Cartels Are an Emergent Phenomenon, Say Complexity Theorists
technologyreview.com
technologyreview.com
He also points out that many fundamental economic principles are flawed, and have been proven to be flawed for years, but economists lack rigor, and would rather live in their "supply meets demand, actors are rational, and the market is in equilibrium" fantasy-land.
Oh, and in about 2006, he warned that there'd be a recession caused by a debt-deflation, just like in 1929. He warned that the government would continually underestimate its impact. He also thinks that the best way to "reset" the system is with a "modern jubilee" - the government engage in massive quantitative easing (which they won't, because they don't realise what's causing the crisis, and how bad it is), and should hand out the free money to tax-payers rather than giving it to banks, as the banks have lost their appetite for risk and won't create new money even if you feed them.
But mainstream economists only study basic calculus, linear algebra, and statistics, not ODEs, and don't believe anything they don't understand. A 50 year old professor (or central banker) isn't going to go back to school and sit in a class with sophomore engineers, just to be able to understand what "complexity theorists" and "econophysists" are talking about, so they just pretend that it doesn't exist. There's no conspiracy, they just don't read or teach anything more mathematically advanced than IS-LM (which is stone age). Their journals don't accept non-mainstream papers for "methodological reasons" (they don't understand basic differential equations), or because it "doesn't sit well with the current theory" (it proves them wrong) and since there's tens of thousands of them they tend to dominate the field.
No, that's not why economics is in the state it's in. The reason economics is full of theorems that only hold in highly stylized worlds is that economics is hard.
Betrand Roehner provides an insightful classification of problems in the natural and social sciences by their level of complexity.[1] Two-body problems (like most of those in classical mechanics) are the least complex. N-body problems in which the interactions are local and the bodies are homogenous (like many problems in statistical mechanics) occupy the next level. At the third level of complexity are homogenous N-body problems with long-range interactions. And at the fourth level are heterogeneous N-body problems; these include lots of hard problems, like regulation of gene expression, morphogenesis ... and pretty much all interesting problems in economics and the social sciences.
Yes, I just said economics is harder than physics. (Well, it's harder than the physics physicists have been able to figure out so far.) And the difficultly of the problems is compounded by the difficultly of doing experiments in economics and the social sciences; you can't measure and tinker with a self-contained version of society in a lab somewhere. Is it any surprise then that most of economics doesn't really tell us anything directly applicable to the real world?
> ... mainstream economists only study basic calculus, linear algebra, and statistics, not ODEs, and don't believe anything they don't understand. A 50 year old professor (or central banker) isn't going to go back to school and sit in a class with sophomore engineers, just to be able to understand what "complexity theorists" and "econophysists" are talking about, so they just pretend that it doesn't exist. There's no conspiracy, they just don't read or teach anything more mathematically advanced than IS-LM (which is stone age).
You could not be more wrong. Economists, at least those from good programs, have great quantitative training. Pick up any copy of Econometrica if you don't believe me.[2] The mathematical tools used by financial economists, in particular, are largely the same as those used by statistical physicists, which is why so many physics PhDs end up working in finance. (Brownian motion, power-law distributions, etc. were all first described in an economic context.) Informed criticisms of the profession usually comes from the other side, e.g., that top economics journals are full of mathematical fireworks and devoid of reference to the real world.
And IS-LM has not been a part of mainstream macroeconomics since at least the early 1970s. It is still taught in some intro undergrad macro courses (precisely because it requires no mathematical sophistication to understand, I suppose), but it is not a part of the graduate curriculum or a part of the way macroeconomists comprehend the workings of the economy. Your citing it illustrates how little you know of what you speak.[3]
1. http://books.google.com/books?id=LEzfZ-Z98V0C&lpg=PA10&#...
2. http://www.econometricsociety.org
3. There was recently some hullabaloo in the economics blogosphere of the place of IS-LM in economics pedagogy. See, e.g., http://marginalrevolution.com/marginalrevolution/2011/10/put...
I'm saying that too.
> You could not be more wrong. Economists, at least those from good programs, have great quantitative training.
Which programs? Is this Ivy League PhD, econometrics stream? Or undergrad? First year physics students learn special relativity. They won't learn the advanced stuff until 3rd year, or maybe 4th, but it's still taught.
But most economics students struggle with "econometrics", which is glorified linear algebra. There's theoreticians who do understand complexity, but none of their work is likely to be widely read. There'll be a few ivory tower academics doing really hard stuff, but it doesn't enter the general body of knowledge.
It's like saying that computer programmers don't understand algorithms. Of course some do, and there's research out there if you want to find it, but generally just they kind of know that O(n) is good, and O(2^n) is bad.
And while economists are pretty good at statistics (at least, the good ones are), very few use differential equations. It's not covered in undergraduate classes, and by postgrad classes it's too late.
And while IS-LM is going out of fashion, it's just being replaced with DSGE. I can't see any appendix on differential equations in the TOC of that textbook you mention, is that because everyone already understands them, or because despite the model being called "Dynamic", it's not? I quote Solow, the guy who kind of invented it:
http://econlog.econlib.org/archives/2010/07/robert_solow_on_...
> I do not think that the currently popular DSGE models pass the smell test. They take it for granted that the whole economy can be thought about as if it were a single, consistent person or dynasty carrying out a rationally designed, long-term plan, occasionally disturbed by unexpected shocks, but adapting to them in a rational, consistent way. I do not think that this picture passes the smell test. The protagonists of this idea make a claim to respectability by asserting that it is founded on what we know about microeconomic behavior, but I think that this claim is generally phony. The advocates no doubt believe what they say, but they seem to have stopped sniffing or to have lost their sense of smell altogether.
Yes, economics is hard. Yes, there's smart people working on it. But by the time it filters down to teaching undergraduates or setting policy, the general attitude is "we'll assume rational behavior and a market at equilibrium, so nothing we say is going to be too controversial".
(trollish rant ahead)
I wonder what to ration of neo-classicals to quants is on Wall Street (where people actually have to make the right calls) compared to the Federal Reserve and the Treasury?
It's basic agency theory - the economists who can't handle complexity take over dumb institutions (like most universities and governments), then insulate themselves against having to learn anything they aren't comfortable with.
did they use it? no. why not? because they felt it was important that they fully understand the forces at play that cause the default rates to move in that direction. you could say that they were worried about models overfitting the data - sure, that's kinda it. but the point is the entire nature of the market can change overnight and kind of go into a different state.
so what does that have to do with economics? well i guess maybe some of the same reasoning is at stake - people who think "we're proved wrong so often we need to retreat back to just those few things we can really rely on being true" and are worried about being "too clever" with all this math stuff.
now its true that cargo cult like application of mathematical formula has been (imho) part cause for many truly dreadful economic theories. but still, they are wrong. if you don't understand the nature of far from equilibrium, complex systems, you'll never even begin to understand the forces that drive the economics of the real world.
So while they don't overfit, they also are't based on any real knowledge of the system.
It seems the other extreme is elaborate overfitting models.
Steve Keen uses fairly simple models, but tries to base them on actual reality (i.e. broad money is created by the banks, not the government, while Ben Bernanke's explanation of the Great Depression apparently focuses on M0, as he accuses the government of not minting enough coins with the gold they had in reserves, with the implicit assumption that the money would be multiplied if the government would only print it. In reality, the money was not multiplied, because the banks didn't feel like lending.)
I aked Steve about this exact problem, [0] the mismatch of decision making using at best questionable economic modelling and the reply was along the lines you suggest. It starts right at University level theory where nobody questions the accepted orthodox theory. What I don't understand is why more mathematicians don't call mainstream economists bluff?
[0] "Why I went" http://www.flickr.com/photos/bootload/5518902314/in/set-7215...
[1] http://www.flickr.com/photos/bootload/collections/7215762379...
It's basic agency theory - the economists who can't handle complexity take over dumb institutions (like most universities and governments), then insulate themselves against having to learn anything they aren't comfortable with. I guess the smarter ones work for Goldman Sachs, or work in university research departments, writing obscure papers that are never going to be compulsory reading because undergrad economics courses don't require hard-core math.
Why punish it? By this reasoning, all you need to do is to give buyers better information so they can react faster.
The problem with gas prices, I suppose, is that you have to drive around to collect them, and once you come to a station, if you decide its price is unacceptable there's a cost to checking out the next one (the time it takes you to drive there).
Seems like there could be a mobile app that tells drivers the gas prices at stations in their vicinity, which, according to the article, should change the system's behavior mode to push prices down instead of up.
Also, that mobile app (or several) exists. But they rely on users reporting the data, so there are significant network effects -- the more people use it, the more timely and useful the data becomes.
Well, I imagine the logic is simple: a cartel--even if it is not the result of collusion--undermines the market (to some extent). Therefore, it is in the interest of a fair market to disincentivize it in some way. It doesn't have to be punishment per se, but some way of trying to limit cartels would be good.
It is an interesting observation/model but calling it a cartel is a little bit over the top. The good/service reaches its market price regardless of whether it gets there from lower or higher prices.
The significance is because of the loss of efficiency, not because of the name/label given, or how the price is determined.
We like free-market economics because they get good results. If there's a market failure and the standard free-market economics aren't applying, like in the case of a cartel, then that's bad and we want to change it. Bad results are bad, good results are good.
It must be nice to live in a world without regulatory capture.
"Want to change" doesn't imply "will do less harm than good" in this world. In fact, "want to do good" is pretty much the basis of all of the really huge problems we have.
Since you didn't make an argument I won't do the same disservice to you. The relevant laws in this case are anti-trust. I support them, generally, as did TR who introduced them. Regulatory capture is a problem to be battled, not a reason to throw your hands up and say "oh well, guess i'll just let myself get screwed by a couple people who managed to lock up an industry through collusion".
That's nice, but the question is whether the realized benefits exceeded the incurred costs.
> Regulatory capture is a problem to be battled
You're assuming that you can win the war. Some evidence would be nice.
There is: http://gasbuddy.com/
As anyone who has studied complex systems understands, the evolution of interactions among even very simple agents is enormously sensitive to initial conditions. (That's often taken as the very definition of chaos in deterministic systems.) So it is almost always wrong to draw strong conclusions about a complex part of the real world based on your computational model of it: vary one parameter a little bit and you end up with something that looks completely different.
This is why agent-based models, like the one employed in this study, are widely despised in economics and the other social sciences despite offering a way to do experiments that would otherwise be impossible in these fields. We simply don't know enough about people and how they interact to build meaningful models, and we probably never will.
For example, an economist might point out that the model presented in this paper doesn't account for the spacial distribution of gas stations and consumers, for buyers' and sellers' expectations of future prices, or for any number of conceivable and inconceivable factors—almost all of which would affect aggregate outcomes in the model, I would bet, just as the authors found that the speed of reacting to price-changes did. (That was the only parameter the authors varied, by the way.) And of course we could model the problem in many fundamentally different ways to start with.
In other words, this is much different (and harder) than building an N-body model of the dynamics of Alpha Centauri, where's there's basically one force (gravity) that operates in the same way on a bunch of things that can all be described adequately by a single parameter (mass).
Their description of their model[1]:
Our model is defined as follows. We consider a population
of N agents, where each agent has two simultaneous roles:
donator (e.g. a buyer) and rewarder (e.g. a seller).
"No need for conspiracy: Self-organized cartel formation in a modified trust game" -- sure, in a purely locally bounded space of action and information. But that is not the reality of life on this planet. (Not for the past 200 years, for sure.)
I believe the same issues plague subjects like theoretical physics.
Then again, we do have the preference-utility and utility-maximization core of microeconomic theory, which appears quite well-accepted. Accepted, in spite of it's inability to reconcile with some of the non-rational parts of human behaviour.
Having said that, I'm quite partial to some of the PDE models (i.e. the works of Dr Yang Xiao Kai). And PDE models have at least one corresponding agent-based model simulation.
I think that at some point, a leap of faith based on intuition is taken enmasse. Although this really goes into the philosophy of science...
Agent-based models, though built on better-founded micro theory, are arguably even harder to get right because agents are adaptive in unexpected ways. And because it's possible to make many models give results that "look right" if you fiddle with the parameters enough, you're just not going to convince anyone that you've built a model that represents the way the world really works.
Nobody trusts a black box. I think that's why most economists still build little mathematical models that try to give some targeted insight into some small aspect of some economic phenomenon. If you can understand how a model works (and, again, you usually don't "understand" simulations—all the interesting stuff is emergent), you feel like you've learned something about how the real world might work. Then you can get down to bickering about assumptions, which effects dominate, etc.
1. http://en.wikipedia.org/wiki/Computable_general_equilibrium
2. http://en.wikipedia.org/wiki/Dynamic_stochastic_general_equi...
I'd say that they are generally exploratory models and pedagogical models. That's why the critic by the original poster is off-target: it hardly addresses the commonly used/accepted metric for economic insight.
Yes, they are starting with a simple and straightforward model and limiting the parameters they vary. But they can produce what looks to be emergent cartel behavior.
That's the important takeaway: a simple model is unexpectedly able to do that. AKA you don't need a complicated model with any number of variable parameters.
[Also, what they've done here is study the statistical properties of the system. You're not going to be able to "predict" or replicate the exact behavior/configuration of one particular instance unless you do know its starting parameters in detail. But looking at the statistical properties should smooth that out.]
Yes, I know. That's why my criticism was directed at the Tech Review article that totally misinterprets the paper, and indirectly at all of the comments in this thread that make the same mistake. The paper does not describe how gas-pump prices change; it does not describe how any actual market actually works. It does not identify time to react to price signals as the key to well-functioning markets; that is an artifact of the highly stylized model employed.
As I wrote in the first line of my comment, the take-away is that cartel-like behavior is possible without direct coordination, not that it is actually happening, ever has happened, or ever will happen. The rest of my comment was meant to point out why we can't draw those conclusions.
EDIT: And, by the way, there are lots of other ways to show that cartel-like behavior is possible without concerted price-fixing. Several authors have shown this in a game-theoretic framework, and there is a whole literature on indirect communication among economic competitors that has arisen out of law-and-economics analysis of anti-trust regulation.
I agree that it's not an actual explanation or a model of any actual market. How both the article and paper are worded is a bit ambiguous on that point most likely for purposes of attracting a bit of attention and grant money. But if you look closely, there's no outright claim in either.
As for seemingly ignoring existing literature in another field, it's kind of par for the course. A lot of "discoveries" made early on in complexity research on sandpiles were already well-documented in civil engineering.
Taking this a bit further, the best way to fight this phenomenon may be to improve buyers' ability to react quickly to the market. I'm sure there's some startup ideas here for specific instances of "market".
The model might apply, but only very slowly. A single tick of the model would be equal to the duration of a contract (i.e., instead of updating prices multiple times per day, as the authors propose, prices would be updated every 1-2 years).
There are two conditions, high number of cycles, and company's reacting faster than customers. As I said there is a high number of cycles because company's can change their price for new contracts every day or hour etc. The second condition that customers reactor slower which does not mean they can't renew at the end of their term just that they don't react as fast to changing prices as companies do.
Personally, I have gone though several cycles of renewing my contract and I must say I don't price shop on the day I sign the new contract. Do you?
By contrast, when sellers react quickest, they are quick to copy others offering poor value for money. This reduces the number of sellers offering good value for money in a vicious cycle that drives prices as high as possible.
Does this model match practice?
i.e. If we consider all industries in which there are price cartels, particularly those that were formed without price collusion, could the existence of those cartels be attributed to sellers reacting faster than buyers?
Also, are we confident of the converse? That there an no examples of non-collusive cartels in markets where buyers react faster than sellers? This seems plausible to.
Our best bet solution is currently regulation. Not great, but it works.
That sentence doesn't make any sense to me. There is no evolution without mutation.
Our best bet solution is currently regulation. Not great, but it works.
Some have very good arguments that it doesn't work, since the regulators are a lever that is most easily pushed by the powerful: http://en.wikipedia.org/wiki/Regulatory_capture
Also "monopolies" don't always have to be in the disinterest of the consumer. If there really is a single company that produces the best product, why punish it? The definition of "product" alone is difficult enough - does Apple have a monopoly on computers? on iPads?
A friend of mine has done a lot of research on the major credit card companies for his hedge fund. He noticed that these companies tend to raise their fees basically in lock-step (when one does it, the others follow quickly).
This behavior seemed counter-intuitive to me -- I would have guessed that the companies would compete to keep fees low. But this actually makes some sense in a world where the seller (the credit card company) reacts much more quickly than the buyer (the fee-paying banks and merchants). Banks and merchants can only switch credit card issuers if they can get their customers to (i.e. the card-holders). Therefore, it's possible for a card company to raise fees without losing market share right away. Other card companies take note and their dominant strategy is to raise fees as well.
Surprisingly and scarily in tune with Zeitgeist Moving Forward [1]
I guess I understand why people would not want to see it put so bluntly (capitalism is flawed. full stop.) but even you in so many words said the same thing. It is in part inherently corrupt and the silent hand of the market is sometimes a sinister one.
In any case, we already know that there are cases where capitalism fails, the best-known probably being the monopoly. We even know that there are certain market conditions that encourage these failure modes (for example, economies of scale encourage monopolies and oligopolies). Recognizing another possible failure mode (non-collusive cartels) and the conditions that trigger it doesn't push capitalism over the line from desirable to flawed. Maybe you already believe that capitalism is flawed, but I don't think that this by itself pushes it over the line.
It permeates markets with perfectly inelastic demand (all consumers are obligated to buy), where the sellers all use a particular pricing strategy, and where sellers have considerably better information than buyers (i.e., when K=1, ~60% of price updates must be seller price updates before cartel behavior begins).
The suggested association of the mathematical model to the model of markets (with many suppliers) where the buyer has a "slow reaction time" results in a price that would, in some sense, be analogous to the price generated by cartels.
The implication is likely that such a market doesn't reach the perfectly competitive price, which is what a simpler model that doesn't consider the buyer's reaction time would predict.
Neither the article nor the paper claims that this emergent phenomenon occurs (can occur) in all, or even the majority of, markets.
Quote: "Under certain market conditions, cartels arise naturally without collusion. This raises important questions over how the behavior should be controlled"
Neither the article nor the paper claims that cartels always "emerge" as "a product of the system". No systems were mentioned, and no definition of product was given.
Neither the article nor the paper claims that cartels do not emerge "as a product of immorality or deceit".
It seems plausible that people might not enjoy a misrepresentation of the authors, who did put in effort (authoring such a paper is never done in a night).
All systems are flawed and unfair to some level, and there are tons of examples for capitalism in particular. The question is to what extent we can mitigate some of the negative effects and which ones we're willing to accept.
If your curiosity is strong enough, you can find a more detailed outline of the concept in an introductory economic text.
Whether that means capitalism is flawed depends on the extent to which you consider the neoclassical utility-maximization explanation of how markets work to be an important part of the justification for capitalism. It probably undermines the utopian libertarian vision of a laissez-faire, unregulated economy in which everything is free and everything is perfectly allocated, more than it undermines the U.S./European brand of pragmatic market economies.
Consumers may complain that this behavior results in higher prices than they would otherwise pay. But I suspect those prices are likely only raised at the margin, to some degree that shifts consumer surplus from consumers to sellers. They aren't being priced out of the gasoline market entirely -- were they so, total sales would fall off signalling reduced demand, and we'd expect prices to fall accordingly. Copying here is sellers following those prices that best capture that surplus. I'd be more worried about collective behavior that restrained market entry by other sellers.
Also: this discussion is focused on the short- to mid-term. The overall price level, even if shifted among possible equilibria by collective behaviors, signals demand to producers. A high gasoline price raises oil prices which increase incentives to find and produce more oil. As that oil comes to market, supply increases and price falls.
tl;dr Free markets promise "best", not "perfect", allocation. Not clear that these tacit cooperations do more than move prices within possible local equilibria so as to capture more consumer surplus for producers. In any case, prices still serve, over the long term, to signal returns to investment in production.
No, classical economic theory doesn't assume this. It derives this as a conclusion under some specific assumptions: demand and supply are elastic, many non-colluding buyers and sellers, good information and low transaction costs. The conclusion of classical economic theory is that under the circumstances I described above, prices will stabilize near a price P satisfying supply(P)=demand(P), where "near" is bounded by the size of transaction costs.
The paper under discussion is also classical economic theory, it just makes different assumptions: it assumes perfectly inelastic demand (i.e., users have no choice but to buy) and poor information/high transaction costs (buyers can only switch sellers infrequently and can't switch arbitrarily). It also assumes that sellers only copy the pricing strategies of others, and never come up with their own (e.g., no aggressive price competition).
It's not a "flaw" in classical economic theory, it's just applying classical economic theory to a new circumstance. It seems like it might be a good model for certain markets - mandatory insurance for example (auto collision insurance or health insurance in MA), display both demand inelasticity (you are legally obligated to buy) and comparison is difficult.
Essentially, you are claiming that North Korea is also a democracy, only without free elections.
If you want to argue that the assumptions made by the article are more accurate in general than the more common assumptions (good information, low transaction costs, elastic supply and demand), be my guest.
I'd make a rule that if you want to sell anything and you have list of current prices in any electronic format you have to publish it online in machine readable format before you are allowed to sell single item of your stock.
Other necessary rule is to establish that for store to be allowed to sell anything it must allow all customers use their cellphones in any way they see fit.
Looking it up this is apparently also a recognized use of the term and so my question was a bit dumb.
Usually these complex systems consist of parts where the individual interactions are thought to be well understood but the aggregate behavior is not (aka non-linear).
In the case of sand, you can pretty much model the forces acting on a single grain/particle using intro physics but you wouldn't expect the long-range force chains you see in static piles or avalanches when more sand is added.