Episode 278: Juhani Linnainmaa: Financial Advisors, and the Cross-Section of Returns

22:15The reason for the MFDA, the stated reason for the regulation was to, after the Great Recession, to restore the face in the financial system. That maybe people were afraid to participate in the financial markets because they didn't understand those markets and they were concerned that there was rampant fraud going on. And so maybe if we have more rules, people have more trust in the system and they are going to be more willing to use advisors. What happened in terms of this regulation was that when it was introduced, according to industry reports, you had lots of consolidation. The competition went down massively.

22:44And so effectively, the supply of advice went down. And so when we look at people in our sample, the use of advisors in those regions that were affected by the regulation first went down significantly. How did it affect equity market participation? Yes. So advisors have a huge effect on equity market participation. So maybe not surprisingly that you take away some of the advisors, you decrease the supply, people are not going to have the advisors. And if they don't have the advisors, they're not going to be participating in the markets. So it's kind of interesting. It's clearly an unintended consequence of the regulation because the whole point was to restore the face, get people to use more advisors, and by using more advisors, participate more in the markets.

23:24But because of the decrease in the competition, decrease in supply of advice, it turns out that the participation went down because of the regulation. Crazy. So how important is the role of financial advisors in getting households to take equity risk based on your findings? We were kind of lucky that the MFDA regulation was rolled out in different provinces at different times. So we can use the staggered introduction of the regulation to try to estimate the causal effect of having an advisor on stock market participation. Again, the concern would be that maybe the decision to participate and the decision to have an advisor are going to be correlated, in which case the correlation might be overstating the impact of advisors on participation.

24:02But in our case, again, we can try to identify what is the causal effect of advisors. Whether you don't participate or participate in markets, and then you have the advisor or don't have an advisor, having an advisor is going to increase the probability of participating between 54 and 90 percentage points so there's a massive difference coming from that one advisor you're not going to be participating if you have an advisor for extra chance reasons you're going to be much more likely to participate in terms of how much you invest in market like do you take put 30 percent of your money in the equities that increases by about 30 percentage points in a sample so you would go from 30 percentage points without an advisor to 60 with an advisor.

24:43So there are economically huge effects. That is wild. There's kind of interesting thing over here. So I mentioned before that there's a concern about the correlation that maybe if people who would be participating in the markets anyway, they are more likely to get advisors. Then if you just look at the, you have an advisor, don't have an advisor, and you measure how likely they participate, that might just be overstating the effect. But in the day, that turns out that the bias goes the other way around. It's telling us that the people who are less likely to participate when left on their own are more likely to seek financial advice and then participate.

25:18So when you cut down the supply, you're really like, if we only care about equity market participation, you're hurting the neediest people the most, the ones who are not going to be participating without advisors. That is crazy. So I just want to recap and collect my thoughts for a second. So The MFDA is created to increase trust in the financial system, which increases regulation on financial advisors. That's rolled out to provinces in Canada at different times, so you can test the effect of what happened. It reduces the supply of advice, which causes people who would have been more likely to seek advice to have a much lower equity share in their portfolios because they needed the advice to get invested in equities.

25:59Yeah, that's right. Wow. That's crazy. How does the benefit of increased equity share compare to the cost of advice? So now we're kind of at the same point as before when we spoke about, well, it's on customization and better performance. Well, what's going on here? One could say that, well, if advisors help get people to stop market, there's going to be some benefit from that one because there's going to be some equity risk premium, maybe 5%, 6 % in terms of long-term estimates. And so if you're going to be investing 30 % more in equities, if you're an advisor, and that's the right thing to do, then you're going to be getting more return on your portfolio.

26:33So if you are like 30 % and you have like 5 % equity premium, that would be a sizable increase in the performance of your portfolio. But that wouldn't be enough to offset the cost of these advisors in our sample alone because the cost is going to be like 2 % to 3 % at this point. So it has to be something more than just equity market participation that's going on here. But differently, having advisors to just put them in the stock market to give people higher returns, the return is not going to be high enough to cover the cost of the advice. So again, it could be some other benefits that people get from having advisors.

27:04But in terms of direction, having people participate in equity markets is a good thing, setting aside the effects of how risky the portfolio is going to be. Do you have any empirical research on what the other services that clients may be getting from advisors would be? Is it increasing their saving behavior? Do you have anything on that? We don't have something where we could compare the not-advise and advise people. We see in the sample of advice people, we see the extent to which they use some kind of automatic savings plans, which might be a good thing, but we don't have the same information for the unadvised people.

27:34So we haven't been able to look at it. Okay. Now the MFDA sample in Canada, we're very familiar with this because we're in Canada. It's kind of notorious for mostly using high fee commission-based funds where like the complex ventures that we talked about are a problem. Based on your findings in this paper, is it reasonable to assume that financial advice that's using lower fee products is net beneficial by increasing equity exposure and whatever else, but not having the 3 % fee? Yeah, I think that sounds broadly speaking correct in the sense that if you think about just investing in any mutual fund and you randomly pick one, if you pick one with the low fee, that's never going to be a terrible investment.

28:15Just like it's very hard to beat the market in terms of gross returns, it's very difficult to systematically lose to the marketing gross returns. And so if the fee is kind of tiny, it's not a terrible mistake to make. So giving those mutual funds to people would be a pretty good idea. But that being said, it brings us back to the problem that we don't really know what's the right thing for people to do, how much risk you want to be taking in portfolios. But if we fix that one and we say that, well, you want to be allocating 60 % of your money into equities, and then you just randomly pick some low fee funds into it, that's going to be a good thing.

28:45And also, if you have the low fees and you have the advisor, they are going to be helping with the level of participation, and that might be a good thing. But again, with the caveat that we don't know what the right number would be. Do people on average 30 % of risk inequities or 90%, what's the right number? How does the duration of the relationship with a client affect the equity participation and equity share? So we looked at that in a different paper. The motivation for that study was this already kind of classic famous study by Chennaioli, Schleif and Vishny, money doctors. The point of that paper, the theory paper is kind of simple.

29:20they view advisors as being like doctors in the money space. The idea being that many people don't understand the stock market. They're kind of fearful to invest in the stock market because it's unknown. If you have these advisors who people can contract and the advisor is going to be holding the client's hands, now they're going to be more willing to participate in the stock market. And when they participate in the stock market, they're going to be earning a higher return. And part of the return that they get, they can hand over to the advisor as compensation. So both parties are going to be better off when they have these money doctors.

29:49An important element of that story would be the trust, that you are fearful of the market on your own. If you have the advisor who you trust, then you're more willing to take the equity risk. So in our study that you are referencing, we were looking at the building of trust over time, with the assumption that if you have an advisor for a week, you are not quite sure what to make of this advisor. But if you have been with the advisor for 10 years, that at least signals that somehow you're probably more trusting of this advisor. And what we find in the study is that the longer the relationship between the advisor and the client, the more equity risk on average the clients are taking their portfolios.

30:25How big is the effect? We looked at this in terms of things like the financial crisis. So we looked at people who go into the financial crisis with the new advisor, that the old advisor had just left the industry and they were with the new one compared to people who had had the advisors for a long time. And we looked at that, what's the probability that you stay invested in the equities throughout the crisis? And we found that there was a 7 % difference in the sense that if you go to the crisis with a person who you don't really seem to be trusting, it's very likely that when things go south, you're going to be selling your portfolio and just giving up on equity markets.

30:59If you have an advisor who you have known for years, you're much more likely to just come out from the other end. So how does the duration of the relationship with a client affect the client's equity participation and equity share? I think it's going to be positive that the more time you have known your advisor, the more you're going to be investing in equities. For example, when we partition people based on how much they invest in equities at the time they first found the advisor, and you look at people who had somewhere between 0 % and 20%, that's going to be running up over 10 years. So at the end of 10 years, those people are going to be having about 30 percentage points more invested in equities that time.

31:33So it's a pretty massive effect in terms of economic magnitudes, going from 0 % to 20 % to having 30 percentage points more at the end. And of course, that's over 10 years. That's still a huge effect though. Yeah. How did getting a new advisor before the financial crisis affect investors' ability to remain invested? Okay. So there are two different effects that we were looking at. One of them would be the mean effect. The longer you know, how much more risk you're going to be taking equities. The other one was about the sensitivity to bad returns. We know from the mutual fund literature, we know from other studies that people, of course, become very fearful and call it quits when they have bad performance in their portfolios.

32:11And so we wanted to understand if knowing somebody, trusting somebody is going to be attenuating the return investment sensitivity in the sense that if you had known somebody for a long time, maybe you are willing to go through these rough times without liquidating your portfolio. So going to your question, when we look at the financial crisis, when we look at people who enter the crisis with a brand new advisor, because they're all advisor with the industry, and we look at people who go into the crisis with an advisor they have known for a long time, there's an 8 % difference in coming out from the other side with your portfolio intact.

32:45So the people who have a new advisor are much more likely to just give up on the stock market during some finance crisis than those who have their trusted advisor. So that's 8 % points measuring the number of people that stuck with it? Yes, that's right. So the problem of quitting is like, let's say, 10 % points for the people with a long-running advisor, it would be 18 percentage points probably quitting for the people who have a brand new advisor. That's amazing. So there's two effects. They're having more exposure to stocks in their portfolio and also are less likely to capitulate when the market crashes?

33:20Yeah, that's right. And both of those effects would be consistent with the money doctors type of story that again, these advisors are providing a valuable service that there's somebody who's holding your hand and telling you that, okay, sometimes stocks go down, you don't want to give up on the stock market when the returns are poor, that in the long run, it's probably going to be a good investment. Yeah, super interesting. Okay, so what do you think that those findings tell us about the role of financial advisors? They tell us that financial advisors are really important in the markets that they do seem to be serving a valuable role.

33:50The different question would be like, well, what do we do with this information? I guess the key lesson would be that the market for advice is just like any other market. And so we want to think carefully about the regulation of the market. We could be pretty heavy-handed. And like we discussed before, you could always say that there's going to be bad behavior, good behavior. Then we have all these guardrails in place to ensure that people get the best possible service so they get the benefits and none of the downsides. But now we're again back at the issue that it's really hard to say ahead of time what proper level and type of regulation would be because there's always going to be the unintended consequences.

34:23It could be cutting down the competition as we discussed before, and that could be having effects that you didn't anticipate. But again, this study is highlighting one of the benefits of having an advisor. So you were asking before, well, if it's not about the customization, it's not about the performance of your portfolio compared to benchmark index, why do you use an advisor? And this would be a partial answer to that one. It would be saying that you have these people who are reluctant to invest in the stock market when left on their own, they are much more likely to be willing to take risk when they have an advisor.

34:53And that is a valuable service, something that we didn't quantify in the other papers. Interesting. Really interesting. And the duration of the relationship matters. Is there any lesson or idea in there about clients not jumping around between advisors? That's a good question. It would be kind of interesting to look more deeply into that one. But then you would have to have data on that quantify exactly why people are jumping around. It goes back to your revealed preference thing that if we just see people jumping around a lot, maybe those people and advisors they have been with are very non-random.

35:24There's some kind of selection going on. But the question you're asking is an interesting one. Right. But you're using duration of the relationship as a proxy for trust. But if a client doesn't trust an advisor, even if they've been with them for a long time, then that kind of makes the effect go away. Yeah. It would be kind of ideal to have some kind of objective measure of trust besides the length of the relationship. So I think you touched on this already, but I want to put a finer point on it. what do these findings tell us about the role of regulation? I was afraid that somebody's going to be listening who's very anti or pro-regulation.

35:55So my first thing is just to say that we don't really know. But what we do know from the other studies that we have done is that, again, regulation can have these bad outcomes that people don't anticipate. Even if you are really well-meaning and you think that, well, there's nothing bad in saying that we want to ban this and this and this behavior, and you're going to be charging more than this in fees and so forth, it might seem like a perfect thing to do. It's very sensible. But then people can be changing their behavior, people can change their recommendations, or people can be quitting the industry.

36:24And all kinds of less than ideal outcomes may happen. So I guess the point is to think about this more as a market and think about how you would be in the stock market. You don't want to just ban good and bad stuff in one dimension and assume that that's going to be helping the outcomes. But yeah, that's just a long way of saying that we just don't know. Let's just be careful. Yeah. Super interesting. I think Burke and Van Vinsbergen had a paper on regulating charlatans or something like that. And it was like an equilibrium. And they found that even if there are charlatans, if you increase regulation, it can be a net negative to consumers.

36:57Yeah, that's right. It was an argument about the competition. So even if you had these bad people, if you put them in the market, they are going to be competing with the good people. And the good people had to be responding by lowering their fees, which are going to be a good outcome for the markets. If you restrict the market to be only the good people, then it's going to be less competition and they can charge higher fees. So it's kind of interesting take on it that charlatans can also have a positive effect because of the increased competition. Yeah, it's crazy. Counterintuitive. Okay. That segment on financial advisors was incredible.

37:26Your research on that is just awesome. So I'm glad we spent some time on that, but I want to move on to another area that you've done a lot of really interesting work, which is the cross-section of returns. So you got this paper, the history of the cross-section of the returns, which is really cool. What did you find when you looked at accounting-based anomalies like profitability and investment in the pre-1963 period? Yeah, as a background to that question, when you specify the 1963, people might wonder like, why that number? Why not 62, 64? And that's really kind of like the key point of the paper.

37:55But when we look at any accounting-based anomalies like value or profitability, we often start the sample in 63 because S &P founded the Compostat in 1962. And so after that one, we can have comprehensive non-selected accounting information for companies. So that's when we can construct these training rules and we can see what the performance looks like. So in this study with Michael Roberts from Wharton, we got access to Moody's manuals going back to early 20th century. And so now we can be looking at the performance of the same rules before 1963. And roughly speaking, when we look at alphas, information ratios, things like that, we find that the efficacy of these accounting-based anomalies is about 50 % of what they are in the in-sample period, the period that the original studies were looking at.

38:40So they're far less powerful in the history than what they are in the in-sample period. I brought this paper up in a discussion recently, and a comment that I got was, well, we can't trust the quality of the data in that pre-sample period. But I think you touch on this in the paper. How did the data quality compare? We look at that extensively, and we look at other papers that have been using historical data for other purposes, and they identify different points in regulation that should guarantee that this data is going to be of high quality. And whether we go all the way back to 1920s or whether we start in 1940s, the conclusion is going to be the same.

39:12So we are pretty convinced in the paper that this is not an issue with data quality. And what do these findings suggest about the anomalies you looked at? Yeah. So the motivation for this study was to kind of try to disentangle two effects. That when you look at something like, let's say, the value premium, and if you think that that's coming from mispricing, you might imagine that as soon as this thing has been discovered and people start trading on it, it's going to make markets more efficient and the value premium is going to go away. But there's also the other possibility that maybe the effect wasn't real in the beginning.

39:40And so people have been data mining, they have found something in the sample, and when you look at performances anomaly in the outer sample period, it's going to be weaker. So two potential effects are going on. Maybe it wasn't real to begin with, it was data mining, but this was stronger because of data mining, or it could be about the actions of arbitrageurs, that something was real, it gets discovered, and people trade against that one. And if you only look at the original sample and you look at the sample that accrues afterwards, so the out-of-sample period, you don't know really which one is going on.

40:09There's a great paper by Maclean and Pontiff that tries to separate those effects. But what we do in this study is we try to look at the other end of the sample. So we are saying that if people didn't know about this in the original sample period, then they most certainly didn't know about that when you go back to 1920s and 1930s. So that's going to be a kind of clean test of the data mining hypothesis. And given that we find that the decline in the performance of these factors, when you go from the in-sample, either to the post-sample or the pre-sample is about the same, 50%, that is telling us that most of the attenuation that we see in the performance of these factors is coming from the fact that they were data mined to begin with.

40:46Not so much that they were discovered and then arbitrage shares traded against them. That's crazy. So you're saying that profitability and investment are data mined factors? So for any one factor, it's kind of hard to say that, well, this is a real one. This isn't a real one. If you take value scheme as an example, value performed really well up until something like 2007, 2008. But it has been in a drawdown since then. But given how noisy the returns are, even today, you're going to be rejecting that the mean return since 2007 has been equal to the average mean return before 2007, just because there's so much variation in returns.

41:21The benefit of our paper is that when we look at a large number of anomalies, they're going have much more power to try to see if there has been breaking a performance. But for any one thing, we cannot say that this was about data mining. We can just say about the averages that either 50 % of the things that people have discovered are not real, or maybe all of them are real, but they're just inflated by a factor of 100%. In the sense that the performance decreased by 50 % when you go out of sample. Yeah, that's crazy stuff. That kind of made me think of the Fama and French's paper on the value premium.

41:52I think that's what you're referring to with value, if you look at their post out of sample period, post sample, they found that the value premium was gone in the post sample period, but they couldn't statistically prove that it was different from the in-sample period, something like that? That's right. They had a paper about that. I think you had the title, right? The value premium. Yeah. Okay. So how would the post 1963, so we take that sample that everybody's familiar with, how would the mean variance efficient portfolio using the market size value, profitability, investment factors, the Fama French five factors, how would that mean variance efficient portfolio have performed if we ported it back and invested it in the pre-1963 sample?

42:34Okay. So there are two things that we were trying to look at here. So if you have some kind of sample like 1963 to 2020, and we see in that data that you have some of the factors that seem to be working, then the question would be, well, how do you optimally invest in these factors? Oftentimes, admittedly, people are pretty careless about how they overfit their data and how much they think that people could have known ahead of time. So oftentimes, people construct like exposed optimal portfolios. They look at what would have been the optimal investments in the market size, value, profitability, and investment to maximize the shop ratio in the in-sample period.

43:06And then they kind of assume that people would have known to make that investment ahead of time. Even though, of course, in 1963, they wouldn't even know about the profitability, let alone know how exactly what the conferences are created from a portfolio. But in the computation that you're referring to, we say, well, let's train this model of investment using all the post-1963 sample, and then let's port it to the pre-1963 sample and see what the performance looks like. So obviously, by construction, in the modern data, after 1963, this would be a great investment. The Sharpe ratio of the portfolio would be about one, which far exceeds the Sharpe for the market of 0.4 over the same time period.

43:44But if you do exactly the same trading rule in the pre-sample, the performance is about the same as that of the market. So having these additional factors with those weights are not really adding any value. Yeah, that's crazy. You talked about investment fees earlier, and if you were paying whatever more for the mean variance optimal portfolio, you're that much worse off than the market. Yeah. Okay. So it's kind of hard to form optimal anti-portfolios using X-post data. Do you have any thoughts on how to think about the optimal ex-ante mix of factors in a portfolio? I think people just need to be much more careful about the aspects of data mining and overfitting.

44:22And you definitely don't want to be creating any portfolio based on the full sample and assume that that's going to be the experience that you're going to be getting. So what we highlight with the computation would be the data mining aspect, that if you look at any set of factors, you probably want to be assuming that the alpha side of sample is going be about 50 % of what you see in the in-sample period. And then when you compute the portfolio, that's going to be a whole different discussion because anytime you bring in the covariance matrix, you're making lots of assumptions about how the factors are going to be offsetting each other, how much benefits you get from that one.

44:51So you want to be pretty cautious. That being said, we are definitely not negative on the factor investing because we do find that there's going to be lots of alpha remaining in the factor investing. It's just that you want to be quite cautious about this one. I believe that in the same Fama French paper that you mentioned, they also point out that factors are really, really noisy in the sense that we tend to think about the US stock market going up, up and up. But if you look at the volatility of the stock market, it's not outside the realm of possibility that you would see a 20-year period in which you're going to be losing money by investing in stocks in nominal terms, just because it's very easy to get bad enough draws just by luck that offset the mean effect in data.

45:31The same is going to be true for all these factors. So even if the value premium is still there, it would have still been easy to see a drawdown like we have seen after the financial crisis. So again, we are not condemning on saying that fact investing doesn't make sense. We're just saying that you want to be kind of careful about it. You don't want to look at the in-sample evidence and say that this is the greatest thing ever. I want to put 500 % of my portfolio with leverage into factors. You want to construct smart portfolios and you have to have some patience with it. Yeah. Interesting. So you're not negative on factors, more negative on over-optimizing.

46:02You wouldn't want to build a portfolio that's heavily tilted toward, I don't know, the investment factor because that one did well in the sample that we have. Is that kind of the idea? Yeah, that's right. Yeah, that's interesting. We use factor investing for client portfolios. We use funds from dimensional fund advisors and they tilt toward all five of the factors we're talking about. is there any way to think about the optimal mix of how much value and how far do you push on profitability and all that kind of stuff? Or is it just kind of a naive, you should have a bit of exposure to factors and that's good enough?

46:34I think that kind of brings us into the realm of statistical learning or machine learning, but you want to be kind of careful about how you train the models or investment portfolios. That if you do something like that in the in-sample period, you want to be splitting that in different folds and trying to see how much I want to be tilting in different directions, not to overfit to the sample. So there are methods for doing it, but people want to be careful about it. Okay. I want to move on to a little bit more on the value premium. You've got an incredible paper on this too. What mechanisms can cause firms to move between growth and value?

47:06So in the first part of this paper, the Composite Value paper, we complete the statistical approach to asking where the value premium is coming from. So like you're saying, well, why is something a value firm? Why is something a growth firm? and I guess there would be three answers to it. You might be a value firm because you have always been a value firm or you might become a value firm because your book value equity went up a lot and your market value equity didn't change or it could be that your book value equity remained the same and your market value equity fell. Again, we are not taking any stance here in terms of economics.

47:33We're just saying that just mechanically, statistically, it has to be one of those things that tells you that you're a value firm today or growth firm today. So change the book value equity or the change in market value equity or some kind of fixed characteristic of the company that you're looking at. And how does the value frame decompose into firms that changed in size and firms that changed in book value? When we isolate that into those three components, that you have always been a value firm, or you become a value firm because you retain more earnings, or you become a value firm because the market value of equity goes down, it's really the last component that drives all of the value premium.

48:06Instead of looking at the total book to market ratio of firms, if you only look at the change in market equity component of that ratio, So you're going to be getting more of the premium with less risk. So firms whose market value of equity declined, and that's how they became value firms, that's where most of the value premium is coming from? Is that right? That's right. And it's not that surprising afterwards. If you think about the sequence of events and the history of the literature, there was the DeBond and Thaler paper in the mid-1980s about the long-term reversals. And that effect was discovered before the value premium.

48:38The idea that if you look at stock returns over the prior five years, maybe skipping over the most recent year because of momentum, you get significant returns on a strategy that buys losers and sells winners. That's the long-term reversals. Fama and French came up with their value premium or the HML in the early 1990s, and they found that there's correlation between these two things, that the reason you get the long-term reversals is that it's also a value strategy. We are seeing something a little bit different because we're not really measuring the stock returns. We are looking at the market cap changes.

49:10And the market cap, of course, can be changing for many reasons. It's going to depend on the dividends. It's going to depend on mergers and so forth. That's going to affect the total market value of the company. But when you put all that stuff together, it's really the change in the market value component that tries to value a premium. The book value of equity component doesn't really play a role. And what does this finding mean for investors who are pursuing the value premium? Again, I want to be kind of cautious because just like we discussed before, now there would be a great temptation that instead of investing with this rule, I'm going to invest 100 % of my money with this rule.

49:40So being the most cautious, you could just say that simply statistically, if you have any predictor that seems to be predicting returns and you believe in it, and you can decompose a predictor in different ways. Like in our case, there would be three components, the historical or the long-term average book to market, the change induced by the book value of equity, and the change induced by the market value of equity. If you use that as one predictor, it's probably going to be less powerful than if you decompose the predictor. Assuming that you have enough data to do so, you can definitely get some value out of that.

50:12Our results that are specific to the US, they seem to be telling us, again, you would have gotten more premium with less risk by using the market value equity component than you would get by using the total book to market. I'm not making guarantees that that finding is going to be open in all instances and all companies and so forth. Yeah. Yeah. That's the thing with financial economics, I think. There aren't many guarantees. What do your findings mean for the theoretical explanation of the value premium? So what we did up to this point was kind of like, look at the things just statistically and try to learn from the data if these three different components behave differently in terms of predicting returns.

50:47But the interesting question is the one that you are asking, well, what's the explanation for what gives you the value premium? And our answer is going to be kind of negative on that one. After the former French paper, there was this cottage industry that started that was churning out explanations for the value premium. So all kinds of different papers that we're looking at, different components of consumption and things like that to understand, well, what is the risk that people see in value stocks to explain the high premium? The way these things work is that you have some kind of theory. The theory is suggesting that there's some kind of empirical factor that you can be constructing, and then that factors can be covering positively with the value factor.

51:20That would be explained that in this dimension, value is more risky. And kind of surprisingly, many of these theories in the work. So even though they have very different explanations, they all seem to be explaining the riskiness of value. But if you think about what we are doing, we are splitting the value factor in two components. There's the good part, the one that gets the premium, and there's the bad part that doesn't get the premium. And it turns out that these models, the factors that they suggest, they do co-vary with the value factor, but they co-vary with the bad part of the value factor, the one about the book value of equity or the persistent value component.

51:53They don't really co-vary with that component that gives the value premium. So they cannot be explanations for by value stocks or by growth stocks. They're grabbing onto the wrong component of the predictor. Really, really interesting. We've got one more question for you just to finish off on a topic. The geekiest segment of our podcast community was very excited about when your papers on this came out. So I hope they'll be happy that I'm asking you about it. What are your findings on factor momentum suggest about pursuing momentum as a strategy? That's one of the key ideas of that paper. So if you think back to momentum, people often say that the momentum has to be behavioral effect, that there's no good reason why something like this would exist in a rational world.

52:36Of course, we can have theories for momentum that would make risk and so forth. But people said that that seems at least some segment of the economic population said that that seems kind of implausible. At the same time, people were pointing out that if you look at the data, momentum doesn't seem to be a free lunch. If you look at the stocks that are winners, those winners tend to be covering with other winners and losers tend to be covering with other losers. So there has to be some kind of commonality to those. But people couldn't find the one factor that would be explaining that. So what we are saying in the paper is that the momentum actually resides in the factors themselves.

53:06You have a large number of factors, like you have the Pharma-France five-factor model, but you have many, many, many factors beyond that. Some of the factors may be useless as unconditional investments, but when they have good performance, they keep on having good performance. And they have bad performance, they keep on having bad performance. So the mechanism that we have here is that if you have momentum in the underlying factors, and you have all these different stocks that load on these factors, then the momentum from the factors is going to be transmitting into the investor stocks. So when you look at a winning stock, it has to be, setting aside the idiosyncrasy component, it has to be that it's loading positively on the factors with good past performance, and it has to be loading negatively on factors with bad past performance.

53:47So you're trading momentum, but when you trade the momentum in investor stocks, you're indirectly trading the momentum with the factors. But that kind of brings up the issue of why momentum is risky and what factor momentum tells us about momentum. Every time that you trade momentum, you're effectively making a different bet on the underlying factors in the economy. When you buy the winners, you're going to be buying a certain configuration of the underlying factors. And when you sell the losers, they are going to be having opposite loadings on those factors, and you're going to be making bet on those factors.

54:16And so obviously, if you buy those winners, it's not surprising that all of them tend to be co-moving going forwards in time, and all the losers are going to be co-moving going forwards in time, because they have the similar factor loadings. It's just that when you now go forward one year and you rebalance those portfolios, the loadings are going to be very different because different factors have different past performance and so forth. But underneath it all, there's going to be this factor structure that's driving the returns on the stocks and the stock momentum, and it's also driving the differences and covering structures of the stocks that we are trading.

54:47Is momentum a separate factor then? Is it a standalone factor? So in our study, when we look at something like the UMD, that's going to be the complete momentum factor using all the equities in the US. And then you try to explain or span that with the fact of momentum constructed from the US equity factors, you find that there's no incremental value in the momentum. That you can capture all of the profits of the momentum factor by trading the momentum in the factors themselves. So our results indicate that there wouldn't be a separate momentum factor. To some extent, it's something that's kind of hard to prove.

55:19So we are lucky that in our sample, the factor momentum that we get from known factors is powerful enough to span the momentum. But you could imagine that if we didn't know about all these factors, that we were back in the 1970s and we had only one or two factors, the momentum in known factors wouldn't be probably enough to span the momentum that we see in stock returns. Because when you buy winners, you are buying factor momentum in all the factors that are out there, the known and unknown. So it's not easy to show that factor momentum spans momentum because we can only construct factor momentum in the factors that we know.

55:54Yeah, that's super interesting. So what are the investment? What are the portfolio management implications of factor momentum spanning momentum? At least it's something about rethinking about how you trade momentum, that you can always trade the momentum in the underlying assets. If you think that that's going to be profitable net of trading costs, it should also be strictly better to do it that in the factor space, because there might be better ways of trading like portfolio stocks than trading the individual stocks and trading the factors. So it should be helping you construct better portfolios if you're leaning towards momentum.

56:26Super interesting. Awesome. Good for our last question, Ben. Yep. All right, professor, how do you define success in your life? I was thinking about that one. I think I only have platitudes for you in the sense that for me, has been mostly about minimizing regret. That is kind of hard to know ahead of time, What do you want to be doing and where you might find success and what you're going to be good at? But I always think about I want to be doing more stuff. I want to try all kinds of new things because that's the only way that you can actually find something that's going to be interesting. And you can really maximize your success at a time because you don't know what's out there.

57:00But if you grab at every opportunity and try to do all kinds of things, it's going to be more fulfilling. In the sense that I don't think we are ever going to be regretting things that we tried and it didn't work out for us. but we are going to be regretting that 10 years ago we didn't do something we had the opportunity to do something but we didn't take the opportunity and then we think that well what might have happened so as i get older and older i'm painfully aware of that one that i always think about well i must try to do more and more both in like work and life so that i'm not going to grow up with regrets that well i should have done that thing over here and also that's i guess kind of specific academia that are many of these people who have been super successful they have done some really important work.

57:39But as they get older, they tend to be doing what comes east to them. They keep on doing variations of the research that they did before. And then they are 60s, 70s, 80s, and they kind of do the same stuff. And that might make them happy, but I'm kind of fearful that that's where I'm going to be ending up. So that's why every single day I try to do something different. Wow. It's really interesting. Yeah. Really cool answer. All right. Yanni, this has been an excellent conversation. We really appreciate you joining us on the podcast. Thank you, this was great. Thanks so much for having me.