This week, Nick and Goldy talk to Doyne Farmer, a renowned physicist and mathematician, to discuss his new book, “Making Sense of Chaos: A Better Economics for a Better World.” Farmer, who is a professor at the Institute for New Economic Thinking, challenges traditional orthodox economic frameworks by applying complex systems theory. Their conversation explores the limitations of mainstream economic models, the importance of incorporating uncertainty into economic thinking, and the potential for complexity economics to provide better guidance for policymakers in addressing pressing issues like climate change and inequality. It’s a thoughtful discussion that explores more effective approaches to understanding and managing complex economic systems.
Doyne Farmer is a renowned physicist and mathematician who is currently a Professor at the Institute for New Economic Thinking at the University of Oxford and the Director of the Complexity Economics program. He is also an author known for his groundbreaking work in the field of complex systems and chaos theory. His recent book, “Making Sense of Chaos: A Better Economics for a Better World,” delves into how chaos theory can be applied to understand and address the complexities of modern economic systems.
Twitter: @doyne_farmer
Further reading:
Making Sense of Chaos: A Better Economics for a Better World
Website: https://pitchforkeconomics.com
Twitter: @PitchforkEcon
Instagram: @pitchforkeconomics
Nick’s twitter: @NickHanauer
Nick Hanauer:
The rising inequality and growing political instability that we see today are the direct result of decades of bad economic theory.
Joe Biden:
It’s time to build our economy from the bottom up and from the middle out, not the top down.
Nick Hanauer:
Middle out economics is the answer.
Joe Biden:
Because Wall Street didn’t build this country, great middle class built this country.
Nick Hanauer:
The more the middle class thrives, the better the economy is for everyone, even rich people like me.
Speaker 3:
This is Pitchfork Economics with Nick Hanauer, a podcast about how to build the economy from the middle out. Welcome to the show.
Nick Hanauer:
Goldie, today we get to talk to one of my absolute favorite people. Doyne Farmer is a renowned physicist, a mathematician, who is currently a professor at the Institute for New Economic Thinking, where my colleague, our good friend Eric Beinhocker works, he runs the institute. And today we get to talk to him about his new book, Making Sense of Chaos, which addresses how complex systems can be applied to economics. And Doyne is just one of the world’s most brilliant and interesting people. He’s had an insanely consequential career in the academy, but also done crazy stuff like applied this really modern mathematics to beating the market, to beating casinos in Las Vegas. He’s a really extraordinary character and so smart, so funny. I can’t remember, have you ever met Doyne before, Goldie?
Goldy Goldstein:
We had him on the podcast with Eric-
Nick Hanauer:
That’s right-
Goldy Goldstein:
… once before.
Nick Hanauer:
… a long time ago. Long time ago.
Goldy Goldstein:
But I’ve never actually met him in person.
Nick Hanauer:
Yeah, yeah, yeah. Just an incredible person. And obviously what Doyne is writing about is very central to our own views of economics and highly adjacent to our thinking.
Goldy Goldstein:
Right, because it turns out a better model of the economy, our intuition is that it supports what we’re advocating for, this middle out approach. When people at the bottom and the middle do better, that’s good for the economy.
Nick Hanauer:
Not bad for it, exactly.
Goldy Goldstein:
Not weirdly bad for the economy. So, we’re all for better models. Considering the current Orthodox models, I don’t know, what’s the technical word for it? They suck.
Nick Hanauer:
Yeah, exactly. Well, let’s talk to Doyne.
Doyne Farmer:
I’m Doyne Farmer, I’m director of Complexity Economics at the Institute for New Economic Thinking at the Oxford Martin School, and the Baillie Gifford Professor of Complex Systems Science at the Smith School for Enterprise and the Environment. And I’ve just written a book called Making Sense of Chaos: A Better Economics for a Better World.
Nick Hanauer:
So Doyne, we go back a good long way. And you’ve been thinking about these things carefully for a very long time. And I think it’s worth mentioning that you were one of the fathers of complexity, complex systems theory and that kind of thinking. For how long have you known that this way of understanding the world was applicable to economics?
Doyne Farmer:
I would say I had the suspicion it was starting in 1987, and thinking about it during the ’90s I became more convinced that it could happen. Well, I would say I’ve had building confidence in this for the last 30 years.
Nick Hanauer:
Right. So during this discussion, we want to tease out how complexity can make sense of how an economy works in a way which is better than the existing frameworks. But I should say that there are things wrong with the existing orthodox economic framework that extend beyond just not accounting for complexity. But can you try to tease out what the fundamental sort of drawbacks of the existing economic framework are, orthodox economic thinking, and how complexity economics is different from that?
Doyne Farmer:
Yeah. So just to say, in my book I tried to refrain from criticizing mainstream economics too much. In part, I just want to be nice and I don’t want to alienate them.
Goldy Goldstein:
Well, good luck.
Nick Hanauer:
Yeah. Good luck with that.
Goldy Goldstein:
Have you met an economist?
Doyne Farmer:
I know quite a few of them. But anyway, I don’t want to …
Nick Hanauer:
We’ll try not to destroy your reputation on this podcast.
Doyne Farmer:
That’s okay.
Nick Hanauer:
But go on.
Doyne Farmer:
But I can tell you, so all mainstream economic models work in the same way and this has been in place now for 70 years. And begin by assuming we all have utility functions, which are scorecards, that we all make decisions that maximize our utility, and we can do this taking everybody else into account. They use that to deduce what decisions we’re going to make, and make a few other assumptions like equilibrium. And then that’s how the model works. So, the obvious problem is I don’t think that’s how we really do things. So, the whole setup relies on an as if argument. It’s as if this is the way the world worked. Now, the other big problem though, besides the question about how realistic that is, is how tractable is it to build a model that has the complexity of the real economy in it? The problem with that setup is that anytime you have to do a calculation that involves what’s the best possible thing you could do, taking into account that everybody else is also doing the best possible thing they could do, it’s super complicated.
And so, as soon as the setup starts to get anything but fairly simple it becomes impossible to solve the equations, and that’s why the models are always kept very simple. In contrast, in complexity economics, we say, “Well, let’s try and model things more or less the way they really work.” So, how does the economy work? Well, we have people going along, every day we get some information, we have some desires, we make a set of decisions, we make them using simple rules, rules of thumb, or maybe we do a little calculation of projecting into the future a bit, and we make our decisions. Those decisions then impact what actually happens in the economy, which then may have some more information flowing in, or the economy make information itself, and then everybody makes their decisions, and that cycle just repeats itself again and again.
So, it’s a totally different way of building the model that has a big advantage that we can make the model, we can put all the things that we need to put in into it, and it has the other advantage that it’s more the way things really happen.
Nick Hanauer:
And one of the things that makes that possible today is compute power, right?
Doyne Farmer:
Yeah. That’s a key.
Nick Hanauer:
And is it fair to say that without the kinds of computational power that we have today, this would be a fool’s errand? Well, maybe not a fool’s errand, but this would not be tractable.
Doyne Farmer:
Yes. I think frankly we could have done it 10 years ago, but we could not have really done it 30 years ago. Or let’s say the models would’ve had to be fairly crude and it would be difficult. Now, if we really needed to or if we had the resources to do so, we could simulate a world with eight billion people and three and a half billion households with 200 million firms, and simulate the whole world. We could do that, that’s tractable. Well, let me just say that’s been an important thing to keep in mind, because people often challenge this saying, “Well wait, this has been around, these ideas have been around since Herb Simon in the early 1960s. Why hasn’t it gone anywhere?”
And I think there are two reasons for that. One is that back in those days it just wasn’t computationally tractable. But the other is that because it wasn’t tractable, mainstream economists at the time, who were smart people, locked into a different way of doing things because it was the best they could do. But then they got locked in, they got stuck doing it that way and came to believe it’s the only way to do it. And my main complaint about mainstream economics is simply not being open-minded enough to let something new under the tent.
Nick Hanauer:
Right. And to be clear, you relay in the beginning of your book some examples of some of the modeling that you have done using this approach. For example, during the beginnings of the COVID crisis, trying to model what would happen to the economy and the world under different scenarios using this agent-based approach. Can you speak briefly about that to give our listeners a picture of what you did, and how it’s different from the other approaches?
Doyne Farmer:
Yeah. So, when the COVID pandemic broke out, I rounded up my best postdocs and graduate students and we went on a crash program to build a model to understand the economic impact. So the way our model worked, well, we did two things. One is we predicted how big the shocks would be. That is, people would not be able to go to work in many cases. Who would not be able to go to work? I had two remarkable ex-students, Penny Mely and Maria del Rio-Chanona, who were experts on something called the ONAP Database at the National Bureau of Statistics. And it contained information like if you’re in a given occupation, how close do you usually work to other people? Well, if you work closer than two meters, you probably won’t be able to go to work. So, we were able to predict who wouldn’t be able to go to work.
We also predicted who would be able to work remotely. So, we predicted that ahead of time. And then we got a study from the Office and Budget Management, what services will people not ask for? Like during the pandemic, airlines stayed open but people didn’t fly. So, we figured out what the shocks would be. Then we built a model that showed how those shocks would propagate around, because the shocks don’t just stop at the entry point. So, we were able to see which companies would be lacking demand, which companies would be lacking the inputs they need to make their product, and which companies wouldn’t have the labor to make their product. So, we could watch the demand shocks propagating upstream in the economy and the supply shocks propagating downstream, and we could look at the way they collided along the way. And so, our model stepped forward day by day.
Every day it would look, does this company have the demand, labor, and inputs it needs? If not, it’s not going to produce what it wants. And of course, if it’s supplying another company and it’s not producing as much as it normally would, that’s going to affect that company too. So, that’s what the model did. So we made predictions, we pass them to the British government. They actually enacted the scenario that we thought was the best one, which was to keep all the upstream industries like mining and food production, and so on, keep all those open but close the downstream customer facing industries, or make them operate without directly interacting with customers face-to-face. And we showed that it was the least bad choice in the sense that not that many people more would die than under a full lockdown, but the hit to the economy would be a whole lot less.
And so, when they emerged from the first lockdown in the UK, that’s actually the policy they took. We don’t know whether it was us that really caused them to do that, but so we predicted what would happen during the lockdown, after the lockdown, and we looked a year later, once the dust had settled and we knew what all the numbers, what really happened, we predicted a 21.5% in second quarter of 2020, and the actual hit was 22.1%. And we predicted all the other things, unemployment and so on very accurately. We predicted how they would change through time during the first year accurately. We predicted the hits to the 55 industries we were simulating pretty accurately industry by industry. So, the model really worked.
Goldy Goldstein:
I’m curious, did you look far enough ahead to predict the surge in inflation?
Doyne Farmer:
Well, what we did, we explicitly said, “We’re only predicting a year out. We think for the first year, inflation is not very likely.” And so we said, “No prediction after that.” But because our model actually didn’t say anything about inflation, it was a surprisingly mechanical model. We didn’t have prices in it. It just showed how the shock propagated. And so, we stuck to what we knew how to do.
Nick Hanauer:
And as I recall, the Bank of England’s prediction was a drop of 30% and you were at 21%, so your prediction was significantly more accurate than theirs.
Doyne Farmer:
Yeah. Now, of course you can get lucky in any given single event, but that’s why we did very careful postmortem to see where did we do it right and where we get lucky? And we did get lucky in a couple of spots, but mostly it just was because we did it right and we could test, well what if this was less realistic? What if that were less realistic? And we could see that anything we were doing that made our predictions worse. So, we’re pretty confident that it was mostly from being right.
Goldy Goldstein:
Had the government adopted different policies, would it have messed up your predictions?
Doyne Farmer:
Well, we simulated five different policies. So, if one of the policies they adopted was close to the five we guessed they might adopt, then I think we would’ve done pretty well. But I can’t say because they didn’t adopt the policy, so that only happened in an alternative universe where our quantum doubles are doing something.
Goldy Goldstein:
Okay. So, maybe the Bank of England had presumed the government would adopt a more austere agenda, and that’s why their numbers went to 30%.
Doyne Farmer:
Well, actually I don’t think that’s the problem. My guess is their prediction was conditioned on something, but I know how their models work and those models don’t have the right machinery in them to do a problem like this. This was a problem that fell into our laps. One of the reasons I did this, because I saw this is an ideal problem for us, because it’s completely disequilibrium. The COVID shock was so hard and so fast that you really needed a model that could track the disequilibrium behavior. The Bank of England’s models are going to be equilibrium models that assume … They’re either models that just use data, like machine learning or statistical style models, which can’t get it because there was no example to train on, or they’re equilibrium models that make a bunch of assumptions that were just wrong.
Nick Hanauer:
Can you just explain to our listeners a little bit more about the difference between an equilibrium model and a non-equilibrium model, and how these things are so profoundly different?
Goldy Goldstein:
Right, because that gets to the heart of complex systems, doesn’t it? They are inherently non-equilibrium systems.
Doyne Farmer:
Yeah. Well, they can go to an equilibrium. We can run our models and they may settle into an equilibrium, but if they do we say, “Oh, that’s an emergent phenomenon,” because it’s not something we knew a priori. So, we don’t assume it, sometimes it happens. Now the key, equilibrium in economics is actually a very slippery term because it means different things in different cases, but it means, roughly speaking, two things. One is it means supply equals demand. And the other thing it often means is that everybody’s taking everybody else into account and they’ve thought it all through, and their strategies are lined up. So, everybody’s adopting a strategy that’s going to be fixed until things change, because it’s effectively the best strategy they could adopt. Whereas in our complex systems models, people are adopting strategies on the fly, making the best decision they can, and they might settle into an equilibrium strategy or they might not. In this model you see, because we were literally marching through day by day and watching the dynamics of the economy change as the COVID shock played out, it was very explicitly disequilibrium.
Nick Hanauer:
Yeah. But of course, one of the things about complexity economics is that it accounts for the fundamental uncertainty in a complex system like an economy, which is very different from the approach that standard economics takes. Can you explain the difference between risk and uncertainty, and how that plays out?
Doyne Farmer:
Sure. So, normal mainstream models do take into account risk. So, they don’t assume that the world, the future is known with certainty. There might be a 40% chance of one thing and a 60% chance of the other. But the way they back out what the economy is going to do to look and see assuming that, what decisions will people make taking that into account, and then what effect will that have on the economy? And that’s risk, because you have to know the possibilities and you have to know the relative probability that one of them will occur. What is the likelihood of each of the known possibilities? Uncertainties, when you either you don’t know the likelihoods of the outcomes, or you may not even know all the possible outcomes, there may be outcomes you haven’t even thought of. So, that’s uncertainty. Now, the reason we argue that our kind of models work better under uncertainty is because heuristics that we use, and heuristics are rules of thumb, are evolved typically in fairly general situations and apply broadly, and often work better when things are uncertain.
That is if you have a choice between pretending the uncertainty isn’t there and making a decision, making some estimate of what you think those probabilities are, versus a method that doesn’t even pretend to know what those probabilities are, oftentimes these simpler methods do better. I mean, the classic example that I mentioned in the book is a funny one, because Harry Markowitz, famous economist, who got the Nobel Prize for his formula for portfolio allocation. If you know the returns on stocks and if you know the correlations between those returns, then it tells you the optimal way to buy and sell the stocks to make the biggest profit with a given level of risk. So when he got the Nobel Prize, how did Harry Markowitz invest his money?
He picked a few stocks he liked and he made equal bets on all those stocks. Why? Because he knew that even small errors in predicting the returns on the stocks would amplify into big errors in the correct allocation. And you on average do better just with a very simple heuristic, make equal bets. So, that’s what he did. And there are many situations like that where simple heuristics actually beat fancy stuff.
Nick Hanauer:
So, I think one of the problems generated by orthodox economic thinking and the orthodox economic framework is that it creates its own set of heuristics, and common sense about economic cause and effect.
Doyne Farmer:
I wouldn’t quite call those heuristics. I think what you’re saying is it has a conceptual framework that creates a certain kind of intuition and belief about how the economy works that led us to things like neoliberalism.
Nick Hanauer:
Correct, correct. I mean, and by heuristics I mean the framework would lead you to believe if you take it literally and seriously, that if you raise the minimum wage there will be corresponding job loss, because you push the system away from equilibrium. And that turns into a policymaking heuristic, which is that we should never raise the minimum wage, which is a tenant of neoliberalism. I mean, in rough terms.
Doyne Farmer:
Yeah, I wouldn’t quite call that a heuristic though. I would call that received wisdom, or something that if everybody believes is true, even though the evidence for it being true is actually pretty weak. Why do they believe it’s true? Because in a really simplified model it shows you that it’s true in that model. And that went across the board for all of the things that neoliberalism, they were all based on oversimplified models. Where when you start complicating the model, you see it’s not as simple as you’ve seen.
Nick Hanauer:
That’s right. And I realize this is really putting you on the spot, but can you describe the best that you can how this new framework, using the perspective of complex systems, can change that collective common sense? Do you know what I mean? Is that a fair question?
Doyne Farmer:
Yep.
Nick Hanauer:
Because at the end of the day, what’s happening in the academy is really important, these models are really important, but what’s most important is what policymakers do, right?
Doyne Farmer:
I totally agree.
Nick Hanauer:
And the policymakers are not going to get PhDs in economics, and they’re never going to build the models, they’re never going to understand the models, they’re probably not even going to look at the models. What’s really important is the collective common sense of the policymaking class. And can you explain how this way of thinking could influence that?
Doyne Farmer:
Yeah. I mean, there’s something people call systems thinking, which is trying to really take all of the indirect effects into account. And I think the complex systems view encourages that. I mean, I think we really need better guidance through the crises that we’re facing. Climate change, of course being a big one, inequality being another big one, and the next financial crash being one that’s looming on the horizon somewhere. And I think if we have models that really more realistically simulate the economy, people will start to pay more attention to what the models say because we’ll accumulate a track record that says, “Hey, we’ve gotten these things right, maybe you should pay attention.” The reason people don’t pay much attention to the models now is because predictions aren’t very good.
Nick Hanauer:
Yeah, exactly.
Doyne Farmer:
So, it’s very sensible not to pay attention. Now, it’s never going to be celestial mechanics. We’re not going to be making precise predictions about what’s going to happen, but we do think we can up the batting average so that we’re right more of the time. But secondly, we create models where it becomes easy to do experiments that you can’t even do with the mainstream models. We’ve been building a micro macro model. I call it a micro macro model because at the micro level it has millions of households, it has tens of thousands of firms, and a government. And so, we simulate the economy day by day, or quarter by quarter, and we simulate what happens. And so, if we see that if we adopt a new policy, “Oh, look. This is making inequality worse. It’s really hitting Black people harder than white people.”
We can see that in our model. Why? Because we have a synthetic population with lots of synthetic poor people, and rich people, and old people, and young people, Black people, white people, et cetera. They’re in the model. And furthermore, we can test policies. In a mainstream model, everything has to get reduced to equations, and the equations have to involve typically money as the main output. You have to be able to reduce the policy to money. Like carbon tax for example, they can do that. One of the reasons they love carbon taxes is because really the only policy the models can directly test, but we can put any kind of policy you want. It’s just code. Code is easy to write. You just plug the code in, see what it does, and you see where the economy goes when you test that policy properly.
Nick Hanauer:
It’s so interesting. I mean, I can’t remember you say it in the book, but one of the profoundly frustrating things about contemporary economics is that every physicist will tell you that force equals mass times acceleration. That there’s unanimity around basic physics, but the simplest-
Goldy Goldstein:
And it’s true.
Nick Hanauer:
Yeah. Well yes, and it’s true. But the most basic economic proposition, which by the way I feel is what will happen if you raise the minimum wage? Is there a simpler question to be asked? Generates just such amazing amounts of disunity. And I think that this is, I guess the problem you’re trying to address.
Doyne Farmer:
Yeah. This is maybe slightly off-topic, but we had a visitor who’s been looking a lot at empirical facts about macro rather than models. So, he gave us a series of lectures, and you’re really going to like this, Nick, because it showed that whenever we’ve lowered taxes on poor people, that the economy, GDP goes up. And why do I think that happens? I think that happens because the economy is normally demand limited. We’re not limited in output because we don’t have enough investment capital, we’re limited because we don’t have enough customers to sell the stuff to.
Nick Hanauer:
Absolutely.
Doyne Farmer:
And so, the minimum wage is going to work in a very similar way. The minimum wage of ah, it’s going out of the hands of the plutocrats, but they’re actually gaining because then those people buy their stuff.
Nick Hanauer:
Right, no. It’s-
Doyne Farmer:
And so, it’s a win-win situation to make changes like that. And one of my goals is to really be able to demonstrate that in an accurate and compelling way.
Nick Hanauer:
So, is it fair to say that your team is building the most sophisticated model that will accomplish these things?
Doyne Farmer:
Yeah, there’s another team at IIASA that’s also building a similar model,
Nick Hanauer:
At where?
Doyne Farmer:
At the IIASA, which is an institute in Austria, a very international institute. And then there’s also the Bank of Canada. There are some researchers at the Bank of Canada that are also pursuing a similar direction, and some people at the Bank of Hungary, and a few people at the Bank of Italy. So, it’s starting to happen. And so, we’re at a good stage where we’re racing each other to do better. And the model at the Bank of Austria, that was really made mainly by one graduate student, Sebastian Poledna, what he showed is that his predictions for Austria were as good as a standard DSG model. Now, you need to do lots of other countries, you need to do more statistical testing, but we’re now starting to see that these models are really competitive with the mainstream models, and we hope to pass them soon.
Nick Hanauer:
So, where are you in building your model?
Doyne Farmer:
Well, the model runs. We’ve run it on 38 countries. We see that it’s giving better predictions than our competitors model. We’re now running it against mainstream models, which is a little harder because we have to figure out how they work so we can make sure we’re doing them justice and creating a good benchmark. But yeah, the model’s up and running. There’s still a lot of stuff when you look under the hood that I go, “Ooh, I want to make that better.” So, we know there’s a lot of room for improvement, but it’s striking that we’re able to produce pretty reasonable results with an effort that’s in its infancy relative to the mainstream. I would guess the cumulative amount invested on this approach is 1000th of the amount invested on the other. And yet we’re already breaking even.
Nick Hanauer:
That’s amazing.
Goldy Goldstein:
I’m curious, how much is the model limited by the quantity and quality of the available data? I mean, it’s hard to collect data on things that are hard to quantify.
Nick Hanauer:
And is that data just not available, or is it hard to normalize?
Goldy Goldstein:
And are we collecting the right data? Or are there things that the government agency should be doing better in terms of their surveys?
Doyne Farmer:
Yeah, so it’s a combination of all of the above. A lot of the data, just many data are not available, but there’s a lot of data that is available that we can’t get our hands on because it’s under a confidentiality wrap and the government won’t let us see it.
Nick Hanauer:
Oh, interesting.
Doyne Farmer:
But now the U.S. is very backward, because the U.S. is still using surveys for the more detailed data that they construct things from, whereas other more sophisticated countries like Rwanda actually, they have a value added tax and they record the counterparties on the value added tax. So Rwanda, Ecuador, Chile, Belgium, Hungary, there are quite a few countries that have a value added tax, and they can actually see every transaction taking place in the formal economy. So, that gives you incredibly rich data. Now of course, it’s hard to get access to that data for obvious reasons. Confidentiality becomes a serious problem, but they have ways of, for example we could run our models on that kind of data without us ever actually seeing the data, just to get the answers. Hence all-
Nick Hanauer:
That’s right. Yeah, you can anonymize that kind of data.
Doyne Farmer:
Yeah, you can anonymize that kind of data. The UK, I mean, one of the exciting things for me with my book is I got contacted by the head of data science at the Office of National Statistics in the UK, and he said, “Wow, I liked your book. I like complexity economics. What kind of data should we be collecting that we’re not?” He went over the data they are collecting, he said, “And we’re going to try and think of ways to make more use of that data.” So, I would say roughly speaking it’s a patchwork quilt. Some countries have very good data, some other countries don’t. The U.S., though the government doesn’t have good data, all the commercial data vendors have the best data about the U.S. relative to other countries, because they scrape the tax filings.
And there are various things companies have to do that you can infer information from. And so, we know a lot about what U.S. companies do. So, it’s a very complicated patchwork quilt that has a lot of big holes in it, but we have ways. B the way, one of the things that we’re trying to do is convince somebody to fund us to actually construct the whole global supply network, who sells what to whom, because we’ve created methods for using machine learning to fill in those holes in reasonable ways. And if we were able to do that, then we’d have the right data to run our models on.
Nick Hanauer:
Do you have a list of data that you wish you had?
Doyne Farmer:
Yeah. I mean, the key thing that we really want and that we can’t get is the supply network. What’s erroneously called a supply chain. I say erroneously because they aren’t chains, they fork like crazy. And that is, as you go back, Ford might have-
Nick Hanauer:
It’s a supply tree, isn’t it?
Doyne Farmer:
It’s it a supply tree, exactly. It’s a supply tree.
Nick Hanauer:
That’s interesting.
Doyne Farmer:
And to really understand the economy you need to understand that supply tree, because as we’ve shown in other work, a lot of what companies do depends on that, and as I mentioned in COVID, whether the company could produce its stuff or not depended on whether its suppliers were giving it what is needed. So, you really need to know that information. You need to know that information also to track carbon emissions for lots of other things. So, that’s the key thing. Then we’d of course, like more information about financial flows. Some things are surprisingly well understood. We have a data set that records all the power and energy companies in the world and has information about 90% of their plants, and a history of what they own through time. Well, we’re building a model that’s just centered around that to help us get through the climate transition. And I’m confident that model’s going to work, because it’s going to be initialized from the bottom up with details about what all those companies did year by year.
Nick Hanauer:
So, just a few more questions. So, what are the implications for managing the climate crisis and policies related to climate change from complexity economics?
Goldy Goldstein:
There’s good news here, right?
Doyne Farmer:
There is good news, yes. That is, we’ve taken a totally different approach to this. The standard models say if we have a two degree target, and there’s a dictator of the world who can just force everybody to take the right decisions, what is the optimal path for us to take to get under two degrees as quickly and cost efficiently as possible? The problem with that approach is that it’s not testable because that’s not the approach we actually take. There is no such dictator, we make other kinds of decisions. And so, if the model doesn’t make good predictions they go, “Well, people didn’t follow the optimal path.” What we do instead is we just try and predict what’s going to happen, and then we try and nudge the world, try and find policies to nudge the world towards a better place.
And that has a big advantage of being testable, because we now have 30 years of data on the climate transition, and we can look and see what’s happened so far. And so, that’s what we’re trying to do. The good news is that we have some new models that will come out soon that show that the S-curves that technology follow, we can make predictions about deployment. And solar and wind, all the indications are they’re going to grow exponentially for another 5 to 10 years and then start to taper off, but they’re going to take over the energy system pretty quickly. And that’s good news because that’s going to help us hit a two degree target.
Goldy Goldstein:
And the faster we adopt this, the faster the prices fall?
Doyne Farmer:
That’s right. The faster the prices fall, the more money we save. Because we have argued, in what I hope is a convincing way, that we’re going to save money in the transition. It’s not a burden actually, it’s a profit making opportunity for business people to take advantage of.
Nick Hanauer:
Yeah, that’s fantastic. So, do you have some more glimmers of hope that the economics field will begin to move in this way?
Doyne Farmer:
Yeah. I’m confident that it will in some form, that is within 30 years these models are certainly going to take over. I think the question is do they do it in 5 or 10, or 20? And then there’s a sociological question of how will they take over? I think first they’re going to take over with private industry. I mean, entrepreneurs are going to be using these models. I’ve started a company called Macrocosm, and our goal is to sell predictions and models to companies. We also want to put them in the hands of policymakers who can use them to make better decisions. And I think when that starts to happen on a major scale, you’ll start to feel pressure to do different things in academia. Now, whether there will be complexity economics departments and normal economics departments side by side, or whether complexity economics will invade, or whether the mainstream guys will start doing what we’re doing and say they did it all along, I don’t know. But I can’t predict that, but I’m sure it’s going to happen.
Goldy Goldstein:
We had Angus Deaton on, was it almost a year ago, Nick? And we asked him this question about whether these new ideas were changing minds in academic and economics. And he did the old quip, “One funeral at a time.”
Doyne Farmer:
Oh. Well, that’s better than I thought. I mean, I’m glad he said that.
Nick Hanauer:
So, couple of final questions. So, we always ask the benevolent dictator question, if you were in charge of everything what would you do, Doyne?
Doyne Farmer:
If I were the benevolent dictator? Wow, that’s a big one. Well, I would collect all this data so we had a proper model to guide us, and I would really test the hell out of that model, and I would follow the predictions of the model. The model’s not built, I can’t tell you with high confidence what the right path is, though I have strong gut feelings, and I also like to separate my role as a scientist from my role as a citizen. My role as a scientist is to make predictions about cause and effect. So if we make this decision, where will it lead us? If we make this other decision, where will it lead us? And then I turn to the citizens and say, “Guys, what do you want to do?” And that’s where I put my vote in. So, I do think that when we see this, so I’m making a gut prediction here, that we are going to see that the kind of things you want to do, Nick, are really good ideas, they’re win-win solutions for everybody. Increasing the minimum wage is something we should do.
Rapid climate transition is something we should do. And if we do it right, meaning we don’t waste money on small modular reactors and boondoggles like that, we will actually profit financially from it. Not to mention making a cleaner, more secure, less volatile energy system that will just do a better job than the one we have, and a much cheaper one. And of course, a non-carbon emitting one, a more sustainable one. So, I think we could put ourselves on that path and get there pretty damn fast. I mean, if I was a dictator of the world I’d get us there in five years.
Nick Hanauer:
Oh, wow. Interesting.
Goldy Goldstein:
Okay, you got my vote.
Nick Hanauer:
Yeah, we vote for Doyne. And one final question, why do you do this work?
Doyne Farmer:
I love it. I love doing science, and I spent, roughly speaking, the first part of my career doing challenges. Can I beat roulette? Can I beat the market? I just felt like I had to do those things just to show I could do them, like climbing mountains. But now I go, “Well, I’m 70. I’m about to be 72. I want to put this stuff to good use. I want to make the world better, I want to provide us with tools that can give us better guidance into the future.”
Nick Hanauer:
Yeah, that’s fantastic.
Doyne Farmer:
Thank you guys.
Nick Hanauer:
Well, this was a terrific conversation.
Doyne Farmer:
It was a fun interview.
Nick Hanauer:
Technical, but fascinating.
Goldy Goldstein:
Well, the wonkier or the better, as far as I’m concerned.
Doyne Farmer:
Yeah, it was a lot of fun.
Goldy Goldstein:
I’ve mentioned this before, Nick, but my approach to orthodox economic models is to compare it to that old joke about the guy looking for his car keys under the street lamp, because that’s where the light is. Right?
Nick Hanauer:
Exactly.
Goldy Goldstein:
So, you could understand 50, 60 years ago, putting together these very simplified equilibrium models based on a narrow set of assumptions, because that’s all they could calculate. They didn’t have the computational power, they didn’t have the data, they didn’t have the algorithms so you could put together these very simplified models. And let’s be clear, equilibrium models are pretty simple because one side of the equation has to equal the other.
They always work. It’s the great thing about an equilibrium model is that it always works, but that’s changed over the past 60 years or so. And for some time now, we’ve had the computational power to do these calculations and to do these models. We also have, by the way, 60, 70 years of evidence that the old models? Well, they’re really not very good at forecasting the future. They are-
Nick Hanauer:
Or anything.
Goldy Goldstein:
Right, they’re wrong. And it’s worse than that, Nick, and this is something which I’m sure I’ve brought up before, in his book, Doyne makes this comparison to weather forecasting, which the weather, it’s a complex system and it’s not very good at predicting 30 days out. It’s pretty damn good at predicting three, four, five, six days out. And the better the data gets, and the better these models get, and the more we train it on actual data, historical data, the further out we can predict with greater accuracy. It’s always a probability. But here’s the thing, the difference between an economic model and a weather model. When the meteorologists get the forecast wrong, it doesn’t change the weather. But when economists get the forecast wrong, it changes the economy, because we change our policies and our expectations and our based-
Nick Hanauer:
And our approaches, yeah. Based on the forecast.
Goldy Goldstein:
Right. If you are predicting high inflation, people are going to think that we’re in an inflationary cycle and they’ll change their behavior.
Nick Hanauer:
Yeah. Or if you’re predicting a recession continually, that will cause people to hold back and create a recession.
Goldy Goldstein:
Or when you’re the congressional budget office and your model predicts that increasing the minimum wage five bucks is going to destroy two million jobs, the policy makers are going to be less likely to raise the minimum wage. And that’s actually, as we know, is going to mean we’re going to create fewer jobs than we otherwise would have had we … So, it’s a self-fulfilling prophecy, and that is the problem with bad models. Economists say, “Oh, well we’re running models. We’re just forecasting.” No, no, no. You are changing the economy with every forecast because it is a complex adaptive system.
Nick Hanauer:
Yes. No, it’s really true. Well, that was a fascinating conversation. Obviously it’s very wonky and complex, and we only scratched the surface, frankly, of the book and Doyne’s thinking. But I do think it’s a really hopeful conversation in the sense that there is real progress on the fringes of economics towards a better way of understanding cause and effect, and just a much more successful and fulsome way to understand how human economies work, and how to make them work better, which is exciting.
Goldy Goldstein:
Yeah, it is. And so, obviously we recommend the book, Doyne Farmer’s Making Sense of Chaos: A Better Economics for a Better World. You can buy it at that big online monopolist, or at your favorite local bookstore. And of course, we’ll provide a link in the show notes.
Speaker 6:
Pitchfork Economics is produced by Civic Ventures. If you like the show, make sure to subscribe, rate and review us wherever you get your podcasts. Find us on Twitter and Facebook at Civic Action and Nick Hanauer, follow our writing on Medium at Civic Skunk Works, and peek behind the podcast scenes on Instagram at Pitchfork Economics. As always, from our team at Civic Ventures, thanks for listening. See you next week.