In Flux: Is the Popularity of Passive Investing a Boon or a Bane?
Popular Strategies Lead to Positive Feedback Loops in the Stock Market. Is there a Way to Speculate Based on the ETFs' Tracking Error?
Hi, thanks for sticking with my blog.
As explained in my introductory post, this section of my blog contains my attempts at distilling thoughts about things I read within the context of other things I am aware of. I do so in order to make them clearer, foster learning, enhance recall and understanding, and adapt my behavior as an investor.
In short, this is a very selfish endeavor, which is why it is free. Also a word of caution, this is is probably open-ended endeavour. My understanding on this topic is evolving and far from complete, i.e. it is in flux.
Thank you again for being the audience, I will appreciate your feedback.
Prelude: What Triggered This Post
Complex Adaptive Systems & Passive Investing
I recently read chapter 3 of John Jennings’ “The Uncertainty Solution” and how the stock market is a complex adaptive system that is made up of agents. These agents (the investors) observe, adapt, and interact continuously. As a consequence, self-reinforcing (positive) or dampening (negative) feedback loops of behavior can emerge causing a market behavior that is in general unpredictable. That is consistent with several chapters of Michael Mauboussin’s “More Than You Know” that I read recently.
WHEN HUMANS INTERACT WITH EACH OTHER IN SOCIAL SYSTEMS, EXTREME OUTCOMES OCCUR THAT DEFY PREDICTION.
Source: The Uncertainty Solution, J. Jennings.
Mr. Jennings explains that sensible strategies at the agent level can lead to complex catastrophic behavior due to imitation, front-running, and taking it to the extreme (ok, the last part is not from Mr Jennings, but from Morgan Housel’s “Same as Ever”). For example, a contributing factor - not the sole cause - to the Black Monday market crash in 1987 was the then-popular portfolio insurance investment strategy.
[…] one widely acknowledged cause is the popularity of a derivative strategy known as portfolio insurance. The scheme involved using options to hedge a portfolio to enjoy gains when stocks went up and limit losses when they went down. It was dynamic trade, meaning that portfolio managers adjusted the hedges as the market gained or lost.
[…]
On an individual basis, it made sense to implement portfolio insurance. Who wouldn’t like to enjoy one’s gains while limiting one’s losses? However, on a system-wide basis, having tens of billions of dollars deployed using the same strategy was disastrous.
As market volatility increased in the days leading up to Black Monday, the portfolio insurance strategy led investment managers to sell holdings to raise money to increase their hedges. This selling generated losses, which caused the portfolio insurance algorithms to require the sale of even more assets to place more hedges. This feedback loop of losses generated still more selling, creating still more losses, leading to more selling, and so on. The next thing you knew, it was Black Monday.
Now, this triggered a recall of something I mentioned in my article on share-based compensation using options: the popularity and widespread adoption of passive investing strategies and their criticism by
.1 I would also like to reiterate from my article “Is SBC using Options Less Bad?” that investor David Einhorn stated that he adapted his strategy to seek undervalued stocks that can “pay” back the investor because of the shift from active stock selection towards price-insensitive passive investing. For reference, please check the interview of Michael W. Green by Adam Taggart about passive investing, and a “Ponzi market”. I quote from my article:When Does The Music Stop?
The interview sounds concerning. Eventually the “Ponzi market” will have to come to a halt and investment flows will reverse, e.g. due to inefficiently allocated capital, a bifurcated and unequal society, or simply when the demographics demand that retirees liquidate their holdings to fund their “Ruhestand”. If (when?) this occurs, the price-insensitive buyers would turn into insensitive sellers crashing the top index holdings stronger than components with a lower index weight. However, the music will probably keep going:
Probably there are strong incentives and motivations to keep the music going. Not just from the wealthy, but e.g. Investopedia indicates that pension funds do also invest in ETFs (among other asset classes). If the ETFs evaporize the pensions may not be backed by assets anymore and there are real-world consequences.
Worker immigration (or a baby boom) would contribute flows into retirement vehicles (I assume pre-dominantly passive vehicles) that could offset demographic pressures. Also, as asset prices are already inflated, retirees would not have to liquidate as many “shares” as years ago to get the same Dollar amount. So: less selling, new workers’ inflows would offset outflows.
As Michael Green mentioned, retirees are more heavily invested in active vehicles than younger people, i.e. their outflows have a lower multiplier effect, see footnote 1 below. (It sounds correct that younger people prefer passive vehicles more than older people, but I have no statistics to back that up).
As argued in the interview above, there are reasons why the music may keep on playing and the funds keep on flowing. Some are demographic (worker migration counteracting a baby boomer retirement wave), and some are political (e.g. pension funds failing due to passive ETFs must be avoided).
The following line of deductive reasoning is a good summary of my understanding before the inspiration for this article struck me.
Arguments
the stock market is a complex adaptive system that is unpredictable
sensible investment strategies are popular with investors
popular investing strategies see net inflows of investors’ funds, unpopular strategies see outflows
passive investing is a sensible and popular investment strategy
passive investing strategy market action is largely driven by net investor flows and the index it tracks
the net balance of flows drives asset prices in the market as passive strategies are price-insensitive actors. In = more buying than selling of assets = prices higher, and vice versa
if too widely adopted, popular strategies can lead to positive feedback loops in complex adaptive systems that can support periods of extreme behavior
any popular investment strategy that stops working (i.e. outperforming) becomes unpopular
the net fund flows can be impacted be exogenous events
Therefore
Passive investing can foster unpredictable market crashes, and severe corrections
Upon market correction or crashes, passive investing becomes unpopular and sees and outflows
Exogenous events can act similar to item 2 by impacting flows.
But Index Components Change! Thus ETFs Rebalance
It just so happens that I wrote about the Index Effect’s subtleness recently. As the mind is an association machine, it probably was still fresh on my mind when reading J. Jenning's chapter this morning.
In a nutshell, the constituents of a market index such as the S&P 500 change regularly based on a set of criteria defined by the company that provides the respective index. Some companies enter, and some leave the index. ETFs that track the altered index will be forced to implement the changes to stay true to their defined strategy.
Could ETF Rebalancing In Response to Tracking Errors Trigger Crashes?
Introducing the Question
Prior to reading said chapter I was already uneasy about passive investing and the quote (a part of the above) is at the heart of it:
On an individual basis, it made sense to implement portfolio insurance. Who wouldn’t like to enjoy one’s gains while limiting one’s losses? However, on a system-wide basis, having tens of billions of dollars deployed using the same strategy was disastrous.
However, the political incentives above may alleviate some of the concerns that somehow there is a way to keep the music playing and the funds flowing.
This morning my thought process was something like this:
I am generally uneasy about passive investing using index-tracking ETFs, but probably it's fine. But what if fund flows revert? Nah, it’s going to be fine.
Interesting… sensible agent-level strategies, if widely adopted, can lead to instability of the complex adaptive system called “the stock market”; the crashes that follow are not predictable as stock market behavior is not predictable.
Wait, … passive investing is popular, and it relies on tracking indices. Did I not write about index rebalancing? ETFs do have to rebalance, i.e. buy and sell according to the index reconstitution. But more generally, wouldn’t a daily fluctuation of a heavily weighed component such as NVidia or Meta have outsized impacts on the buys and sells of ETFs? Do ETFs have to track this daily? Couldn’t this force a bubble or a sell-off? Especially if the ETFs would impact each other's buying and selling due to the changed index weightings? So, a spiral or doom loop of heavy trading among ETFs as they adjust to the others’ updated weightings triggering new updates? Brain hurts.
My Perplexity on ETF Rebalancing
(pun intended) To appease my mind, I searched with Perplexity.AI and it appears that ETFs have different strategies with respect to how often they adjust their weightings to the index they track. The triggers mentioned are
When the underlying index’s constituents change (the Index Effect).
Rebalancing in regular intervals (daily, quarterly, semi-annual) - a strategic decision on how closely the passive ETF should be tracking the index
Corporate actions - e.g. a merger of two index components held by the ETF (per definition) may trigger a rebalancing (and an index reconstitution, I think)
Threshold-based, if the weighting of an asset held in the ETF deviates more than a specified amount from its weighting in the index
Side Note: I should note that I am no expert on passive ETF investing and their strategies and that the recent new article by
does not mention variations in how ETFs handle their tracking error. From that article it appears that ETFs directly track the index even if equities’ weightings within the index change (I guess that would imply daily). Maybe Perplexity.AI is mistaken? Or maybe it is a less important factor, but it is the focus of this article, so please bear with me …Mitigated By Timing Differences?
Beware - I am a computer scientist, not an economist, so maybe this line of reasoning is completely off.
If ETFs implement different strategies for the timing of their index tracking, different ETFs’ rebalancing should fall on different dates. As a consequence, ETF rebalancing will not have a very strong positive feedback loop2 (the co-occurrence of the ETFs’ rebalancing actions is low). I try to explain that in a simple example:
Let’s assume ETFs A and B track the same index but are not rebalancing within a particular short period (whatever that period length would be). Let’s also assume for the sake of argument that an equity in that index dropped significantly in relative weight. Then, when ETF A rebalances it will sell significant amounts of that equity and purchase others in that index to reduce its tracking error. ETF B, however, is not yet rebalancing and thus adhering to the previous asset index weighting. If ETF B receives investor funds inflow it will even buy that equity in accordance with the old weighting. That means that ETF B’s purchasing should dampen the selling pressure on the equity caused by ETF A. I think the timing difference might have a dampening effect (a negative feedback loop, i.e. good) so that there is no positive feedback loop of “selling-begets-selling” if ETF A and ETF B both sold the equity at the same time. Sooner or later, of course, ETF B will rebalance and sell the equity, too.
In the real world, also the popularity of different ETFs with investors will play a role as it skews the distribution of Assets Under Management in the ETF universe. That should lead to differences in the impact of different ETFs rebalancing strategies.
Where am I Now on Passive Investing?
Based on the above, I try to create my current picture of passive investing and its role in the markets:
Broadly speaking, the complex adaptive system “stock market” is impossible to predict as explained in John Jennings’ “The Uncertainty Solution” and in “More Than You Know” by Michael Mauboussin. Millions of investors interacting can cause positive feedback loops (i.e. reinforcing observed behavior) giving rise to bubbles & manias, collapses & crashes.
Interestingly, the new article mentions that Michael Mauboussin is starting to investigate passive investing, too. In general, it appears there is more research interest to look into the impacts of passive investing.
Nowadays, passive ETFs are significant factors in the price discovery of stock markets. However, they are price-insensitive, so they buy or sell assets irrespective of price as opposed to an active investor or an actively managed fund. Instead, they depend on the net fund flows from investors. The magnitude and direction of investor fund flows matter and create positive feedback loops, driving up or depressing prices. On top of that, index reconstitution and regular ETF rebalancing to reduce the tracking error impact buy and sell decisions, too.
I understand that currently ETF fund inflows outweigh outflows, e.g. 401K retirement plans allowing employees to programmatically save into ETFs. However, if (when?) large cohorts start withdrawing funds and fund flows reverse, e.g. baby boomers entering into retirement, the associated reversal of flows would turn ETFs into forced, price-insensitive sellers. These would depress equity prices.
The timing differences of different ETFs’ rebalancing strategies can have a dampening effect, potentially avoiding asset prices from an unfettered positive feedback loop (to the upside or the downside).
The differences in ETFs’ AUM reduce this dampening effect, I think: small funds cannot mitigate the impact of big funds.
In a broad sell-off or FOMO frenzy, the dampening effect of spread-out rebalancings across ETFs will only help so much. However, in a crash or a bubble, I would expect passive investors to alter their flows anyway, which then causes price-insensitive actions by the ETFs (up or down, depending on the flows).
I wonder if analyzing the tracking error of a sizable ETF before its next rebalancing will provide insights into when and how many equities will be bought and sold. Possibly, speculators could “play” the rebalancing schedule dates (above my pay grade).
So all in all, will the popularity of passive investing strategies facilitate big, broad, bad outcomes like other popular strategies did (e.g. “portfolio insurance”)? I am undecided but cautious about passive investing. It may or may not cause index and asset mayhem - or at least reinforce trends driven by fundamental investors or speculators. I will focus on what I can control, investing in businesses with healthy fundamentals at reasonable prices.
Note that not only Mr. Green is wary of passive investing flows messing with the stock market structure, but e.g. in the recent Value After Hours S06 E33 episode Seawolf Capital’s Porter Collins and Vincent Daniel endorse or at least recite his concerns. I also note that there is more academic research on the effect of fund flows, e.g. paper 1, paper 2, paper 3 (referenced by
in the above mentioned interview with ).As I am a computer scientist, this reminds me how networked systems (and APIs) use timing differences to avoid overload situations. For example, in case of a problem with a server, it is common for clients to back off and retry after a configured period plus a random additional delay. This way, each client will use a slightly different waiting time before reconnecting. Thus, the server (when functioning again) will face fewer simultaneous requests by clients attempting to reconnect and is less likely to be overloaded. A similar phenomenon can also be observed in power networks, e.g. at home. When a circuit breaker trips and cuts power to appliances it can be necessary to disconnect some of the devices before closing the circuit again. Otherwise, all appliances will start up the same instant and may cause an overload spike that trips the breaker again.