The paradoxical efficient market hypothesis (2024)(3quarksdaily.com) |
The paradoxical efficient market hypothesis (2024)(3quarksdaily.com) |
To quote Richard Bookstaber, "The principal reason for intraday price movement is the demand for liquidity... the role of the market is to provide immediacy for liquidity demanders. ...market crises... are the times when liquidity and immediacy matter most. ...the defining characteristic is that time is more important than price. ...diversification strategies fail. Assets that are uncorrelated suddenly become highly correlated, and all positions go down together. The reason for the lack of diversification is that in a high-energy market, all assets in fact are the same.... What matters is who holds the assets." (from A Framework for Understanding Market Crises, 1999)
Was the market drop an accurate reflection of the value that would have been destroyed by those tariffs, discounted by the probability that they would have been enacted as drafted? Nobody knew then, and I maintain that nobody even knows now. That was not the calculation that was being made.
The price of something and the value of something were never expected to be the same. What's the value of food? If you have none you die, so the value is quite high, but the price is much lower than that because there are many competing suppliers.
And the price of a large class like investment securities can easily change all at once if there is a large shift in supply or demand.
I would pay anything for air if I needed it, but I will gladly sell air in my yard for $1/m^3 because that air is worthless to me.
Is air priceless or worthless?
That is why price != value as most people think of it.
This is a (un)surprisingly deep rabbit hole.
Human Society and the Global Economy by Kit Sims Taylor, Chapter 6: Theories of Value
https://www.d.umn.edu/cla/faculty/jhamlin/4111/2111-home/val...
While I agree with you (quite firmly: it’s a great starting point to put on the table to challenge orthodoxy in this space), and think you’re agreeing with the parent comment, it is a fundamental tenet of mainstream economics and the political arguments of neoliberal (aka current mainstream) policy that [price == (market averaged) value], or at the very least [price ~= value].
Another interesting line of argument is to explore things that are valuable that don’t typically get a price: for example household labour, or love and friendship (at least directly: I’m sure a Friedman acolyte would reduce all relationships to exchange and reframe gifts and acts of love as investments).
As an aside for the parent comment: thanks for sharing this, it’s one of the top category of comments/quotes I’ve seen on HN in being useful, insightful, and challenging of conventional understanding in a way that improves understanding and future prediction.
The main question, I suppose, is why correlations were so high after the tariff announcements:
- In some cases, the high correlations are probably due to the markets being directly affected by the announcements: both commodities and equity are affected, and they got more correlated, which makes sense.
- In some cases, the high correlations are probably due to liquidity demand rather than markets being directly affected by the announcements: we would not expect cryptocurrencies to be directly affected by US tariffs, but they ended up correlated with equity markets anyway. That's probably because people needed to sell off their cryptocurrency to cover equity losses.
Thus in this case, it's again probably a bit of both.
Great paper. Thanks for referencing.
Trump had been threatening tariffs for the campaign and mentioning them before. There wasn't that much new information that should have caused the plummet.
Also I will point out that it's more like the avoidance of information that caused some of it Nvidia's stock plunged on an announcement that went something like:
Sentence 1: we are putting tarrifs on Taiwan Sentence 2: except semiconductor related goods
It as if the market participants read sentence 1 and very few of us read sentence 2.
The EMH would assert that a casual observer like me wouldn't see the price gap between the time it took for people to read sentence 2. But it took several business days...
https://www.risknet.de/uploads/tx_bxelibrary/Bookstaber-Unde...
This is an oversimplification of how professional investing works.
The vast majority of the dollar value of markets isn’t governed by immediate profit seeking behaviour - it’s people trying to manage money in the context of a real business. Pension fund money is the largest “pot” in the markets at any one time.
Pensions funds aren’t incentivised to maximise returns in any particular quarter/year. Instead, they’re looking to manage risk and ensure consistent returns in the very long term.
Therefore, the “value” they place on various assets is different to what a trend fund or retail investor is thinking about. The price at which they would buy/sell is different.
The market value might “reflect” that information but it could easily create a situation in which short-term, strictly returns-motivated investors might value an asset much more than pension funds or vice versa. That creates opportunity for both to do a non-zero-sum trade and both “make money”.
I’ve seen it elsewhere in this thread but it’s simply not the case that the “markets are a casino”. The vast dollar value of the market is about sharing risk and providing liquidity.
The global bond market are at least 1.5 times the size of the equities market(s).
yes some markets are basically a casino but they’re tiny in comparison.
Let's say almost everyone believed in the Efficient Market Hypothesis (EMH). Then, trading would decrease significantly, since most people would think that stocks are already fairly priced. That means the few people who trade would move the market significantly, based on whatever idiosyncratic value-theories they had.
But then the EMH believers would see wild moves in the market and stop believing in EMH. They would start trading more to gain profits.
And as more traders participated, the market would behave more and more like the EMH were true. Eventually, the market would stabilize. Prices wouldn't swing so much. This would increase the number of EMH believers.
It would be interesting to survey belief in EMH among traders. If my model is correct, the percentage of EMH believers should be roughly constant, or at least oscillate around some optimum value.
So here, Phase 1 would be a market dominated by EMH believers who passively invest. In phase 2, speculative “noisy” traders start to exploit this landscape to profit. In phase 3 there’s a crisis or period of high volatility. The old complacent EMH strategies suffer losses and become extinct. Then no doubt in phase 4 the market moves to some new equilibrium with new strategies dominant!
So in this AMH theory what you describe is a natural process of evolution.
EMH is about the tendency of the market to be efficient over time. It is purely of academic interest to dream up hypothetical scenarios where everyone is equally rational and informed, etc. There are degrees of efficiency and information, and it's useful to talk about this to try to understand how real markets work and can be made to work better.
Like the classic economist joke where they ignore a $100 bill on the ground: "It can't be real. If it were, somebody else would have already picked it up."
But we can all gamble on what it is worth!
So stockholders are like roulette pill holders. Everyone just bets on where the pill will fall. Few are luckier than others. Some smarter know whether the roullete is rigged and have better chances.
Either way, the owners of a successful company are going to want to profit from it, which will make the shares valuable. Of course, investors know this, and so the share price tends to track current value of expected future earnings even without the company taking direct action to distribute profits.
Some subset of shareholders. For example: Meta Inc. and their Class A vs B shares, GOOGL vs GOOG, etc.
> A Keynesian beauty contest is a metaphorical beauty contest in which judges are rewarded for selecting the most popular faces among all judges, rather than those they may personally find the most attractive.
This explains why informed investors know TSLA is worthless, but they also know that the retail market as a whole thinks it's as precious as unicorn tears, so it is priced accordingly.
For example people who kiss their dice will likely put money on red.
the hypothesis maintains that
stock prices reflect all relevant
information about the stock
What if it takes 12 months of hard thinking to draw the right conclusion from the information? Are there many investors who go to such lengths? Are they all thinking at the same speed? And if not, what does that tell us about the EMH?
Google released DeepDream in 2015. My feeling is that with enough thinking, one could have predicted where image generation is going in the next decade and that language generation would go a similar route. And that this will lead to a high demand in Nvidia's GPUs. But that thinking would not be instantly. It would take months or years.
You can think of it like a form of compression: it condenses an incredibly complex, chaotic system into something we can reason about. That simplification makes it powerful and insightful, but it also means that a lot of nuance and unpredictability are lost in the process. In contrast, a physical law can be calculated precisely and consistently, while market behavior is always shaped by human psychology, uncertainty, and imperfect information.
It is a form of informational efficiency, but it does not necessarily follow that prices are even statistically correct. The market can be irrational for longer than you can remain solvent.
The conclusion is that with a sufficiently large number of actors in the market all seeking profits by trying to find misevaluation of stock prices, the excess profits of any individual actor will (assuming they all have access to the same information) converge to zero.
Its less a paradox and more a matter of game theory. Every investment firm which gives up trying to look for alpha (believing it is fruitless) means the remaining firms have more opportunities to find stocks with available information not reflected in the price. There's no paradox here: each individual actor is incentivized to participate in order to not miss out on that potential for excess profits, and the net effect is the EMH.
(And looking at how traders work, it's all about finding a strategy that no-one else has found and executing on it. Once two competitors with similar resources know the strategy, it ceases to be particularly profitable, which to me seems to be pretty in line with the EMH)
This is the paradox.
EMH is unfalsifiable at best and tautological at worst.
Information publicly available doesn’t mean anyone can process it all. Every actor is operating off a different subset of information.
Lots of intentionally low information investors (inhabitants of indexed funds) demand stock or supply stock, pushing prices in directions unrelated to value changes, due to index list changes and rebalancing events.
Investors, of all magnitudes of wealth, have unending personal or private idiosyncratic reasons for the timing of many investments or sales, besides individual asset return optimization.
The value of a stock rises and falls as its absolute expected return rises and falls relative to the changing returns of the rest of the entire market of investment vehicles. Everything impacts everything.
All these shifts happen over varying time frames.
Almost all relevant market facts are time varying, often with turbulence and ambiguity.
The fast moving investors most influential in setting prices, must model the whole market’s 2nd order and even 3rd order reactions (by similar actors) due to feedback effects and dynamics.
Sudden market wide changes trigger waves of low analysis buying and selling. Compounded by the higher order risk this creates to leverage, annuity responsibikities, hedging, and many other amplifiers of behavior.
The efficient market hypothesis is an interesting and enlightening thought experiment. A reduced dimension toy/sim market.
Not a credible model.
Not even if every single participant was frantically and relentlessly re-valuing and re-balancing at the margins to a firehose of comprehensive market information.
It’s the same stupid people who say the market is a random walk. Oh yeah, if it was a random walk, then why do earnings reports even matter? Companies could just lose and gain whatever they want, and stocks would just fluctuate randomly.
Here’s the truth. It’s called Rice’s theory of opportunity. It says that there is a golden window on the order of a few weeks to a few months where the signal-to-noise ratio has the least attenuation. This is because it avoids the initial transitory periods, the real-time gap between the knowledge existing and the knowledge spreading to people with enough resources to make a difference.
I feel like the stock market is pretty divorced from fundamentals at this point i.e. speculation makes it more like a Keynesian beauty contest (picking stocks you think other people will think are valuable).
> The Grossman-Stiglitz Paradox is a paradox introduced by Sanford J. Grossman and Joseph Stiglitz in a joint publication in American Economic Review in 1980[1] that argues perfectly informationally efficient markets are an impossibility since, if prices perfectly reflected available information, there is no profit to gathering information, in which case there would be little reason to trade and markets would eventually collapse.[2]
* https://en.wikipedia.org/wiki/Grossman-Stiglitz_paradox
So the more efficient markets are, the hard it will be to find "alpha" (returns), and so more people will stop trying. But as more people stop trying, markets will become more inefficient, in which case people can find alpha again, which encourages more participants.
It's always worth remembering trade involves use values as well. We don't only trade for asymmetric profit, and there are things like hedging which include a yield where both can acknowledge future risk, and price accordingly.
I'm probably ignorant of some magic economist reason why the words are fluid and don't mean what I think they mean: this always seems to be the case talking economics from the stuffed armchair.
Another take on this is that we can agree to facts and disagree to consequences. Same information, different conclusions.
Farmer et al. 2013. The Inefficient Market Hypothesis
https://www.amse-aixmarseille.fr/sites/default/files/_dt/201...
The current insider trading rules only prohibit actions, and does not prevent inaction.
As an example, you could imagine that an insider were going to sell their portfolio of company issued shares, but because of insider info they have about a current project that would give rise to a price hike, they may choose to sell _later_ (or not to sell at all). This means the liquidity of the market is now less, and thus, raises the price vs the counterfactual world where said insider _did_ sell. All without revealing any information about the actual insider project.
Slightly more seriously his assertion:
>Alternatively stated, the Efficient Market Hypothesis is true if [...] a sufficiently large majority of investors believes it to be false.
is flawed. They could believe it false but still make a mess of the valuations. Which can cause real world problems if profesional investors put your pension money into webvan or other bubble stocks and then there is a cash shortage after the dot com bubble burst. Of course they are wiser now and won't make such errors with AI.
Market prices are derived from supply and demand. A heavy determinant of demand is income equality. Another is interest rates. These are nothing to do with, in general, a particular stock.
It’s so obviously false to anyone trading or even watching stocks that serious discussion by academics just adds weight to the accusation that they don’t know what they’re talking about. We need a new, more serious, science of economics.
How do you think it’s useful? Can it be used to make predictions about the future, for example?
This, very well summarized.
I would nuance (but not disagree with) your comments on pension funds though. The thing is PF do not invest themselves, they usually are, or delegate to, funds of funds, which in turn decide on allocation based on the desired risk profile. It could very well happen that the total allocation is the sum of a multitude of individually short term investments, as long as these are diversified enough. I would concede that in practice that is not really feasible though.
These risks profiles are numerous, diverse, and ultimately idiosyncratic. People often forget or don't know about all these risk constraints, because they work in a fund that is bound to a specific risk mandate.
For instance, depending on how your investment vehicule is structured (the regulatory enveloppe through which you sell your fund, which ultimately determines to who you can sell, how you can advertise, how profits are taxed, etc), you can have liquidity constraints (e.g. clients should be able to redempt daily, weekly, ...) risk parity constraints (e.g. per asset class vol budgets, to be respected daily, weekly, etc), exposure budgets (e.g. country, sector, beta, ...), counterparty risk (e.g. minimum number of managers to allocate to, or clearing houses, or custodians), idiosyncratic risks (e.g. an insurance company will need to be neutral against natural disasters, healthcare exposure, etc), ESG, etc
Say you want to know the 400 trillionth digit of pi. We have all the information needed right now to know how to compute it. But you don't know what the actual digit is yet. The information isn't available and won't be until you set your supercomputer on it for some number of months. Having the information necessary to derive other information isn't the same as having the derived information.
If there is some information about a future stock price that could theoretically be computed after months of work, that's still not information that currently exists, and therefore is not currently reflected in the price. If no investors go to the lengths to get that information, it'll continue to not affect the stock price. It's not violating EMH because it's not information that exists yet.
That seems to be directly the opposite of the common definition of the EMH, which emphasizes how the market reacts to new information. And not how it produces information. For example in TFA:
"the market rapidly responds to new information"
Wikipedia starts the "Theoretical background" with an example on how information becomes widely available to all investors, not how one fast smart thinker generates it:
Suppose that a piece of information about the value
of a stock (say, about a future merger) is widely
available to investors.
It's not required to be all of them. Suppose that it indeed isn't, but the ones who do that work for investment funds who control significant pools of money.
Now the investors in two or three of those places do the research and conclude that some company is about to start doing well and their share price is currently $50 but is about to be $150. So they start buying it, and keep buying it until it gets up near $150. Which happens pretty quickly because they control enough money to use up all of the short-term liquidity at the lower prices and the majority of the shares are held by people who aren't even paying attention and therefore don't try to sell when the price starts going up. Once the price gets to that point they don't buy any more because it's no longer selling at a discount.
Then the company actually starts doing well to the point that everyone can see it but the price hardly moves because it was already priced in.
That would mean that the p/e-ratio of a company would rise sharply long before the profits set in. And that rise would be called "mysterious" by the general public. And then only when the profits set in, the p/e would come down.
I can't see that in Nvidia for example:
https://www.macrotrends.net/stocks/charts/NVDA/nvidia/pe-rat...
The price roughly rose along the earnings. Even though the foundations for generative AI became clear in 2015.
I think the idea behind EMH is that this probability is priced in, at any point in time. It just so happens that longer term probabilities are discounted as more volatile, thus impacting less the present price.
Suppose we have a stock and a bunch of investors. All of the investors have some set of information implying a value v. Except, one investor is smarter and finds an edge (rise of AI or whatever) which implies a value of 2v.
That investor will buy the stock any price up to 2v. The rest of the investors will be happy to sell at any price above v. Given unlimited money the price should stabilise at 2v very quickly.
However there are lots of real-world caveats like, not everyone has an infinite money glitch.. and there are probably second-order and third-order effects like some hedge fund notices the pattern and does XYZ which influences price... options make price a function of expectation of price, then the price of options is driven by the expected price of the same options near execution date... idk
And the immediacy comes from the large amount and speed of the transactions. It does not require that these participants sus out the correct value from information - they could've actually just guessed.
To paraphrase William Gibson: the information may be available, but it is not evenly distributed.
It's why (e.g.) hedge funds use satellites to get information on company activities:
* https://newsroom.haas.berkeley.edu/how-hedge-funds-use-satel...
* https://internationalbanker.com/brokerage/how-satellite-imag...
It's takes resources (time, money, etc) to gain an advantage, and it's only do it because they think some extra bits of information will allow them to know more than The Market in general / their counterparties to get a better conditions on a trade or options.
Why do you think insider trading became illegal: some folks have that information before others simply because of their job/position. There was a case of someone knowing something early, because information can only travel as fast of the speed of light, which some "beat":
> Last Wednesday, the Federal Reserve announced it would not be tapering its bond buying program at 2 p.m. ET. The news takes seven milliseconds — about the speed of light — to reach Chicago. But before the seven milliseconds was up, a few huge orders based on the Fed's decision were placed on Chicago exchanges.
* https://www.npr.org/sections/alltechconsidered/2013/09/24/22...
* https://www.motherjones.com/kevin-drum/2013/11/final-frontie...
EMH is saying people that if people think they can make money, they will spend the resources to get an information edge to accurate price what a commodity is 'worth', either higher or lower. If you better know what it 'should' be, then you can devise a trading strategy (buy/sell/short/long) to get one over your counterparty.
Momentum investing is a thing:
* https://www.investopedia.com/terms/m/momentum.asp
* https://en.wikipedia.org/wiki/Momentum_investing
A number of people make / made money when The Market became "divorced from fundamentals": see The Big Short.
* https://en.wikipedia.org/wiki/The_Big_Short_(film)
Just remember: "The market remain irrational longer than you can remain solvent." — Keynes, https://www.goodreads.com/quotes/603621
It's instructive to compare "Crowdfunding" which took off with Kickstarter ~15 years ago, with "Equity Crowdfunding", which gets tried again and again, and has not a single success story to its name.
Kickstarter was made to fund artistic ventures, and for the first years, they were strict about only allowing that on their site. The idea was to reduce risk for e.g. people trying to bring their favorite band to the area for a concert.
On old Kickstarter, you only pledged to a project if YOU want the product/outcome for its own sake.
However, in "equity crowdfunding", where backers are tempted with a share in the profits of a venture, you should, if you are smart, try to ignore what YOU want. Your own wants are a source of error here: as a fan of the band, you're likely to overestimate its appeal. You should play the Keynesian beauty contest and try to guess what others want.
Kickstarter understood the difference very well. In the early years, they banned such things as "reseller's tiers". Some people would support e.g. a boardgame with pledging for five copies of the game, betting on its success and hoping to resell four of them. That brings the KBC factor in again, and Kickstarter thought that it would eventually lead to the site being flooded with the things everyone thought everyone else wanted, rather than the things they actually wanted.
There's a whole scam industry dedicated to exploiting the gap between what you want and what for its own sake and what you want because you think others want it: MLMs. MLM victims get tricked into a loop where they on one hand convince themselves that the product is great because they hope to sell it, and on the other convince themselves that the product will sell because it's great.
Trump made many empty threats his first term, so many didn't believe he would follow through to the degree he's done this term.
> The EMH would assert that a casual observer like me wouldn't see the price gap between the time it took for people to read sentence 2. But it took several business days.
This is a good point and surely sounds like an effect of liquidity demand. The same investors had to dump Nvidia also to pay for other losses and that briefly removed liquidity providers who wanted none or the volatility, until things calmed down a little.
How the information gets produced is irrelevant to the EMH. Whether it's obvious or takes hard thinking, either way, once investors obtain the information, they will trade based on it, and that will move the stock price.
And that is trivially true
And you'd need to know back in 2015 that Nvidia specifically would be the big winner from AI. They don't even manufacture their own chips. Intel also designs chips and GPUs, but if you bet on them in 2015, you'd have lost money between then and 2025.
You have to look at the volumes involved: if there are tens of millions of shares of a particular stock moved everyday, a single event that involves 100,000 shares is going to be lost in the noise.
There are always people who think they know better (if they didn't think so they wouldn't be trading), and they may make crazy-appearing trades. Lots of the people in The Big Short were viewed as 'lunatics' ("You're betting against the housing market?") that turned out to be right. But also remember that there are people who think the world is flat.
> The price roughly rose along the earnings. Even though the foundations for generative AI became clear in 2015.
It's also why you hear the talking heads on television say things like "…this has already been priced in.".
Where that really happens is with startups and younger companies. Some company is currently making negative dollars but a few people have figured out that they're likely to be big so their share price is up before their earnings are.
And suppose you somehow actually knew what every major company's earnings would look like in every year from 2015 to now. Do you invest in Nvidia in 2015? Or do you invest in Netflix in 2015 and Tesla in 2019 and so on and not bother with Nvidia until just before the hockey stick?
Now, it may be that the risk is one that other people care about but you don't. Some investors might be more sensitive to short-term volatility than others, for example. But a weak EMH can at least give a framework for thinking about these kinds of decisions. (Like whether to value tilt, for example. Or whether to invest in active or passive funds.)
It also gives you a framework to think about in which cases it is more or less likely to hold. Small, illiquid markets are the least likely to be efficient, in my opinion and experience. My only actively managed investment is in such a market.
Yes: most invested dollars will be unable to beat the market, on a risk adjusted basis and net of costs.
I suppose there are edge cases where they will instead attempt to profit by convincing the board to pay the CEO a trillion dollars, but even that kind of thing probably only flies if the stock price is also going up. (I wouldn't have thought to include that exception at all some years ago, but at least one salient example has proven this possible, if not likely.) So I could see a case for not trusting the valuation of companies that behave in that particular manner. Where the CEO is effectively the controlling shareholder, especially if they have shown a willingness and ability to inflate their own compensation.
If you buy 5 apples from me for $5 then two things must be true: 1. The value that those 5 apples have to you exceeds the value that $5 have to you, at least at this very moment. Otherwise you would hang on to your $5 instead. 2. The value that those 5 apples have to me is less than $5 have to me, otherwise I would hang on to the apples.
The price of those 5 apples at this moment may be $5 but that doesn't reflect the value they have to neither me nor you. It's not the avereage either, necesarily. The only thing we know is that the value of them to you is higher and to me is lower.
That isn't common but that doesn't mean it could never happen.
That is value. It is any benefit they capture which they would not otherwise.
The "Efficient Market" can be seen as an Eternal Steady State, neglecting all transient signals.
Trade very much is zero sum, at least some of the time. Prices are set by power disparities, not by abstract concepts of relative value.
One of the many problems with mainstream econ is that gloms together a whole set of unrelated interactions in a single crude concept of "price."
In reality share dealing, corporate wage bargaining, negotiations between farmers and supermarket chains, lemonade stands, and tourists haggling with craft vendors on vacation are all completely different kinds of interactions.
They end up with a price because they're mediated in currency, but their differences are far more interesting and economically revealing than their very superficial similarities.
> How often is "You should pay me much more, it will make both of us wealthier" a winning argument when asking for a raise?
A great deal. If you don't pay me, I will quit, and you will be worse off, so it's a win-win proposition for you to give me a raise. That's what labor market competition is about.
> They end up with a price because they're mediated in currency, but their differences are far more interesting and economically revealing than their very superficial similarities.
Hence microeconomics, labor economics, financial economics, and the various other mainstream economics disciplines that do try to split those hairs.
This is perhaps the broader point, which is that to the degree economics acknowledges non-priced value it’s a hand wave to “there’s some economic surplus here otherwise these people wouldn’t reach agreement to exchange”.
But to the degree that senses of value are the motivating factor behind economic exchange it’s oddly absent from the discussion. I get the reasons: philosophical inquiries about value and aesthetics are a lot more challenging to work through than concepts of utility reflected by measurable actions, but this goes to the overall point about the limitation and overreduction of current mainstream economics, and the criticism of people like Graeber of its politicised and somewhat arbitrary intellectual grounds.
To the degree mainstream economic reduction is useful to support understanding that’s fine - it absolutely allows reasoning and insight in many scenarios, especially microeconomics - but to the degree that decision-makers double down on it (especially macro) despite it being incomplete or absolutely wrong in certain environments and contexts because it promotes certain power structures and power plays that is highly problematic, especially when people jump to its defence out of misplaced loyalty to a school of expertise.
> What's the value of food? If you have none you die, so the value is quit of high, but the price is much lower than that because there are many competing suppliers.
The first calories of the day, the ones that prevent you from dying, have a very high subjective value - but you pay them at the value of the 3000th calorie of the day, the extra drop of ketchup on your fries, which has a very little value.
And thus of course average value x volume is very different from (marginal value of last unit) x volume.
For mainstream economics, this is true in a very specific technical sense; all averages lose information, and the "market average" is a very particular form of average that doesn't behave the way most people think of an average behaving—particularly, it is not like a mean, the normal "average" that people think of, that is sensitive to changes in any individual values, it is somewhat like a median in that it is insensitive to changes in existing values that do not cross the "average"; e.g., if you take an existing market for a commodity with a given clearing price, and reduce, by any amount, the value of the commodity to any proper subset of sellers who would sell at the current market clearing price, the market clearing price does not change. The assessment of value across the market has decreased, but the output of the particular averaging function performed by the market has not.
F. Lee, 1998, Post-Keynesian Price Theory
https://assets.cambridge.org/97805213/28708/sample/978052132...
Also see Nicholas Kaldor's Economics Without Equillibrium which can be read in an afternoon.