Discussion Topic |
|
This thread has been locked |
k-man
Gym climber
SCruz
|
|
|
Jan 16, 2009 - 03:07pm PT
|
Woody, haven't you heard? This is the dawning of the age of Aquarius.
|
|
TradIsGood
Chalkless climber
the Gunks end of the country
|
|
|
Jan 16, 2009 - 03:29pm PT
|
Read the book.
|
|
hobo_dan
Social climber
Minnesota
|
|
|
Jan 16, 2009 - 03:56pm PT
|
I tried and it bored my ass
|
|
Ed Hartouni
Trad climber
Livermore, CA
|
|
|
Jan 16, 2009 - 04:40pm PT
|
I haven't read the book, but there was a recent article in the NYTimes interviewing the author, having followed him around a bit...
The "modeling" done in economics to quantify investment risk was not news to me, and it is apparent upon a very brief study that those models have some very limiting assumptions, a physicist would say "domain of validity." The assumption that economic fluctuation can be modeled by a Gaussian probability distribution function ("pdf" not the document!) has obvious limits... and the parameterizations used to evolve the economic time series is similarly limited.
While the risk models do well to quantify a relatively stable situation, once the situation begins to behave beyond the model, the results of the model will not help you decide on how to respond to that situation.
I draw two conclusions from the current situation: 1) that models not based on "underlying" theory must be used very carefully and 2) the behavior that the economy just underwent will result in better parameterizations that will be incorporated into future models.
Ultimately, however, I despair that economics will ever have an underlying theory which can be used to build a "physical model" with predictive ability.
This is in stark contrast to other, controversial, models discussed on STForum (and other places). I am thinking about climate models. Climate models are more and more becoming the application of physical parameters to the atmosphere-ocean-land systems rather than a paramterization. As we understand the science that is relevant to this system, and make the physical measurements required for the system (the physics/chemistry of cloud nucleation, chemical reaction rates, light absorption rates, etc) the models depend less on the past answer and more on the underlying mechanisms.
The question becomes: has all the science been included?
The answer is by testing the output of the model to reproduce observation, and observation is the derived from recent climactic data as well as proxy data from the historic record. We would test the consistence of the model, and see what its deficiencies are, produce a more sophisticated model and test again.
As this has been going on for decades, the resulting models are better and better, leading to the belief that the models could be used to understand regional climate... Because these models are based on the underlying science, and not on a parameterization of history, the rest on a very solid base.
Further, many of these models can be applied to other atmospheres which can be observed and those observations understood as a result of changing the initial conditions. This could not be done with models based on a parameterization of earth atmospheric behavior. Because the underlying science does not depend on the specific atmosphere it is operating in...
The goal of a general economic model would be to have an underlying "science" that could predict the economies of extraterrestrial societies... as far as I know, there is little about economic theory that would be applicable... unless the economists stole the ideas from ecology, their meta-science...
|
|
WoodySt
Trad climber
Riverside
|
|
|
Topic Author's Reply - Jan 16, 2009 - 04:57pm PT
|
From the layman's--mine--point of view, I don't have much confidence in the "experts" dealing with the present economic crisis. I sense they have all fingers and toes crossed. The world wide economic structure now existing is far more complex and possibly unstable than at any time before in history. Actions now applied are based on past models which don't comfortably fit the contemporary economic emergency. We're sailing on truly foreign seas.
|
|
rmsusa
Trad climber
Boulder
|
|
|
Jan 16, 2009 - 05:01pm PT
|
* I haven't read the book, but... *
I have. It doesn't have a lot to do with models, except in a philosophical sense. Read the book.
|
|
TradIsGood
Chalkless climber
the Gunks end of the country
|
|
|
Jan 16, 2009 - 05:31pm PT
|
I will repeat the suggestion...
Read the book.
Not to pick on Ed here, but Gaussian? Definitely not. Gaussian is symmetric, prices are not (They stop at zero.) They do like the diffusion equation though - at the simplest levels. But economics is no more "scientificable" than predicting the average annual temperature of the earth's atmosphere in 2100, because they both depend on politics and all manner of other chaotic inputs in highly complex ways (humanity).
Here I do pick on Ed, but for a different reason. There is an asteroid moving around the sun whose position, momentum, etc. we know with fairly high certainty. And we know the physics even better. So why is there any uncertainty of its position in 2036, within plus or minus one earth's diameter?
The answer is because when you iterate forward from time T to time T + 27 years, the uncertainty is huge compared to its uncertainty at T. On top of that, we do not know what the solar output will be over that time, the cumulative effect of which matters.
The book is not only about economics, but questions, knowledgeably, the models used by economists whose fame got them "Economics Nobel Prizes".
Along with that book, get the one I have mentioned previously on disasters with complex systems. No simple models can explain the current economic crisis.
|
|
Ed Hartouni
Trad climber
Livermore, CA
|
|
|
Jan 16, 2009 - 07:10pm PT
|
Here I do pick on Ed, but for a different reason. There is an asteroid moving around the sun whose position, momentum, etc. we know with fairly high certainty. And we know the physics even better. So why is there any uncertainty of its position in 2036, within plus or minus one earth's diameter?
because we don't know the position and momentum well enough to limit the error of a 27 year extrapolation to smaller than the earth radius. However, the error is incredibly small and the uncertainty can be calculated, rigorously.
We know better as it gets closer. However, I don't believe that we are worried about any large asteroids because we know were they are, and we know a lot about their orbits. So that is not an example of what I believe the subject of "The Black Swan" is: unknown unknowns (perhaps TiG would say unknowable unknowns)... and the Gaussian model is an estimate of the fluctuations of prices... not the price itself but the "noise" in the system.
But I don't know what was written in detail... I'm trying to get my hands on the book.... my criticism is of economic modeling, not of the book, and it isn't so much a condemnation of the modeling, but of those who would use the models to justify their irrational behavior.
"The model made me do it" isn't a good excuse for these models...
The alternatives to models is essentially "Wild Ass Guesses" or WAGs.... throwing hands up and saying "it's too complicated" is really just a cop out. The models work well in a limited area of application. And that is better than nothing.
|
|
Toker Villain
Big Wall climber
Toquerville, Utah
|
|
|
Jan 16, 2009 - 07:15pm PT
|
I'm with Dan.
Its in my library, but it seems like he's trying to milk what should be a magazine article into a book.
Too much repetition, bored me to tears.
|
|
paganmonkeyboy
climber
mars...it's near nevada...
|
|
|
Jan 16, 2009 - 09:12pm PT
|
ok - i haven't read the book, but the underlying problem i see with any sort of economic modeling is how do you model the greed or altruism of the uber-rich ? the actions of the few set the tone for many, one person or corporation can skew the whole thing...
think about it - how can you account for the greed and corruption at enron in a model ? wall street today and the *cough blowme* bailout? i'm not sure you can...
|
|
TradIsGood
Chalkless climber
the Gunks end of the country
|
|
|
Jan 16, 2009 - 10:44pm PT
|
Ed,
OK, respectfully I would not suggest that my meager understanding from years ago of relativistic quantum electrodynamics would match your complete mastery of the field, experimental or theoretical.
Even if I could, off the top of my head, toss out the equations and even solve a few simple examples today, I would not claim to have the same knowledge as somebody who did this kind of stuff continuously for years.
If you think about it for a few minutes, you are smart enough to know that if the measurement is not Gaussian, the variation of the measurement can not be Gaussian either. (If price is not Gaussian, the variation of price can't be Gaussian.)
You are right about the extrapolation of the orbit. That is exactly my point. There is an uncertainty of position and momentum of the asteroid, the planets and other bodies involved, as well as the solar radiation.
My point with that example of the asteroid is that the uncertainties are known with respect to positions and momentums of the various masses fairly well, the solar flux variations are unpredictable, but the variation (historically) is somewhat known. From this, we think we can extrapolate forward only 27 years and estimate to the hundredth or thousandth of a percent the probability of a collision with the earth.
This is about as perfect (simple) as science gets. We are in agreement that economics is not predictable in the same way.
Hopefully some day you will realize that not only are the physical uncertainties important to atmospheric physics, but for long term atmospheric physics, unknowables are, including politics, but also including the unknowables like whether fusion will ever happen and when, and whether batteries will ever be rechargable in less than 5 minutes.
:-)
|
|
Ed Hartouni
Trad climber
Livermore, CA
|
|
|
Jan 17, 2009 - 03:34am PT
|
you are smart enough to know that if the measurement is not Gaussian, the variation of the measurement can not be Gaussian either. (If price is not Gaussian, the variation of price can't be Gaussian.)
this is not true at all. If your model of the world predicts the "average" price at some time t, for a set of parameters, p we can write the average price P as:
<P> = <P(t;p)>
that is, we have a model of the average price. Now if the fluctuation of the price is distributed in a random fashion with a standard deviation σ, then the probability that the price at some time t is Q would be given by the Gaussian probability:
℘ = 1/√(2πσ²) x exp[(Q(t)-P(t;p))²/(2σ²)]
the variance of the price is not related to the value of the price... but to the underlying mechanism that drives the variance. In this case we constructed it as a Gaussian, even though P(t;p) can be something quite different.
The problem is, P(t;p) has no "fundamental" physical origin in economics. In atmospheric science, or in celestial mechanics, the equivalent of this value DOES have an underlying physical theory. The degree to which unpredictable events like the solar activity can effect the prediction of the asteroid's trajectory can be used to set an uncertainty in the position of the asteroid at some time t, we would usually refer to such an uncertainty as a "systematic" uncertainty to distinguish it from a "random" uncertainty due to stocastic processes, or finite measurement, etc.
Because the effect of the solar activity on the asteroid is calculable, if we knew what the sun was going to do we could calculate the change in the asteroid's path. Because we don't know exactly, we can form a bound on the paths which represent possible paths, given some model of solar activity... maybe taking the historic minimum and maximum activity for the time period we are projecting the asteroid. Maybe something else. But because the effects are knowable, we can estimate our uncertainty.
if you don't know the underlying physical processes, you have no way of calculating the effects. That is state of the economic models, as far as I understand them.
|
|
TradIsGood
Chalkless climber
the Gunks end of the country
|
|
|
Jan 17, 2009 - 10:06am PT
|
When the price hits zero, the fluctuation can only go in one direction - it is not symmetric. As the price approaches zero, the randomness is similarly constrained - still not symmetric, and so forth.
The fluctuation can't be random!
|
|
Ed Hartouni
Trad climber
Livermore, CA
|
|
|
Jan 17, 2009 - 11:02am PT
|
I agree with you completely TiG, when the price is so low that the Gaussian pdf doesn't make sense, you have to find the correct pdf to describe the variation of the price. If you don't know what the price depends on, your done, you can't... you might measure it a lot of times and then come up with something...
So if the fluctuations were driven by a very few, discrete, independent events, we would describe the pdf as a Poisson distribution, which is asymmetric and has a finite probability of having a value of zero. That might be the correct pdf...
But the economic models, as they were written, failed long before the pdf became an issue.
|
|
Karl Baba
Trad climber
Yosemite, Ca
|
|
|
Jan 17, 2009 - 12:37pm PT
|
"When the price hits zero, the fluctuation can only go in one direction - it is not symmetric. As the price approaches zero, the randomness is similarly constrained - still not symmetric, and so forth. "
I dunno but these days, it seems when some prices hit zero, some subsidy or bailout is pumped in and for practical purposes you have a negative number
I know that's probably outside our models but the fact that economic models keep failing so miserably means they deserve some skepticism.
What do people say? Economics is the business version of Religion.
Peace
Karl
|
|
WoodySt
Trad climber
Riverside
|
|
|
Topic Author's Reply - Jan 17, 2009 - 01:55pm PT
|
Beware "stagflation". It may be on the way, and we'll all die. Everywhere I look, darkness looms. I think it's time to hide in the closet.
|
|
Redwood
Gym climber
West Sacramento CA
|
|
|
Jan 17, 2009 - 02:17pm PT
|
There might be a significant difference between modeling in the areas of physics and chemistry on the one hand and economics on the other. Physics and chemistry are pretty far away from what might be called a political power nexus. Whereas economics is intimately related to it. That means if anyone were interested in maintaining or directing the course of power -- particularly by means of deception -- that economic theories and models (especially those presented to the outside) might be tainted with distortions necessary for the control effort.
Everybody who considers these affairs hopes they're wrong these days, it seems. I hope I'm wrong, too; but I believe in the collapse model. More: we are at the end of an age. When the asteroid smacked into the earth way back when, causing the extinction of the dinosaurs, that was the end of the age (the Mesozoic). This is going to be like that, or even more so.
|
|
jstan
climber
|
|
|
Jan 17, 2009 - 03:45pm PT
|
I have long thought economics might be modeled as are phase changes. A mixture of statistical mechanics and thermodynamics. It is all driven by perception and that is what has to be modeled. Collective psychology. But it is all going to be fast from here and there is no time for proving out a model.
Right now I am seeing the first questioning as to whether the resources of the Federal Government may become exhausted. It looks like we need to start hedging that eventuality. California is well along that road right now.
Unemployment continues to be the benchmark IMO.
|
|
|
SuperTopo on the Web
|
Let us know!