January 30, 2025

Episode 342 - Matthew Ringgenberg: When Do Anomaly Returns Happen?

January 30, 2025

Matthew Ringgenberg is a Professor of Finance at the University of Utah. Prior to joining the University of Utah, he was an Assistant Professor of Finance at Washington University in St. Louis. His research focuses on the role of institutional investors in financial markets. Specifically, he has examined how the actions of short-sellers, hedge funds, mutual funds, and exchange traded funds interact with various frictions to affect real economic activity and the formation of asset prices. His research has been published in the Journal of Finance, the Journal of Financial Economics, and the Review of Financial Studies and has been cited in The New York Times, Bloomberg, and The New Yorker. He currently serves as an Associate Editor for Management Science and the Journal of Financial and Quantitative Analysis.

Prior to his academic career, Professor Ringgenberg worked as a consultant for Charles River Associates in Chicago. He earned a bachelor’s degree in Finance and Economics from the University of Wisconsin in 2003, a M.S. in Economics from the University of North Carolina in 2009, and a Ph.D. in Finance from the University of North Carolina in 2011.

Today we are joined by the Professor of Finance at the University of Utah, Matt Ringgenberg to discuss everything related to anomaly returns. Matt’s research – mainly centred on the actions of short sellers – has been published in all the major journals including the Journal of Finance, the Journal of Financial Economics, and the Review of Financial Studies. We begin with the definition of an asset pricing anomaly before learning about the anomalies that Matt’s research is primarily focused on. Then, we unpack anomaly returns and how they relate to anomaly signal information, what causes anomalies, the risk versus mispricing debate, and the barriers to accessing financial data that allow anomalies to persist. We also weigh Matt’s research against its anomaly-denying counterparts, assess anomaly behaviour before and after publicly available signal information, explore models that help to predict future anomalies, and learn more about the economic mechanism underlying asset pricing anomalies. To end, we dive into Matt’s paper, ‘The Loan Fee Anomaly’ and explore the relationship between cross-sectional predictors and market returns, and Matt explains why long-term happiness is the only true marker of success.

Key Points From This Episode:

(0:05:07) Matt Ringgenberg defines an asset pricing anomaly and describes the anomalies his research is focused on.

(0:06:27) When anomaly returns appear relative to the release of anomaly signal information.

(0:07:57) How the annual forming of portfolios in June affects anomaly returns.

(0:08:50) The cause of anomalies, and the risk versus mispricing debate on anomaly returns.

(0:10:35) Unpacking the barriers to accessing financial data that allow anomalies to persist.

(0:13:41) How Matt’s rebalancing approach could affect anomaly-denying research.

(0:14:37) Applying his work to valuation-based anomalies and to investors capturing anomaly returns in live-traded portfolios.

(0:16:04) How anomalies behave before anomaly signal information is publicly available.

(0:17:48) Exploring the models that can be used to predict future anomaly signals.

(0:19:05) How anomaly premiums traded on predicted signals compare to trades on actual information release dates.

(0:19:37) Understanding the economic mechanism underlying asset pricing anomalies.

(0:24:38) Dissecting one of Matt’s short-selling papers, ‘The Loan Fee Anomaly’.

(0:32:51) The relationship between cross-sectional predictors and market returns.

(0:39:11) What Matt hopes to pass on to his students in his Introduction to Investments course.

(0:40:48) How Matthew Ringgenberg defines success.

Read The Transcript:

Ben Felix: This is the Rational Reminder Podcast, a weekly reality check on sensible investing and financial decision making from two Canadians. We are hosted by me, Benjamin Felix, Chief Investment Officer at PWL Capital, and Mark McGrath, Associate Portfolio Manager at PWL Capital.

Mark McGrath: This is our first recording of 2025.

Ben Felix: That's right. We had a few pre-recorded in the can at the end of last year, so we were saying we were feeling a bit rusty.

Mark McGrath: Yes. I was in Mexico for the past two and a half weeks. I just flew back. We're recording this January 7th. I just got back late last night, and so I was like, “Ah, this is not going to be a good one. I won't be on my best behavior. I'll be tired and everything but.”

Ben Felix: You're on the ball. You're great.

Mark McGrath: I felt really good about it. It’s one of those things where you don't do it for a while, and then you do it, and it's like the best one you've ever done kind of thing. I'm not saying this is the best one. But sometimes, you just need a break and a refresh, and things come back to you pretty quick.

Ben Felix: Totally. Today, our guest was Matthew Ringgenberg. He wanted to be introduced as Matt, but his name is Matthew. He's a professor of finance at the University of Utah. Some of his papers have been discussed in the Rational Reminder community. That's not why we invited him on. It was just a nice coincidence that we'd already reached out to him. He's a frequent co-author of one of our recent past guests, Davidson Heath. We invited him on after we talked to Davidson, and he's got some recent research on anomaly returns that are just fascinating. That's what we talked to him about for quite a bit of the time.

Matt's research focuses on the actions of short sellers, which we talked about a little bit. It wasn't a main focus of our conversation, but he's got a ton of interesting research there. Hedge funds, mutual funds, and exchange-traded funds, and how they interact with various frictions to affect real economic activity in the formation of asset prices. His research has been published in all the major journals; Journal of Finance, Journal of Financial Economics, Review of Financial Studies. He currently serves as an Associate Editor for Management Science and the Journal of Financial and Quantitative Analysis, which are two major journals in this field.

The research that we talked about that I found just fascinating to read and to discuss with Matt, you can tell me your thoughts in a second, too, Mark, but on the timing of anomaly returns. Anomalies are the return premiums that people have heard us talk about in this podcast a lot like systematic return premiums that we see in the data. But Matt looked at when those anomaly returns actually happen relative to when the information that would allow you to identify that anomaly is released. He used the example a few times of the asset growth anomaly.

As soon as you have the information available to measure that anomaly, the return premium tends to happen very close to that information release date. What that implies is that if you're not implementing your portfolio very, very quickly based on that information, it’s really hard to capture these anomalies. That's one main takeaway.

Then another really interesting thing is that in academic research, they typically form portfolios in June of every year, and that completely ignores what I just said. It completely ignores the timing of information release, which Matt finds affects anomaly returns pretty significantly. When you look at research that says, "Well, no, anomalies are gone now," or they've decreased significantly, Matt's research can revive them. Before, if you go and measure the anomalies closer to that information release date, all of a sudden, things get a lot more interesting.

Anyways, we talked about that quite a bit. We also talked about the loan fee anomaly, which is another really interesting anomaly. We talked about some research on cross-sectional predictors and whether they contain systematic information that can be used for basically market timing. Do cross-sectional predictors contain information that can predict the time series of at returns?

He teaches an investments course. We talk to so many professors. We don't usually ask them about the classes they teach, but I just noticed that Matt teaches an introduction to investments course. We asked him about the main lesson that he tries to leave his class with in that course, and he had some great answers. What did you think?

Mark McGrath: No, it was fantastic. We talked a bit about predicting those anomalies in advance, too. I think some of his answers there were – of all the cool things we talked about, to me, that was actually one of the more fascinating things is predicting anomaly returns before they happen and how they might do that. To your point about the timing of the release of information, I won't spoil it, but there's some really, really interesting insights from out there.

Overall, great episode. Obviously, he’s very eloquent, and he explains things in a very easy to understand way, which is really good for me because half the time I'm doing these episodes, and I'm acting like I know what I'm hearing. That's not always the case. But I think in this episode, I understood a good chunk of it, so that was nice.

Ben Felix: Yes. He explains things really, really well. There's the anomaly signal. Returns happen really close to that, so that's interesting. That's one paper. But then the one that you just mentioned, that one was absolutely mind-blowing. There are signals that anomalies are based on, so that information gets released. You have to trade very close to that to capture the premium. But you can also predict that signal before the information actually gets released. That's the second paper on can you predict anomaly signals before the information actually exists, and is the premium still there? If you do that and they find that it is, I think he said it's even stronger with the predicted signal.

The implications, though, for portfolio management but also for market efficiency, for the explanation of why we see return anomalies, are they data mining? Are they risk-based? Are they based on missed pricing? His research has really, really interesting implications for all of those topics.

Mark McGrath: That was a great episode.

Ben Felix: I could go on and on, but let's go to our episode with Matt Ringgenberg.

***

Ben Felix: Matt Ringgenberg, welcome to the Rational Reminder Podcast.

Matthew Ringgenberg: Thanks for inviting me. It's an honour to be here.

Ben Felix: Really excited to be talking to you. All right, so let's start this off with a question that I know our listeners are going to love the answer to. What is an asset pricing anomaly?

Matthew Ringgenberg: I feel like you led that in with a lot of pressure. But at a high level, I would say an asset pricing anomaly is a trading signal that predicts future stock returns. More specifically, I would say normal returns, and I think that abnormal part is important.

Mark McGrath: What types of anomalies did you study in your research on the timing of anomaly and returns?

Matthew Ringgenberg: We mostly focused on accounting-based anomalies. Think about variables that show up in, say, a firm's annual report. They were the types of anomalies where you could just go to an annual report and find the information you needed to actually do the trading strategy.

Ben Felix: Can you talk about why you didn't have valuation-based anomalies in the research?

Matthew Ringgenberg: Well, a lot of the valuation-based anomalies, by their very definition, involve some stock price measure. The real purpose of our paper was to try to understand how the release of information about a particular trading strategy related to the specific returns of that strategy.

The tricky thing about valuation ratio, think about something like the price-to-earnings ratio, I know when earnings comes out. I can see and measure that date. But when does the price come out? It basically is continuously updated. It’s a little bit harder with some of the valuation measures to really think about when we first learn the valuation measure.

Ben Felix: The next question is the result of the paper. The result is, man, mind-blowing stuff, and its implications are, too. When do anomaly returns show up relative to the release of anomaly signal information?

Matthew Ringgenberg: One of the key punch lines of the paper is we basically show that if you really are graphing the returns to different strategies, what you see is that most of the return is earned right after the information that you need to start the strategy is first released. Basically, you take something like the asset growth anomaly, which is formed by looking at a firm's assets from their financial statement. As soon as that asset's information comes out, that's when you see the strongest return predictability. Then the farther you get away from that date, the more that result just decays away.

Mark McGrath: Wow. Okay. How is the timing of anomaly returns trended over time? Well, I think that's the other big result we have in the paper is that if you look over the last couple of decades, the returns to different trading strategies have been moving earlier and earlier. Or put differently, they get closer and closer to when the information was first released. In that sense, I think it's an intuitive result. Once information is released, traders race to be the first to profit from it. What we basically show is they're getting faster and faster at doing so.

Ben Felix: That trend in anomaly returns being closer to the information release date is basically information is moving faster now than it used to.

Matthew Ringgenberg: Exactly. I think it's actually two things. One, the information itself is being disseminated faster. Two, people are processing it and trading on it faster.

Mark McGrath: Makes sense. How does the academic convention of forming portfolios annually in June affect anomaly returns?

Matthew Ringgenberg: Well, I think this is one of the interesting side effects that came out of our paper is when we really started looking at when these strategies earned their returns, we realized it was super important to look at the period right after the information first came out. But then we noticed that if you go and actually look at what academic studies do, they almost always form portfolios once a year, usually in June. It actually turns out for a lot of these anomalies, June is way after the information was first released, sometimes weeks, sometimes months. In the extreme case, it can be almost a year. What that basically means is that the existing academic studies are often using really old stale information, and that makes anomalies look weaker than they actually are. If anything, these academic studies were understating the return predictability that was really in the data.

Ben Felix: What do you think the results suggest about the cause of anomalies?

Matthew Ringgenberg: I think that's another interesting question, and it's something academics have been debating since the first anomaly was discovered way back in the 1960s. In some sense, there's a group of academics who think anomalies are not actually there. They might just be artifacts of looking at the data too much. There's a group of people who think anomalies really are there. I think our results actually are supportive of that latter view because the fact that you actually see returns concentrated in that period around the information release is suggestive of something about a connection between returns and that release. In particular, I think what it shows is that it takes time for information to get fully impounded into presses. That's what an anomaly really is. It's delayed processing of information.

Ben Felix: That's like a mispricing story for anomalies. Does that relate at all to the risk versus mispricing debate on anomaly returns?

Matthew Ringgenberg: This is a very complicated debate, and so I'm a little hesitant to take a strong stance on it. But in some sense, I think a lot of stories about mispricing. Imagine that the asset growth anomaly is really just a proxy for mispricing, so the asset's value tells you something about mispricing. Well, you'd expect that stocks with a certain asset level would just earn different returns than stocks with a lower asset value. There's not necessarily a reason to believe that you'd see this return pattern that we document right when the asset information first comes out.

Now, there's a subtler angle to that, which is, of course, if I learned something new about the riskiness of a stock, then you might actually get a similar prediction. But I think in the most part, our results suggest these things are not a risk-based explanation. They really are just delayed processing of information about the anomaly.

Mark McGrath: Interesting. What about barriers to accessing financial data? What are the current barriers to accessing financial data that would allow anomalies to persist?

Matthew Ringgenberg: Well, this is one of the other things. I've been going down a rabbit hole to study more about the history of financial markets, and I won't speak for you guys or your audience. But at least for me, I think I used to have an under-appreciation for how much financial markets have changed over the last, say, three or four decades.

Just go back in time a little bit. If you were a trader in the 1970s and you said, "I want to trade on this asset growth thing, how do I actually do that?" In the 1970s, it would have been incredibly difficult. Back then, if you wanted a firm’s 10-K and you weren't already the owner of this thing, you would have had to wait until they physically mail their annual report to the SEC. The SEC would have gotten this thing, taken a couple weeks to process it. They would have put it on the floor of one of their main libraries. They had a library in Chicago and New York and DC. You would have had to have sent an analyst to go physically get this 10-K and look at it. It was incredibly hard to get this information back in the day.

Today, we have the information revolution. That stuff's a lot easier to access. But still, if you really want to get this data, which we call point-in-time data, before everyone else gets it, the only way to do so is to pay a third party who actually collects and processes all this data. Of course, they understand the value of this. There's a price to do this. Even in the information era where we think all this information is free and instantly available, there are nontrivial costs to getting it. Then on top of that, even once you get this information, you've got to process it. You've got to figure out how to use it.

Ben Felix: Yes. That's really interesting. We work a lot with dimensional fund advisors. One of the things they talk about is they've built all these internal processes and systems for doing that. You need that capability to do this without getting killed by information access costs. Does that make sense?

Matthew Ringgenberg: Exactly. I think that's exactly right, and I think this is one of the things where it's really easy to say, "Hey, in the information era, information is free.” I think that's an oversimplification. I think in a lot of these cases, there are still serious frictions that make it not impossible, but make it challenging for information to be processed as quickly as you might think.

Ben Felix: I believe it. We pay for Morningstar Direct and YCharts. Stuff like that's not cheap, and that's not even going to give us timely access to data to do stuff like this.

Matthew Ringgenberg: Exactly.

Mark McGrath: Well, not only that, but once that type of information becomes super low cost or free, then it'll be new information that's harder to access that people will want to use for trading strategies.

Matthew Ringgenberg: Exactly.

Ben Felix: Your point, Matt, on how things have changed, since the seventies, I was reading this morning a paper in the Financial Analyst Journal from 1976, and it was an interview with Ben Graham. He was talking about how much harder it's gotten to do individual security analysis to get an edge since 40 years prior to that, so the 1930s, because now there's so much research being done. Then the same things happened again over the next 40 or so years.

Matthew Ringgenberg: I think that's exactly right. Markets have gotten more and more competitive every decade since Ben Graham first started doing his work. In some sense, that's a great thing. It helps capital get more efficiently allocated in the economy. But it also means it's harder and harder to get an edge if you're a trader.

Ben Felix: I don't mean to call somebody out here, and that's not what I'm trying to do. We did an episode with Andrew Chen, who I think you mentioned in your paper, and he talked about how anomalies are gone, basically. How would your rebalancing approach, closer to information release dates as opposed to the academic convention, how would that affect research like Andrew’s?

Matthew Ringgenberg: I don't think Andrew and I actually disagree on very many things. I think at a high level, if you look at just anomalies as they were originally discovered in the academic papers that first wrote about them, they're not very strong anymore. I think part of that is, as we show, if you follow the original strategies, you end up rebalancing using information that's weeks, if not months, if not a year stale. By then, most of the alpha's gone.

I think in that sense, if you take the strictest definition or the strictest look at these anomalies, a lot of them don't look very good. But our point is, I think, subtly different, which is if you were to actually optimize and try to trade these things as fast as you possibly could, they’re still very much in the data.

Mark McGrath: You mentioned valuation-based anomalies earlier. How would your findings apply to valuation-based anomalies like price to book?

Matthew Ringgenberg: That's an interesting issue. Unfortunately, there's not a very simple answer to that question. In some sense, as we said earlier, the trick with valuation-based anomalies is since they involve a price-based measure, I don't have a discrete way of measuring when the information first comes out. That makes it harder to apply our methodology to those types of things.

That having been said, I think there's a whole realm of research that could still be done to further explore this idea in those contexts. If you think about things like price to book, the price number, I don't know when that first comes out, but the book value number I certainly do. You could apply our methodology to the book revelation, so when that information first comes out.

Moreover, you might also be able to argue that there might be certain times where things like the price measure are more salient or more visible to investors, and so it’s possible you could find a similar result around those events. It's not something I've looked at. But, again, I think there's a lot of interesting stuff that could be done.

Ben Felix: That is really interesting. What do you think the implications of your findings in this paper are for investors who are hoping to capture anomaly returns in live-traded portfolios?

Matthew Ringgenberg: In some sense, I think anyone who's seriously trading in markets knows this already. They're ruthlessly competitive, and our results basically say to do this, you need to be competitive. You need to be one of the fastest to react. If you are trading on these things three months after the information first came out, you're too late.

Ben Felix: Makes sense. We talked about anomaly returns after the information is released. How do anomaly returns behave in the period before anomaly signal information is publicly available?

Matthew Ringgenberg: That's something I've been thinking about a lot lately. In my original paper, as you said, we looked at when information first came out and then looked at how returns moved after that. The more I thought about that, the more I thought about this race to be the first to trade, I started thinking, “Well, how do you actually be first?” It occurred to me that for some of these information releases, I might actually be able to predict what the information itself says.

We actually then started looking at this, and we started realizing that for a lot of accounting releases, it's very easy to predict what the financial statements are going to say, sometimes three months, sometimes as much as a year prior to when the firm actually finalizes these things. Then what we've started doing is going and looking at how the stock returns move in that period prior to the release of information.

I think the astonishing thing there is we find a very strong pattern. Much like the pattern we see after information releases, we basically see that prices drift in a predictable manner prior to the release of some of these signals.

Mark McGrath: Wow. Okay. Then I think you partially explained this, but what explains the market's ability to incorporate information before it's public?

Matthew Ringgenberg: That is a fascinating question. The fact that the returns are exhibiting this predictable drift suggests somebody else is thinking the way that my co-authors and I were thinking. It kind of suggests that someone else is out there actually doing this predictability exercise. I suspect what's going on is we're not the only ones who realize that I can predict the asset value a quarter in advance. I think there are traders out there who are putting their position on earlier and earlier and earlier as they try to profit from that competition to be the first, the race to be the first.

Ben Felix: What models can be used to predict those future anomaly signals?

Matthew Ringgenberg: The interesting thing is we initially weren't sure how hard it would be to predict some of these signals. It actually turns out that for most of them, almost any predictive model that you use does pretty darn well. In general, I'm a fan of parsimony. We tried to start with the simplest possible model, and the simplest possible model here is just a martingale model, which is just a fancy way of saying, basically, if I want to know the best guess for tomorrow, I just use the value I have today.

It turns out for a lot of variables, that works pretty darn well. In some sense, if you think about it, think about something like the asset growth anomaly, firm's assets levels are pretty darn persistent, quarter to quarter. If you were one of the high-asset growth value firms last period, chances are you're going to be again next period.

Mark McGrath: Which anomaly signals are the most predictable?

Matthew Ringgenberg: Consistent with what I was just saying, the most predictable ones are the ones where the accounting information, when you look at it quarter to quarter tends to be the most persistent. Things that involve assets or sales or revenues, those are the types of variables where if you were to just plot a graph of them from the firm's annual reports, you would see a pretty steady pattern. What that means is if you're trying to predict the anomaly signal, it's usually not that hard.

Ben Felix: How do anomaly premiums traded on predicted signals compared to those traded on actual information release dates?

Matthew Ringgenberg: What we actually find in our paper is that trading on the predicted signal – let's say you get in basically one quarter before the signal is released. Trading on that predicted signal adds maybe something like three percent of alpha to the overall return strategy. I don't know if you consider this huge or not, but it's not trivial.

Mark McGrath: I'd say that's pretty huge in our world.

Ben Felix: Yes, definitely.

Matthew Ringgenberg: Agreed.

Mark McGrath: What do your results on predictable anomaly signals say about the economic mechanism underlying asset pricing anomalies?

Matthew Ringgenberg: Yes. This is where I think it gets a little bit more interesting and a little bit more subtle. The honest truth is I can't even decide what I think in total on this. In some sense, I'm tempted to say it suggests anomalies are more anomalous than we even previously thought. Here, basically, academics have been sitting around for the last 40 years puzzling over the fact that there are certain accounting variables that can be used to predict future abnormal returns.

Now, I'm telling you, actually, it turns out I could have predicted that anomaly itself, and somehow that's still not being accounted for. I'm tempted to say it suggests that maybe markets are a little bit less efficient than we thought. That having been said, one of the other things we also see in our paper is this result is getting weaker and weaker over time. It does look like market participants are doing this. As they do this, prices are getting more efficient.

Ben Felix: Interesting. Okay. Same thing that's happening with actual information, release data anomalies, the premiums are reducing over time.

Matthew Ringgenberg: Exactly. I mean, that's one of the things we see. Early in our sample, if you were to go back 20 or 30 years and try to trade on these predicted anomalies, you would have actually made a very large amount. In more recent period, what we basically see is that as every year goes by, the returns to these strategies move earlier and earlier and earlier. In fact, for some of the anomalies in our sample now, if you really want to make money, you can't even trade one quarter early. You need to trade two quarters early or three quarters early.

Ben Felix: Wow. The premiums are decreasing, and you have to be earlier and earlier to capture them.

Matthew Ringgenberg: Exactly. Again, I think that's consistent with this idea that a lot of what's going on with these anomalies is, basically, it takes time and money and effort for traders to process information and actually develop these strategies. They are competing on this. The more competition there is, the more the returns to these things basically get arbitraged away.

Mark McGrath: It's interesting you say that. On the one sense, it seems like it's less efficient over time. But the way you've described it in my head seems to make it even more and more efficient over time. Given that these signals are showing up even earlier and earlier and earlier, I need to be predicted in advance. It almost seems like the information needs to be predicted, and it is being predicted, which seems to me makes things incredibly efficient, or at least to your point, incredibly competitive.

Matthew Ringgenberg: I think I would agree with that, and that's where I said in In my own head, I've got conflicting views on what all of this means. Like I said, in some sense, the fact that anomalies predict returns is itself puzzling. The fact that I can predict the anomaly and that even predicts returns further is itself puzzling. But that having been said, it does look like these effects are getting arbitraged away as time rolls on, and that is broadly consistent with markets being pretty darn competitive.

Ben Felix: It sounds like markets are maybe less efficient than we thought because anomalies are anomalous, but they're getting more efficient over time.

Matthew Ringgenberg: I think that's the exact way I would say it. Well put.

Mark McGrath: Wow. What are the implications of your findings for investors hoping to capture anomaly returns in live portfolios?

Matthew Ringgenberg: Again, the real key to this is lots of people are trying to trade on these anomalies. If you're using the standard information that everybody else has access to and you're trading when everyone else is trading, it's too late. What I think we show is there's actually ways of using publicly available information to develop signals that might not be completely obvious to all market participants. Doing so can help you generate a trading edge.

Ben Felix: Really, really interesting stuff. It is nicely complementary, I think, to Andrew Chen's research. You look at the standard academic data, and yes, maybe this stuff doesn't look so great. But you've uncovered some really interesting stuff about the timing of information and how that affects anomaly returns. It really does drive home the point that you can't have a quarterly rebalanced strategy or something like that if you're trying to capture anomaly returns.

Matthew Ringgenberg: Yes. I think that's right. Again, as the old joke goes in finance, if there's a $20 bill on the ground, someone's going to pick it up. I think this is the case here. There's so much research now on asset pricing anomalies that lots of people are looking into it. If you really expect to generate alpha from this, you're going to have to do something different. You're going to have to be either faster or have better information.

Ben Felix: Or better information processing capabilities.

Matthew Ringgenberg: Exactly. Or better processing capabilities.

Ben Felix: Yes. Really, really interesting. It really makes me think about – I don't mean to keep plugging Dimensional, but their Founder, David Booth, says something along the lines of, "The ideas are free. Everyone knows the stuff, but implementation is really hard."

Matthew Ringgenberg: I think that's the other thing, too, is from my own work on this stuff, things like transaction costs are rarely really taken seriously in academic papers. The devil is in the details on this stuff. When you're talking about a realm that is so hyper-competitive, the implementation is really the key. All these anomalies are well-known now. To actually profit from these, you have to be an unbelievable expert at implementation.

Ben Felix: All right. I want to move on. I only have one paper of yours on short selling. You have a ton. This is a major area of your research, so I almost feel bad having just one paper in here. But it's one that fit best, I think, with the general theme of our podcast. I look through all of your papers on short selling. It's incredible stuff.

Matthew Ringgenberg: Short selling has always been near and dear to my heart, so I'm happy to talk about it as much as you want.

Ben Felix: Your papers on this are incredible, but we're going to ask about the loan fee anomaly, though, which is one of your papers on this topic. I'll ask the question, but maybe you can also describe what equity loan fee means just for our listeners. How well do equity loan fees predict returns in the cross-section?

Matthew Ringgenberg: Sure. Let me back up and just provide a little bit of context. If you want to short a stock, what do you do? So what you do is basically you have to go borrow the share from somebody who already has it. You've been talking about Dimensional. So if I want to short a stock, maybe I go to Dimensional. I borrow shares from them. Then what I'll do is I'll take those shares. I will sell them into the stock market. I hope that maybe over the next month, say, the stock price falls. The stock price falls. I buy those shares back. I return them to the person from whom I borrowed in the first place.

Now, you might say, "Why on earth would anyone lend their shares to me, the short seller?" Well, it turns out basically I pay them a rental fee. If I get shares from Mark, I have to pay Mark a little bit of a fee every day to incentivize him to give me those shares. That fee we call sometimes the short selling fee or the equity lending fee. That's what we examine in this paper. And what we basically show is the stocks that have really, really high equity lending fees do predictably bad going forward. High lending fees today tells me you're likely to have very low returns over the next month.

Mark McGrath: Really interesting. And why would those expensive to borrow stocks have lower expected returns?

Matthew Ringgenberg: I can think of a couple different mechanisms through which you'd get this result. One, short selling is costly, it's risky. And as a result, short sellers tend to be informed traders. What do I mean by that? The fact that short sellers are exposed to a lot of risk and that they have to pay a fee to do the trade and extra fee to do a trade means they rarely short just for the fun of it. They do a lot of homework on average and they tend to short when they're very confident in their beliefs.

And so one way you can think about this equity loan fee is what economists will call a revealed preference argument. If you see a stock with a really expensive loan fee, that means I have to pay Mark a lot of money to borrow his shares. If I'm still willing to do that, it must be that I'm pretty darn sure the stock is going to go down. The fact that I'm willing to pay a fortune to borrow the shares tells you I'm pretty darn confident in my belief. And that's one of the things we see.

The other thing that I think is a more subtle argument is these fees are a transaction cost. And in any case where there are transaction costs, that actually makes it hard for other traders to arbitrage away mispricing. The paradox to our paper is, I tell you, the high loan fee stocks are the ones that underperform the most, and our result is incredibly strong on that. But what's the catch? Well, what our paper says you should do is you should go short the high loan fee stocks. Those are the precise ones that are really expensive to short.

Mark McGrath: That's really interesting because I think if you follow the GME saga and everything that was going on there, a lot of retail traders, at least from Reddit, were specifically seeking out these highly shorted or expensive to borrow stocks in hopes of a short squeeze.

Matthew Ringgenberg: And that's the other side of the coin. And that's also why I think my research and others have shown that short selling is a very risky business. Short sellers are exposed to a lot of unique risks. And so for that reason again, they tend to only short when they're very confident in their expectations about the future.

Ben Felix: How persistent has the loan fee anomaly been over time?

Matthew Ringgenberg: A lot of asset pricing anomalies have periods where they do really well and periods where they do less well. So if you really look at a lot of different anomalies, a lot of them say break in crisis times and do really well in booms or vice versa. The interesting thing about this loan fee anomaly is it basically seems to work all the time. I would say it's the most persistently consistent anomaly that I've ever come across.

We look at this in our paper. I think it's in table five. We have this plot where period by period we show you exactly how the loan fee anomaly does relative to all the other anomalies that have been documented. And one of the remarkable things that you see is it's almost always the best anomaly. And when it's not the best anomaly, it's still one of the best anomalies. It really does seem to be consistently a very strong predictor.

Mark McGrath: It survives its inherently high transaction costs?

Matthew Ringgenberg: Yeah, that's an interesting question. As I said, in some sense, it might not be surprising that loan fees predict returns because they are themselves a transaction cost. That makes it hard for you to arbitrage away the predictability. But what we find actually is that, in a lot of cases, these things survive transaction costs. So even after you pay the loan fee, it seems to predict returns.

Ben Felix: That's really interesting. Why do you think that it's been so persistent?

Matthew Ringgenberg: That's a more difficult question. In some sense, the fact that there's a transaction cost component means it is expensive to arbitrage away. I don't think I'd ever expect it to go away completely. The question that's more puzzling is, why hasn't it at least gone away to the point where the predictability is roughly equal to the transaction cost? And the answer there is I honestly don't know. But I will say I think there's a lot of other frictions that make people hesitant to short sell beyond simply the transaction cost. Short sellers, as I've shown in my other papers, they're exposed to recall risk. They're also exposed to political risk, and legal risk, and all sorts of other things that I think just make people less likely to short.

Ben Felix: Just because you've got some great papers on this, you've baited me, can you talk a little bit more about the unique tail risks of short selling?

Matthew Ringgenberg: Let me go back to my earlier example. Again, if I want short of stock, I want a short GameStop, what do I do? I need to borrow shares from an existing long investor. I go to Mark. I say, "I'm going to pay you a rental fee. Let me borrow your shares." It turns out that, in almost all cases, this agreement between Mark and I is an overnight agreement. I take Mark's shares. I sell them to you. I hope the price falls.

But let's say I expect that the price will fall sometime over the next quarter. So I have a one-quarter holding horizon in mind. Mark wakes up tomorrow on the wrong side of the bed and just says, "You know what, I'm done with this." And it turns out he can just call me and demand those shares back at a moment's notice.

Now for a short seller, this is a really, really serious risk. Why? Because basically it could mean you have to close your position before you wanted to. And I think the real catch here is when is Mark as a long investor more likely to call these shares back? Well, when the price has gone up and he wants to sell at a gain. That of course to me is the worst-case scenario. So now I have to go out and cover these things. I have to buy back the shares at a higher price. And so that actually incurs a loss for me.

Mark McGrath: Yeah, it makes sense. How can practitioners or investors use this information in managing portfolios, especially in a long-only context?

Matthew Ringgenberg: Well, I've done some trading on this myself, so I don't want to give away all the secrets here. But I'll say a few things. I mean, I do think there's a lot that could be done to incorporate this information into a portfolio, even a long-only portfolio. And so on the long-only side, I would say a surprising thing my co-authors and I have noticed in the data is that there are a decent number of investors who seem to own the high loan fee stocks, sit on them, and not lend them out to short sellers.

To me, this is the obvious no-brainer here. If you're going to own these things – one, if you're not constrained in terms of your portfolio and you don't have other reasons to own these things, you should get out of the high loan fee stocks. But if for some reason you have to own them, then at a minimum, you should be lending these out to the short sellers so that you can at least recoup some of this predictable price drop in the form of equity lending fees.

Ben Felix: That's very interesting. I wonder what clientele effects there are in high loan fee stocks? Is it unsophisticated retail investors that are holding them a lot of the time?

Matthew Ringgenberg: Well, not necessarily, actually. And I think there are a lot of different reasons why people hold high loan fee stocks. I mean, as we've said before, there are a ton of other asset pricing anomalies out there, and it could be that they're holding them for some other reason. But I also think a decent number of these might be investors who have some benchmark portfolio or even a specific index that they track. And so they're locked into holding these things.

Ben Felix: Interesting. That makes sense. Even a total market index fund is probably a good example.

Matthew Ringgenberg: Exactly. I mean, there's a certain number of stocks that are just going to have high loan fees each period, and they're going to have to own these things. My point is just, if you're going to own these things and you really have to, you might as well lend it out and get that extra revenue from the securities lending side.

Ben Felix: All right. We're going to move on to another paper on the relationship between cross-sectional predictors and market returns. It's a really interesting idea. Can you talk about the difference between a cross-sectional return predictor and a time series return predictor?

Matthew Ringgenberg: This is a little bit more technical, but here's how I would broadly explain this. Think about what we were just discussing, equity lending fees. My other work looks at the cross-sectional aspect of equity lending fees. Stocks with high equity lending fees tend to do worse than stocks that have low equity lending fees.

And when I say cross-section, I mean at a point in time, if you just look at the cross-section of stocks, that's the result you see. So cross-sectional predictors are the ones that predict the cross-section at one point in time. There's a whole different category of return predictability that we call time series. And this is where we're looking not at one point in time, but over time. And so what we do in some of my work is we try to look at basically when do cross-sectional predictors contain information. And do they contain the same information as the time series predictors?

Mark McGrath: So, how are they linked, the cross-sectional and time series predictors?

Matthew Ringgenberg: Again, it's a little bit technical, but one of the things we really look at carefully in the paper is that it turns out cross-sectional predictability and time series predictability can be linked, but they don't have to be. And the way I'll explain it is, again, go back to my equity lending fee example, stocks with high loan fees predictably earn lower future returns. It is possible that it is the idiosyncratic, the firm-specific piece of the return that I am predicting in that example.

If that's the case, then when you add up all this information across every stock in the economy, there's going to be no return predictability for the overall time series. Why? Because if you're predicting just the firm-specific piece, some firms every quarter have positive news, some firms have negative news. And when you add them all up, they tend to cancel out. What we show is basically the relation between the cross-section and the time series depends heavily on whether the cross-section is loading on the firm-specific return or if it's loading on the market component.

Ben Felix: That's a really nice way to explain it. It's the title of the paper that makes a lot of sense. Cross-sectional predictors, do they contain idiosyncratic or systematic information? If they contain systematic information, then they should be time-series predictors as well.

Matthew Ringgenberg: Exactly. And that's exactly what we find.

Ben Felix: Really cool. How did you choose the cross-sectional predictors to study in this research?

Matthew Ringgenberg: Well, in that paper, we basically started with just a general set of almost every cross-sectional predictor that had been discovered to date. We took a database of basically 100 of these cross-sectional predictors from leading academic papers.

Mark McGrath: And how did you organize them? Evaluation, opinion, existing literature?

Matthew Ringgenberg: What we really wanted to do was to understand whether or not these cross-sectional predictors in general contained market-wide systematic information. And so we started off by just looking at all of them. And what you find, of course, if you look enough at data, by random chance, you're almost always going to find some pattern. And then the real question is, "Is that just dumb luck, or is that really actually eliminating some economic force?"

And so we looked at all of them. We found some seem to have something. Some don't. What do we learn from that? So then we started saying, "Okay, what if we could organize these things into more manageable subsets to try to see if things that had characteristics in common also had that same result?

So what we broadly looked at is we looked at all the valuation ones. Why look at those? These are, again, things that have some ratio of market prices relative to fundamentals. Well, you might think that the fact that they have a market price in them suggests that there's some systematic information. We wanted to check just that group.

And then we also looked at a couple of other groups. The opposite of the valuation group is to look at what I would call the opinion group. These are things like analyst upgrades where it's just some analyst opinion about the valuation of the stock. We looked at all of those as a group. And then the last thing we did is we took all of the cross-sectional predictors that had been shown to have the strongest return predictability, and we said, "Let's just focus on the best ones and see if the best ones contain information about time series predictability."

Ben Felix: How well do the cross-sectional predictors tend to perform in predicting the time series of returns?

Matthew Ringgenberg: The punchline of this is not so well. What we actually find is, again, there are some that work really, really darn well. But when you really look at all 100 of these things, you find roughly about as many good ones as you'd expect by a random change. So that's where I said, again, one of the challenges whenever you look at a large amount of data, is that some patterns will show up just by done luck. So we find that some of these things really do seem to have systematic information, but not very many of them.

Mark McGrath: And what adjustments do you need to account for data snooping?

Matthew Ringgenberg: Yeah, I think this is something that the academic literature has been thinking about more and more lately, and especially this whole literature on asset pricing anomalies and return predictability, is how do you adjust for data snooping? And you mentioned Andrew Chen earlier. He and I have had some debates about this. And I think that the honest point is that the literature really hasn't converged on the proper way to do this yet. But what we do in our paper is we basically make an adjustment that corrects for the number of things we examined. That should actually make sure that the data snooping bias from the fact that we snooped through a bunch of different variables is corrected.

Ben Felix: Do your findings basically say that if we take cross-sectional predictor, like price earnings or whatever, that it's probably not going to be great for predicting time series returns?

Matthew Ringgenberg: I think one of the takeaways from the paper is that very few of these cross-sectional predictors contain systematic information. If they are working, it's because they're loading on the idiosyncratic component of returns. And so most of them just – if you really wanted to use them for market return predictability, they're just not going to add much.

Mark McGrath: I guess you just answered that. What are the implications for investors who might be tempted to time the market with these?

Matthew Ringgenberg: Again, this is like your market efficiency question from earlier where I feel like I'm going to give a wishy-washy answer that never really answers a question. In some sense, I think I'm skeptical of anyone who says, "I can strongly predict the market." Markets are incredibly competitive, and we know that. And so it's very, very hard to predict the market. I think our results suggest very few of these things can be used to predict the market.

That having been said, we do see some of them contain a little bit of information. And if you trust me, I generally do think it's possible that some of these things can help you, at least at the margin, improve your forecast of the market.

Ben Felix: You teach an Introduction to Investments course. I'm curious to know what are the most important lessons that you try to leave your students with by the end of the course?

Matthew Ringgenberg: I guess I'll say two related things. The two things I stress with my students, Thomas Sowell famously said there are no solutions, only trade-offs. And I think this is one of the ideas I really try to impart on every student I see. For most problems in the real world, you don't get to do some simple calculation and then get a nice answer like 42. Most problems involve trade-offs and you've got to dig down and understand what the trade-offs are. And if you don't see the trade-offs and you don't understand the trade-offs, you're not ready to make a decision. That's one of the key things I try to stress with everyone.

But since I teach an investments course, the closely related idea to that is I really hope everyone understands the risk-return trade-off. I think that's one of the most fundamental ideas in all of financial markets. And so at a high level, riskier choices require something extra to incentivize you so that you're willing to choose to take the risk. And I hope everyone understands that. Certainly, I hope all of my students understand that.

Ben Felix: Two great lessons.

Mark McGrath: Yeah, that's a great answer. I know you're talking about that with respect to or in the context of investing mostly. But Ben and I are financial planners as well and I think it applies universally to many fields, I'm sure, but specifically with financial planning, beyond investments. Great answer.

Matthew Ringgenberg: And I even joke with my students that, in my household, I'm not even allowed to do the grocery shopping anymore, because if you send me to the grocery store, I sit there and think risk-return trade-off for every item on the shelf. And it ends up taking the hours to even pick out a simple grocery cart full of stuff.

Ben Felix: That's pretty funny.

Matthew Ringgenberg: But I do think the risk-return trade-off is a very powerful and important idea.

Mark McGrath: Absolutely. This has been an incredible conversation, Matt. And We've got one more question for you, and that's how do you define success in your life?

Matthew Ringgenberg: That's a tough question. I'm going to give a very academic answer since economists often solve optimization problems. I think that the honest way I think about this is I want to maximize my quality-adjusted years on this earth, and so that means you want to be happy and you want to be happy for as long as possible. I think success is anyone who finds long-term happiness.

Mark McGrath: Great answer.

Ben Felix: All right, Matt, this has been a great conversation. You got a ton of great research. The stuff on anomaly timing is for us specifically because we manage portfolios using funds that try to capture some anomaly premiums. Your research is highly relevant and really interesting. And I don't know if worrying is the right word, but it's something to see how the trends have changed over time and how important information processing and access to information continue to be, even as we talked about earlier, in this age where information feels like it's free.

Matthew Ringgenberg: Yeah, completely agreed.

Ben Felix: Awesome. All right. Thanks, Matt. It's been great.

Matthew Ringgenberg: Thank you. It's been an honour.

Mark McGrath: Thanks.

Is there an error in the transcript? Let us know! Email us at info@rationalreminder.ca.

Be sure to add the episode number for reference.