bio: Harvard Business School
Marco Sammon is an assistant professor in the Finance Unit at Harvard Business School. He teaches FIN2 in the required curriculum.
His research is focused on asset pricing. Currently, he is working on several projects regarding the factors that affect the incorporation of information into stock prices, including the rise of passive ownership and financial journalism.
Professor Sammon holds a PhD in Finance from the Kellogg School of Management and a BA in Quantitative Economics from Tufts University. Prior to Kellogg, he worked in Supervision, Regulation and Credit at the Federal Reserve Bank of Boston.
Can we really understand the impact of passive ownership on the US market? Marco Sammon is an Assistant Professor in the Finance Unit at Harvard Business School. During this episode, he joins us to share deep insights into the complex and counter-intuitive nature of the index fund revolution. To kick off our conversation, we discuss some of the challenges associated with getting a true understanding of the scope of passive ownership across the US. Distinguishing between different approaches to investment, we begin to unpack Marco’s paper with Alex Chinco, titled ‘The Passive-Ownership Share Is Double What You Think It Is’. We touch on the relevance of Grossman Stiglitz in 2024, pricing and reconstitution, and the ins and outs of employee stock and compensation. Using the case studies of huge global firms, we consider how to best accommodate passive demand. Lastly, as an index investor who does not own index funds, Marco shares his opinion on whether index funds have had a net positive or negative impact on financial markets. Tune in today to get a more dynamic view of the complex world of index funding and investment.
Key Points From This Episode:
(0:00:45) Index funds, index and passive investments, and why Professor Marco Sammon is perfectly positioned to unpack these concepts.
(0:03:36) The challenges of understanding just how big passive ownership is in the US market.
(0:08:16) Distinguishing between partial investment, direct investment and passive funds.
(0:10:14) Important concepts on the closing auction and which indexes Marco focuses on.
(0:15:50) Defining the Grossman-Stiglitz framework and its validity in 2024.
(0:20:36) Evolving ideas around pricing and reconstitution over time.
(0:23:05) Why indexing is a fixed-point problem and how to measure market efficiency.
(0:32:19) Nuances of security demands around indexing and how it differs from other investors.
(0:38:02) Employee compensation and reverse causality as illustrated by Marco’s friend.
(0:42:10) Why it is important to distinguish between equal-weighted and value-weighted.
(0:44:13) How huge firms like Facebook and Tesla accommodate passive demand.
(0:48:19) Conditions that affect the responsiveness of firms in accommodating passive demand.
(0:51:13) The ‘Dead Reckoning’ metaphor to describe how we can know who is clearing the market.
(01:02:22) Marco’s thoughts on whether index funds have had a net positive or negative impact.
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're hosted by me, Benjamin Felix, and Cameron Passmore, Portfolio Managers at PWL Capital.
Cameron Passmore: Welcome to episode 322. And listeners may or may not know this, but we kind of wing the intros. We don't prepare at all other than the bio of our guests. And, today, we have an incredible guest. And my setup is quite simply. If you like thinking about stuff, if you're interested in index funds and you like hearing to someone that brings the energy and brings a lot of mental horsepower, man, you are in for an incredible episode today.
Ben, you found and invited Professor Marco Sammon on, who is an Assistant Professor of Finance at Harvard. And, wow, this index fund revolution, we've talked about this a lot over the past six years. But he goes into stuff that really, really makes you think. And this stuff is complicated, counterintuitive. I know we dipped into some of these topics with Mike Green recently. And, wow, what a conversation.
Ben Felix: Yeah. I agree. Totally incredible conversation. Marco's got research on other topics, too. But we really focused on his stuff related to index funds. We talked about how much of the market, of the US market in this case, is actually owned by not index funds but strategies that are indexing. And we talked about this with Marco. But you can have an institution, for example, that is an index investor but does not own index funds. How do you measure that? Well, he figured out a way to measure it. And we talk about that. How much of the market is actually indexed? Which is an interesting question.
We talked about how passive ownership has affected one measure of price informativeness. And we talked about – and I say this during the episode, too. This is probably my favourite paper of the ones that we talked about. Who actually clears the market when passive investors trade? If an index fund is buying or selling, who's on the other side of the trade? We got paper on that. It's just mind-blowing and fascinating.
And then, also, we talked about how index funds are involved in primary capital market transactions. When a firm goes public and issues stock for the first time, how are index funds involved? And how do they affect IPO pricing? And how do they effect how much capital firms actually raise?
Cameron Passmore: Employee exercising options. I never thought about that before.
Ben Felix: Who clears the market when passive –
Cameron Passmore: Who clears the market? It's wild.
Ben Felix: That paper is so, so good. And then we also asked Marco in his opinion whether index funds have had a net positive or negative impact on financial markets. I won't spoil the answer to that one. It's at the end of the episode though.
Cameron Passmore: And speaking of spoiling, make sure you stick around for his definition of success.
Ben Felix: Yeah. He's got a great answer to this exact question.
Cameron Passmore: He's a good answer. Very good answer.
Ben Felix: He's got his PhD in finance from the Kellogg School of Management at Northwestern. A BA in quantitative economics from Tufts. And prior to Kellogg, he worked in Supervision, Regulation and Credit at the Federal Reserve Bank of Boston. Yeah, great episode and great energy.
You mentioned that, Cameron. We were chatting after we talked to Marco before we recorded the introduction and you said something like imagine having a professor like that. That's a career-changing kind of professor. I agree. He clearly would be a great teacher. I'm sure he is. The way he answers questions is incredible.
Cameron Passmore: And this is his first ever podcast appearance.
Ben Felix: Yep. Love that.
Cameron Passmore: Love that. Yeah.
Ben Felix: Yeah.
Cameron Passmore: Okay, Ben. With that, let's go to our conversation with Professor Marco Sammon.
***
Ben Felix: Marco Sammon, welcome to the Rational Reminder podcast.
Marco Sammon: Thank you so much for having me on here. I'm really excited to talk about passive ownership.
Ben Felix: Awesome. We are very excited, too. To start on that topic, how much of the US Stock Market is owned by index funds?
Marco Sammon: If you want to pay attention to the words, you said by index funds. And remember, index funds are actually quite a broad thing. There is stuff that we think of like passive ownership, like VTI, which holds the whole market. But then there is stuff like COWZ, which hold the 100 most profitable stocks in the Russell 1000. And that was so successful they launched CALF, which owns the 100 most profitable stocks in the Russell 2000. Right? And if you add up all things that track an index, that's about 16% of the market.
Cameron Passmore: And how does this differ from, say, 20 years ago?
Marco Sammon: SPY I think was launched in 1993. And so, that was the first ETF. And before that, Vanguard had some pretty small index funds like – track the Vanguard 500. And so, I think about 20 years ago, index funds owned about 2% of the market. And it's grown a lot since then.
Ben Felix: Now, you kind of alluded to this, is index fund ownership a complete measure of the total assets that are managed passively?
Marco Sammon: Not necessarily. And I think that Alex Chinco and I have this paper The Passive-Ownership Share Is Double What You Think It Is. And we make the argument that there's a lot of other passive vehicles. For example, if you look at Utah pension funds 13F filing, every institution that has more than 100 million has to list every security they hold, every equity they hold. And you'll notice they hold a thousand stocks. And these are the Russell 1000. Right? When the Russell 1000 drops some stocks, Utah pension drops it, too. When Russell 1000 adds them, the Utah pension adds them. And so, you can imagine, this is a form of direct indexing. They on their own, rather than pay a management fee to BlackRock to hold IWB, they do the Russell 1000 on their own.
I also use a robo-advisor. My robo-advisor replicates an index for me. And so, that's another form of indexing. And there's also shadow indexing, which we could talk about more I think later. There are people who say they're doing something active but they might have some portion of their money which is invested passively. And so, when you're going to count up all that stuff, the number is going to be much, much higher.
Ben Felix: Can you talk about the challenges with measuring how much assets are held by passive investors?
Marco Sammon: Right. For example, Utah pension funds in F doesn't come. And I don't mean to call them out here. This is just one salient example when Alex and I were working on the project.
Cameron Passmore: It's a well-known example though. The Nevada one, that's a well-known example, too. Right?
Marco Sammon: Right. But I think Nevada, the one – they have that really famous New York Times article. But I think they're literally buying the ETFs. And so, that, it will be easier because then we'll just count up how much money is in the ETFs they hold. I think the challenge is – there's a few different challenges. One, when you look at a 13F for, say, Utah pension, there's no tag that says, "Oh, we're holding the Russell 1000. You're going have to figure that out yourself." The second thing is if you look at, say, a 13F for a financial advisor, right, maybe for you, you're doing some individual replication of the Russell 1000. Maybe for me, they're doing an individual replication of the S&P 500. Maybe for Cameron, it's the NASDAQ 100.
And if you try to kind of look at across all of us added up inside one institution, it's going to be hard to figure out, especially given overlap in indices, who's doing what. And the third thing is – and this is something actually I could talk about later is that it turns out using holdings data, there's a lot a lot of issues. John Shim and I have actually spent a lot of time. And we could talk later about this. And we've talked about some other projects I'm working on. The holdings data is actually kind of tricky to work with. It's disclosed at low frequency. There are delays in reporting. And so, trying to figure out who's tracking what from holdings data is actually kind of hard in of itself. It's a separate problem. And so, there's a lot of challenges to measuring how big passive ownership is.
Cameron Passmore: Can you talk about the you take in estimating the total passive share in your paper on this topic?
Marco Sammon: Right. Alex and I said, "Well, if the holdings data is going to be hard to use, what other data can we have that we have a little more certainty of?" And so, we said, "Well, what if we went through the idea and thought, if you were tracking an index, how would you trade?" If you were tracking the S&P 500 and a stock gets added to the 500 or a stock gets dropped from the 500, how would you trade?
And the logic was, well, the dollars traded in that stock would be equal to how much AUM you have tracking that index times the index weight. This is one equation. One unknown. Right? We observe the dollar volume. We know from – you can get data from Russell, or S&P, or whoever on what the index weight is. And we can literally just divide them and get an estimate of the AUM index.
Now, if you want to be even narrower, Alex and I went down to the TAC data and we said, "Look, if you really wanted to minimize tracking error, you would trade in the closing auction just like all the index funds do," to make sure you hold the stock at the price that it's added to the index. And so, you could again even further just say, "Well, dollars traded in the closing auction, in the stocks that are added and dropped on the day it's added and dropped, divide that by the index weight to get an estimate of AUM index."
Ben Felix: In this paper, how are you defining what a passive investor is?
Marco Sammon: There's going to be a few different things that are going to kind of get lumped together. And we can loop back to what we talked about earlier. If you think about the index funds that we talked about, if you're an S&P 500 index fund, you are going to be classified as passive under our definition because you're very, very likely to trade in the closing auction in the stocks that get added and dropped. The tracking error on the really big passive funds is essentially zero.
We're also going to classify someone who has – maybe Utah pension. If they have someone trading on their behalf. If they're trading in the closing auction, we're going to classify them as passive. Now, to clarify here, you might say, "Well, why doesn't Utah pension, if they know –" remember, Russell discloses a month in advance essentially. And who's going to go in and out is fixed even earlier than that. Why don't they trade earlier or later? That's a great question.
If you look in the data, Alex and I kind of have the picture that motivated this whole thing which shows there's no runup in trading before. Everyone trades in the day of. And no kind of slow unwinding afterward. And this is what you might actually predict in some like old models of trading. You want to trade where the liquidity is. If all the big index funds are trading there and everyone knows the big index funds are trading there, why not all trade at the same time? But that's another people. Anyone who trades in the close. If Utah pension trades in the close.
Now, suppose you're an active manager, suppose you have 100 billion AUM, you have 50 billion of good ideas and 50 billion you don't know what to do with. Well, maybe then, to minimize tracking error against your benchmark, you're also going to – when they add Tesla, and Tesla's going to be an index weight of 1.8% or whatever it was when it got added. Again, when Tesla gets added as an active manager, among that 50 million in your portfolio that you are using to track your benchmark, we will classify among that 50 million that you are passive. I guess it's those three things. People who have a portion of their portfolio allocated. People who are replicating index directly. And then actual passive funds.
Ben Felix: You mentioned the closing auction a couple times. Can you just explain for listeners what that means?
Marco Sammon: What it means or why we're focused on it, I guess?
Ben Felix: Just what it is. I don't know if it's a common term that everyone's going to be familiar with.
Marco Sammon: At the end of the day, kind of like any other auction, once you get close to the end of the day, you can submit different types of orders to trade in an auction. And so, two important ones to think about would be a market on close order, which says, "In this closing auction that's going to happen at the day, regardless of what the price is going to be, I want to buy a certain number of shares or sell a certain number of shares." You could also have a limit on close order, which says, "In this auction –" and this is going to be important for determining the price, right? "I am willing to buy these many shares at this price at the end of the day."
And so, an index tracker would be willing to put in the market on close order, say, for a stock added to buy it. Why? Because the closing auction price, which is the price that comes out of this auction, is going to determine the price at which that stock is added to the index. If they put in a market on close order, they're going to have no tracking error with respect to that add because they're going to get the stock at the closing auction price.
Cameron Passmore: Which indexes do you look at for your estimates?
Marco Sammon: We focus on a subset of indices that we have data on. And, of course, there are a whole bunch of indices. Like we said, we're not the COWZ 100. We don't have in our data set. We have the Russell 1000 and 2000. And the reason is, and we can come back to this later, it turns out, dividing a big number by a small number can create problems.
And so, we have to get the weights really, really, really right. If you think about a stock being added to the Russell 2000 or added to the Russell 1000, that calculation has to be dollar volume divided by the weight. And so, that weight is a really small number. And to make sure our estimates are precise, we have to divide by the right weight.
And so, we look at the Russell 1000 and 2000 because we have daily weights directly from Russell on what the weight of every stock is at every point in time with a high degree of precision. Similarly, we have pretty good data on the S&P 500 and S&P 400 for index weights. And we have the float adjustments, too, are important.
If we go back to Tesla, I think when they were added, it was like around 80% of the cap was considered float. If you had guessed their index weight without the float adjustment data, you would have massively overstated what their index weight would have been. Again, we have really good data on float adjustments directly from S&P, on the S&P 500 and 400. That's, again, the mid-cap and the large-cap.
Finally, we have the NASDAQ 100. QQQ is a really, really big ETF. We wanted to make sure we picked that up. And so, we just focus on those indices because we had really good data on the index weights and the float adjustment factors to figure out what the right number to divide by is.
Ben Felix: Okay. The headline finding, based on that approach, what is the total passive ownership share of US stocks?
Marco Sammon: That number, with just those indices, we have that 33.5% of ordinary common shares traded on major exchanges. To be clear, this is like not focusing on like SPACs and MLPs. Focusing on ordinary common shares like Apple and Google traded on NASDAQ or traded on NYSE. We have that it's about 33.5% with just those industries we looked at.
Cameron Passmore: Wow. And how much bigger does that number get if you include other indices?
Marco Sammon: If you think about like Vanguard, they started switching in 2014 and completed the switch to using CRSP, Center for Research and Security Prices indices. And Vanguard has the largest mutual fund in the world, VTI. They have like a huge number of other very popular cap-based, growth and value-based index funds. If you just add Vanguard to our number, you're getting a number of 38.5%. And that's not even including the COWZ and the CALFs of the world that would easily get you over 40%.
Ben Felix: Yeah. Super interesting. You gave the bucket of active managers who might be allocating passively will be included in this number. How do you know like a lot of that number isn't that?
Marco Sammon: This is one of the reasons it's important to get the weights so accurate. Because one way that Alex and I are going to check our estimates is that, at any given point in time, especially was something like the Russell where there's a huge reconstitution, there could be 10, 20 stocks that are going to get added to the 2,000. There could be 10, 20 stocks that are going to get dropped.
And so, at any given point in time, for each stock that is added and dropped, we will have an estimate of the AUM tracking that index. And so, then what we could say is, "Look, suppose all those numbers are really similar." The AUM tracking. The S&P 500, say, is 5 trillion. Everything we have is right around 5 trillion. I'm just making the number up. Then you would have to argue that the active managers were all buying in proportion to the float-adjusted market capitalization of each add and drop or selling in proportion.
And if you have 20 stocks being added and you're going to tell me that all – in perfect proportion. Well, if they're doing that, I'm going to call them indexing. You can call it whatever you want. But if you're buying everything in proportion to float-adjusted market capitalization, you're indexing. I'm sorry.
Ben Felix: That's pretty funny. How consistent is the estimate? For example, if you look at different stocks on the same trading day, are they going to imply a similar estimate of total passive ownership share?
Marco Sammon: Just going back to that estimate, we have a table – I think it's table five in the final version of the paper. And let's just pick one index. Let's pick the S&P 500. We have that S&P 500 trackers own at the end of the sample in 2021 17% of the ordinary common shares traded on major exchanges. Our estimated standard error on that is 80 basis funds. Our within-event variation, around that 17% is like plus or minus 80 basis points on average. And so, I would say – again, remember, we're dividing a huge number. Multiple trillions by a small number. And we're getting that kind of precision. It suggests that the trading among stuff added and dropped on the same day is pretty tightly clustered.
Ben Felix: Yeah. That's really cool. What do your finding suggest about the validity of the Grossman-Stiglitz framework for how investors choose between active and passive investing?
Marco Sammon: First of all, let's zoom out and say what is the Grossman-Stiglitz framework? The Grossman-Stiglitz is a super famous paper from 1980, I think, which made this point about why markets can never be perfectly informationally efficient. The idea is, if everyone is learning, there's always going to be some person who's like, "Well, prices are going to be really efficient. And learning takes effort."
I love to tell this story. When I was in college, I was the head of the investment club. And we had managed this little portfolio for the endowment. And one of the guys who was in the club with me worked for an unnamed active investment manager. And his job was to drive around rural Massachusetts and count up RVs in RV dealerships. Because the manager believed that RV sales like in real-time, you didn't have to wait for earnings, he could predict RV sales by looking at how many RVs were in RV dealerships. However, like this was costly. They had to pay him. They had to pay other people to collect this information. And so, if information gathering is costly, once a hundred people are driving around Massachusetts and gathering the RV, you don't want to also do that.
And so, there's three types of investors in Grossman-Stiglitz. There's the informed investors who drove around Massachusetts and the signal. There's uninformed investors who are still sensitive to prices, right? They don't want to buy when prices are high. They want to buy when prices are low. Very simple. And then there's noise traders. Basically, people who you could imagine. Suppose you work at an RV dealership, you get some employee comp in RV stock that's correlated with your labour income. And so, you sell sometimes when you get some RV stock.
And so, in Grossman-Stiglitz, basically, I think what your question is trying to get at is like the decision to be informed or uninformed depends on what you think everyone else is going to do. If you think everyone else is going to be informed, you maybe decide, "Okay, I'm going to free ride." If you think no one else is going to be informed, you're like, "Well, now's the time to learn a lot and really profit from my informational advantage."
And I think what Alex and I are trying to get at a little bit in our paper kind of speaking to this is like a super classical framework, is that, "Well, if measuring passive ownership is hard, the idea of Grossman-Stiglitz is what's called an REE," a perfectly rational expectations equilibrium. Everyone knows everything. But if not, everyone knows what the true passive ownership share is, how can everyone decide perfectly in equilibrium how much they have to learn relative to how much everyone else is learning? That's kind of how we think it speaks to that.
Cameron Passmore: It's really interesting. I'm curious, if passive owns so much of the market, why don't stock prices move more on the reconstitution days?
Marco Sammon: Well, maybe I'm telling too many stories here. But this is one I like to think about, right? On Thanksgiving, are turkeys expensive or cheap?
Ben Felix: Expensive.
Marco Sammon: Really? Many stores, as far as I've seen – I haven't got to a plethora of stores. But in fact, many turkeys are usually offered as a discount. And they're usually relatively cheaper on Thanksgiving. Now, why is that? Because the whole year, stores know Thanksgiving is coming and they can freeze turkeys. Because turkeys can be frozen and preserved. Frozen for a significant amount of time. They can freeze the turkeys. Bring out all the frozen turkeys on Thanksgiving. And they're not that expensive. Now, if I was to say Valentine's Day, are roses expensive or cheap? You say roses are expensive. Why? Well we can't freeze roses a whole year. That doesn't work.
And so, this kind of leads to some other work that I've done with Robin. And maybe we're going to talk about that later. But there's two possibilities. One is people are just really good now at freezing the turkeys and bringing them out when they're added to or drop from indices. Another possibility is that the price effect has just moved somewhere else in time. Maybe back in the 80s, when stock got added to the 500, prices went up on that day. Now, there's a desk at every investment bank and every hedge fund trying to predict index additions a year in advance. They're very slowly building up that stock of turkeys and pushing the price up the whole time. And so, by the time it actually gets added to the index, we don't see anything. But that's because we're looking in the wrong place.
Cameron Passmore: I might use that turkey example going forward. That's a good story.
Ben Felix: I did not know that the prices of turkeys were cheap on Thanksgiving. I guess I've never bought a turkey on –
Cameron Passmore: There's our takeaway, folks.
Ben Felix: I learned something today. Thanks.
Marco Sammon: When I first started grad school, this professor who had been there a really long time says, "You know, you guys are going to go to grad school. You're going to learn a lot. But don't use your graduate degree to assert knowledge about everything. You know one thing narrow." And so, I know a lot about indexing maybe. But I definitely don't know a lot about turkey supply and demand and pricing factors. And so, I'm sure someone will correct me that like, "Oh, at my Market Basket, turkey shot up 20% on Thanksgiving." So don't quote me on that.
Ben Felix: That's funny. Just continuing on the idea, and I think this is the paper with Robin that you mentioned, the idea of reconstitution and how it affects stock prices. Can you talk about how the price effects of index inclusions and deletions have changed over time?
Marco Sammon: Yeah. But, first, I want to give a little bit of a history lesson, I guess, and just go back in time to the 70s. Back in the 70s, economists and finance people thought a lot about how prices should respond to pure demand shocks. In the sense that like, "Well, my friend who gathered information on RVs and then buys a stock, he has information. And so, when he buys it, you could imagine that prices should adjust."
But an index fund buying a stock based on a mechanical rule like the Russell 1000 or 2000, it doesn't necessarily have information. And so, Sharpe has this famous paper from the 70s saying that you can buy, with no information, as much of a stock as you want and the price should never move.
Now, in the 80s, Andrei Shleifer has this really famous paper called the Do Demand Curves for Stocks Slope Down? And he showed that when stocks get added to the 500, prices go up. Saying a totally non-informational demand shock, possibly non-informational demand shock moves prices a lot. And then people refined that result. They show, "Oh. Well, it doesn't actually happen when it gets added to the index. It shows that it happens when we announce that it's getting added to the index." That was in the 90s we figured that out. And like, over, "Okay, do demand stocks really, really matter?"
Well, what Robin and I show that the effect of being added to many indices. And the S&P 500 is their main focus. But we also have in the later tables that Russell 1000, 2000 and various other indices have essentially gone to zero over time speaking about the return from the announcement to be very precise. The return from the day before the announcement to the day after the implementation, those effects have gone to zero.
Cameron Passmore: What do your findings suggest about why the index effect has declined?
Marco Sammon: This goes back to what we've kind of been hinting at a little bit before is one possibility is that – I read this article. I think it was in Bloomberg in the New York Times suggesting that one of these large investment Banks had a 20-person team that was making like $200 million a year front-running index editions. They're not waiting until the announcement to buy. They're buying ahead of time. And so, this suggests that the old way of measuring it by looking from the announcement to the implementation is just looking in the wrong place. That's one possibility.
Second possibility is that – and something we talk about the paper, is that S&P does not want – and don't quote me. But my understanding is that S&P doesn't want a lot of volatility around these events. They don't want it to be that, every time a stock gets out of their drop, prices go really, really wild. That also might make it hard for the people who are paying them for the right to license their index names to track indices if we have huge volatility around these events every time.
And so, what Robin and I find is that, over time, we've had more and more migrations. In the modern era, when a stock gets added to the 500, a lot of times now it comes from the 400. And, importantly, the $400, tracking dollars, has also been increasing over time. The way index funds work – and this is something that I think is kind of interesting. In a weird way, indexing is kind of a fixed-point problem. The weights are determined ex-post. We take the total float-adjusted cap, we take the AUM of your fund, and you buy a constant percentage of the float-adjusted cap of every constituent.
If 400 funds own about 7% of every constituent's float-adjusted cap, and 500 funds, which use the same float adjustment own about 7%, when you move from the 400 to the 500, those two demand shocks offset and the net demand shock is zero. And if the net demand shock is zero, why would prices move?
And, finally, you can imagine that over the last 30 years, if you look at some of the data from the 80s on just even simple stuff like bid-ask spreads, or you look at how much prices move five minutes after an order, markets were much less illiquid. I mean, we had pieces Of eight based on like the pirate code. We didn't have decimalization until – I think around 2002 was when we fully had decimalization. Markets have just become a lot more liquid and people understand these huge events are coming. Russell, I think has like a huge party every year. I've talked to people who work there who like they celebrate this like it's a holiday. These things have become focal. And markets have likely adapted to how focal these huge index events are.
Ben Felix: That's so crazy. The migration thing is super interesting. It's like more assets being in index funds reduces price effect because stuff's just migrating from one index fund to the other.
Marco Sammon: Exactly. You move from one to the other. Literally, the demand shots could perfectly offset. And, actually, we've seen sell-side reports. Many investment banks for their clients will publish lists of possible adds to the 500s. A merger happens, a slot opens up. And then they're like, "Oh, these are the most likely adds."
They also have a notional expected demand shock in these reports. And if you look at these reports, you'll see that the notional demand shock for the migrations is tiny. Market participants, going back to the Grossman-Stiglitz kind of rational expectations equilibrium, everyone knows what the migrations are. Everyone knows – if we go back to our other thing, they're both going to submit market on close. Market on close. Market on close buy. Market on close sell. Those things are going to offset and have very little effect on prices.
Ben Felix: Yeah. What's crazy what would you say the implications are of the findings in this paper for market efficiency?
Marco Sammon: Now, let me ask you a question. What is market efficiency?
Ben Felix: It's a great question. I don't have the answer.
Marco Sammon: Cameron, you want to take a try at it?
Cameron Passmore: I will let you take it, professor.
Marco Sammon: Well, I think – and maybe we can talk about this more. Because I've worked on this before. I actually think measuring price efficiency, whatever that means, is really hard. And one of the reasons it's really hard is that it kind of assumes that you have some way to know what a stock is worth at any given point in time.
And there's, I guess, two things I'd say kind of on this front. One is, suppose, I tell you go around, you figure out, you drive, you know how many RVs they sold. If they sold 10 more RVs than you thought, what is that translate into the value of the company? Is that a persistent shock to RV sales? Is it a temporary shock to RV sales?
I think even people whose job it is to predict how much stocks are worth, that's a really hard job. And it's really hard over the long run. And the second thing is the horizon. Suppose you believed even back in the 80s that like when a stock gets to an index, prices go up and then they go back down. Okay. For two weeks, prices were high. And then two weeks later, prices went back down. Okay. Maybe it was inefficient. Or prices moved away from fundamentals for two weeks. But is the company now suddenly not functioning? Are they not investing? Is it having real implications because prices were temporarily distorted? Is it any different than if, I don't know, Elon Musk needs to sell 100 million shares of Tesla and that pushes the price down a little bit while that demand gets absorbed? This is a really kind of tricky area, I guess.
Ben Felix: Hmm. That's like the joint hypothesis problem, right? Where any test of market efficiency is a test of your model and a test of market efficiency?
Marco Sammon: Right. That's a really, really good way of putting it. And I don't want to call people out. But I'll just give a really great example of this. There are two really famous papers. They're both published very well. I don't want to give too much information. One of them measures efficiency by regressing future fundamentals on current prices, essentially. And what that's saying is, "Well, if prices are a better predictor of future fundamentals, that suggests that the market is more efficient." That's kind of intuitive, right?
Another paper by two people I know, which has a different model of the data-generating process of the world, flips the right-hand side and left-hand side and says, "Okay, do fundamentals predict prices?" And they're saying, "Well, the logic is that, well, if fundamentals are the key driver or should be the key driver of prices, we want to figure out how much the variation in price is determined by fundamentals and not vice versa."
We've now taken the same variables and swap the right-hand and left-hand side. This is kind of like trying to measure supply and demand by regressing prices on quantities and then quantities on prices. In some model, it makes sense to do one. In some model, it makes sense to do the other. It can't make sense to do both, especially if you believe that you need shifts to the supply curve to estimate the demand curve. You need shifts to the demand curve to estimate the supply curve. It's actually a much bigger problem even. But, yes, you put it very well with the joint hypothesis problem.
Cameron Passmore: Wow. How has index fund ownership affected price informativeness?
Marco Sammon: Index fund ownership I think is good. Let's be precise. I have a paper on this that I'm looking narrowly at things that are defined as index funds. And we have to also be really careful about what I mean by price informativeness. And so, in that paper, I am zooming in on the incorporation of earnings information ahead of earnings announcements. Just to be clear, I'm defining the rules of the game over a very narrow time window before a very clear release of information.
Now, before we get into the answer, I think it's important to highlight some of the trade-offs like the economic mechanisms that actually might be in play. I think people think, "Look. Well, if we think back to our Grossman-Stiglitz style framework, and we have a bunch of people who are essentially saying, "I'm going to index my money. I'm going to put it in Vanguard," they're not driving the car around Massachusetts."
And so, as fewer and fewer people drive the car around Massachusetts, maybe ahead of earnings announcements, less and less of that information gets incorporated ahead of time. And, therefore, about that earnings information, prices maybe are a little bit less informative. That's one part of the story.
But there's other parts of the story. For example, suppose you're an active manager. Suppose you want to take a bet on Facebook. Now you have ETFs that make it easier to hedge, right? You can go long, Facebook. Short, some tech sector ETF. Rather than having to go out and short a bunch of individual stocks. It might help you isolate that debt. Or a lot of passive funds lend out their shares. They're allowed to lend out a certain fraction. And so, suppose some bad news is coming, well, if we've increased the lendable supply of shares or we've decreased the relative cost of shorting, maybe now bad news can get into prices more efficiently because we can short easier ahead of bad news. And so, if we just think the Grossman-Stiglitz, it's really simple. But I want to highlight that like there's a lot of different forces.
One other thing to think about is that there's different types of information. There's systematic information and there's idiosyncratic information. And so, if you think about the RV dealership, maybe we have some information about that one RV company. But it also could be if like consumer sentiment is low and people aren't buying RVs, that's systematic information. And maybe having an RV ETF lets you bet on the systematic information in the easier way. And so, we have to be really, really precise. I'm talking about firm-specific information, a height of earnings announcements.
And I find that it does decrease the earnings information in prices ahead of time. But, again, there's work that has shown everything else increasing systematic information, increasing long-run information. I'm really making a narrow claim about less earnings information being incorporated into prices ahead of time.
Ben Felix: When earnings are released, prices contained less information about those earnings at that time. Yeah, okay.
Marco Sammon: That's the very narrow point that I have zoomed in on in that paper.
Ben Felix: I remember we talked to Ralph Koijen a while ago on this podcast, and he talked about findings in one of his papers suggesting that the shift to passive has not affected price informative. Do you know why your findings are different?
Marco Sammon: Well, I think it goes back to the point you raised before, is that – first of all, I guess the thing that I like about looking at earnings announcements. And I'm working on a new project on this. And so, hopefully, you'll have me back or we'll talk about it another time, that suppose you have a one-cent earnings beat versus a two-cents earning beat. We don't know maybe how much more prices should move for the two-cent earning beat than one-cent earning beat. But we know that that's good news. Similarly, if you don't beat earnings, we know that it's bad news. And so, we could think about like a response function to earning surprises, like that's pretty clear what we should do with good news, buy. Bad news, sell.
And so, I actually don't even have a theoretical model in the paper. I'm really just looking at this idea of like, "Well, if the news was going to be good, prices should have went up ahead of time." My thing is very simple. Going back to the 60s, like Ball and Brown classic finance stuff. They have a model and they're also using one of these measures where they're regressing long-run future earnings on prices today. And we've said before that even really smart people can disagree what the right-hand side and left-hand side of that regression should be.
And so, when you're measuring – also, they're measuring it over a much longer horizon than I have. That original paper looked up to five years afterward. I'm looking at a span of 22 days. And so, it could go back to we said before, it could be that passive has a short effect. And it could be once that news comes out, because it's easier to short stocks, because it's easier to incorporate systematic information, for all the other reasons, once the news comes out, people could trade it into prices faster. And so, the effect is really not a long-run effect but a short-run effect.
Ben Felix: Man. I want to move on to a paper that – I enjoyed reading all your papers. But this one, man, is so cool. To start off with that one, can you talk about how does the index funds demand for securities differ from that of other investors?
Marco Sammon: Right. Index funds are really special, in the sense that like if you give a dollar to SPY, they don't get to go like, "Hmm. Apple looks cheap. But Facebook looks expensive. Let me buy a little bit of one and less of other." They have to allocate it based on a mechanical rule.
Now, of course, your decision to allocate to SPY could be because you think the securities in the 500 are undervalued or whatever. But conditional on you giving a dollar. And this is going to be important, because our whole thing in this paper is about flows. Conditional on giving a dollar, they have to allocate it in the mechanical way.
And going back to Ralph, and Xavier Gabaix's work, and Moto Yogo's work, they have this whole big thing about the inelastic markets’ hypothesis. And if you want to think about passive, they are completely inelastic. That market on close order that's completely insensitive to prices, that is completely inelastic demand. And that's one way that it's this rule-based inelastic demand is really special in response to flows, especially.
Cameron Passmore: Can you explain the concept of market clearing?
Marco Sammon: This is actually something funny. I have another project looking at newspaper articles going back to 1900 and trying to understand what caused large moves in the stock market going all the way back. And if you read an article from like 1925 or 1929 even is a great one, when the stock market's going down, the Wall Street Journal journalist – and this is, remember, 100 years ago. We're not ragging on them. This is 100 years ago. They might say something like, "Well, the market went up because there were more buyers than sellers." It makes sense, right? People are buying, prices go up. But, well, it may make sense. It can't be. For every share that someone bought, someone had to sell them that share. And so, the idea of market clearing is going to be, for every person who bought a share, someone had to sell it to them. For someone who sold a share, someone had to buy it from them. That's kind of like the fundamental premise of the exercise we do in this paper.
Ben Felix: Yeah. It's so interesting. Who's clearing the market when there's a demand shock from index investors?
Marco Sammon: Okay. We have to answer this. Again, maybe I'm being a little bit pedantic. But we have to be really careful about answering this. Because there's two different ways to think about clearing the market. One is going to be a level effect. On average, who's clearing the market? And one's going to be a slope effect, which is when passive investors demand a little bit more, who's clearing the demand for the marginal share?
We find on an equal-weighted average basis, on average, it's firms. And so, what do we mean by that? How does the firm clear the market? Our estimates, especially when index funds are net buyers, suggests that when index funds buy, 1% of a firm's shares outstanding, firms issue roughly 98 basis points of shares outstanding. When an index fund is a net buyer, firms are issuing almost one-for-one on an equal-weighted average basis.
Now, I know you're thinking, "Marco, wait. Something's wrong here. The market has to add up." We know over the last 10, 20 years, buybacks in dollar terms have been really big. We know over the last 10 to 20 years that index funds have been buying stocks. And so, it literally can't be that on a value-weighted average basis that index funds were net buyers and firms were net buyers. And that's true.
However, there are two nuances there. One is, we find that on a value-weighted average basis, at the margin, our estimates are pretty similar. Meaning when index funds demand more than average, firms initially buy back less. And then, eventually, supply shares. But for the level effect, it's different for value-weighted. There's been outflows out of active mutual funds. There's been outflows out of certain things like hedge funds and those things. On an average basis, yes, that's where they came from. But on the margin, we find that firms, even on a value-weighted average basis, firms are still the most responsive group.
Cameron Passmore: That means when index funds have a little bit more demand, firms are issuing shares and index funds are buying them. Firms, that's like Apple's issuing stock and the index fund is buying it.
Marco Sammon: We have to be a little careful. I guess I'm being a bit imprecise here. We have to be a little more precise. That is a measure of net issuance. That includes all types of issuance, including employee compensation, the exercise of warrants, the exercise of convertible debt, SEOs, changes in buybacks.
And so, I think that we call all that firms. But to be more precise, some of that stuff is decisions firms made in the way past that could be happening now. I have a friend who works at Google. She gets some stock options. She was granted those on a 4-year basis. And so, if index fund buying comes, index fund buying pushes the prices. She goes, "Oh, wow. This is amazing. I'm going to exercise my options. I'm going to have some Google shares now." Then Google doesn't get a choice. Like, "Oh, well, we don't give you the shares." They promised in the past that they were going to give the shares. And she gets the shares. See what I mean?
It's not necessarily that they're saying, "Oh, my God. We're printing these shares so that index funds can buy them." It might be – actually, when John and I decomposition into these different channels, we find that this like employee compensation part is the largest part. And that, again, it may not necessarily be that the firm at that point is making an active decision. Other stakeholders in the firm.
Actually, to give you a little more. You said, professor, that I can explain. I teach this corporate finance class. And we have this idea of the market value balance sheet. Rather than an accounting balance sheet, it's like the market value of the assets. The market value of the claims.
You could imagine that employees who have stock options, or stock grants, or PSUs, RSUs, MSUs, they're actually all on the right-hand side of a market value balance sheet. In the sense that they do have a claim on the cashflows of the firm. However, that claim only materializes when stock prices are high, when their options are in the money, when they get granted the stock. And so, they go from being on the market value balance sheet to the accounting balance sheet.
And, also, if you look at that kind of figure two in our paper, we have this really cool kind of – and I wish I could pull up a graphic right now. This really cool kink shape where we show that, basically, when index funds are net buyers, firms respond a lot. When index funds are sellers, they don't.
And if you think about, again, our mechanism of employee compensation, if index funds buy a lot and push up prices, employees can exercise their right to get shares. If index funds sell a lot and prices go down, employees can't force their employer to buy back shares. They have a one-sided compensation contract. And that may explain that super option-like feature we observe in figure two of the paper.
Cameron Passmore: So interesting.
Ben Felix: On causality though, how do we know the direction? In your example with your friend, she exercises her stock options. Index funds need to increase their holdings in response to issuance, right?
Marco Sammon: Just to explain a little bit more. There's two things here. Or at least two things here. One is a reverse causality. Sorry. Which I think is what you're getting at, right? The idea is if a firm does, say, an SEO, a seasoned equity offering, and they sell 20% of shares outstanding. I'm picking a big number. But let's pick a big number to make it easy.
Now, suppose, previously, index funds held 10% of that firm shares outstanding. Let's suppose, also, there's no float adjustments to make it easy. They're going to have to buy 10% of that issuance to maintain their 10% ownership of the float-adjusted cap by the firm. You're saying, "Well, Marco, what you're actually getting is a firm issues a lot of shares. Index funds need to, in response to issuance, buy shares." And so, you're going to get a reverse causality where it's actually the issuance that causes the buying. Not the other way around. That's one problem for [inaudible 00:39:23] identification wise.
Second problem for [inaudible 00:39:25] identification wise is going to be omitted variable story. The idea is, suppose, that there's some new technology shock. Suppose that, now, generative AI is getting commercialized. Everyone loves generative AI. Well, all the firms that are exposed to generative AI that can leverage that are going to say issue equity to buy GPUs. They're going to pay equity to their employees to get new people to work there. And at the same time, people like you and me who want to be in on it might buy ETFs that are exposed to it, say.
And then we'll have – at the same time, the firm will be issuing. ETFs are buying. But it's really because a common omitted variable. And so, we're going to try to attack the reverse causality and the omitted variable problem at the same time with an instrumental variable strategy.
What are we going to do? The first thing is we're going to focus on flows, like I highlighted before. Why flows? Because the flows part is not necessarily going to be subject to this specific reverse causality you talked about. And the second piece is we need to identify shocks to passive demand that come from flows that are not related to fundamentals.
And so, this AI revolution is like a fundamental shock. It's about how you were doing as a firm. Or how you're expected to do as a firm. And so, I don't know how much detail I should go into. But I'm going to try to distill down the ideas. The intuition of this instrument we have, which is going to be basically we're going to try to predict changes in passive demand that are uncorrelated to your own fundamentals. And that come from flows.
And so, the way we do that is we actually leverage something interesting that we notice, which is that there's a strong flow performance relationship in passive funds. First of all, at some level, this is weird. When you think about, "Oh, this active manager is doing really well. I'm going to give them money because I think they're skilled."
When the Russell 1000 does well relative to the 2000, it's not because IWB was better at stock picking. It was because those stocks happened to do better. It's not like you're giving them money. You're getting access to their skill. But that's true in the data. And, actually, there's a lot of heterogeneity across funds. And that's true in the data.
And then what we exploited is, "Look. Well, maybe my company is in the COWZ 100. Your company is in the COWZ 100." The COWZ 100 actually is pretty sensitive. Well, performance. And you and I are unrelated. We are going to look at basically the co-holdings that are not in the same industry, not in the same momentum, value, size, kind of like factor exposure buckets. And all these unrelated stocks do well. This leads to flows into your fund, which leads them to buy you passively.
And we're going to do an instrumentals variable where we're going to instrument or observe passive demand with this exogenous part of passive demand. And we're actually going to find almost identical results. We're going to find that, even in this IV setting, where we really narrowed the scope of the flows we're looking at, our point estimates are nearly the same.
Cameron Passmore: Wow. How important [inaudible 00:42:06] to the growth of assets managed by index funds?
Marco Sammon: Again, there, we need to be a bit careful about equal-weighted versus value-weighted. Something interesting we found out. Net issuance, meaning just changing shared outstanding. On an equal-weighted basis, 80% of firms in a given year over the past 20, 30 years have done net issuance. Equal-weighted average basis. On a dollar-weighted average basis, it's net buybacks.
And so, we could say that, "Look. John and I's estimates suggest that, overall, for every one percentage point of a firm bought by passive index funds and ETFs, firms on the [inaudible 00:42:39] marginal, firms issue, say, 65 basis points. And that comes down to about 35 basis points from employee comp, warrants. 25 basis points from SEOs. But that's kind of misleading because those are really lumpy. The average SEO is like 10, 15% of shares outstanding. And then the rest is coming from adjustments and buybacks. We have to be careful about that.
The second thing we have to be careful about is that, even though about 20% of firms have buyback programs, if we narrow in on those firms, those firms’ buyback programs are much more responsive to passive demand than firms that did not, say, in the past year or two years have a buyback program at all.
And so, to say how much actually came from issuance, it's really tough to exactly nail it down. And, also, there's a counterfactual. Well, if firms weren't issuing and passive demand was coming in, prices might have adjusted up a lot more, which would have changed incentives. And so, I don't think that John and I really have the exact counterfactual for all those reasons to answer that question. But those are at least the factors you might need to think about if you were trying to answer that question.
Ben Felix: You threw it a bunch of numbers just now. Can you say where are the shares provided by firms to meet passive demand actually coming from?
Marco Sammon: We do this decomposition exercise. And I'll just kind of break it down a little bit slower. That main figure one and figure two in the paper, we're looking at an overall measure of issuing that includes all changes in shares outstanding. Now, the ways that a lot of people think about is this comes from buybacks where the firm literally is going out and buying shares, adjusting the supply. We hand-collect data from Bloomberg on seasoned equity offerings.
What's actually really interesting if you think about this, Tesla sold shares directly to the market before being added to the 500. That's kind of a very clear example of the firm accommodating passive demand. Facebook did a traditional SEO I think just a day before or a few days before being added to the 500. We're going to hand-collect data on all those shares being sold by the firm, the market. That's a second source.
Then we have this residual category. Now, it turns out that, as far as I understand, you do not have to report to the SEC, at least in the 10Ks, how every possible change in shares came about. Now some people do. Nvidia, if you look at their 10K, they'll have like, "Okay, we have this many restricted stock units. This many preferred stock units. This many employee options. This many were exercised."
And so, John and I are in the process of scraping all that data. And for a subset of firms, we do have the option-based comp, share-based comp. We can kind of do an even finer decomposition. But for now, let's lump together all of that kind of forms of issuance that come through basically employee comp, PSUs, RSUs, MSUs options. And then warrants and convertible debt.
And so, if you think about like for example a company like Carnival. Carnival, during Covid, issue a lot of convertible debt. Their stock price recovered after Covid. And they ended up having to issue a significant amount of equity. However, not everyone does. And so, this is another tricky thing that John and I are working on, is that you might hedge that. You might have an investment bank on your behalf provide the shares to the people who you promised the shares to with the convertible debt. And then that doesn't lead to issuance. And so, you can't just like go in and use convertible debt right away. Because that might be hedged. For now, we have a very simple other group that captures all these different features.
And so, among our baseline total effect, a little more than half is going to come from this employee comp, warrants, convertible debt. About a third is going to come from SEOs. But, again, we have to be careful with that because that's very tail-heavy. SEOs are kind of rare. But when they happen, they're big. That's that. And that also might be more subject to the reverse causality. At least if we're running an OS regression, the reverse causality problem there is probably larger. And then the rest is buybacks.
Cameron Passmore: Does this imply that compensation of skilled labour income is tied to index funds buying up shares in companies?
Marco Sammon: Right. This is a really good question. And I think, one, when I first started thinking about this, and I've talked to people about the project, it's like, "Oh, if you're a VC investor and you invest in a company, when the company goes public, maybe that's your opportunity to exit." And maybe there's this other weird thing going on that like one employee's option to exit is that they effectively – of course, some intermediation has to happen. When my friend works at Google, gets her shares, she doesn't call up Vanguard and go like, "Okay, guys. Here are the shares. I'm sending them to you." Some intermediation might happen there.
But you could say that her skilled labour is being bought out by Vanguard or whoever else. I think that's one thing. But I think that we have to zoom out and think about more broadly how important is this relative to other factors. Nvidia is a great example. Nvidia's stock went up a ton. And I heard this story kind of casually that like – Nvidia, as far as I understand, you can exercise your stock options anytime within two years of them being granted.
And so, the price went up a ton. And a bunch of people who had worked there for a while had these options. They exercised them. And then house prices in San Jose went up a lot because these people who were suddenly much wealthier than they were before were able to afford much nicer houses in San Jose.
And so, when we go to your question of how does this link to the compensation of skilled labour income, we have to think about how variation in prices is due to like how much a variation in prices, and specifically positive or outperformance, is due to index fund buying versus, say, fundamental outperformance. Nvidia didn't go up as much as it did because it got bought by index funds. Nvidia had a product that a lot of people wanted and would be really, really useful.
And so, while to some degree, yes, maybe index funds might push up prices around a little bit. But I think that at the end of the day, fundamentals are likely much bigger driver of prices than index fund demand. At least casually. I don't have a quantitative estimate of that. But that's my intuition.
Ben Felix: Yeah. That's interesting answer. Can you talk about the conditions that affect the responsiveness of firms in accommodating the passive demand?
Marco Sammon: If we think about the most basic advice you could give my MBA students is like you want to buy low, you want to sell high. And so, you could imagine that firms – and Jeremy Stein had this 1989 paper in the QJE. And it kind of hints at this idea that firms ultimately are the arbitrageur in their own shares. Who knows better how much a firm is worth than the firm itself?
But firms obviously don't have unlimited flexibility in doing this. They can't go out and just buy whenever they want. They can't, for example, buy a certain part of the day. Buybacks can't happen at certain times. There's a lot of rules. But firms could be the ultimate arbitragers in their own shares.
John and I say, "Well, if we think about this simple advice, when would you want to sell shares in the face of passive bond?" Well, you would want to sell shares when your shares look relatively expensive. And so, in the paper, we look at kind of two very simple measures of expensiveness. Excess earnings yield and book to market. And we find that it looks like firms are much, much more likely to accommodate passive demand when their shares look expensive. When they look like a growthy firm. When their excess earnings yield is low.
Cameron Passmore: Who clears the market under conditions when firms are less likely to respond to prices?
Marco Sammon: This is something – I know you guys had Mike Green on the podcast before. And Mike and I actually talked about this. When I was working on this paper, early on in the paper, I sent it to Mike. And he noticed right away. I told you, we have this kind of hockey stick picture, which says that when index funds are buying, firms are issuing. But when index funds are selling, firms are not buying back. And so, that means that when index funds are selling, someone else has to clear the market.
Now, we in our data set try to do as very careful accounting exercise as we can for a bunch of different groups of investors. We look at active funds, passive funds, foreign funds, pension funds, insurance companies. We have all these different individual investor groups. However, because of market clearing, for every buyer, there has to be a seller. And there's some groups we can't see, like foreign institutional investors, retail investors. Institutions smaller than 100 million.
And we find that in the case of aggregate index fund selling, it's this other category made up of retail investors, foreign institutions and small institutions that tends to clear the market. Now, John and I's thinking about this is that, "Well, given that, on average, index funds have been buying a lot over the past 20, 30 years over our sample, maybe stuff that has really big index funds selling is stuff that's leaving the investable universe."
For example, maybe you do a tax inversion and you become a Canadian firm. Index funds sell. But then maybe many US institutions have mandates against holding those stocks. We're going to need someone else to clear the market. But when I showed this to Mike, he was like, "Oh, man. This is kind of scary. All the people who you think might be clearing the market index funds are selling or not. We need this other group." And that was something that I think is kind of something in the paper we don't push too hard, but we've been trying to amp up a little more, is we need a different group to clear the market when there's big index fund selling.
Ben Felix: Wait. Do we know who's clearing the market?
Marco Sammon: Do you know what dead reckoning is?
Ben Felix: No.
Marco Sammon: Dead reckoning, as far as I understand, is like a type of navigation system for missiles. There's like this joke that a missile knows where it is by knowing where it isn't. And it knows where it isn't by knowing where it is. It's basically like a relative positioning system. We know that we don't have market clearing coming from above the institutional groups, and mutual fund groups, and insiders, and short sellers that John and I can track down.
Ben Felix: But it's coming from somewhere.
Marco Sammon: Yeah. And that's why I said, most likely, we think it's small institutions, foreign institutions or retail investors. These are the groups that John and I cannot track. And so, mechanically, it has to be among these groups. But we can't say, "Oh, it's retail. Oh, it's foreign institutions." Because we literally, for many of these groups, cannot check their holdings.
Ben Felix: Okay. I mean, you brought up Mike Green. How do you think the market would respond to significant net selling by index funds?
Marco Sammon: Well, I guess what I would ask Mike is that how is that different than significant net selling by, say, active funds? There's this very famous paper in the Journal of Finance 2007 by Koval and Stafford that shows that like there was something that happened – it's called Fire Sales and Mutual Funds. Basically, because of something that happened, a bunch of active mutual funds had to sell a bunch of shares and prices went down.
And so, the first thing I would say is there's something special about index funds selling a lot of stocks versus, say, when CALFers reduce their allocation to equities from like 60% to 40%. Is there something different than that? I don't know. That's the first thing. And then the second thing I'd say is that, yeah, if you take our estimates seriously and you take our papers seriously, it would need to be that other groups step up.
Now something I didn't highlight is that everything we've been talking about so far is really focused on results due to flows. Now, in 2% of firmed quarters, we have stocks that are out of their drop from indices. If you're, for example, dropped from the S&P 1500 universe, have massive index fund selling. We find that in these cases, hedge funds and financial institutions actually do step up a lot. Active funds do step up a lot.
Then the second question is, "Okay, suppose we believe passive selling is special, what is the reason for the passive selling?" If it's an anticipated passive selling due to indexing rebalancing like we've talked about, these salient events that people prepare for for months, then I'm not that worried. We know in the data that these other groups.
I guess the thing that I don't quite have that you might imagine might have a relatively larger price impact is if we had a very large flow-driven shock to passive selling. Then it might be that that has a significant price impact. Because a lot of people who tend to trade the same way as passive might also be selling amplify that demand shock that might have a larger effect on markets.
Ben Felix: A lot of things would have to happen at the same time.
Marco Sammon: Yeah. It's flow-driven. It's a lot. It's large. Because, also, remember that hockey stick. For small passive selling – we have these really cool bin scatters in the paper in the appendix. The financial institutions group actually kind of has a little u-shape. And for small passive selling, financial institutions, which include head funds, do accommodate the market a little bit. But once passive starts selling a lot, they don't necessarily for flow-based passive selling. Again, it would have to be significant flow in the same direction. Financial institutions don't have the capital. It would have to be like a perfect storm of a lot of stuff. And then maybe that could be not great for security prices.
Ben Felix: Also, like you said, that's not necessarily unique to index funds, right?
Marco Sammon: Yeah. Exactly. If everyone decided – CALFers and everyone else decided to reduce their equity allocation by 20%, I think prices would probably go down.
Ben Felix: How active are index funds in the market for IPOs?
Marco Sammon: Okay. I have a new paper on this. I'm impressed that we're talking about this. I literally I think put this my website two weeks ago. And so, this is a paper with Chris Murray. Now, Chris Murray is a guy who has worked in the indexing world for many years. He brings a lot of real-world expertise. And it's been awesome working with him on this project. And I should also highlight, I should say that the last project was with John Shim. I don't know if I said his name. John and I have had a blast working on that project. I should really highlight. Again, John deserves a lot of credit. We worked on that together. I really want to say that. Okay.
Chris and I worked on this project. Chris, because he worked in this industry, he brought up to me something interesting, which is he says, "Look, there was this 2017 rule change." Remember, I said in 2014, Vanguard switched to using CRSP indexes. Now, CRSP starting in 2017 had this new rule where they said, "Look, any stock above the small-cap threshold, we're going to add to the index 4 days after the IPO in the closing auction." And, now, well, we talked about, what do index funds like to do? Index funds like to minimize tracking error.
Now, VTI is a very, very, very large fund. Those Vanguard cap-based funds are very large. If you're above the small-cap threshold, you're going to be added to many of these things, the growth and value. And so, we find that for fast-tracks, CRSP-tracking funds are buying around 7% of the float in those IPOs four days after the IPO.
Cameron Passmore: How do firms with fast-track index inclusion perform relative to their non-fast-track peers?
Marco Sammon: Here, we got to be careful. Because a lot of IPOs are a little underpriced. There's this well-known IPO pop. And so, that number I think in our sample is around 22%. That's already like a very large return on one day. We have to be careful. We find that even controlling for that, controlling for the relative strength of the IPO market, controlling for how small caps did and a whole bunch of other factors, we have a difference in difference-inspired design.
We find from the IPO itself, the IPO initial pricing, to when the CRSP tracking funds buy, they outperform by 15 percentage points. Not percent. Percentage points. If it was a $10 stock, it would go up by another $1.50. Now not everyone can access that price. You would have had to have gotten in the IPO to access that 15 percentage points. However, like we talked about, this is all perfectly predictable.
Based on T plus 0 close, we know what your eligibility is going to be. From the T plus 0 close to the T plus 4 close, it's five percentage points. It's still a substantial – even just based on totally public information outperformance for the fast-track stocks.
Ben Felix: How does it affect the IPO pricing and the capital that gets raised by firms?
Marco Sammon: If we go back to our Grossman-Stiglitz and our rational expectations equilibrium, suppose I'm book building an IPO and I think to myself, "Hmm. Well, I know that CRSP-tracking funds are going to buy 7% of this four days later." And now, also, suppose I'm an intermediary, I have access to getting to the IPO. I know that I can sell some number of shares in the closing auction on T plus 4 to these mechanical buyers. Well, then maybe I think, "Well, I can maybe price the IPO a little bit higher. I can sell a little bit more shares."
We actually find – and we think this is kind of a coincidence. And I'm happy to talk about this offline. Because I was just talking to Chris about this yesterday. Between the pricing higher in upsizing, they raise about 7% more capital, which is very coincidentally aligned with the 7% buying. We think that's kind of a coincidence for a few reasons. I don't want to get into that right now. But, basically, they can raise 7% more capital in the IPO.
Cameron Passmore: Wow. How do fast-track IPOs perform after index inclusion?
Marco Sammon: We said there's this big 15 percentage point outperformance in the IPO itself to the T plus 4. We actually find that, over the long run, their total performance looks similar to the stuff that didn't get fast-tracked. And so, one way to think about that is CRSP has some rules for what gets fast-tracked. What gets not? How big it is? What percentage of the shares are floated? What the domicile of the company is? There's a whole bunch of rules.
And it suggests that over the first 180 days, overall, the total performance is similar. It doesn't seem as though the fast-tracks are systematically better than the other ones. It just happens that they get a big demand shock. That creates some temporary price pressure. Now, remember, you IPO'd four days ago. If you try to buy 7% of any stock, you're going to move the price. If you try to buy 7% of a stock four days later even with world-class execution, you're going to move the price. There's a little price pressure that kind of reverts. But we're not saying it's anyone's fault. This is like you're following the rule of an index provider. And that creates price pressure.
Ben Felix: Firm IPOs meets the fast-track criteria. Price gets bid up. And then it comes back down. And then in the longer run, there's no effect.
Marco Sammon: It looks similar. It doesn't underperform in the long-run. If anything, it's still slightly higher. You could imagine that once you get added to VTI, unless you delist, VTI is not going to sell you. Maybe there's a little bit [inaudible 00:59:22] significant. Maybe you're a little bit better over the long run. But it's like almost no statistically significant difference by say 180 days after. Prices go up. Prices come down. But over the long run, the companies – if you talk about, back to my point, market efficiency. Any long run measure will suggest that it doesn't look like there's some long-run distortion in prices caused by this.
Ben Felix: Are index funds buying high though, in that case? How does this end up affecting index fund investors?
Marco Sammon: Right. There's a few things. One is, suppose, they are buying high. Well, if they put it in a market on close order, they will have no tracking error, right? Because the price that the stock gets added to the index is going to be the price in the closing auction on T plus 4. Even if they're buying high, they are doing what they say they are doing, which is minimizing tracking error.
The second thing about that is that you could argue, I guess, that one alternative would be, say, why don't we just let them add it in the IPO? Well, then you wouldn't be buying high. That's tricky. Because you can't add a stock that's not at the public price. But then you might say, "Well, why don't we just buy later and wait?" If over the 180 days or 60 days, performance looks similar, why don't we wait?
And I think here's the trick, and this is something Chris highlighted to me that's actually really important. There's two problems. Problem one is the bubble in the rug problem, I call. You ever had a rug in your house? You're trying to lay the rug perfectly flat. You have a little air bubble. And when you press on the air bubble in the rug, it just comes up somewhere else.
And so, if they have to buy 7% of a company 30 days later, that could easily create price impact 30 days later. Or going back to my project with Robin, in anticipation of buying 7% 30 days later, prices might slowly drift up ahead of time. Now, suppose they somehow, in a stealth way, were able to acquire the shares. Say, they actually got into the IPO but no one knew about it. Now here's the problem. T plus 4 gets added to the index. All the intermediaries who thought the CRSP trackers were going to buy, sell in T plus 4 close.
The CRSP trackers are not there because they've already bought their shares. Prices go down. The stock gets added to the index at a low price. You bought at 100. It gets pushed down by the price pressure to 90 bucks. Now you've created tracking error in exactly the wrong way.
First, to highlight. Even if they're buying high, it may be actually a stable equilibrium. In the sense that, by trying to deviate from that, you may actually make it worse for yourself by creating tracking error or creating unwanted price impact in one direction. I think another thing – and I don't want to push this too hard. But there's work by Antti Petajisto on like the hidden cost to index funds investors of index additions and deletions. And so, you could imagine that, right, that if you do believe that this is something intermediation cost, that you're essentially paying the people who get in the IPO and sell it to you as an index fund investor four days later, 15 percentage points for the right for them to do that intermediation. That cumulates to around four or five billion over our sample. There may be a hidden cost. But, again, given the equilibrium we're in, I think it's actually tricky to think about a way to avoid that. I haven't totally figured it out yet.
Ben Felix: Yeah. Interesting. It could just be one of the costs of being an index investor.
Marco Sammon: Yes. You're a price taker. You're submitting market on close orders. Exactly. It's just part of the rules.
Ben Felix: Okay. Based on your research on index funds, do you think index funds have had a net positive or negative impact on, I guess, two things? Financial markets and people investing in index funds?
Marco Sammon: Okay. This is a great question. I think, at some intuitive level, some of the very big benefits of index funds for people like me and you is that I get access to low-cost diversification. We teach there's no free lunch. There is a free lunch. Diversification is like a free lunch. That's amazing. And getting access to low-cost diversification across stocks, asset classes, that's amazing.
And, also, I actually think, there's an old literature showing that people, retail investors used to make a lot of mistakes. They would trade too much. They trade within the day. Now, if you buy VTI and hold it in your portfolio for a long time, I believe that you'll probably be better off than if you were like buying and selling the penny stocks that you were trading before, right? There's obviously a lot of positives.
And then you could say, "Well, Marco, your research shows that index effects push around prices. But maybe those effects aren't super persistent." Index funds create reduced price informativeness but like only for a certain amount of time. And so, I don't want to say that the passive investing has no effect on markets. But I think that we don't necessarily have yet enough evidence about the real effects.
I want to really understand, how does this affect firms' investment decisions, right? John and I say firms clear the market for passive demand. For every dollar less that they had to pay their employees because they use stock options, what do they do with that dollar? Do they invest it in NPV-positive projects?
Here's a scenario. Suppose you're a firm that has a really, really, really good project. You have no way to raise financing through traditional channels. You can't do an SEO. You can't raise debt. And then, all of a sudden, the fact that you have these mechanical buyers buying your shares allows you to relax your financing constraints and invest in NPV-positive projects. That would be great. That would be growing the economy. We're growing the pie. Everyone's happy.
On the other hand, if they have no NPV-positive projects and they don't want the cash just sitting there, the example was like back in the 80s when oil prices were high and the oil companies were buying private jets and art collections. And the bathrooms in the executive offices were outfitted with literal gold. That would be a bad use of the fundraising from – well, I don't know. Bad. I think that would be NPV destroying for everyone except the executive's decision there.
That's the super promising next step that I would love to answer and would love to see more evidence on would be what are the real effects of this? I think price is being a little bit less informative ahead of earnings announcements. For people trading, that's a huge deal. But for firms raising capital and making investment decisions, that's not super obvious.
But like I think this new thing with John and with Chris, if you're raising more capital in your IPO, that's amazing. But if these are all secondary shares that are going to early VC investors, that's very different than if it's like the firm isn't raising more money to invest in projects. This is the next thing that I'm really excited about. I haven't answered it yet. But I would love to work on this.
Cameron Passmore: You certainly do seem to be excited about this, Marco. And we have one final question to cap this amazing conversation. How do you define success in your life?
Marco Sammon: I'm going to give a bit of a long answer. And this is something that it's going to sound like a joke but it's actually not. Is that okay?
Cameron Passmore: Fire away. Go for it.
Marco Sammon: There's this like old psychology theory called the Kübler-Ross Model of Grief. The five stages of grief that has been doubted, disproven, whatever. But I actually think that this is a very good summary of how academia works. Remember, there's five stages. There's denial, anger, bargaining, depression, and acceptance. And I actually think this mirrors exactly the journey of any academic paper. And because this is the journey, you have to actually just be okay with going on this journey as opposed to defining success as the last step, which is going to be acceptance.
Let me walk you through. This is just my own thinking. This is totally Marco opinion. When you first start working on a project, people will say something to you like you shouldn't write this. You have some new idea. It's crazy. I'm telling you, firms clear the marker for passive demand. People are like, "Oh, no, no, no. You shouldn't write this. This is super speculative. This is never going to work out." And this is akin to denial. It's like just like the first step. It's denial. It's doubting the validity of the idea.
The second step is it can't be true. You write the paper on firms clearing the market for passive demand and people are like, "No, no, no. That's impossible. I know SEOs are rare. I know they've done net buybacks. There's no way." This is akin to anger, right? You're met with the anger of people and the harsh criticism and skepticism of your work.
Now the third part is I knew it already. After everyone tells you it's not true. Then it's like, "Oh, my God. It's obvious." Of course, passive comes. Passive pushes up prices. People have option like claims in the firm. I knew it already. And, in fact, you're not citing this work, this work. This is like bargaining. You're like, "Well, if you want this to get in, it's okay. But you got to cite this, this, this, this."
The next step in terms of I knew it already is actually even stronger. It's actually it was my idea. I'll make fun of Mike Green here. Mike is a friend of mine. Mike, if you're listening – I hope you're still listening at this point. When Mike was on this podcast a few months ago, Alex and I wrote this paper. And he says, "Well, Marco and Alex wrote this paper in response to a challenge that I issued them when I was discussing their paper at a conference."
In fact, not only did Mike know that passive ownership was bigger than we thought. It was his challenge that forced us to lead this. It was his idea to look into how big the passive ownership share. And this is depression, right? You feel really depressed because you worked on the project for two years and someone else is saying it was actually their idea.
And the final step is acceptance. I have a project that, for eight years, I worked on it. Submitted to journal. Rejection. Journal. Rejection. Journal. Rejection. Revise. Revise. Revise. Finally, you come to the point where it's literally accepted, which is akin to the last stage of the Kübler-Ross model, which is acceptance. And as you can see, this cycle is actually quite challenging. Because you have to go through people telling you you shouldn't write it. It can't be true. I knew it already. It was my idea.
And so, by the time it's accepted, the paper is actually not – it might even not be super relevant. A lot of time has passed. You may not care about it anymore. It's not your passion project. And so, if you define success as acceptance, at the end of that cycle, this is going to be a really, really hard job. And you have to enjoy all the other steps of the process.
And what I think success would be, and I'm not saying I do this or I've internalized this, I think what success would be would be being okay and accepting actually each step along the way of the process. Because that's what makes this job fun. I get to talk to y'all. I get to go talk to really smart people. I get to talk to Mike, and John Shim, my colleague, and everyone here. It's like that's what this is about. And that's what success is, I think, at least in this profession.
Cameron Passmore: Wow. What an answer. Marco, this has been amazing. Thanks so much for coming on. Learned a ton. Fantastic. Thank you.
Marco Sammon: Thank you. Yeah. This was amazing.
Ben Felix: Awesome. Thanks, Marco.
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.
Participate in our Community Discussion about this Episode:
Papers From Today’s Episode:
‘The Passive-Ownership Share Is Double What You Think It Is’ — https://www.hbs.edu/ris/Publication%20Files/double-what-you-think-it-is%20may%2023_3c1ae213-5aec-407d-b656-13e3822f0b8b.pdf
‘On the Impossibility of Informationally Efficient Markets’ — http://www.dklevine.com/archive/refs41908.pdf
‘Capital Asset Prices With and Without Negative Holding’ — https://ideas.repec.org/p/ris/nobelp/1990_003.html
Do Demand Curves for Stock Slope Down? — https://www.jstor.org/stable/2328486/
Links From Today’s Episode:
Meet with PWL Capital: https://calendly.com/d/3vm-t2j-h3p
Rational Reminder on iTunes — https://itunes.apple.com/ca/podcast/the-rational-reminder-podcast/id1426530582.
Rational Reminder Website — https://rationalreminder.ca/
Rational Reminder on Instagram — https://www.instagram.com/rationalreminder/
Rational Reminder on X — https://x.com/RationalRemind
Rational Reminder on TikTok — www.tiktok.com/@rationalreminder
Rational Reminder on YouTube — https://www.youtube.com/channel/
Rational Reminder Email — info@rationalreminder.ca
Benjamin Felix — https://www.pwlcapital.com/author/benjamin-felix/
Benjamin on X — https://x.com/benjaminwfelix
Benjamin on LinkedIn — https://www.linkedin.com/in/benjaminwfelix/
Cameron Passmore — https://www.pwlcapital.com/profile/cameron-passmore/
Cameron on X — https://x.com/CameronPassmore
Cameron on LinkedIn — https://www.linkedin.com/in/cameronpassmore/
Marco Sammon — https://marcosammon.com/
Marco Sammon on X — https://x.com/mcsammon19