Do Short Term Flows Permanently Affect Share Prices?

To any of us that remembers an economics class in high school or college, we tended to think about upward sloping supply curves and downward sloping demand curves; and then shifts in those curves, or changes along them, affecting equilibrium prices. If new buyers enter a market, the demand curve shifts to the right, and the new market-clearing equilibrium price (where we have as many buyers as sellers) is higher.[1] If new sellers enter, the demand curve shifts to the left, and the price falls.

However, in 1958 Merton Miller and Franco Modigliani proposed that demand curves for stocks could not be affected by supply and demand for a firm’s equity. Sure, there could be short term price pressure, but it would dissipate, and the price would return to equilibrium. A company was worth what it was worth, no matter how many shares the company had outstanding, and whether or not it had one investor, or one million investors.

It kinda makes sense, right? In other words, the demand curve for stocks was flat.

This observation was one of the building blocks for the foundation of finance, and it (alongside other contributions to financial theory) won each of Miller and Modigliani a Nobel Prize.[2]

Yet in 1986, academic Andrei Shleifer asked if it was possible that demand curves for stocks maybe weren’t horizontal, and if they indeed might slope downward.

Before making his case, Shleifer summarized conventional wisdom at the time. Indeed not just conventional wisdom but the framework upon which most theories of asset pricing were built. This was his synopsis:

‍“Several important propositions in finance rely on the ability of investors to buy and sell any amount of the firm’s equity without significantly affecting the price.”

He continued:

“Most of the common elaborations of the efficient markets hypothesis predict horizontal (or “nearly horizontal”) demand curves for stocks, (where) the stock price is an unbiased predictor of underlying value, maintained through the workings of arbitrage. To the extent that stocks have close substitutes, that underlying value is not significantly dependent on supply...thus the (excess) demand curve for a security is (nearly) horizontal.”[3]

Then fast forward to 1997, Shleifer (along with colleague Robert Vishny) wrote another doosie. This one was about the likelihood of there potentially being a limit to these arbitrage opportunities, thus suggesting – among other things – that share prices could stay in this passive-flow affected disequilibrium for an inconveniently long time. Just a few years later, Wurgler and Zhuravskaya (2002) built on that and drew similar conclusions about the lack of arbitrage opportunities (for securities without close substitutes) specifically upon index inclusion.

So, even way back then, folks in finance had been thinking about this question of whether or not flows affected security pricing. And not just Shleifer. Harris and Gurel (1986) themselves built on earlier work from Scholes (1972). Many others, including Holthausen, Leftwich and Mayers (1987) and Loderer, Cooney, and van Drunen (1991) added to the early literature as well.

The issue very much intrigued me too, and in 1998, my friend Craig Dawson (now head of PIMCO Europe) and I wrote a paper in business school, jointly submitted to future Nobel Prize winners Eugene Fama and Richard Thaler asking similar questions about fund flows, the rise of passive investing, and whether or not they affected security pricing.[4]

I’d read Shleifer’s papers, and admittedly went into our own paper looking for some confirming evidence. I know, that was a severe violation of the golden rule, but at least I was aware of my behavioral misgivings. 😊

Anyway, I believed that fund flows, and specifically passive fund flows, must be impacting share prices for S&P 500 constituents somehow, especially if index funds were gaining more and more share of the total pot.

Craig was more objective, and perhaps more cynical, and as it turned out we didn’t find anything significant enough to be unexplainable. But folks smarter than us (well, smarter than me, not Craig) did.

Before going any further, I should point out that finding that index fund flows affect share prices isn’t necessarily a violation of the traditional tenets of asset pricing. Folks have proposed efficient market (or “risk”) stories about information asymmetries, market segmentation, and even liquidity itself; and suggested that flows might reasonably impact share prices without violating a requirement for a horizontal demand curve.

Bob Merton (1987) wondered if the price of a stock could increase with its investor base, and if addition to an index would increase its visibility, and subsequently its share price. Amihud and Mendelson (1986) asked if S&P 500 membership increased liquidity, then if that liquidity created a price premium for these stocks (or reduces a penalty for illiquidity). Denis, McConnell, Ovtchinnikov, and Yu (2003) suggested that addition to an index might convey positive information about the company itself. There were many other similar studies.

But none of them helped to explain the impact of flows on prices for securities that were already in the index. In other words, if a security was already in the index, and there was a continuing shift out of active strategies, and into passive ones, this new demand would not have conveyed any new information to folks, nor made them more liquid, or visible.

For Shleifer himself, the efficient market “information” stories were contrived. Mendenhall and Lynch (1997) made similar conclusions[5], and suggested that the effect of passive flows (on new additions) created price-effects that were, to some degree, permanent. Indeed, event studies by the likes of Greenwood (2005) and Kaul, Mehrotra, and Morck (2000) showed that the liquidity explanations were just not sufficient to explain the implied price elasticities of demand. Morck and Yang (2001) did a paper that was closer in spirit to what Craig and I did, and suggested evidence that the popularity of indexing was affecting share prices.

As you know, I had always suspected that these flows did impact prices, and could do so for a lot longer than Merton Miller ever would have conceded. But at the same time, I also believed – and still believe – that eventually share prices return to what they are actually worth. It’s the old Ben Graham voting machine and weighing machine thing. Sure, the market is a voting machine in the short run, but in the long run, we eventually find true, fair value.

And people have done some work on this too. Petajisto (2009) suggested that a stock’s addition to the S&P 500 index might affect things for a very short term. But there have also been papers (e.g. Chang, Hong, and Liskovich 2014) suggesting that there can disequilibrium for even up to a few months.

Importantly – from my perspective – however long this “mispricing” lingers, if these academic studies are correct, the effect must extend beyond index names. The academic focus on indices is just a historical convenience because we have a lot of data about index membership, weights, additions and deletions. If demand curves do slope down for index names, they must slope down for all equity securities.

So the question, or at least my question, is how long we have to wait for the voting machine to become a weighing machine.[6] How long do we wait for that price pressure to subside, and for prices to return to a reasonable equilibrium representing fair value? Do we have to wait minutes, or do we have to wait for months? Or is something else going on?

And with that embarrassingly-long introduction, I want to present a new working paper that is getting a lot of attention. Hopefully my preamble encourages normal human beings (aka those not propelled to spend days and nights poring over finance papers) to click on the link below, and understand how this might be a big deal.

It’s been written by some very smart guys (Xavier Gabaix and Ralph Koijen), with other similarly clever folks helping them along the way (including Andrei Shleifer). It extends well beyond the academic work by those mentioned above, and goes next-level. Essentially, the paper suggests that it is only flows that matter, and that the price changes manufactured by these flows are not just temporary – lasting from minutes to months – but are huge, and permanent.

It’s a long paper, and despite having a whole lot of math, it flows. Heck, it is even elegant. Here is a link.

You should draw your own conclusions, and try not to be affected by your own beliefs going in. That’s what I am trying to do.[7]

Believe me, as mentioned above, I really want to believe in downward sloping demand curves for stocks (at least the temporary ones). I’m an active stock-picker, and we all know the Sinclair quote about the likelihood of understanding something if our salary depends on not understanding it. I don’t want M&M to be right.

I’d like to think that prices can get out of whack for some period of time, and in that window, the nimble, unbiased, fundamental stock picker can take advantage of overreactions and underreactions. If they don't, if the M&M propositions truly hold, and if there are no behavioral reasons why prices might be wrong for a period of time, then I don’t have a meaningful job.

However, if this G&K paper is right, and it is only flows that matter (that is that they are exogenous and price-making) then not only are the M&M theorems upended (and many of our other foundations of finance) but the result is the same for me. I don’t have a meaningful job in a world where only flows matter either.

Along with my irrelevance, there are many other real-world implications of the paper that could be extended to conclude that:

• Outside of taxes, there actually is a big difference between dividends and buybacks.

• Share buybacks could be responsible for a significant proportion of a stock market rally.

• There actually are “sidelines”, that theoretical place from which cash can come into the market, but the exact same amount of cash doesn’t get moved off by the sellers.

• Arbitrage isn’t needed to correct “mispricing” because there is no such thing as mispricing.

• Bubbles can be permanent.

• The pricing of stocks can be permanently divorced from the fundamentals. In other words, Gamestop and AMC aren’t necessarily “overpriced.”

Now, before getting too worked up, I believe that there may something that the authors need to address.

In all of its elegance, I think that there is a whole lot of math going on that mostly determines correlation between flows and price in the short term. Yes, it could be that these phenomena indicate a steeper (more inelastic) demand curve in the short-term than we all thought, but there is a major omitted variable here.

That variable is the underlying cause, or reason for these flows.

In my mind, it could be changes in expectations of some thing or things that drive the flows. It could be macroeconomic things like interest rates and inflation, or changes in preferences for certain factors (like value or growth or quality or junk) it even could be that new firm-specific fundamental information is digested.

These are variables that would simultaneously drive both flows and price. So, yes, they are correlated; but of course they are. Flows are not “driving” the price, they happen as the stock moves.[8]

There is something in finance called “the overnight return puzzle”. It’s fascinating until you think about it. I’ve attached a link to a post we wrote about the phenomenon.[9]

Basically, all of the increase in equity markets over the years happened overnight. None of it happened from open to close (in fact, the market gave back overnight gains during trading hours). There are all sorts of – in my mind – weak explanations for it (specialist activity, algorithmic trading, short-sale constraints, investor heterogeneity).

To me, the best explanation is that “overnight” is when the news happens.

Earnings reports happen after hours (or pre-market). So do changes in guidance. So do merger announcements. Because of this new information, the bid-offer spread can move dramatically from the close of one session to the open the next morning.

And it doesn’t require a single share to trade.

So, if "overnight" is when all the return happens, then flows don’t matter.

Sure, those bid-offer spreads and indeed the first tick will reflect where the market thinks equilibrium will be, but is that “driven” by the flows, or driven by the information that drives the flows? I think the latter.

So yes, flows may matter in the short, and maybe even the intermediate term. As Graham said, the market is a voting machine. But it eventually weighs.[10]

Accordingly, flows just aren't the primary motivator of long-term stock returns. They may accompany them on the journey, but they aren't sitting in the driver's seat.

 

‍

‍