http://informationtransfereconomics.blogspot.com
Did anyone have a look at this econophysics?
I think it's not good economics. But I don't know how others think. I may be a stupid being.
It sure looks like he's just kinda fitting moving average trends to level data.
I want econ to be able to learn from other disciplines, but guess communication is hard. Probably would be better if he just posted his code and somebody translates it into econ. It looks like what he's doing in the end is really simple.
@341f
I use the notation Y ~ X to say that "Y is in information equilibrium with X" (I've also used p:Y→X) which basically means that:
p = dY/dX = k (Y/X)
in equilibrium. However sometimes this isn't true, and we have:
dY/dX ≤ k (Y/X)
And it is really simple! Just like how equilibrium thermodynamics is simple. I just fit data to the solutions of that differential equation. Or expectation values of solutions to that differential equation with the partition function
Z = Σ_i exp(-k_i log X)
What I've done is mostly just Econ 101 type stuff -- trying to reproduce basic results of economics. I'm not one of those "economics is wrong LOL" people. Ensembles of utility maximizing agents will produce an information equilibrium model. However there might be some insights that I hope could be useful -- for example, selecting which Arrow-Debreu equilibrium is realized via a maximum entropy argument.
Maybe it's not useful. I'll have a proper paper out soon (hopefully in the next month) so we'll see how it does in peer review.
@c7a1
It may well be useless, but it can't be absolute nonsense because it's mostly just supply and demand.
@ c7a1
I can only conclude you haven't actually read any of it before declaring it garbage. The derivation of the mechanics of supply and demand is fairly simple and requires only a bit of high school calculus:
http://informationtransfereconomics.blogspot.com/2013/04/supply-and-demand-from-information.html
@ c7a1
I can only conclude you haven't actually read any of it before declaring it garbage. The derivation of the mechanics of supply and demand is fairly simple and requires only a bit of high school calculus:
The words demand and supply are being used in the post but they don't appear to refer to anything that an economist would recognize as demand or supply. To assume that the demand side is the sole source of information which is transmitted to the supply side is also just not how information works in markets, it is two sided despite what Austrians might think about the matter. It is not clear from the math how quantities relate to prices and how you would relate quantity demanded back to consumer preferences. So I think c7a1 is correct when he says that there is no supply or demand in this in the sense that an economist understands supply and demand. Calling things supply and demand doesn't make them that.
@1e92
Calling things supply and demand doesn't make them that.
I completely agree with that. However, it's not that I just defined some variables X and Y as supply and demand, but rather that I derived something that is isomorphic to a supply and demand diagram. Variable X shifts in one way holding Y constant and a derivative falls; it shifts the other and the derivative rises. Vice versa for Y. There is then a 1-to-1 map from from this system to supply and demand.
Now maybe I am missing something in that 1-to-1 map on one side or the other -- I'd like to know what it is.
To assume that the demand side is the sole source of information which is transmitted to the supply side is also just not how information works in markets ...
This is one where the colloquial definition of information and the information theory definition tend to generate confusion.
There is a probability distribution P of demand instances on one side and a probability distribution Q of supply instances on the other. If they do not match, that is information loss in the sense of the KL divergence D(P||Q).
If I show up at a store to buy a toaster (P) and there is no toaster there (Q), then there is information loss D(P||Q). If I am the only person in the world who will buy a toaster (P) and there are two there (Q), that is information loss D(P||Q). If I am the only person in the world buying a toaster (P) and it is in Australia (Q) then that is information loss D(P||Q).
The probability distribution firms selling toasters are trying to figure out is P, the demand distribution. Their guess is Q and if Q ≠ P, then there is information loss. When that happens, on average, toasters sit idle or there aren't toasters available to buy. In that sense I(Q) < I(P) and we should think of P as the information source.
This is different from the way economics seems to treat the information flow in the market -- and it may well be wrong or useless. And I could be even be wrong about it being different. It is what is new about this information theory approach to economics.
http://informationtransfereconomics.blogspot.com/2015/03/the-price-system-as-communication.html
Just so you are aware, this is not, in fact, a serious forum for discussion of new academic ideas. This is a forum about memes and trolls about the economic academic profession.
I'll give you one tip though: Go away from trying to predict things and show how good you are at predicting things, but detail exactly and precisely what new insight you will bring to the table in RELATION to the existing literature.
If your insight is that you can predict things better, that is in one sense valuable, but not overly so.
Start by saying what _exactly_ your model adds to the existing field of economics (best in terms of this field). With that, you'll have a much simpler time in gaining traction.
And now back to our regular schedule of trolling the s**t out of this guy
For example, selecting an AD eqm. You have to go into the literature on that, and then best write one paper on just this.
If you paper ends up being: "Here is how I revolutionize everything about economics", then for better or worse you'll only get published in an econophysics journal, among several other papers claiming the exact same thing.
jrsmith237, are you aware of Duncan Foley's statistical approach to equilibrium?
https://ideas.repec.org/a/eee/jetheo/v62y1994i2p321-345.html
There are a few follow-up papers on it. There are also maybe ten relatively recent papers, published in "orthodox" journals, that use entropy and mutual information in connection with other types of problems.
@3758
I have read several of Foley's papers; he has a lot of interesting stuff and that approach is probably the closest to the one I've been working on. The major difference is that his approach still focuses on utility maximization (as opposed to entropy maximization) which is essentially a particular case of a more general information equilibrium framework. (A theory of agents that maximize utility is a particular case of an information equilibrium model.)
There is no free lunch here, though. The more general info eq framework doesn't tell you as much about dynamics. With utility maximization you can say a lot more about fluctuations, but in the real world agents might not maximize utility.
One particularly interesting insight Foley has about why thermodynamics and economics are different is that it's due to physicists looking mostly at reversible processes while economists look at irreversible processes (no one voluntarily tries to undo their utility gains).
[As a random aside: matching models also seem to be a particular case of information equilibrium.]
@81d1
The additional information comes from the increased number of bits required to describe the allocation of an increased number of goods sold/demanded. If there are 3 widgets and two agents, there are 4 ways to allocate the widgets (3 + 0, 2 + 1, 1 + 2 and 0 + 3). If there are 4 widgets and 2 agents, there are 5 ways. That's an increase in the information content of an allocation of about 1/3 of a bit: log_2(5) - log_2(4) = 0.32.
And if you'd like, you're welcome to come and ask questions (or put snarky comments) on my blog linked above.
@3758
I have read several of Foley's papers; he has a lot of interesting stuff and that approach is probably the closest to the one I've been working on. The major difference is that his approach still focuses on utility maximization (as opposed to entropy maximization) which is essentially a particular case of a more general information equilibrium framework. (A theory of agents that maximize utility is a particular case of an information equilibrium model.)
There is no free lunch here, though. The more general info eq framework doesn't tell you as much about dynamics. With utility maximization you can say a lot more about fluctuations, but in the real world agents might not maximize utility.
One particularly interesting insight Foley has about why thermodynamics and economics are different is that it's due to physicists looking mostly at reversible processes while economists look at irreversible processes (no one voluntarily tries to undo their utility gains).
[As a random aside: matching models also seem to be a particular case of information equilibrium.]
@81d1
The additional information comes from the increased number of bits required to describe the allocation of an increased number of goods sold/demanded. If there are 3 widgets and two agents, there are 4 ways to allocate the widgets (3 + 0, 2 + 1, 1 + 2 and 0 + 3). If there are 4 widgets and 2 agents, there are 5 ways. That's an increase in the information content of an allocation of about 1/3 of a bit: log_2(5) - log_2(4) = 0.32.
And if you'd like, you're welcome to come and ask questions (or put snarky comments) on my blog linked above.
You are too smart to be researching economics.