We are so happy to bring you all our 200th episode, and who better to have on the podcast on this auspicious occasion than the legendary, Professor Gene Fama? This is one of the most jam-packed episodes we have ever recorded, with Gene providing concise and thought-provoking answers to our many, many questions. After delving into the foundations of Gene’s work and philosophy, covering market efficiency, and its competing theories, Gene entertains our queries about a wide range of ideas and models, and generously shares the decades worth of wisdom that he is so widely known for. We also find time to talk about retirement plans, inflation, cryptocurrencies, and the influence of machine learning. Towards the end of our conversation, our guest touches on some more personal ideas about productivity, his career, his partnership with Ken French, and what success means to him at this point. For a landmark episode, with a true hero of the evidence-based approach to investing, make sure not to miss this.
Key Points From This Episode:
Gene, what does it mean for a market to be efficient?
Well, a simple statement is that prices reflect all available information.
What are the main implications for investors if markets are efficient?
Well, you can’t expect that activities like picking stocks are actually going to generate superior returns for you. So you get the risk adjusted return appropriate to the risk level that you take with your portfolio, but you can’t expect more. That’s another way to fish markets hypothesis. So your risk adjusted and expected returns are basically zero.
So what are the empirical tests that support market efficiency?
I warn you in the beginning. I chuckle when it’s bad news. So if you hear chuckling, you know bad news is coming. So the strongest evidence from your perspective is that if I look at actively managed portfolios, what I find basically is the distribution of returns around zero for excess returns are normally distributed around zero before fees and expenses. After fees and expenses, it’s a big negative sum gain for those who go into that active management. So the distribution of outcomes looks a lot like what you’d expect by chance if there were no ability to pick investments that have above normal risk adjusted returns. So that’s the strongest evidence I think from the perspective of investors.
Does market efficiency imply that returns are random?
No. Well, depends on what you mean by random, that the deviations of returns from their expected values, where the expected values are function of the risk of the security that the deviations of the returns from the expected values are zero.
So what are the biggest empirical challenges to market efficiency?
Oh, so I don’t know if it still is, but in the past there was momentum, was the biggest one. Aside from that, well, if you really want to push it to the limit, the fact that insiders make money on their trades is the violation of market efficiency as far as those investors are concerned. So the insiders clearly have information that isn’t already in the market price and they can profit from it. What’s kind of surprising is that their average profits are so low, they’re about 1%.
You made a quick remark there, but I want you to elaborate. Why would momentum no longer be a challenge to market efficiency?
I don’t know. I haven’t seen any updates of the evidence. That’s all. So the evidence goes back maybe 10 or 15 years. So I don’t see what it looks like since the last paper was published, easy enough though, Ken French on his website has momentum portfolios. So you can go in there and check the last 10 or 15 years anytime you like.
Did the empirical challenges like momentum, does that change the way that investors should behave?
Not really because it’s such a short term phenomenon and it’s such a high trading cost phenomenon that there’s not really any way to take advantage of it.
So is there an efficient markets explanation for what happened with GameStop?
I don’t follow these individual little aberrations if you might say. But remember now, efficient markets is a model. We call it a model because it’s not reality. It’s an approximation. Models are approximations. It’s an approximation that works quite well for almost everything you want to do in investing. But sometimes there are aberrations. So I didn’t follow the scheme stuff thing very closely, but apparently that was an aberration, it was a small stock that went crazy.
Yeah. That’s about the story. Did the ongoing flows into passive funds pose any potential challenges for market efficiency?
Well, you can’t have 100% of the money going into passive funds because then there’s nobody there to trade to make the market efficient. So the people who actually have information that other people don’t have, you want them to stay in the market and use that information. So the real question that nobody’s never answered is how many of those people are there out there? How many does it take to make the market efficient? So most of the trading by these investors just offsets the dumb things that other active managers do. So active management doesn’t always make the market more efficient. Sometimes it makes it less efficient because people make bad bets. So the informed people have to offset these uninformed people make the bad bets. So it takes more informed to offset the uninformed the more uninformed there are.
Interesting. Do you think that the inelastic markets hypothesis changes anything for the relationship between flows and pricing?
I don’t know, that’s a hot real one. My colleague two doors down from me, Ralph Koijen is working on that, right? But I want to see how that evolves to see what its investment implications are. So basically that’s a different point. The point there is that the demand for individual securities is not flat at a given price. Trading actually, when people move into a security that actually has a permanent effect on the price. Now, they haven’t really finished testing that out, the initial stuff looks very challenging, not for market efficiency, but for the idea that trading doesn’t have a big effect on prices. So we’ll see how that all works out, but it’s very interesting new line of work.
So interesting. So Gene, what are the shortcomings of the CAPMs or the Capital Asset Pricing Model as an asset pricing model?
I spent my early life on… Long time ago I was one of the initial testers of the model and for the first 10 or 15 years that the model was around, it did pretty well on the data. And then the so-called anomalies started to pop up where people who were uncovering things that were inconsistent with the models predictions. And little by little, basically at one point, Ken French and I wrote a paper that said there were just too many anomalies here, this model is dead. So the basic problem in the end was if you look at a long period of data, the relation between market betas and average return is basically flat.
And according to that model, it shouldn’t be flat. Average returns should increase with market beta. And there’s not much evidence of that in the data. So that’s the first order implication of the model and it doesn’t stand up very well. So asset price, it would’ve been great if that model really stood out to the data, because it’s such a simple model, you can teach it to almost any students in 15 minutes and they get the story. And then another one looks a lot more complicated than that.
How did you and Ken choose the size and value factors to create the three-factor asset pricing model?
We chose them based on the fact that these were at the time, the two biggest anomalies for the CAPM to deal with. So the CAPM couldn’t explain small stock returns, and it couldn’t explain the difference between valuing and growth stock returns. So we said, okay, we’ll add to the market portfolio these two factors that will basically absorb those two effects.
So where does the three factor model struggle to explain the differences in returns?
Well, momentum, the thing we started with, blows it up. So nothing explains momentum except momentum. So if you want to explain momentum, you put in a momentum factor, otherwise you haven’t got a chance.
How did you guys choose to add profitability and investment to the five-factor model?
Well, there’s some justification for that in terms of just normal valuation theory, which says if you hold constant and other variables, then you should observe a positive relation between profitability and expected returns. But that’s a holding constant is important there, it’s not a one dimensional story. So we put that in there based on that. We also put in an investment factor. So when you consider all these things together, what you should see is that average returns very positively with profitability and negatively with investment. And there’s evidence of that in the data, but the investment part of that is weak. So my hope is to have less factors that you need rather than more, because the simpler the world is, the easier it is to deal with it. So I’d be really happy if those two factors dropped out of the story, because there wasn’t much empirical support for them. I’d be real happy if it turned out that in the long term, the CAPM worked really well, because then our lives will be a lot simpler in that case. But so far it hasn’t worked.
Is the five-factor model still an empirical model?
Definitely an empirical model.
Okay.
It’s got some weak underpinnings in valuation theory, but I’ll emphasize the weak.
So are high profitability and low investment firms riskier than the low profitability at high investment counterparts?
If you tell me that two firms with the same profitability and one does more investing than the other, and they have the same price, well, then one of them has to be riskier if the market pricing thinks personally, so that’s the notion.
Where does the five-factor model fall short? Is it still momentum?
It’s investment. The investment dimension of that is kind of shaky. And profitability, well, that’s somewhat better. But we haven’t looked at that since we wrote that paper, must be almost 10 years now.
So why isn’t momentum in the model?
Well, because I can’t tell a rational story for it. So if I can’t tell a rational story for it, well, it’s just a violation of market efficiency. So there are asset violations, you don’t want your asset pricing models to be tautology basically. This has been a problem. People have lost interest in asset pricing because of the proliferation of factors. So people come out with papers where there are 100 factors. Of course, when you put them together, you find out that there really aren’t 100. A lot of them are more or less the same thing, but that kills all interest in asset pricing because becomes too flexible at that point.
So you don’t think all of the factor research that’s happening right now is a good thing for asset pricing?
No. No. Well, I think it’s stopped actually. I think people have stepped back and said, hey, is this really interesting or not? How are we going to shovel our way out it, if it isn’t?
What do you think about the Q factor model, if you’ve looked at that?
So the Q factor model is basically value, isn’t it? Is it a price to book or something like that?
Yeah, it’s got, it’s-
I remember now. If you look at the investment business, 90% of it is marketing. So they come up with a new name for an old idea which basically lots of what goes on as research in the investment sector.
And to piggyback on that, how sure can we be that factored premiums are not simply the product of data dredging?
That’s a really good question. So in our stuff, what we do is, when we come up with a model based on US data at a particular time period, then we take it out a sample for a different type period. So when we originally did the three-factor model, for example, that was based on data I think it was starting from 63 onward. And then what we did, we went and hand collected the data that we needed going back to 26 so we could test it out of sample. And then we said, okay, that’s out of sample in US. Now, let’s look at foreign markets and see if we see the same thing. So we’re looking for robustness basically, how to sample the stuff that confirms what you observe in sample. And for that model, we found it everywhere basically. The same to the five-factor model, we found that pretty much everywhere, too. It’s much more difficult to go backward in time because you don’t get good profitability data if you go back much past COMPUSEC going backwards.
Gene, can you talk about that time, going back and building that data set, going back to, I think it was the ’20s, because you didn’t have the data back then that we have now, how big a deal was this?
You had to go by hand into the books. The books existed with the income numbers in them. They weren’t in a machine readable form, but CRSP Center for Research in Security Prices at the University of Chicago had been collecting the stock return data, going back to 26 from the very beginning of those files. So we had the stock returns. We just didn’t have the supporting accounting information. So that was collected by hand. Not my hand, thankfully.
You mentioned the added sample testing. How important is the theoretical work to make sure that it’s not data dredging?
You would like to have a good theoretical model that encompass these things for the size and value factor, it’s not there. Well, the value more so than the size, the size factor doesn’t have much theoretical on the penny. That should be encompassed in other things, as I said, profitability and investment, there’s some foundations for that in valuation theory, but they kind of, to say the least weak. It’s not a fully specified model in the same way that the CAPM has. So now Bob Merton, back in 1973, he basically gave us the architecture for a multifactor model and how you develop them. He just didn’t put any names on the variables. He said this is a form of such a model. Any model you develop will show up in this form, you have to put names in the variables, because putting the names in the variables is the high factor. That’s where we were going with the five-factor model basically.
So that’s the ICAPM, is it possible to know what those state variables are that investors are worried about?
No. Well, possible in what sense though? I mean, can you go into their minds and take out what dimensions of returns or special interests or disinterest, what things do they have positive tastes for and what things do they have negative tastes for, and are those things general? I mean, because everybody have positive tastes for one thing and negative tastes for another. So it’s not that easy. Bob Merton was one of the smartest guys, if not these smartest guy I’ve ever known and he didn’t even attempt to do it, he did not even take a crack at it, he just gave us the mathematical framework and said run with it guys.
Okay. I see what you mean in your papers when you refer to them as unknown state variables now. That makes a lot of sense. Want to move on to expected returns for a bit.
Okay. Sure.
What do you think makes sense to use as an estimate for expected stock returns, just market returns?
Okay. That’s a very good question because I don’t know what to use except for the historical average return. The problem is historical average return is the number whose deviation from the two expected value has a big variance. You just don’t get a lot of information, even with a huge sample of data about what the true expected market return is. So I think the market return from back to 26 to now return in access for rate has been in the neighborhood of maybe four to 5%. But the uncertainty on that numbers means that two standard deviations away could be much closer to zero or much, much higher. Even though you have now almost 100 years of data on this, you still don’t get a very precise estimate of the expected value. That’s a fact of life in investing that there’s just no way to get around to that, to handle it in any better way. We just don’t know the expected premium of stocks over bills, for example.
And what about the expected factor premiums?
Same thing. Because as long as you have stock returns in there, the variance associated with them is going to be very high. So the expected values of any premiums that you put in are always very uncertain, no matter how much data you have. Or another way to think about it is you’ll never get enough data to know that for certain you’ll get a positive, expected, premium. Even if I tell you the expected value of the premium, you don’t know that in any finite simple, you will get that because the variance is so high.
Yeah. And we don’t know the expected value. So it’s a-
So it’s a double premium, right?
We touched on randomness earlier in an efficient market. Do you think long-term investors should think about returns as random or as predictable? Long-term investors.
Predictable in the sense that I think stocks have higher expected returns than bills, predictable in that sense. It’s not predictable in the sense that I know for sure that stocks will do better than bills over any length of time. It becomes more likely the longer the period, but it still never said. So I don’t know if that answers your question or not though.
I’m thinking John Cochrane, for example, talks about long term predictability and that in the very long run stocks are a little bit less risky than you’d expect if they were completely IID.
Yeah. Oh, okay. Right. So there’s some negative auto correlation that’s built in there that lowers the variability long term relative to short term. The other correlation numbers themselves are estimated with a lot of uncertainty. So you can’t really get a precise hook on that either, but he’s right on that.
Interesting. So if you’re thinking about long term returns, it’s really, IID and use historical as the-
Yeah. So what it looks like the reason it’s not IID, at least him and I wrote a paper on this too, and John there too. It’s not the same paper, but the paper we wrote basically said if expected returns vary through time, but they’re mean reverting. In other words, they don’t go off to infinity plus a minus. They tend to come back to a constant mean, then you’re going to overwhelm, if I look at long periods, I’m going to observe some negative auto correlation generated by this variation in the underlying mean. And the way the empirical work, this goes back to the early ’90s, I think. The way the empirical turned out, that seemed to be a good story for the behavior of stock returns. There was never anything in that, that was a message for investors. Because you’re talking about variation and the underlying expected value that’s really not so big relative to variation around the expected value and with a ton of uncertainty about estimating the process that generates time varying expected value.
So let’s shift to portfolio structure. Is there a single optimal portfolio for all investors like in the market with mean variance portfolio theory?
Well, if I think about market clearing, markets have to clear, everything has to get held. So what that says is that an aggregate, this is like a definition. Investors hold the market portfolio where the market portfolio is not just stocks, it’s everything, and it all gets held. So that’s the central portfolio of every asset pricing model, every asset pricing model starts with that and says deviate from that according to your taste for different dimensions of risk. But the central portfolio is basically this overall market portfolio. So that’s a good place to start for any investor, I think.
You mentioned the market portfolio, is the stock market a good proxy for the theoretical?
No. Because there are too many other assets out there. So I got to got to bring the bonds in, too.
Okay. So the global stock and bond markets is a better proxy for the market?
Right. And then I got to start asking myself, what other investments do I have access to? And should those be part of the market? So there’s some uncertainty about what I should do about government bonds. So our government bonds, an asset or a liability. You and I are, we can go long government bonds, but we are really on the short end too, because we’re going to be the ones that pay them off. And that’s clear that the net supply of government bonds from our perspective, is that anything other than zero because we were on both sides.
What about other assets like private equity or alternative investments?
Right. In principle, everything that could be put into your portfolio is part of the market. Now the question is, do you really have access to those things in an efficient way, in the sense that you can do it with relatively low costs? So we don’t have good models to answer that question. The other thing that’s really bad now, some of my colleagues work on, Steve Kaplan in particular, what is the expected return on private equity? The data don’t give you a good answer to that because they’re so self-selected, you only get to see the ones that survived pretty much. So you don’t get to see how much money was put in there that blew up and was totally lost. And that’s very important, very important. If I were on your side of the table and I had to advise investors what to do. I don’t know what I’d do about private equity, because I don’t think the data are good enough for me to give you a good answer.
So why is the cap-weighted market portfolio, a good starting point for investor portfolios?
That’s what the population has to hold an aggregate, that is the market as far as the population is concerned. In aggregate, we have to hold all the assets out there, cap-weighted. You can deviate from that, but when you do, you don’t have the market portfolio anymore.
What determines that? What determines when an investor should tilt their portfolio away from the market?
Taste, new attitudes towards different dimensions. I think of them as different dimensions of risk, but attitudes to different dimensions of risk are what do it. In Merton’s perspective, it’s basically her attitude towards whatever these underlying state variables are that generate premiums in various dimensions.
So is it just taste or can it be outside risks like labor income and stuff like that? Or is that a taste?
Well, our asset pricing model aren’t too good about putting labor income and consuming its correlation with asset returns, that was in the ’70s. People were worried about that. And basically they threw up their hands. They said, we know how to put this in there. There’s another un non-traded asset, nothing you can do about your human capital. You’re stuck with it pretty much. But you want to know the correlation of your human capital returns with the other returns in your portfolio and take that into consideration in their asset decisions. Well, the way we finance that was, we said, well for most people return, the human capital is uncorrelated with everything else. So you don’t have to consider the correlation nature very much. Basically, that’s the way it stands now. I’m not sure that’s satisfactory though. I’m not sure that’s satisfactory. Surely for example, you do not want to invest a lot in the stock of the company you work for because you’re likely to go if that stock goes.
Right, John Cochrane’s done some interesting work on that recently or a kind of summary work in his paper portfolios for long term investors that we actually talked to Sebastien Betermier who’s done some very interesting empirical work on how investors change their portfolios based on labor income. And it looks like they actually do what you’d expect, theoretically.
I’ll look for that. What’s the name?
Sebastien Betermier we can send you his papers. It’s very good. I want to touch on international investing for a minute. I’ve heard you say in other interviews that for a US investor, you don’t really need to worry about international investing. And if I remember correctly, the reasons were there’s expropriation that doesn’t show up in the historical data. So the data is better than what you can actually get. And that US stocks are no more volatile than global stocks. So therefore a US investor probably doesn’t need too much international diversification. My question is, does that change for someone in a country like Canada, which is a much smaller portion of the global market?
Yeah. Great. That’s a good one. Sure, it does. Sure, it does. I mean if Canadian investors only invest in Canadian stocks, it’d be really heavy in mining stocks. Right? So basically one industry concentration will be pretty high. We US Americans are very narrow in that perspective. So this was a statement for US investors, not for Canadian investors. Canadian investors clearly should be looking at investments, at least in the US. So whether the US or whatever expropriate, Canadian investors seems unlikely. Now, people look at expropriation risk as if it’s not there anymore. This kind of stuff just doesn’t happen. Well, I bet there’s a lot of expropriation that’s going to take place right now between US, Europe and anybody doing business with Russia. And the problem is, nobody cares about investors. Investors get expropriated, each side always expropriates the other sides investors, but they don’t fix it after the war. It stays expropriated even if you win. So that’s the risk of international investing and it’s not gone. It’s not gone. I mean, there’s nothing more poignant right now than that actually.
We mentioned that doesn’t show up in the data. Is there any way to see? I’ve looked and I haven’t found any papers or anything. How do you find the historical-
Also, I think Steve Ross and Roger Robertson maybe there was another guy involved. Way back when, what they did was they said, look, there’s this risk that nobody takes into account, that markets actually close entirely. So during the second world war, for example, lots of markets just closed. And then they came back after the war. So they went and looked at, well, suppose we were holding the overall market portfolio back in whenever, what happened to us in the meantime, when these markets closed, how did we end up? And they had a paper on that, that was a long time ago in the ’70s maybe. I don’t know what would happen if you updated that. I haven’t seen an update of that, but people have worried about that, that you only get data because the markets are open. And when they close, you don’t have the data, so you tend to ignore that those periods. But an example I like to give you, is I think Argentina was the second biggest market at some point in the past. And that market has closed multiple times since then.
I’m curious, Gene, has your own investment philosophy changed through your career?
No, my problem is I never intend to retire. So my portfolio is going to cover my retirement. It’s basically my charities that I got to suffer if I don’t make good decisions, and my kids of course, but I’m a really a sloppy investor. I don’t change my portfolio very often.
I want to circle back to asset pricing models for a second. We finished that part of our discussion talking about how size doesn’t have any real theoretical basis. Why does it still gain a place in the models?
Because there’s clearly lots of covariation in the returns on small firms that’s different from what you observed for large firms and we haven’t updated this for a while. In the past at least, that seemed to show up in differences in average returns expect showed up as differences in average returns looks like it was possibly differences in expected returns, statistically. It’s pure empirical. I have no justification for many there.
All right. I want to move on to inflation. You have a paper from many years ago on this, but I’m curious to hear your thoughts. What assets are hedges against expected and unexpected inflation?
Now. So there are these index bonds. That’s about as close as you can get. They’re index government bonds that haven’t been very popular in the past, and there’s a limit on how much of those you can buy. I think that’s probably why they’re not so popular, but that’s as close as you can get as an index bond. So I wrote papers back in the ’70s for a period of time when very short term government debt looked like it was a good hedge against expected inflation. As soon as I wrote those papers then thereafter that hasn’t worked. So for example, ever since the financial crisis, interest rates have been near zero, inflation has been going all over the place and the interest rates have stayed near zero. So short term bonds haven’t been a good hedge against expecting inflation. Now I think we’re going into a very interesting period coming up and now about. Wow, this is a different topic, you probably don’t want to get into that because I’ve been waiting for a long time about what would happen when we actually came up against the period when there was serious inflation.
And we seem to be doing that. Not that I wish that on anybody, but I just wondered what would happen when we came to it because I don’t see that the Fed has the tools to really deal with that. I think what happened when they went to the QE business is they decided that the QE business was more important than controlling inflation because inflation was very low. But now they’re faced with inflation and their only tool is to raise the short term interest rate. Now I wrote a paper several years back that said, I’m not sure that Fed even controls the short term interest rate because when it puts out lots of reserves, they basically better pay open market interest on those reserves otherwise the banks won’t hold them. They’ll try to get rid of them and they’ll have a hyper inflation. So they’ve been paying reserves on them.
And I think they haven’t been setting that rate. That’s the rate that’s dictated to them by the market. So I had a paper that basically was trying to document that. And then it said, well, how far out does any influence go of the Fed? It was very short. I mean the term structure at the intermediate long hand had a mind of its own, had basically nothing to do with the Fed funds rate. So what does varying the Fed funds rate?
They can only go in one direction as far as I can see. They can go up. They can’t go down. If they go up, fine, banks will sit on the reserves, but how much will they have to go to actually slow down economic activity? How do we put it differently? What firms take the short term overnight rate as they cost the capital? I don’t know any. How sensitive violates to that rate? Maybe not at all. This whole assumption about how they’ll control inflation with that huge balance sheet that they have is really untested. We’re going to test it now. We’ll have one observation a year or two from now on this process. Sorry that was tangent.
No, no. I want to keep going on this tangent. So no worries. What can the Fed do? Can they do anything?
Well, the question is if they raise the Federal funds rate, how far do they have to raise it to have any effect on inflation? That’s a wide open question. We don’t have any data at all on that because this QE business is a new regime. The Fed was always operated in an environment where there were no free reserves basically. And now you get, I think about $9 trillion worth of pre reserves out there. So we’ve never had this regime, so we don’t know what it will take to make it work, what it will take in terms of raising this short term rate. The big discussion is, is that an eighth a quarter? Well, I don’t think that’s anyway near, would it have to push it up to have any effect. It might be 10%. That’s extreme, but wouldn’t surprise me that they had to bring it up so that the real rate was positive. It wouldn’t surprise me at all because historically the real rate’s been fluctuating within plus or minus 1%, zero.
Can you elaborate a little bit on before QE? So before the ample reserves regime, what the Fed would’ve done to stop inflation and why they can’t do that now?
Yeah. So what they would do is they’d cut back on reserves. They’d make it more difficult for the banks to win and that would in principles, so real activity and pull inflation down. So the idea was, well it’ll cause a little recession and that’ll do it. They weren’t terribly good at that because if you go back to the late ’70s and early ’80s, we had inflation running near 20% for a couple of years. So they were never very good at this game.
So what do you say to a typical retiree who might have a 60/40 portfolio, they’ve done their planning properly, so given what you said about inflation, what would you say to them?
Well, I’d say hope that the 60 part of that isn’t hurt by the inflation. So far, this about hasn’t been bad for stocks. In the past, high inflation has been negatively related to stock returns, but that hasn’t been true in this experience. So they might be protecting the stock part of that portfolio, but you’re stuck, you got to hold the assets that are out there. That’s all you can do. So you can worry about it, but it doesn’t help a lot. I hate to check about that one, but it’s true.
Yeah. That’s difficult advice to give.
That’s why I say I’m glad I’m on this side of the table, not on your side.
So you talked about QE and how the Feds got themselves in a bit of a pickle now, can the Fed cause inflation?
Oh well, in the old days they could cause inflation because they could put on a lot of reserves. They weren’t paying interest on reserves, so the banks didn’t want them. So the banks would expand their balance sheets in order to get rid of the reserves. And that could heat up the world in such a way that you got a lot of inflation and vice versa. So that was the idea in the old days, is that the small changes in the monetary base reserves plus currency, would’ve a bigger effect on inflation, but that’s gone. That’s gone.
Is that still a lending channel in that case?
That case was a lending channel. Right.
Okay.
You don’t have a lending channel now. Well, they’re hoping you do. They’re hoping that by raising the Fed funds rate way above what the equilibrium would be, that’ll get banks to sit on the reserves. They won’t try to lend them out. So reduce economic activity, but we’ll see, we’ll see. But this is a different world now, you have all these FinTech companies out there. They’re not even part of the system that are doing a tunnel lending.
Yeah. That’s a good point. Okay. I want to move on to theory versus practice because obviously you’ve been working well theory and empirical work, academic work versus practice I guess. You’ve worked in academia for a very long time. You’ve also worked with dimensional, implementing these ideas for a very long time. What’s the biggest challenge in translating your academic work into live investment products?
Initially it was we didn’t know whether these things would carry over, whether you could actually implement them. So for example, when a couple of students at Chicago here came up with a small firm effect, most of the academic profession said, yeah, that’s in the data, but you’ll never get it because you’re going to get wiped up by the bid-ask spread on the small stocks. So forget about the small stock premium. And it turned out that wasn’t true. So the slow trading and small stocks basically didn’t pay the bid-ask spread. Dimension basically established that through its own trading. So nobody talks that way about it anymore. And then let’s say the value premium, you worry that if too many people get into that, maybe they can kill it.
So if they’re getting into it because they think it’s a profit opportunity rather than a different risk factor, that could kill it. So that remains to be seen I think. We’ll never have enough data to know the answer to that, but that’s one of the issues involved there. So another way to say it is where we started. There’s so much uncertainty involved in the outcomes from investing that it’s difficult to extract the signal from the noise. It’s difficult to tell what’s the real stuff going on underlying what we see every day, given it’s what we see every day or every year or whatever is buried in a lot of noise.
So even if value is theoretically a risk premium, if people believe it’s a profit opportunity, even if it is riskier, the premium can still go away?
Sure.
Oh wow. Hadn’t thought about that.
Sure. Well look, if I were misled and thought that stocks were much less risky in the long run than the short run, I could kill the stock bring over bonds too. If enough people believe that, it’s not special, the value or smaller or any of that, if I don’t really understand the differences that the long term does not erase uncertainty, that’s in the short term, unless you get a lot of negative correlation in there, then that’s kind of an inefficiency actually, did you kill a risk because of false beliefs?
But the risk would eventually show up but I guess we don’t know that, if the risk eventually showed up, they might-
Well, so the way we thought about it originally was that value stocks are riskier in the sense that, if I look under the hood, what I find is that those companies, they have been badly run or whatever, they are in industries that are declining. So that’s the real risk that’s involved in taking on value stocks is that it’s not a healthy end of the economy. So the question though is, should they carry a risk premium? It seems to have been there in the past if people are not concerned with that sort of risk because everybody that works for those companies should be concerned with it. But if otherwise, if that’s not enough to turn people away from those stocks, then you’d expect that premium to go away. But who knows? It’ll take a long, long time of data before we know the instance of that.
And I guess if value is a proxy for the unknown state variables in the ICAPM, then the risk could show up at times that people don’t want to do.
Right. Right.
What have you learned from working with dimensional that you maybe wouldn’t have learned through academic research?
When we were small, we didn’t have a lot of money to invest, and the markets were different. The market microstructure, which is the end of the market that can cost you a lot of money because of trading costs, that was a much less sophisticated business than it is now. So now they have all kinds of trading approaches to try to minimize the trading costs of the portfolio. So seeing all of that evolved, that’s been really eyeopening. And there’s a whole block of literature in academia about market microstructure and experiences shown it all to be basically most lots of it to be basically hard wash. I mean, they’re on the wrong tech. There’s a difference between your trading costs, if you do slow trading or if you do fast trading. So that’s been something we learned a lot about. Basically there’s always learning and implementing something that you’ve done with data, but you’ve never done in practice.
So what’s the slip between the data and the practice, is there any? How do you go about it so that you don’t create unforeseen blockages somewhere in the process? So I think dimensional has been very good at that. The company now is much more technically sophisticated in terms of how to deal with markets, how to deal with almost everything than it was in the first, let’s say five years of existence. The first five years of existence, you can take the global company home every night on a very small tape. Very small floppy disc, forget it now.
So as a follow up, Gene, what’s it like to look back and see your academic work actually implemented in practice?
It’s kind of satisfying. So I don’t take a lot of credit from that. I think my generation came along at a time when there was nothing in academia or at least didn’t exist, basically. So my generation basically opened a field up and that was great to be involved with all the people who did it, but we were kind of lucky in the sense that there hadn’t been anything before that. So it was like fishing in a barrel, you just threw the line in and you always came up with a fish. The current people coming into finance have a big body of stuff they have to master before they can actually think about doing research in the area. So they’re a little bit hog tied relative to what we were in the old days because the downside of that is we’re now all old.
So when you were fishing in that barrel, did you know what a big deal this was going to become? Did you have a sense?
No, no. Absolutely not.
Really?
Yeah, absolutely not. Look, we were young people trying to do academic research that would eventually get us tenure. So we didn’t really know whether this would go, plus, as Mike Jensen always said, I’m amazed if people pay us to do stuff we would do anyway. So talking about academic research and that’s basically too, I mean the people who do it basically love to do it, isn’t really a job.
So did you see at all the evolution of indexing and Vanguard and the book trillions by Robin Wigglesworth, did you foresee this at all back then?
Well, in my view, that all took too long that the evidence was there in the early ’60s that this was the way to go. And it took a long time before that had a big impact. When Ken French did his presidential address at the American Finance Association, basically in that whatever was 50 years period since the beginning of the research in the early ’60s, late ’50s, the world had gone from 0% passive to I think it was 20% passive at that point. And now it’s up to 50, but it still went far from 100. So I don’t know, to me that seems slow.
Do you think anything useful for a lack of a better word has come out of behavioral finance?
I do. Say I have trouble with this because what do you mean by behavioral finance? All of economics is behavioral. The issue is whether the behavior is rational or irrational. So what we call behavioral finance sounds basically looking at the world as if behavior is irrational. Now, that’s from the behavioral people. My good friend Thaler for example, they acknowledge that this is kind of a nihilistic game, but basically they have no advice for investors because they think whatever advice they put out there, everybody’s irrational. So they screw it up. So basically, they end up at the same place we do. They say no, just index everything because you’re too dumb to do anything other than that.
So way back when I wrote a paper, it’s one of the most highly cited papers in the general finance was about market efficiency and the challenge for behavioral finance. And I said, okay, you guys have been criticizing this, but that’s all you have, without efficicnet markets you have no area, because that’s all you do is criticize efficient markets. It’s time for you to develop asset pricing model of your own that we can all turn around and test and even criticize. And to this day they haven’t done that. So to this day, it’s still just the criticism of efficient markets.
One of the things behavioralists talk a lot about is bubbles. What do you think about bubbles in the context of market efficiency?
Well, I canceled my subscription to the Economists because during the financial crisis, the word bubble appeared in almost every issue in such a sloppy way that I couldn’t stand it anymore. So I wanted to know what they considered a bubble. So my view, a bubble is something that has a predictable ending, that you can make money predicting the other bubble will evolve. If it’s just accumulation of random numbers, it looks like a big hump. Well, okay, fine. I don’t call that a bubble, so I’ll tell you a famous story. So it was a famous agricultural economist at Stanford. This is way back before efficient markets really came out and he thought his colleagues could see patterns and data where they were in it.
So what he did was, he took a random numbers generator and he accumulated the numbers. So you got lots of variation, but it was all just random. And he brought it into the faculty lounge and he showed it to his colleagues and he said, it took them about 15 minutes to come up with stories about what episodes and prices those were. The message really was, they’re seeing things that don’t exist. This is just all randomness. And that’s basically what I say about people who talk about bubbles and markets. You got to tell me how to predict the endings of these things, otherwise I don’t call them bubbles. I just call them randomness.
Does the proliferation of increased computing power, artificial intelligence machine learning, does all of that make it easier for active managers to earn alpha?
I thought we were going to get through this without that question because I’ve never done this doc. I’ve never done one of these and that question hasn’t come up. So we’re consistent. And the answer’s been the same for, I don’t know, 40 years now. I mean, so I came online when computers were first coming around. I was one of the first ones in the University of Chicago to use the old 709 that they brought online. But anyway, my answer always is, in principle, we have a lot more information. We get it a lot faster at least, maybe it’s the same information, but we get it a lot faster than we used to. And we have ways of distributing it that we’re unknown 50 years ago.
But you can’t see the tracks of that in the behavior prices. You can’t see that’s had any noticeable effect on whether the market is more or less efficient. So that’s been the answer to that question for about 50 years. We just don’t know. My view of that is we don’t know because the market has always looked pretty efficient. It doesn’t look more efficient now. It doesn’t look less efficient now. It’s always looked pretty efficient. So it’s nice to have all this information and to get it quickly and cheaply, but it doesn’t seem to have improved markets that much, they all look pretty good.
I think I’ve heard your former student Cliff Asness talk about machine learning as a supercharged version of the anecdote that you told about the agricultural economist, finding patterns and randomness.
Right. Right. So you’re very right, there was a story that I now remember because there was a period of time when with computers coming around, the people in physics were having difficulty finding jobs. So they thought finance was going to be an easy field and they’d come in and develop models to predict markets and prices. And that was a huge failure. It didn’t work, but they’ll try again.
Always, I’m sure.
Sure. Right. Artificial intelligence is just that. I’m pretty sure is the key word there.
And it would be competitive too, that’s the way I’ve always thought about it, is if there’s an AI that’s good enough to earn alpha, then someone else is going to build a competitive AI that-
That it’ll kill itself. Right.
Yeah. Right. I want to move on a crypto for a little bit. You were on a Bitcoin podcast that I listened to, it was from back in 2015 and I’ll quickly summarize your position from that interview. You said that Bitcoin’s an accounting system for exchange that may be useful to drug dealers because it’s somewhat anonymous, but it’s otherwise no different from a volatile checking account. Has your thinking changed at all since then?
Well, this is what I say. I like this area anyways because there’s so much to talk about this, it’s garbage. So you can distinguish between the medium of exchange, like the cryptocurrency itself and the mechanism that does the exchanging. So it’s the blockchain, people often don’t distinguish between the blockchain and let’s say the cryptocurrency, but they are different. So I could put the cryptocurrency into the exchange mechanism that the Fed runs among banks. And I can put reserves into the blockchain if I wanted to, so you got to distinguish between those two. Now the problem with, let’s say with Bitcoin, if you go back to the old monetary theory, what it said was if something’s highly variable in real terms, it’s not going to survive as a meaning of exchange. Simple way to think about it is, firms don’t want to do business in a medium of exchange that itself can put them out business, just its own variability can put them out of business.
So if you look now at the people who “take bid Bitcoin for transactions”, what you’ll find is they take it, but they don’t hold it. They get rid of it almost instantly. They just sell it. Now that’s one thing. The second part of it is, okay, what gives Bitcoin it’s value? If it’s not really being used as a medium of exchange, well, then it should really have no value. And it’s volatility would kill it as a medium of exchange. Now these stable coins that people are talking about, they’re better because they’re basically linked to the dollar. I think they recognize this problem. But at that point, they’re like bank reserves. They’re just linked to that dollar. I don’t know, is the Fed going to love that kind of competition in there? And is it really credible that a private issuer of reserves will always be ready and able to exchange it for currency on demand? I don’t know. That’s a tough one. That’s a really tough one.
How much currency do you keep in the background and obviously to make that credible? So we’ll see. Now, that’s separate. So that’s the medium of exchange. The method of exchange is something else. You could certainly improve on the method of exchange that the Fed uses to clear transactions. There’s no reason it should take two days or even a day to clear transactions through that system. I mean, it’s all just a computer. I wrote this paper like 25 years ago that exchanges in the computer range should be instant. They should be able to trade reserves instantly across the system. There are efficiency of improvements that could take place in the mechanism.
Now, blockchains don’t look to me like that kind of system. They’re incredibly hogs. They’re incredible hogs in terms of the electricity they use, because not have somebody overseeing the system, like the Fed oversees that if the system among banks becomes very expensive and it uses incredible energy. So it doesn’t look to me like that system has much of a feature to it, but we’ll see. We’ll see. But I think there’s a lot of junk that gets talked about in terms of cryptocurrency. I don’t think people who write about it really understand what it needs to be in order to survive.
What do you mean by junk? What’s the kind of junk?
Well, the kind of junk is they don’t sit around and say, well, why does this thing have any value at all? The answer to that from monetary theory would be, if I don’t use Bitcoin to execute transactions, it has no real value. It’s just numbers, there’s no real use for it. So bank reserves have value because they are an electronic means of exchange. A very efficient means of exchange. Well, maybe the blockchain is that, but if that’s it, then you get the problem that this thing has a highly variable real value. So it shouldn’t survive as a means of exchange on that basis alone. So again, these are the things I don’t think people… I don’t hear anybody talking about that. I don’t think people coming up now actually learn monetary theory.
I think that the argument from the Bitcoin community would be that it has value because it is censorship resistant. So even though it is highly inefficient because it has trustless or not centralized trust consensus, which is expensive and it’s volatile, people who can’t operate within the existing financial system because they’re criminals or they’re in a country that doesn’t have infrastructure.
Yeah. That produces a demand for it, basically. You got to have a demand for this as a medium exchange. So illegal transaction produces demand for it. But then my question is, how much of that do you need to give it substantial value? How much do you have of the illegal transaction trade do you have to get to make that work? I don’t claim to know the answer to that, but that’s the question implied by that line of defense. Now, you got to tell me something more about what it would take to make that survive.
Yeah. And we can’t know that. A paper came out recently that found that I think 90% of Bitcoin transactions are not economically meaningful.
They’re just people trading it.
Yeah, exactly.
That’s not telling. So you could have zillions of those transactions, just people trading, but they still could be a large amount of absolute transaction goods being exchanged in the background and still that be very big, so that doesn’t really answer the question.
So what do you think is driving the incredible rise in price at Bitcoin past seven years is up like 11,000% or something?
Right. Right. That’s fine. I mean, I’m not buying it, but it’s volatility. Its high price is very impressive. And the volatility is equally impressive. Goes up and down 30% in short periods of time, that’s really impressive.
You mentioned for reserves or for say the dollar that it’s an efficient means of electronic transactions, and that makes it valuable. Is that what gives the dollar value?
You got to limit the supply and then you got to have people willing to trade in it, willing to execute transactions in it and then you have value. So the dollar is a fiduciary currency. There’s nothing there except the fact that the supply is limited. So this is taking us back to the QE period again. So this is another thing that really bothers me is that in the old days, we had a supply of the monetary base, which is basically currency plus reserves and neither of them paid interest. So they looked identical.
Now you’ve got currency plus reserves and the reserves basically pay market interest. The banks can go back and forth between them on demand. That’s written into the Federal Reserve Act. So you don’t have the supply of a medium of exchange, which the Fed has control over because the reserves are now just another interest bearing asset, another asset that’s bearing market interest. So the fact that it’s exchangeable for currency, currency and principle could be used to control the price level. But when it’s exchangeable freely for reserves, that goes out the window, supply is no longer fixed by anybody.
Do you need to have a fixed supply for the dollar for example?
You need to have something controlling that supply, if you want to control inflation with it.
Interesting, because the Fed controls or tries to control interest rates, but not necessarily the monetary base.
Yeah. It came up on controlling the monetary base when it went into QE, the price of QE was giving up control of the monetary base.
One of the things that again, Bitcoin people would say, is that Bitcoin is better than a fiduciary currency like the dollar because the supply is mechanically fixed. Is that a good property of a currency or of a money?
Well, Milton Friedman always said that, that was a good property of a currency. He was always in favor of the Fed limiting the supply of currency plus reserves to a fixed rate of increase every year, basically the expected expansion of long term expansion of the economy. And that’s it, no more. So that’s basically the same statement that knowing how much is going to be out there and how much will be out there in the future is very important if you want this thing to be the medium of exchange. Now that’s a big problem with a fiduciary currency. They often blow up because governments can’t resist throwing them out there and spending it. So that is an advantage. Having a fixed supply is an advantage. The huge disadvantage is the variability in real terms of the value of that supply. That’s the cost of using that as a medium of exchange.
Right. But in the current system to try and have a stable, real value… Oh, we don’t currently have stable, real value?
We have Bitcoin.
Oh yeah. Sorry, back to the dollar.
Dollar you know, so far. Right? But we’ve had periods where if you go back to the ’70s the periods where the value of the dollar wasn’t that unstable. It was just going down all the time fast, that can kill a currency, too.
What do you gain by fixing the supply? Like maintaining real value long term, like the gold standard kind of argument?
No. So if you maintain the total supply of what people transact in, what monetary theory would say is, the real value of that should go up through time if the economy expands because you have less of it to use in transactions. So you basically create more by increasing the value of it, but the value increase on its own. So you expect the price levels to go down in that case.
Was deflationary, is that bad though?
No, not necessarily.
Very interesting. Just want to shift gears to your career, Gene. And we’re curious, who are the most influential figures in your early academic career?
Merton Miller and Harry Roberts, easy. Merton Miller, probably everybody knows. He was one of the founders of finance, especially the capital structure and the finance. Harry Roberts, very few people know, but he was the one that gave me my upbringing in statistics, and how you go about doing meaningful statistical work and how you look at it. So those two were the most important.
Asking this to you because you’ve been incredibly productive throughout your career. How many hours per day do you think the brain can handle thinking work?
Okay. That’s a very good question. It took me a long time to figure that out. So I would say you have about four hours a day. So I say over my academic life, I do my work in the morning and I do other people’s work in the afternoon because in the afternoon I’m burnt out. I can’t do the original stuff. My productivity per unit time goes way down. So then what I found out later in life is… This is when I took up golf. What I found out was you can take those four hours any time. It’s not important that you get them in the morning. You can go play golf in the morning, get the four hours in the afternoon, but then nobody else gets your time.
So you tackle something hard for four hours every day?
Yeah. Every day, seven days a week.
Right. Your work ethic is legendary.
Well, I don’t know.
I mean, we’ve heard Ken French talk about you calling him on Christmas Day before.
Well, remember now people were asking me, why did I go into academics? I say, because I would’ve control over my time and I wouldn’t interfere with my athletic interests. So initially it was tennis. I played tennis every day for a couple hours. And now I’m an old guy. So at age 63, I switched from tennis to golf, but that takes a lot of time. Basically academia is a way that you can squeeze your working around your other activities.
What do you think explains the unbelievably productive relationship that you’ve had with Ken French?
Well, we have some very similar work habits, so I know that I can call or email him at any time, he’ll be working at about the same time that I’m working. He does more hours than I do. He’s able to survive sleeping five hours a night. If I sleep five hours a night, I’ll kill people the next day.
Me, too.
And we were way different in that respect, but we have similar interests, too. That’s good, me and bad. We’re similar, but we’re all so different. So you got to be different along a lot of dimensions otherwise there’s nothing more than some of the parts that comes out of the relation. But I work with other people in the past that didn’t work the same long hours, the same consistent hours. And I figured out I couldn’t work with them because our work times didn’t intersect enough.
So our final question, Gene, how do you define success in your life?
Okay. That’s a good question. Well, I mean, I guess the primary thing is having a family that turns out to be something you’re really proud of. I’ve been really successful on that. But my wife gets most of the credit for that, not me because I’ve spent too much time working and she’s the one that did all the work on the family side. So that’s very important. The second thing very important when every young person comes to me, I say, look, you’re going to spend at least a third of your life working. It’s important to find something that you really like to do otherwise those hours are going to be basically torture. That’s the first part of that advice. The second part is, find something that not too many other people want to do so you can make pretty good money out it. You don’t want to be a ballet dancer, for example, because very few people succeed in that area. You need to do lots of areas where that can be a problem, that’s the advice I can give to people.
Terrific. Well, Gene, this has been a real pleasure to see you and for you to participate in our 200th episode. So thanks so much for your time.
Sure. My pleasure.