers of the program to charities and to try to think more creatively about how do we deal with things that need to be done but are very difficult to measure. If you had to give some advice to someone that wanted to give to charity and wasn't sure how to be the most effective they can be with their available assets, and they don't necessarily have a full-time job to go and look and check out charities. What's the best place a person can give effectively? Well, I think that's a personal choice, to be honest. I think that's down to the preference, so some people want to give to very small charities that can't seem to get money from anywhere else. Maybe they're doing, you know, controversial work, or, you know, trying to help, I don't know, reform pedophiles or something that's very, you know, socially undesirable. And you don't want to give other people like to give to the oxfams and the uniceps of the world. To my mind, if I could do anything to change people's minds about charities, it would be look for charities that are efficient fundraisers and give them the check, write them the check, and tell them to go use that money to fundraise more. Far too often, the perception is people give money and they think, "Well, I want this money to go help the kids. I don't want this money to go fundraise." But the problem is, if I give a dollar to a charity, probably that dollar can go to help a kid somewhere if it's a children's charity. But if that dollar can be turned into fundraising and then bring in three more dollars, well, isn't that better? I mean, it leveraged my money to bring in even more money into the organization, which can then go to the kids. That's a really interesting point, yeah. But most people shy away from that. Most people, they don't want charities that have high administrative costs. "Oh, they're paying their CEO so much money." Now, these are giant organizations. Oxime is a 500 million pound a year organization. You need the Harvard Business School graduate. You need really talented management there and you have to pay for that. And so when people say they don't want to give to organizations because they spend too much on fundraising or spend too much on administration, I get frustrated. I get a little frustrated. Absolutely. So, I wanted to kind of touch back again on the questions of no historical data. It seems like it comes up in the charitable world as well that I see economists often are fortunate to fall upon coincidental experiments. You know, like something happened that no one intended and they can go kind of study it retrospectively as though it were a test. But trying to roll out an actual policy or plan change, whether it be an economic policy or a giving strategy, people are very fickle about wanting to experiment because you'll have an effect on real people's lives. How does one balance the amount of time or effort you invest in doing experiments or trying things out, or maybe spending 10% doing something new versus investing in the kind of proven system that's been working? Maybe not perfectly, but it is what it's been. That's an interesting question. I think it's down to sort of preference of the economists that we're looking at, or the preference of the researcher. They need to be an economist. Some people put a great deal of faith in lab experiments in economics or in social psychology. Other people are very critical of those. Some people put a lot of faith in the field experiments and other people are critical of those. It also comes down to funding, so field experiments are very, very expensive. It's sometimes hard to get your experiment through an ethics committee because you can be affecting people's lives. And other people like to look for the natural experiments because I feel like that's the most natural way to look at the impact of a particular policy. You're with the policy and here would be the effect of that policy. I think that's really down to the preference. I think people tend to specialize so you'll have sort of experimentalists and you'll have more conometric-driven people that just look at existing data. It's good to have that diversity and it's good to have different people specializing, I think. But I focus more on using, in my own work, I focus more on using data that already exists just because I sort of sign up to the philosophy that let's go find what people have actually done and not manipulate them and just see what they're doing in their daily lives and try to figure out something from that. Yeah, there are any data sources in particular that you think are very rich and untapped that yourself and/or potentially listeners might be able to go there and find something novel or interesting. So two things I'm working on at the moment use slightly, I don't know, unconventional data sources. One is using chest data to study competition between men and women. Interesting. We're taking data from existing chest tournaments over the past, how many over years, and looking at how men and women compete against each other and trying to understand whether they're, so do they exert different levels of effort when there's inter-gender competition? Do they pursue different strategies? And there's some good work out there already that started to ask those questions that we were sort of inspired by. And then the other one we're using data from the NBA, from the National Basketball Association to try to decompose productivity into different components to say how much of your productivity is a function of how talented you are, how much effort you exert, and how much experience you have, something like that. And then we'll be able to look at monetary returns to different components of productivity. But sports data is great, I think sports data is great for asking economic questions, especially in labor economics, because everything is measured, everything is measured. And more and more are the sports view data and tracking where the ball is and where all the players are and how fast they are. All that stuff is going to be an absolute goldmine for studying this stuff. Now the critique of that is, so you find something for the NBA, what does that have to do with the factory worker from Seattle? But those sort of questions of external validity always come up and you have to address them and be aware of them. But there's so much data out there right now, it's fantastic. As someone that's interested in data and analyzing data, there's so much stuff out there. In the UK, if anybody's based in the UK, it listens to your podcast. It's called the UK Data Archive, and they have tons of stuff that you can just download and start playing with and start having to look at it, so so much of it is free. And a lot of times, your ideas can come out of just having to look at a data set. Oh, there's a variable there. I didn't realize they asked that question in the survey. I wonder if we could give you something with that. Yeah, absolutely. Yeah, so it's a great time to be a data scientist right now, I think. Definitely. Both of the NBA and the chess example are really fascinating. If listeners wanted to know more, what's the best resource? Patients, wait for the papers to come up. But if you Google, say, chess, and it's called stereotype threat, there's some papers that come up or chess in gender. There's lots of stuff out there on chess and gender of people looking at different aspects of it. That would be the best place for the chase. In the NBA, that work is ongoing, but the paper will be out hopefully early next year as a working paper. Well, let me know when. If you're inclined, I'd love to have you back on to talk more deeply about either of those topics. Yeah, sure, absolutely. So lastly, I'd like to ask my guest to provide two recommendations. The first is the benevolent reference, something that you think the listeners would enjoy your benefit from, but you have no direct connection to. And the second is the self-serving reference, something that ideally you derive direct benefit from exposure here on. Okay. I guess the benevolent recommendation, there's a book that I recommend everybody should go read. Unfortunately, he's not going to benefit the author very much because he's dead, but it's called The Conquest of Happiness by Bertrand Russell. It's a fantastic read. It's a self-help book for people that hate self-help books. But it's a wonderful read. He was a brilliant guy. Absolutely. I really, really interesting. Good to remind yourself of what he's saying in that book every few years. And in terms of a selfish recommendation, I don't really know. I had a think. You mentioned it in your English for me. I'm an academic. I don't have any books published. I'm on Twitter if you want to follow me on Twitter. What's your Twitter handle? I'm @awesomenomics. So if people want to follow me on Twitter, I tweet about economics. There's nothing on there that's really my opinion or anything. I just articles I found or research that I found that I think is interesting and just try to share interesting ideas that other people have and not try to make it about me. Excellent. But that's sort of the most selfish thing I could think of, really. All right. Well, I'll be sure to follow you. And I imagine most other listeners will as well. Yeah, that'd be great. Well, thank you everybody. This has been great. Thank you so much for your time. I think we had a really interesting chat. Yeah, me too. Thanks a lot. It's been, yeah, it's been interesting. [MUSIC] [BLANK_AUDIO]
