[HN Gopher] The Sonnenschein-Mantel-Debreu Results After Thirty ...
       ___________________________________________________________________
        
       The Sonnenschein-Mantel-Debreu Results After Thirty Years
        
       Author : gone35
       Score  : 27 points
       Date   : 2025-01-19 00:38 UTC (2 days ago)
        
 (HTM) web link (citeseerx.ist.psu.edu)
 (TXT) w3m dump (citeseerx.ist.psu.edu)
        
       | troelsSteegin wrote:
       | It's an economic theory for characterizing market dynamics. The
       | paper's conclusion: "Thus many of the problematic outcomes from
       | SMD theory remain entrenched."
        
         | 1oooqooq wrote:
         | like every single social science therory dressed as economic
         | research truth, the slight harder look at it crumbles all
         | assumptions.
         | 
         | so tiring.
         | 
         | best line on that paper: Much of the intuition that would seem
         | to follow from the SMD results is still intact. Rationality, as
         | Arrow said, does not have aggregate implica- tions, but now, we
         | have to add, if only aggregate data are available.
        
       | senkora wrote:
       | I didn't know anything about this and found the article hard to
       | understand at first, but thankfully the Wikipedia articles on the
       | subject are pretty okay:
       | https://en.wikipedia.org/wiki/Sonnenschein%E2%80%93Mantel%E2...
       | 
       | The very abbreviated TL;DR of this post is:
       | 
       | 1. We are taught that markets look like this:
       | https://upload.wikimedia.org/wikipedia/commons/8/8c/Supply-d...
       | 
       | 2. By applying rigorous math ("general equilibrium theory"), it
       | turns out that there is no reason that markets can't look like
       | this instead:
       | https://upload.wikimedia.org/wikipedia/commons/a/ab/SMD_Dema...
       | 
       | 3. OP's article summarizes how economists have reacted to this
       | problem.
        
         | hackeraccount wrote:
         | For other reasons I'm reading Adam Smith's Wealth of Nations
         | and this seems like nothing that Smith would object to. He'd
         | say there's a natural price that that the market price
         | gravitates around - very much like that second graphic.
        
           | bckr wrote:
           | The point is there are multiple "natural prices" that the
           | market price may gravitate around. If Adam Smith thinks
           | there's only one "natural price", then he's wrong, and you
           | probably want to read with that idea in mind.
        
       | cs702 wrote:
       | The Sonnenschein-Mantel-Debreu Theorem states that the excess
       | demand curve for a market economy of rational agents _can take
       | the shape of any function_ , subject only to three fairly general
       | conditions: the function must be continuous, must be homogeneous
       | of degree zero
       | (https://en.wikipedia.org/wiki/Homogeneous_function), and must
       | satisfy Walras's law
       | (https://en.wikipedia.org/wiki/Walras%27s_law).
       | 
       | Inconveniently for all economic theories, the implication of this
       | _mathematical proof_ is that excess demand and supply curves _can
       | take any shape_ that meets those three conditions. The nicely
       | sloped supply and demand curves we 've all been shown in
       | Economics textbooks are basically _figments of economists '
       | imaginations_.
       | 
       | Moreover, the textbook supply-demand curves we've always been
       | shown are for only one good. The curves for all goods in a market
       | economy are high-dimensional.
       | 
       | High-dimensional supply and demand curves can have multiple
       | equilibrium points, and there is no guarantee that those points
       | will be optimal or even good.
       | 
       | In other words, for a long time we've had proof -- proof! -- that
       | "free markets" are not guaranteed to converge toward good
       | outcomes.
       | 
       | Markets can get stuck in crappy equilibria.
        
         | shermantanktop wrote:
         | Ooh, now do the Laffer curve!
        
           | cs702 wrote:
           | I can't. There's basically no theory behind the "Laffer
           | curve." IIRC, it relies on two assumptions. Assumption #1: If
           | the income tax rate is 0%, the government will collect zero
           | income taxes. This assumption is basically a tautology. No
           | one can disagree with it. Assumption #2: If the income tax
           | rate is the maximum possible, 100%, the Laffer curve assumes
           | the government will collect _zero_ income taxes, because
           | taxpayers will find ways to avoid paying _all_ income taxes.
           | This second assumption is questionable, because taxpayers
           | might successfully avoid paying only _some_ taxes. Even if
           | you agree with both assumptions, there 's no agreement as to
           | what tax rate would maximize government tax collections.
        
             | shermantanktop wrote:
             | In other words, nobody can draw the curve, but people
             | assume it is a simple shape with a single maximum...which
             | is why I made my silly request.
             | 
             | But if it were anything other than a talking point, I'd
             | expect the possibility of complex Laffer curves with
             | multiple local maxima.
        
         | bjourne wrote:
         | Don't you mean homogeneous of degree one?
        
           | cs702 wrote:
           | Sorry, I wrote the equation for homogeneity of degree one by
           | mistake. I edited my post to the link to the Wikipedia page
           | on homogeneity instead. Thanks.
        
         | mturmon wrote:
         | Not a disagreement with your take, which was helpful to me, but
         | just your parenthetical aside.
         | 
         | Homogenous of degree 0, which is key to the conditions of the
         | results in OP, means f(tx) = t^0 f(x) = f(x)
         | 
         | [https://en.wikipedia.org/wiki/Homogeneous_function]
        
           | cs702 wrote:
           | Thank you. I inadvertently wrote the equation for degree one
           | homogeneity, from memory. I replaced the equation with a link
           | to the Wikipedia page.
        
         | emmelaich wrote:
         | Surely it's one thing to say that "curves can take any shape
         | that meets those three conditions" and yet still hold that any
         | observed or realistic actual curves will only take certain
         | shapes - as long as the consumers and producers have the right
         | information.
        
           | kibwen wrote:
           | _> as long as the consumers and producers have the right
           | information_
           | 
           | Honest question: what is "the right information"? Unless you
           | want someone choosing what is and is not the right
           | information, which feels like market distortion, then "the
           | right information" becomes "all possible information". And in
           | practice, we as consumers don't even have a thousandth of
           | "all the information", let alone a hundredth, let alone half.
           | (For example, where can I look up the details of _the entire
           | supply chain_ of this phone in my hand, for every single
           | material input it requires, to know what degree I 'm culpable
           | in supporting slave labor, which is something that would
           | influence my purchasing decision?)
           | 
           | And then even if we gave someone all the information, or even
           | some useful approximation of all the information, are humans
           | even capable of using that to make decisions that are
           | informed enough to allow markets to approach any sort of
           | rationality? Given finite time, isn't it possible that the
           | amount of time required to make a single informed decision is
           | longer than a lifespan?
           | 
           | So then the question is, if useful amounts of information are
           | neither practically available nor practically actionable,
           | what might that imply about our absolute faith in market
           | mechanisms that require the right information to operate
           | efficiently?
        
             | emmelaich wrote:
             | By "right" I mean roughly correct. To your other points I
             | suspect that close to correct is often enough and is more
             | correct over time. But this is where research should be
             | directed I suppose.
        
             | btilly wrote:
             | I can make it worse.
             | 
             | We make decisions based on partial information, processed
             | in a probabilistic manner. The logically right way to
             | reason involves the update of a Bayesian network - each
             | belief ties to each other belief. But updating a Bayesian
             | network - even approximately, is an NP complete problem.
             | 
             | So we not only can't make a correct decision, we can't even
             | try to approximate it in reasonable time. We can only apply
             | crappy heuristics. Like, "Are you with me or against me?"
             | Hence political alignment replaces our decision making.
             | Even for apparently pure economic decisions.
        
             | aaroninsf wrote:
             | "Right" is alas undefined and undefinable,
             | 
             | not least as "consumers" idon't make economic decisions wrt
             | to some a priori set of N weighted dimensions within which
             | a gradient of desirability is defined, or in fact by _any_
             | set of such dimensions.
             | 
             | They mostly make discretionary decisions wrt to a set of
             | influences triggers habits and cues most of which are not
             | about information or rational processes.
             | 
             | A hobby horse is that an increasing number of these are
             | manipulated by those in a market with the resources to
             | identify exploit them.
             | 
             | This is not uniquely the case for those making
             | discretionary spending decisions... but it for those of us,
             | that dissatisfaction--or dis-satiation--seems to
             | perpetually "find its level." (I.e. the post-orgasmic
             | clarify of the satisfaction of the craving, soon evaporates
             | like the dew.)
             | 
             | The phenomenon of perpetual unsatisfiable desire for the
             | Next Thing is so ubiquitous and accepted we don't even
             | attend to it any longer; and so baked in that 2500 year old
             | Greek myth gives us well-known parables (c.f. the myth of
             | Tantalus).
             | 
             | Related: so, GLP-1 drugs like Ozempic have had discernible
             | macroscopic impacts on "consumption" across many economic
             | categories, enough so that the Economist magazine had a
             | story IIRC on unease in industries most obviously based on
             | addictive behavior.
             | 
             | I am curious whether there has been a partiuclarly-
             | discernible impact in "expensive hobbies" defined by
             | constant pursuit of the Next Better Thing e.g. audiophiles,
             | collectors, constant-tinker hobbies like mechanical
             | keyboards, modular synthesizers, etc..
             | 
             | The battle in the heavens of our psyches may be waged for
             | the next period between GLP-1 agents on the one hand, and
             | exploitation of ML on the other, uniquely powerful tool as
             | it is for identifying and tuning efficacious triggers for
             | addictive behavior.
        
             | cratermoon wrote:
             | > if useful amounts of information are neither practically
             | available nor practically actionable, what might that imply
             | about our absolute faith in market mechanisms that require
             | the right information to operate efficiently?
             | 
             | See e.g. Akerlof, George A. (1970). "The Market for
             | 'Lemons': Quality Uncertainty and the Market Mechanism"
             | 
             | Akerlof shared the 2001 Nobel Prize in Economic Sciences
             | with A. Michael Spence and Joseph E. Stiglitz for work on
             | the impact of information asymmetry on markets.
        
             | Veserv wrote:
             | It implies that you do not get absolutely 100% perfect
             | allocation where everybody is maximally happy according to
             | their constraints which is what "efficiently" means in that
             | context. That is also a useless and impractical standard to
             | hold any system to as any optimization problem is
             | impossible if you have missing non-degenerate
             | constraints/information.
             | 
             | In practice, the question becomes what systems result in
             | the best outcomes given normal conditions and information.
             | As a example, we see that total command economies are
             | dramatically worse than markets at normal amounts of
             | information. Alternatively, how can conditions be changed
             | in normal fashions that can be consistently enforced to
             | maximize system outcomes.
             | 
             | Things are further confused by the fact that we have a very
             | poor understanding of even the implemented systems. What
             | aspects are important, what are not? For instance, we can
             | see that the modern Chinese economy is clearly more market-
             | driven than previously, so the increased amount of market
             | economics has been a dramatic improvement. Clearly, they
             | were previously on the "too little market" side of that
             | equation. In many ways, their markets are more
             | "unrestricted" than the USA, but in others less, are they
             | more market-driven than the USA or not? Are the ways in
             | which they are important, are the ways in which they are
             | not important? About the only thing we can really
             | categorize are directional changes from a previous status
             | quo and maybe we can categorize transformative impacts
             | after the fact.
             | 
             | It is these sorts of questions and observations that
             | matter. Quibbling about how a system is imperfect in a
             | wholly impractical limit to argue against practical matters
             | is just silly especially when those same problems are held
             | by literally everything.
        
           | cs702 wrote:
           | The problem is that a lot of people treat those _assumptions_
           | as if they were sacred, fundamental laws of the universe.
           | 
           | There are a lot of business people and government officials
           | for whom the notion of "free, unfettered markets" is akin to
           | a religious goal. Look around you.
           | 
           | There are NO guarantees that markets will work well. They CAN
           | lead to horrific equilibria, like inequality so extreme that
           | large groups of people cannot afford food and shelter without
           | government assistance, or financial crises that seem to come
           | "out of nowhere," with too many possible culprits that could
           | be blamed (greedy speculators, greedy financiers, complacent
           | banks, idiotic regulators, government-sponsored entities,
           | etc.), depending on who you ask.
           | 
           | I say, let's be humble, open-minded, willing to try new
           | things, and willing to change our views when the facts
           | disagree with them.[a]
           | 
           | ---
           | 
           | [a] In the words of Lee Kuan Yew, who led the transformation
           | of Singapore from one of the poorest to one of the wealthiest
           | societies on earth:
           | 
           | > My life is not guided by philosophy or theories. I get
           | things done and leave others to extract the principles from
           | my successful solutions. I do not work on a theory. Instead,
           | I ask: what will make this work? If, after a series of
           | solutions, I find that a certain approach worked, then I try
           | to find out what was the principle behind the solution. So
           | Plato, Aristotle, Socrates, I am not guided by them...I am
           | interested in what works...Presented with the difficulty or
           | major problem or an assortment of conflicting facts, I review
           | what alternatives I have if my proposed solution does not
           | work. I choose a solution which offers a higher probability
           | of success, but if it fails, I have some other way. Never a
           | dead end.
           | 
           | > We were not ideologues. We did not believe in theories as
           | such. A theory is an attractive proposition intellectually.
           | What we faced was a real problem of human beings looking for
           | work, to be paid, to buy their food, their clothes, their
           | homes, and to bring their children up...I had read the
           | theories and maybe half believed in them.
           | 
           | > But we were sufficiently practical and pragmatic enough not
           | to be cluttered up and inhibited by theories. If a thing
           | works, let us work it, and that eventually evolved into the
           | kind of economy that we have today. Our test was: does it
           | work? Does it bring benefits to the people?...The prevailing
           | theory then was that multinationals were exploiters of cheap
           | labor and cheap raw materials and would suck a country
           | dry...Nobody else wanted to exploit the labor. So why not, if
           | they want to exploit our labor? They are welcome to it.... We
           | were learning how to do a job from them, which we would never
           | have learnt... We were part of the process that disproved the
           | theory of the development economics school, that this was
           | exploitation. We were in no position to be fussy about high-
           | minded principles.
           | 
           | Source: https://www.amazon.com/gp/product/0262019124
        
             | emmelaich wrote:
             | Sure let's be humble. I think alleging that people holding
             | the notion of free markets is a religious goal is not a
             | humble opinion.
        
               | cs702 wrote:
               | I did not allege that everyone holding that notion treats
               | it like a religions goal.
               | 
               | "A lot of business people and government officials" [?]
               | "everyone."
               | 
               | Please don't attack a straw-man.
        
               | emmelaich wrote:
               | They're your words. I didn't say everyone. You didn't say
               | everyone. Where's the straw-man?
        
         | Etheryte wrote:
         | I think it's a pretty crude oversimplification to say it's a
         | proof when the whole idea hinges on the whole rational actors
         | axiom. Sure, it might be a proof with that axiom, but it
         | doesn't say much anything about the real world.
        
           | cs702 wrote:
           | I'm not sure I understand what you're saying. The whole
           | edifice of "markets converge toward general equilibrium"
           | rests on the assumption that people are rational, i.e.,
           | maximize utility -- which is economist jargon for "people are
           | trying to do better for themselves." If you take away _that_
           | , the whole theoretical edifice justifying markets collapses.
        
         | cratermoon wrote:
         | Another inconvenient truth for the theorem is that "a market
         | economy of rational agents" is a mathematical fantasy and
         | doesn't describe any real world market ever.
        
         | Veserv wrote:
         | It has been proven that global utility optimization can be
         | equivalent to solving a arbitrarily complex, arbitrarily non-
         | linear, arbitrarily quickly changing, function. Great. There
         | exists exactly zero techniques for solving all such systems
         | even if the function was not unknown.
         | 
         | That is proof that "free markets" are not guaranteed to
         | converge toward good outcomes because it proves that every
         | decision/optimization process that is believed to be even
         | theoretically possible is not guaranteed to converge toward
         | good outcomes. Singling out "free markets" as if they are even
         | somewhat uniquely susceptible is just plain ridiculous.
         | Everything is thoroughly screwed in the presence of arbitrarily
         | complex functions.
         | 
         | As we have now seen and proven that everything is doomed in a
         | arbitrarily malicious universe, it is only prudent to re-focus
         | on problems under normal conditions and figure out the
         | practical question of what systems produce good, high quality
         | outcomes in a normal implementation over normal conditions.
        
           | cs702 wrote:
           | As I wrote elsewhere on this thread
           | (https://news.ycombinator.com/item?id=42783130), the problem
           | is that a lot of busy people treat the "practical notion"
           | that markets produce good, high-quality outcomes as if it
           | were a fundamental law of the universe. It's not. There are
           | no guarantees. Markets can converge to horrific equilibria.
           | Let's be humble, open-minded, willing to rely on markets when
           | they work, willing to try other things when they don't.
        
             | kelseyfrog wrote:
             | This fundamentally changes the debate on market
             | interventions. It's now possible to argue that the current
             | equilibrium is only meta-stable and that interventions that
             | change the supply and demand curves can potentially kick us
             | out of one meta-stable equilibrium and into another more
             | efficient one.
             | 
             | I wish we had more empirical supply-demand curves so we
             | could refine our priors on supply and demand curve shapes,
             | but that's a topic for another day.
        
           | cherryteastain wrote:
           | Free market economics is "singled out" because, at least in
           | the Anglosphere, it is the only type of economy considred to
           | reach an (eventual) optimum by itself.
        
         | vtashkov wrote:
         | For even longer time we have a REAL proof (not mathematical
         | wordplay) that whenever, without exception, someone takes a
         | swing at free markets it leads to people dying. IT has always
         | been a disaster and people always have been exiting those
         | societies where that happens towards more free societies. You
         | yourself have plenty of options to go somewhere with less free
         | market - Cuba, Venezuela, North Korea, Russia, even China. Have
         | YOU emigrated to such a country?
        
           | lern_too_spel wrote:
           | You also have the option to go somewhere with even more of a
           | free market, like Somalia. Have you emigrated to such a
           | country? It turns out that societies can write laws for
           | mutual benefit.
        
           | cs702 wrote:
           | Oh, I'm a capitalist. I believe Churchill was right when he
           | said that capitalism is the worst economic system ever
           | invented, except for all the others that have been tried
           | before.
           | 
           | I never said we should do away with a market economy. All I
           | said is that market economies are not guaranteed to converge
           | to optimal or even good outcomes.
           | 
           | Markets CAN lead to horrifically bad equilibria (extreme
           | inequality, financial crises, recessions, depressions, etc.).
           | There are no guarantees that they won't.
           | 
           | So let's be humble, open-minded, willing to rely on markets
           | when they work, willing to try other things when they don't.
           | 
           | Please don't attack a straw-man.
        
           | semi-extrinsic wrote:
           | > whenever, without exception, someone takes a swing at free
           | markets it leads to people dying
           | 
           | Do Scandinavian socialist democracies not count as an
           | exception?
           | 
           | On the flip side, I think you will equally find that whenever
           | free market solutions are applied to public services -
           | healthcare, water, electricity, police, etc - people also die
           | in large numbers. And although wealthy people generated by
           | the free markets are philantropic, it's not in proportion to
           | the harm inflicted on people. Only thing is, we ignore those
           | disasters because they are inconvenient for the narrative.
           | 
           | It seems to be obvious that certain things (like public
           | utilities) should just be controlled directly by the
           | government, without any free markets involved, as long as
           | your objective function is to optimize the average life
           | quality of your citizens.
        
             | readthenotes1 wrote:
             | Can you give some examples on actual free market solutions
             | are applied to public services?
             | 
             | The closest I can think of are the metros of Tokyo free
             | government takeover, but I am not sure how free those
             | actually were of government regulation.
             | 
             | One of the struggles I have is that typically these public
             | services may be carried out by private companies paid for
             | by the government, but that is just begging for corruption
        
       | twic wrote:
       | It would be really helpful to see an example of one of these
       | naughty excess demand curves, along with the utility function for
       | each agent. For an economy with a small enough number of
       | commodities and agents that it might be comprehensible, say two
       | or three.
        
       | kelseyfrog wrote:
       | So the nicely sloped supply and demand curves are economic
       | equivalent of spherical cows and as much as we shouldn't design a
       | milk industry around spherical cows, we shouldn't design an
       | economy around a nicely sloped supply and demand curves.
        
         | cs702 wrote:
         | For anyone here who doesn't get the reference to spherical
         | cows, it's from an old joke:
         | 
         | > Milk production at a dairy farm was low, so the farmer wrote
         | to the local university, asking for help from academia. A
         | multidisciplinary team of professors was assembled, headed by a
         | theoretical physicist, and two weeks of intensive on-site
         | investigation took place. The scholars then returned to the
         | university, notebooks crammed with data, where the task of
         | writing the report was left to the team leader. Shortly
         | thereafter the physicist returned to the farm, saying to the
         | farmer, "I have the solution, but it works only in the case of
         | spherical cows in a vacuum."
         | 
         | Source: https://en.wikipedia.org/wiki/Spherical_cow
        
       ___________________________________________________________________
       (page generated 2025-01-21 23:03 UTC)