 Well, if we look at history, this isn't true. If you have largely capitalist economies based heavily on free markets or whatever, they produce good results some of the time. Some of the times they produce not so good results. Some of the time they produce terrible results like the Great Depression or the recent crash. Some of the time they produce results which threaten the very existence of the society that they're in. So if you want to get a model that will reflect or make possible or incorporate the actual good news, bad news about market economies, you have to have sets of realistic assumptions about market economy. If you're allowed to make fairytale assumptions about market economies, right? We know the future, the markets work perfectly, everything's good, nothing bad can happen. That's the only way you can get the conclusions that the profession is seeking, which is markets work perfectly. You can't get that with a realistic theory. Right. But as you're arguing in the paper, the next part of this argument of Milton Friedman's is that the proof of the pudding is in the eating. That is, you have to look at the empirical evidence. And if the empirical evidence proves the theory or is consistent with the theory, then you can make the inference that the theory is correct. So can you talk about the empirical evidence in support or not in support of efficient markets theory apart from pointing to one crisis or another? Can you talk us through a bit of that? Sure. What should we say about that? First of all, you can't use just the predictions of a model to test the realism of the model because if you want to test predictions, let's say in the early 1980s President Reagan came in and we had supply-side economics and said if you cut taxes on corporations and the rich, then the economy would grow very fast. So what do you need? You need to predict what will happen if this was in 1982, say what's going to happen in 83, 84, 85, 86, 87. So what happens if you get data from those the next decade and try and see whether or not the economy has grown because of the tax cut? Well, basically what you'll have to do is say, well, maybe it didn't grow, but that was because of other reasons, bad things happened. Or maybe it did grow, but that was because of changes in the economy. So you have to have a complete set of all the things that might change, economists use the assumption setter as parabas to say if everything doesn't change, this would happen. But everything always changes. So essentially you can't use simple prediction, you have to use econometric testing. But econometric testing is unreliable for a number of reasons. For one thing, we now have advanced technologically enough so that if I have a hypothesis about something changing will cause something else to change and I want to test this statistically, I can run three or four million regressions to test it. And just by luck or random chance, some of those regressions will in fact be consistent with the hypothesis. So that's no good. Secondly, there's something, it's just maybe a little subtle, but in terms of the methodology, there's something called the Duem coin problem in hypothesis testing. That is, you have a theory, a neoclassical theory, mainstream theory, which basically says that every agent will do the best they can under whatever circumstances they are presented. So then you have to put in the circumstances they're presented. So suppose you want to say that if you're going to cut taxes, then you're going to get increased economic growth. So you have to then create a model. You have to create what people's expectations are. You have to put all kinds of things in the model, legs into the model and whatever. And none of these things, the expectation formation, the legs, the quality of the production functions or whatever, none of that's part of the theory. So you have ideas from theory joined with all kinds of assumptions about expectations or legs or something or other. If the thesis turns out to be rejected, it may not be because the thesis is wrong, it may be because the subsidiary hypotheses are wrong. So econometrics can't just bear the weight of this. So what we actually need to do is to use all the available ways of trying to evaluate whether theory is consistent with reality, including whether it's hypotheses or main assumptions are consistent with the world. We need to study the institutions and facts of the world and use those. And yes, we can use some econometrics as well. Now, what some economists would say is, OK, a lot of what you said is true, but this is the only methodology we have. This is the only way we can study the world. But in your paper, you say that's not true, that there is an alternative tradition of studying these things that we can apply to financial markets. Could you say a bit about what you call the Minsky-Keynes approach? Yes. There's an alternative methodology, which let me just associate with Keynes at the moment. In the paper, I have some citations from Keynes to kind of back this up, which is what your mother might tell you to do. If you say, I want to study how the world works, and she probably would say, well, the first thing if you're going to build a theory would be you should probably get the facts of the world right. So the most important or maybe the single most important thing is that the assumption set that you have ought to be at least demonstrably sensible or reasonable or have some relationship to what you study. If you want to study financial markets, and you want to study agents operating financial markets, you should assume that people don't know the future. But they have to make decisions anyway. So they have to come up with some ways to guess about the future, and they're going to use kind of rules of thumb and certain tendencies to extrapolate what's going on in the recent past into the future and to have more or less confidence about this or whatever, that a realistic set of assumptions is the foundation of a useful theory. And so we should test the usefulness of a theory by the realism of its foundations and econometrics and any other tests that we can use as opposed to ruling these all out and relying solely on econometrics which can't bear that burden. So Keynes' main innovation or his most important assumption is that we don't know what's going to happen in the future. So he says about such matters, about the things that will happen in the future, what will be the profits in the future, what will happen to financial markets in the future, he says, I'm quoting, we simply don't know. But as people have to make decisions, we have to do something about this. So that the basic assumption set we would associate with a Keynesian or later on, Hyman Minsky picked up on this in the 20th century in the latter part of the 20th century is to say that what we have to think about is the following in markets. People don't know, but they have to guess. In order to guess, they tend to more or less to think that what's happening in the recent past is going to happen in the future unless something changes their mind. If they make these guesses by extrapolation, this expectation is formed by extrapolating the trends that are going on and it turns out to be useful like it works, which happens in a boom. People become more confident that the extrapolations or expectations that they're forming are actually useful or accurate or good. So how much risk someone is willing to take, for example, will depend on how optimistic they are about the future through these expectations. But also how confident they are in their optimism. So as a market begins to pick up and a bubble begins to form, people begin to extrapolate what's going on. They think things are going to grow, the prices are going to grow, people aren't going to make losses. They become much more optimistic. They become more confident in their optimism. More people want to buy, fewer people want to sell, prices get driven up. What about risk aversion? Is that just exogenous? People are just born with a certain degree of risk. Well, if you start to take what seemed like some risk in an economic expansion, you buy some stocks or you put more stocks in your portfolio than bonds and more risky things than safe things, then it all works out. Then you're going to think this kind of risk thing is pretty good. Either I like the risk or there isn't as much risk as I thought. All of that's indigestly developing kind of in the market. The people in a financial boom who make the most money are those who are most aggressive and those who begin to borrow money to make investment. And so they're the people who make the most money and they're the people who get richest and people want to emulate them. So I think I should do what they do. So people then begin to say, OK, so I want to take more risk and I want to borrow more money. And through not anything that happens from outside the system, but from what's happening in the system as it moves through time, indigestly, as we say in economics, by its own forces. We can see where economic booms come up, where financial booms come up, where things get overpriced, where people take too much risk, where households borrow too much to finance things, invest in things that are actually too risky and can't sustain themselves. Where financial institutions actually begin to take too much risk, which happened in this last thing. So this is the kind of vision you get once you get rid of the assumption that everyone knows the future and it's stable. Once you get rid of the idea that people's attitudes towards risk has just given and doesn't change. Once you get rid of the idea that people can't default on debt, that's not risky. When you get rid of all that, you begin to get theories which say there's an indigestly generated instability in financial markets, which is the reality that we actually observe for hundreds and hundreds of hundreds of years. It's an amazing feat to have observed centuries of instability in financial markets and come up with a theory that said they can't be there. Thank you very much. You're welcome.