Doan Farmer (2:58)

Thank you. It's nice to be here.

Gregory McNiff (3:00)

Dolan, why did you write Making Sense of Chaos and who is the target reader?

Doan Farmer (3:04)

Well, I just kind of wrote it because I felt like I had to. I wanted to communicate to the general public what I feel like is a problem for us as a society. Economists are not very interested or responsive to the few that I have, and so I tried to make it accessible to as wide an audience as I possibly could at a high level.

Gregory McNiff (3:30)

Your book discusses the application of complex systems to economics, which you characterize as a revolution in progress whose time has come. What is complexity economics and how does it differ from the standard economics we hear about today?

Doan Farmer (3:44)

Yeah, so maybe it's simpler if I just say very briefly how the standard economics works. Standard economics, and my book's about economic theory, not about just data and economics or econometrics as they call it. But, but in a standard economic model, you, you assign each agent a utility which is like a scorecard for what they like, and then you give them a model of the world which might be saying they can under. They understand everything. That's the typical one, which is called rational expectations. And they understand everything and they can do any calculation about anything. And then you assume supply equals demand and you write down the mathematical equations and you solve them and you solve for the action each agent will take, knowing that every other agent is maximizing their utility too. And so it's a bit of a not how I think about the world. And it's also computationally really untractable. As things get complicated, it becomes impossible to calculate what each agent should do. So complexity economics, in contrast, takes a bottom up approach, stimulating the economy or thinking about the economy from the bottom up. As agents who, who use bounded rationality, they use their, the limit limiting limited capacity that we have to think to make decisions about the world. And we do so on an ongoing basis. Every day we get up and make, you know, reconsider our decisions. And so, so it models, in the models, what happens is that instead of having mathematical equations. We have a computer simulation in which we have virtual agents who each time period make a decision based on the information they have. Their collective decisions affect the economy as a whole. And then that generates new information which combines with whatever other information is flowing in and they make more decisions and you just go around and around that loop as we do in real life. And that's, I think one of the main points is to get models that are closer to the way we do things in real life.

Gregory McNiff (6:10)

Yeah, that's obviously a key theme that we're not trying to model optimization or how a rational person would behave as if, but candidly as is, as real life works. And I think you cover that theme very well in your book. In the beginning you talk about complexity economics addressing quote, big messy real world problems Covid the economy, financial crises. Is it fair to say it's more of an application for macroeconomic issues or does it, is there a microeconomic in terms of the house and the firm that has that complexity economics would have applications for.

Doan Farmer (6:45)

Well, it really combines micro and macro and that we, we try and understand macro from the bottom up, meaning by understanding micro. So the simulations in an agent based model of the economy will typically have in either individual people or individual households, individual firms, possibly banks, possibly the government. But, but it simulates the behavior of individual actors interacting with each other.

Gregory McNiff (7:18)

You introduced some terms to describe complex economics and you just hit on one. Bounded rationality. I was wondering a, if you could talk about that and maybe two or three others like chaos, verisimilitude and emergent phenomena. They seem to be important.

Doan Farmer (7:34)

All right, thanks. So, so you may have to remind me as we go along. So let's start with bounded rationality. This is the term that dates back to Herb Simon, Nobel prize winning economist, sociologist, management science, AI, founder of AI. He, he did an amazing number of things in the 60s and he was fighting, pushing back against the idea of rational expectations, saying that's not how real people work. And so he, he uses the term bounded rationality to refer the fact that, you know, we're not dumb. We, we can think through some things. We can, we, we use our brains to make decisions in very useful ways. But we're not infinitely smart and that's the boundary part. And arguing that that makes a difference. It makes a difference in a couple of ways. One is that the models become more tractable because since you're just modeling boundedly rational agents, you can model their decisions with boundedly rational decision making objects in the computer that don't have to take everything in the world into account. And so you can model much more complex situations. You know, a typical mainstream economics model, there's maybe two or three agents. We could have millions of agents. So. And I think it's an idea whose time has come because in Simon's day, computers were a billionth as powerful as they are now. And we now have the computing ability, the data and refined methods to do what we need to do to make this a success. But it's not a trivial job.

Gregory McNiff (9:21)

No, absolutely, you've got it. The first one, spot on. Bound to rationality. I was going to ask you about chaos, and I think it will you define as a mechanism for creating disorder in an orderly way. What does that mean, a man or woman?

Doan Farmer (9:37)

So chaos is the phenomenon where nearby points become nearby states move far apart. In the future, nearby states move exponentially apart. And this means that you can make a measurement. And in a measurement, you never know exactly which point you're really dealing with because your measurement has finite accuracy. So there's many different states of the world that are consistent with your measurement, even though they don't seem very different when you make them, because it's after all, the best approximation you could get to a point. But if things are exponentially moving apart over time, they can separate so that after sufficient time, they may be scattered all over the realm of possibilities. They may all become far apart from each other. And so it's a geometric property of models that is very prominent in the weather, where the weather, small things can cause big changes to the weather. And there are many other things. I think in politics we see examples of this all the time. If, you know, hanging chads had not been president in Florida, he might have seen completely different outcome in Iraq. So this tiny little thing happening in Florida had huge implications for people in Iraq. And. And so that's one property of chaos. The other property is that you get what we call endogenous motion, meaning motion that's generated from within the system itself. So the weather, for example, is changing all the time, and you never have any two days exactly like each other. And that's because the weather is inherently chaotic and it's constantly changing as a result. And it changes on its own. You know, imagine sitting next to a mountain stream. The rocks are all fixed in place. There's nothing special happening, but the water is still churns and froths and splashes just because that's what the chaotic dynamics of water does. So now back to bounded rationality. How does that tie in with economics. Well, why do we have recessions, booms and busts? Sometimes like Covid, they come from the outside, but other times, like crisis of 2008, they happen spontaneously by the economy acting on itself. And so that's an example of endogenous motion. And, and I, the reasons are a bit esoteric, but I, it's a good argument that it's the only possible cause of the endogenous motion of the economy. And I think it comes from bounded rationality. We overshoot, we undershoot, we, you know, we don't make perfect decisions and we don't really understand where things are going to take us. 2008 crisis, some people understood that we were in a housing bubble, but nobody really understood that the what that would do to credit would cause a global recession that would dramatically affect the real economy. We just was too complicated for us to foresee. We could have foreseen it if we have better tools, but we didn't have those tools.

Gregory McNiff (13:06)

Absolutely. And expanding on that, another common theme is this emergent phenomena which the economy exhibits. Could you talk about how these agent models are better at capturing that relative to the standard economics? And what is emergent phenomena?

Doan Farmer (13:23)

So emergent phenomena are phenomena situations where the property of the whole is qualitatively different than the components of the whole. So the brain does things that you couldn't conceive of an individual neuron doing. Ant colonies do things that an ant couldn't even come close to doing because ants interact cooperatively with each other and even have the colony as a sort of primitive pheromone driven brain, our intermediated brain. So emergent phenomena are surprising things that happen. Chaos is another example. You can take very simple mathematical entities and couple them together and get chaos, even though the underlying equations are very simple. You get patterns on the screen that look really complicated. So how do we understand emergent phenomena? Well, we've learned that we really need to think about them from the bottom up because they emerge from the interactions of the low level components. And the challenge is to put our finger on what it is that makes those qualitative differences. The economy as a whole is an emergent phenomenon because through our specialized labor, we create a standard of living that's far, far beyond what we could ever achieve interacting if we were all on our own. And so almost miraculous what we can do working together that we could never do if we were just individuals working separately.

Gregory McNiff (15:04)

Yeah, I'm curious. I know we're obviously contrasting standard economics with complexity economics, but I think at one point you suggest Adam Smith viewed the economy as an emergent phenomenon. Is that accurate? Would you. But you said there's some kernel of emergence in the standard models.

Doan Farmer (15:22)

Yeah. Oh, totally. I mean, I think Adam Smith deserves credit for being the first or one of the first to observe that the phenomenon of, that an emergent phenomenon was possible. There's actually, it turns out I, after my book came out, somebody pointed out to me there's a person named Al Ghazali who in circa 1000 AD in the middle east made similar statements that, you know, pointing out that even if you just look at a loaf of bread, you may see the work of a thousand people and creating that loaf of bread.

Gregory McNiff (15:58)

It's interesting to see him laying the groundwork for that. And even a thousand years ago is absolutely amazing. So these are definitely themes I'm sure any type of community has addressed, how to organize and structure their economy and certainly make predictions. I want to hit on two more terms you discussed. One is verisimilitude. And then the second one is this idea of endogenous shocks, shocks that happen inside the economy, and particularly how complexity economics treats that relative to standard economics. So I guess maybe if you could a describe verisimilitude and why that's important.

Doan Farmer (16:35)

Yeah. So verisimilitude is an observation that there are many economics models that the mechanism of the model seems very different than the mechanism of the underlying economic phenomenon. And one example being the theory of rational expectations that, you know, economists, you know, not stupid, they're aware that people aren't rational. Even before there's now a whole field of behavioral economics that studies the ways in which we're not rational. But, but Milton Friedman originally made the argument that even if, even if real people aren't rational, sometimes they behave as if they were. And so he argued that you could model pool. Even if real pool players don't understand the physics of pool balls, they may play pool as if they did. And he's got a valid point. Sometimes that works. But I think oftentimes in economics, the models, when researchers invoke as if arguments, it's really a reflection of something fundamentally wrong with a model that leads it to wrong conclusions. Or oftentimes these models only work in very specialized limits. And so verisimilitude is the idea that you should trust the model more if the components of the model closely resemble those of the real world system you're trying to, to replicate. If the logic is, the logic of the system is the same as the logic somebody would use. And explaining to you, how does that system work? And so the idea is to replace as if with as is. And I believe that that can lead us to more realistic models of the world that we'll be able to rely on better.

Gregory McNiff (18:33)

Great. And I want to discuss that more because you give a number of examples. But last compare and contrast question. Could you talk about the role of endogenous shocks, which shocks that are from the inside, how standard economics treats it, and how complexity economics agent models, or as is approach, would handle endogenous shocks?

Doan Farmer (18:56)

Well, so this goes back to my discussion about chaos. Chaos generates endogenous motion. And this is something that people, you know, it's surprising, really shocked everybody when we discovered it because, you know, people previously assumed if something looked complicated, it must be that it is complicated. And whereas chaos can generate complicated things that actually aren't fundamentally complicated now. So the economy behaves that way too. You know, the economy is a non linear system that's interacting with itself and sometimes it's responding to outside events as it was during COVID and sometimes it's responding, it's actually creating internally driven events as happened in the 2008 crisis. And mainstream economics has a hard time dealing with the latter. The models, because they're based on rational expectations and because they neglect some of the realism that complicates the world and makes it harder to see the future, they don't take our bounded rationality into account. And I think our bounded rationality is really important feature because it tends to cause us to collectively overshoot and undershoot and deviate from what we might have done if we'd really known where we were heading. And so these ideas all connect together to form a different view about how the economy works and how we should go about modeling it.

Gregory McNiff (20:43)

And as you say, it is a relatively new discipline, but you provide several case studies in the book Covid the Housing Crisis, Leverage and the Financial Bust. Could you briefly talk about how predictive or how accurate these models were in helping us understand what were the causes? I mean, I think in one particular case you were trying to decide in the housing crisis, was it lower interest rates or lending policy that was the largest driver or catalyst. And the model that you and your colleague actually helped us understand that much better than again, the standard model. And I know there are acronyms for the standard model, but can you talk a little bit about success rate of the.

Doan Farmer (21:29)

Sure, sure, happy to do that. Maybe to start with the one I'm most proud of, our Covid model that, you know, we threw together based on different bodies of expertise that existed in my group had a literally around the clock effort for a couple months to try and model what the economic impact on the UK would be. And, and we were quite tickled because we predicted in advance of things happening what the economic would be right on the right on the money. And we even did pretty well industry by industry. We got the time trajectory of the effects pretty pretty close to right. And when we did a postmortem, we could see what things we did contributed to making that happen. We had a realistic model of the way production happens in industries and we modeled it as a dynamic phenomenon where in our model, day by day we would look at each industry and see does it have the labor inputs and demand it needs to produce its good and just let that dynamically roll forward in a way that's very different from the way mainstream models work. So that's a great example and that's been being developed into generalized set of models for dealing with supply chain shocks. Covid, the economic impact of COVID was essentially a big supply chain shock and it was a production shock too. But, but so we're building models that allow us to really think about the impact of supply chains and I think those are going to become sort of standard tool over time. Our housing model that you mentioned is another good example. It's perhaps the first example of a quantitative agent based model of the economy where we set out to model in detail the housing market in Washington D.C. and so we got, you know, records of real estate transactions over a 15 year period in Washington D.C. we had records of all the mortgages, we knew year by year how many people moved in or out of Washington in each demographic group. So we have detailed data about, about Washington and about housing choice. You know, we had, there's a government survey where, you know, they asked 40,000 people, what kind of house do you have? How did, what made you decide to buy it? Ask them a lot of questions about things that were relevant for us. And so, so we just tried to duplicate what happens in a housing market. So you know, you don't buy and sell houses in a stock exchange. You want to sell your house, you find a real estate agent, you find a similar house that sold in the past. You mark your house up a little bit, it doesn't sell after a month. You mark it down, keep marking it down until it hopefully sells or you decide I don't want to sell it at that price. So we had a simulation where individual households did that to just that household decided want to Sell, they would consult the real estate agent, would find past sales within the model. And, and, and similarly, if you wanted to buy a house, a buyer would begin by going to the bank and finding out whether they could get a loan and if so, on what terms. So then we would model put the buyers and sellers together, have houses bought and sold at specific prices, and then each household would follow the mortgage that they had through time. And so by having this detailed simulation the economy, first of all, it's an example of a model with a high degree of verisimilitude. We simply built the model by looking, by asking how does this work? And you know, having bought a couple of houses, I kind of, kind of knew how it worked. But we, you know, we talked to realist, we talked to everybody. We could figure out how it worked. But then it, as you, as you mentioned, because we were modeling the details of lending policy, we could model one world in which everybody got 30 year fixed interest loans as they did in the old days, and where the loans were only given to people who clearly had the resources to pay back those loans versus a world where we had subprime mortgages and people getting funky loans where they paid almost nothing for three years and had a big balloon payment. So, so we could really look and see how that changed the world. And we saw very strongly that if we had just stuck with the old fashioned 30 year loans at fixed interest rates, no subprime lending, we wouldn't have had much of a housing bubble. It was really caused by the subprime lending and the funky mortgages attached to many subprime loans.

Gregory McNiff (26:43)

I want to drill down that question, particularly the success of your models versus the Fed model. And I think you point out the Fed was woefully inadequate in running scenarios that led to this breakdown, the housing crisis. And you have a very interesting line in your book that basically you say macro models used by banks assume we live in a world of risk rather than one of uncertainty. Can you unpack that? Why is that different?

Doan Farmer (27:13)

Yeah, so in, in the technical sense there it goes back to a economist named Knight in the 1920s. Uncertainty. He's making a distinction between risk defined here to be a situation where they're pro there things aren't deterministic, you don't know what's going to happen, but the unknown things, the random components, you know, roughly the probabilities attached to them and, and importantly, you know, the possible, the set of possible outcomes. In a world of uncertainty, you don't know the probabilities attached to Uncertain events, you don't know if they're likely or not very likely. And you may not even have the full list of all possible outcomes. I mean, how many people would have predicted a decade ago that we would be sitting with Donald Trump in the White House? And the kind of government we have right now, whether you're on the right or the left, I don't think anybody was predicting that, or a tiny few were. They were quiet about it. So we face uncertainty all the time. And it's important to find ways of modeling the world that can cope with uncertainty, because that's the way it is in the real world. Traditional economic models only assume risk. That is, they assume that the things you don't know, you at least know their probabilities and you know, the list of possibilities. Whereas, you know, the way we make decisions in the real world, we've developed heuristics that allow us to cope in situations where there are a lot of uncertainty. Example. A common heuristic is imitate the best. If you're. You're dealing with an uncertain situation, you're in a crowd of people, you don't know what to do. Well, look around. Does somebody else seem to know what they're doing? Imitate them. I mean, it can lead you off a cliff, but it's probably a better strategy than doing nothing. And. Or trial and error is another great example. Try something. If it doesn't work, try something else. If it works, stick with it. And that kind of evolutionary strategy, people use it all the time in business and elsewhere. I mean, you may use informed choices, you may use your reasoning to inform the choices you make and the things you try. But. But trial and error is a great heuristic. So it's one of many. These are one of many examples of how we make decisions under bounded rationality. And I think it's important to understand that in the way we actually navigate our way through the world.

Gregory McNiff (29:58)

Yeah, I wanted to ask about heuristics. First of all, you say in certain cases, they're more reliable than the models. And I think you use the term gut science. And two, as I mentioned before we started the interview, I have an MBA from Chicago, and it candidly feels like standard economics, the University of Chicago would be, if not ground zero, pretty close to that. And we were taught heuristics there. So my first question is, is heuristics common ground between standard economics or complex economics? Or is standard economics slowly starting to adopt this behavioral approach, maybe pioneered by.

Doan Farmer (30:34)

Yeah, good question. So first of all, business School isn't really quite the same as economics. You, you know, people may even sit in economics departments and teach in a business school and some very conservative ideas come out of business school. But business schools, you know, you're training businessmen, businessmen. Heuristics are an important part of decision making. The place where the difference happens is in mainstream economics over in the economics department where they're trying to build theories for how the world works and where they're very reluctant to base models on heuristics. They, they feel like for a model to be trustworthy it has to incorporate human reasoning and heuristics makes them nervous as a weak way to do that. Now that said, there's a branch of economics called behavioral economics where they've documented heuristics that people use and they do study these things, but that's not well integrated in their economic theory at all. So in other words, any good economist will acknowledge that heuristics exist but, but they have not understood how to deal with them in their models. Whereas we do this all the time. So that's our approach, takes heuristics seriously and uses them inside the models.

Gregory McNiff (31:58)

We, we touched on the background complex economics and I absolutely want to get into it because it's fascinating from a cross disciplinary approach approach, it seems like it was born out of physicists. It has a strong mathematical bent and we should get into the biology and chemistry. But to your point, complexity economics starts with the data and develops a theory. Whereas the economists, basically they're starting up starting with the theory and then testing it in the economy. Is that so the differences in approaches.

Doan Farmer (32:27)

Well you're certainly right that as, I mean I'm a physicist by training and a physicist immediately says well let's take a close look at the data before we start making theories. And there's another movement I've been a part of called econo physics where it was quite striking the difference the physicists took to understanding the financial system and the economists took and this is a key element. I once overheard a few, you know, prominent economists at a meeting where there were both econom physicists and mainstream financial economists shaking their head going well these physicists really come at things differently. They want to look at the data first and then build them all. They don't understand why that's not the right way to go. Well, that's a point of compension. I mean in their defense, economists go the other way because they say social systems are noisier than physical systems. The data may not show us very much. We need to approach the data with a well structured model in order to be able to make progress. I hear the argument, but in the end I think it's important to attack things from both sides. And I've always found exploratory data analysis is an important first step to get a feeling for what's really going on.

Atblinds.com Advertiser (33:49)

Mint is still $15 a month for premium wireless. And if you haven't made the switch yet, here are 15 reasons why you should. 1. It's $15 a month.

Doan Farmer (33:58)

2.

Atblinds.com Advertiser (33:59)

Seriously, it's $15 a month. 3. No big contracts.

Doan Farmer (34:03)

4.

Atblinds.com Advertiser (34:03)

I use it. 5. My mom uses it. Are you, are you playing me off? That's what's happening, right? Okay, give it a try. @mintmobile.com Switch upfront payment of $45 for.

Mint Mobile Advertiser (34:13)

Three month plan $15 per month equivalent required New customer offer first three months only, then full price plan options available, taxes and fees extra. Cmintmobile.com Running a business comes with a lot of what ifs, but luckily there's a simple answer to Shopify. It's the commerce platform behind millions of businesses including Thrive Cosmetics and Momofuku. And it'll help you with everything you need, from website design and marketing to boosting sales and expanding operations. Shopify can get the job done and make your dream a reality. Turn those what ifs into Sign up for your $1 per month trial@shopify.com specialoffer.

Gregory McNiff (34:48)

Yeah, no, that makes sense. And just based on this discussion here, given the success complexity economics has had and the fact that Herb Simon won the Nobel, you would think mainstream economics or academia would be more open to embracing it. But I get the impression from your book there's still tension or they still remain two separate camps. Could you talk about why that might be? It seems like, yeah.

Doan Farmer (35:17)

So first of all, there is tension and there are two separate camps. You know, I have colleagues who are prominent mainstream economists who champion what we're doing, but there aren't that many. You'll note. I had a review from Larry Summers on the back of my book. He's surprised me and you know, I sent him the book just as a courtesy because I mentioned his name and he read it and gave me substantive feedback and said he thought I was onto something. So there are some prominent economists that are pushing back against their own mainstream colleagues, but not many. And, and so from the rest we get either we just get completely ignored or we get a lot of pushback. And, and I think there are several reasons for that. Going back Herb Simon and Herb Simon's day, there wasn't the computing power to carry out the agenda. Economics went the other direction. Herb Simon is somebody economists will have heard of, but a young economist. They're not going to read much Herb Simon stuff in their training. You know, economists, it's very different than the way economists are taught to do things. They're taught a completely different method for building a model. It's hammered into their heads. You've got to keep it simple. You've got to take into account the fact that people can think and in particular their taught models based on rational expectations. The textbook's full of them. The mainstream models that are used at central banks are still like that, even though there's a whole branch of economics called behavioral economics pushing back against that idea. And then the third reason is, I think human nature. If we're right and they're wrong, the legacy of existing mainstream economists will be diminished and something else will take over. They'll have people in their departments doing completely different things that seem very foreign to them. And so it understandably makes them uncomfortable and makes them feel threatened. They have to get well out of their box to appreciate it. And so it's only more open minded people and people who are comfortable with where they are, I would say, that usually embrace this. Yeah.

Gregory McNiff (37:35)

It does feel like the time is coming though, for complexity economics. Like you said with computing, we really haven't talked about AI, but that will certainly play a role there. And then this disciplinary approach, and I want to get into that because I'll be honest, your book is fascinating from start to finish, but the piece where I really thought this is interesting is how you compare the efficient market hypotheses to sort of a market ecology. And you say complexity economics views economy as, quote, an adaptive complex system. Biological concepts such as metabolism, ecology and evolution are very useful for thinking about it. And this sort of flies in the face of fama's efficient market hypotheses, which I think it's fair to say is really all the information is embedded within the stock price and any additional information will lead to a random walk. Could you talk about why you view the economy as a metabolism and how that helps your thinking about how it should develop or grow relative to the EMH random walk approach?

Doan Farmer (38:41)

Sure. So I think of the economy as the metabolism of civilization. So because what does the economy do? It takes in goods and services. Sorry, it takes in natural resources, combines them with human labor and sunlight and oil combines them to create goods and services that we want or think we need. And it's much like metabolism in a biological organism. What does metabolism do? It takes in material from the environment, decomposes it, and works on it to extract the energy and building blocks it needs to replenish itself. So there's a strong analogy there. And in fact, I point out in the book that the equations one uses to study the two are essentially the same, with the names of the variables changed. So. So the metabolism is this remarkable thing that allows us to exist as a civilization. But taking that another step further, how do we do that? Well, we're in a. We have an ecosystem of specialists. We all do our specialized roles, and as we spoke before, that combines to create something that's far beyond what we could create as individuals. And so it's useful to take the tools from ecology, because, after all, what are species? Species are specialists. If you, you know, grass is a specialist at taking sunlight and earth and turning it into grass and water and turning it into grass. Zebras are specialists at turning grass into zebras, and lions are specialists at turning zebras into lions. And so to think about the way an ecosystem works, you really have to think about the specialized role of each species and the interactions. And then you realize there are very indirect interactions, that if you kill all the lions, you may run out of grass because the zebras will overpopulate. And so you really have to think about all the components of the system. So complexity economics embraces that ecological perspective when thinking about the economy, and I give several examples in my book. And then finally, evolution. Well, evolution in mainstream economics is used to justify efficient markets. Selection drives us quickly to efficient markets, but in biology, it does just the opposite. It's used to understand our diversity. How did we end up with diverse species, and what is the process of change through which that happens. And so I'm arguing that we need to think about those things and recognize that even in finance, the path to efficient markets is slow and noisy and constantly being disrupted by new developments. And so the efficient markets theory works okay for some purposes, but it doesn't work at all well for others. And instead, we really need to think in more ecological terms, where we think about the role each different types of investors are playing. Value investors play a very different role than trend followers and momentum traders or market makers. So the financial system as a whole is the outcome of all these different specialists interacting. And in that vein, it becomes much easier to understand why markets malfunction.

Gregory McNiff (42:26)

Dolan, would you say that the economy evolves?

Doan Farmer (42:30)

Yes, the economy is evolving all the time. I mean, we're in the AI evolutionary period right now where large language models are causing rapid change in the economy. But it evolved. I mean technology is an important part of that evolution, maybe even the most important part. But it's always evolving in other ways too. Our institutions are evolving. You know, 200 years ago we didn't have central banks. Playing the role they play now makes a significant difference. So we didn't have sophisticated financial instruments that we have now. We didn't have mortgage backed securities until the late 1990s and we can see what that did.

Gregory McNiff (43:19)

You made a great point about the equations in the book you cite. I believe it's the Latke Volterra equations which refer to predator prey systems and your own equations in a market ecology paper were very similar to that. Is that correct? The way it sort of.

Doan Farmer (43:36)

I was able to derive equations that are essentially the same as the Leca Voltaire equations by just by looking in that view of the financial market strategies. Pray in quotes on other strategic praise can be a bad word. But as my lion, zebras and grass example gives, pray are also very important and striking a balance and making things work. Right. And, and so I, you know, in developing my theory of market ecology, I make it clear how different strategies interact with each other. A trend following strategy first of all creates opportunities for other trend following strategies. If you're a long term trend follower, you help short term trend followers, you, you enhance their profits, you can even derive that. And, and trend followers create opportunities for value investors because they drive the market away from fundamental value giving away for the value investors to make money. And so we, we need the whole ecology of financial participants and, and they have to be in balance for the financial system to work. Right.

Gregory McNiff (44:57)

I want to move on to your analogy as the economy is a network. But as an mba, I would be remiss if I didn't ask you the question you seem to have gotten constantly on your academic career is how good are these models for predicting the stock market? Could you briefly talk about that and your experience with Prediction Company? I probably should have mentioned that in my opening but you actually have real world experience in the financial markets. You know, basically predicting and investing on a day to day business and day to day basis have a very successful track record.

Doan Farmer (45:31)

Yeah. So I'm unusual in leaving academia for eight years and then coming back in and but my eight years outside I was one of the founders of Prediction Company which used, you know, state of the art statistical methods and computation to detect patterns, consistent patterns in financial time series and place. Bets on them. So there, you know, we practice, we processed massive amounts of data and we made bets on the idiosyncratic movements of stocks relative to each other. So making a lot of. None of our models were very good, but we made enough bets that we made very, very consistent profits in spite of that. It's like flipping a coin that's biased to be heads 52% of the time. It's pretty risky if you just take one bet. If you flip that coin a million times, it becomes a near certainty that you're going to win. And so we made extensive use of that principle. Now I'm embarked on a new enterprise. I have a new company called Macrocosm. And we're approaching things in a different way now. We're simulating the economy. So rather than just looking at the streams of data and trying to detect patterns, we're actually trying to understand how it works through complexity economics models and make bets in markets based on having an understanding of what happens. And I'm excited about this. I mean, it's new, I think, I think it will be effective, but it's. It. You know, in this case, I don't want to keep the models a secret. I want to make sure that the models ultimately get in the hands of decision makers who we want to let use them, you know, at a pro bono basis in order to provide the world with better advice. We're trying to use the financial markets as the money cow to build the sophisticated models that I outline of, you know, the Google maps of the economy. And so we want to use it as a vehicle to build those models and put them in the hands of people who can help us by using them.

Gregory McNiff (47:54)

That sounds like a very admirable goal and thank you for that. I know in the book you talk about the COVID model, and unfortunately, the number of deaths there, any type of business cycle, can result in significant job losses in the millions in this country alone. So just understanding these dynamics better can have a profound impact on a number of people's lives. So that seems very worthwhile. We talked about the market ecology approach relative to efficient market hypotheses. You also suggest the economy can be viewed as a network in which networks are essential building blocks. Could you briefly talk about that analogy?

Doan Farmer (48:30)

Sure, sure. I mean, that network is, is a mathematical simplification we make in complex systems to get an idea about the structure of something, because sometimes just understanding its structure is very informative about how it works. And so a network consists of nodes and, and links between the nodes. And so Those nodes could be banks, and the links between them could be loans the banks make to each other. The nodes could be people and their balance sheets and links they make to other balance sheets, which could be contracts like my mortgage, or they could be transactions that we make with other people. And so there are many examples of networks that we use in many different ways as a conceptual tool to think about the structure. How connected are things? Are they in clusters with islands and weak connections in between the islands, or is everything very tightly connected? Do nodes tend to connect other nodes that are like them, or do they connect to nodes that are different than they are? And those things make fundamental differences in the way the network works? So, and networks are, in a sense, the skeleton on which you build agent based models of the type that we use. In complexity economics, we talk about the.

Gregory McNiff (50:02)

Parallel swing, market ecology, and again, the overall market. I was surprised as well. There's a relationship between fluid, I guess fluid turbulence and financial turbulence. Could you talk about that relationship?

Doan Farmer (50:17)

Turbulence in hydrodynamics refers to a state in which a fluid, like water, is what you would intuitively call turbulent. It's churning around. Different parts of the water are moving in different directions. If you're moving through the water, you get randomly buffeted around, as you do in a plane. When the pilot says we're experiencing turbulence, suddenly something hits you. You can't predict. And this is an emergent property of the weather. It's why sometimes you can have a hurricane another other day, you have a complete calm. That's part of the bigger, bigger picture of turbulence. Financial markets act much the same. It's, in this case, it's individual financial actors trading with each other, boundedly rational. So they're overshooting and undershooting and causing things like a flash crash. My colleague Jean Philippe Bouchot has been documented. We have things like the flash crash happening all the time based on no apparent news. It's just spontaneously generated by the financial markets themselves. And they're even. As I discuss a little bit in the book, the statistical properties of financial markets are strikingly similar to the statistical properties of fluid turbulence. They're essentially identical. And you can use models used to predict volatility in a fluid flow to predict volatility in a financial market. The exact same model works and makes very good predictions about those things.

Gregory McNiff (51:52)

Yeah, I found that astounding. And I think we have a discussion about the scaling laws across financial data as well as in, I guess, how that might relate to certain discussions of fluid turbulence. Is that accurate? These scaling Laws. Yeah, it was very interesting. Moving on, I also wanted to ask you, you address a number of applications for complexity economics. Perhaps top of mind these days is climate economics. Could you talk about how complexity economics can help us understand climate economics and make the transition more, I guess, less painless and more seamless?

Doan Farmer (52:29)

Yes. You know, we're undergoing a rapid change. We're far away from equilibrium, we're moving into uncharted territory. And yet the devil's in the details of how should we make the transition, which energy sources should we invest in, which political agreements should we make, what kind of regulations are achievable and effective. And so we're working to understand all of those things. So yeah, and I think complexity economics is well suited to help chart the course.

Gregory McNiff (53:12)

Okay. Another theme I wanted to touch on, and we've touched on this in this conversation, is this cross disciplinary approach with complexity economics. We've talked about biology, physics, I think there's even some psychology involved with heuristics. And you've pointed out how these lock of Volterra equations mirror certain behavior in financial markets. And you also, I think at your time at the prediction company came across this idea of a universal law for market impact. Was that a surprise? The sense that again there was sort of the equivalent, I think you have it as the inverse square law of gravity and physics corresponds to the market impact. Are you surprised that there are these sort of ironclad rules in these disciplines that seem to mirror each other and disciplines that don't seem necessarily related?

Doan Farmer (54:00)

I'm surprised and I'm not surprised. I mean as a physicist, the physical world is full of remarkable regularities that somehow all fit together to give us this beautiful world that we live in. And economics, our economic interactions have some things in common. And, and, and, and so I'm not surprised there are some areas where things like this emerge. It's a very beautiful thing to see, really. And as a scientist, a lot of fun to try and understand properly. So yeah, and it's a good example where there are things in economics that really are like the laws that we have in physics. It's a remarkably good quantitative law of markets. Yeah.

Gregory McNiff (54:54)

Again, we've touched on the growth in computational power and certainly AI. Do you expect those two tools to drive significant progress and complex economics going forward?

Doan Farmer (55:06)

Yes. I mean AI is going to drive progress in all fields of science and economics among them. I think the models we're trying to build, we put AI agents in a very clearly defined context. And you really need, there's just not enough data to do in economics, what has been done in weather and other fields where we have just tons and tons of data to train neural networks on. In economics we just don't have as much data. And so I don't think we're going to get there in the same way. We really need to do pursue more of the approach that was taken when the game of Go was beaten, which is to actually have the agents play the game with each other. In this case, we put agents in a field where they have competitors and where the economic rules are spelled out and where they try and find strategies that allow them to compete. So we use evolution to effectively train agents that can operate well in the economy. So that's a whole unexplored or only beginning to explore the possibilities for doing that right now.

Gregory McNiff (56:26)

Yeah, that sounds exciting. Last question. When do you think a complexity economist or economic specialist will win the Nobel Prize? And who or she do you?

Doan Farmer (56:38)

I can't. Maybe in 15 or 20 years. I think within the next five to 10 years we're going to see real breakthroughs in complexity economics because we're going to start beating standard metrics that the mainstream has set for itself, like accuracy of predicting the economy. And, and once that happens, we'll start to get more mainstream attention. And, and, and then I think it's just a matter of how long it takes the social process to play out. You know who? I don't know. I mean, Brian Arthur would be a name that would come up on people's list. Jean Fleet Bouchot, my colleague Stefan Turner. Perhaps, but we'll see.

Gregory McNiff (57:27)

On that note, that concludes our interview. Again, the book is Making Sense of Chaos, A Better Economics for a Better World by Doan Farmer. Doan, thank you so much for joining us today and running what is clearly a very thought provoking and just fascinating book.

Doan Farmer (57:41)

Well, thank you. It's a pleasure.