Slashdot Mirror
Why Economic Models Are Always Wrong
mayberry42 writes "Did you ever wonder how and why professional economists often seem to get it wrong in terms of predicting consequences or policies accurately (or even at all)? Or how very few even saw the current economic collapse? This article provides an interesting, if obvious, reason as to why economic models are effectively always wrong."
Most economic models are based on "how we would like people to act" rather than "how people actually act". Much of the time, the model works, but they fail when people act in irrational ways.
Simples.
"She's furniture with a pulse"
I believe these two words effectively summarize the article. Either simplify the models to decrease the number of parameters, or collect more data. There is also the fact that the models are probably time varying, which makes data collection much much more difficult of course.
I wonder if that could apply.
... is not a science. The legal structure of money, the way prices work in a one way fashion, and private ownershp are all political all the way through. Now this may piss off Americans but there are alternative ways to organize society whether they like it or not. Human beings tend to be people of their era and they often have a profound lack of imagination, the black and white right/left thinking I see from people already disqualifies them for not even having the courage to analyze or think about the structures and societies in which they find themselves, the false notion that it is either THIS/THAT, BLACK/WHITE is having given up critical thinking and analysis for good.
I think the article is missing another key factor - the fact that people abuse the system to their benefit.
Starbucks, Harbuckle of Breath.
and I saw the current economic collapse coming.
Sometime back in 2006 (2005?) I read that the average american had spent more than they earned (i.e. borrowed on credit cards/etc.)
I then thought to myself: "self, start saving now because there is a storm coming."
Another indicator: housing prices were going up and up and up and all my friends were trying to convince me that buying a house is an investment (because that's what they were told) and that I should get a good one because I'd be able to turn it around for a profit. Since when is a house supposed to be an investment and not just a place to live in?
So small changes in inputs can produce big, unpredictable changes in the output of complex systems? It's almost as if a butterfly flapping its wings could affect the weather!
They should find a snappy name for this marvelous discovery. Something like "chaos theory".
Confucius say, "Find worm in apple - bad. Find half a worm - worse."
Many, many, many people saw the economic collapse.
I was reading plenty of blogs on the housing bubble, housingpanic.com, et etc, describing the preposterousness of "liar loans", subprime this, and idiocy that, and the crazy valuations.
The New York Times even had a plot of the inflation-adjusted Case-Schiller price index which was enormously above any prior peak. During 2006 and 2007 and 2008.
The notion that "nobody" saw it is simply propagandistic truthiness baloney. I personally didn't profit, because I was much too early shorting the mortgage companies & home builders and got stopped out---the bubble was too powerful.
The real crime is that a small number of very powerful people had an exceptionally lucrative interest in NOT stopping it, because they were getting ginormous paychecks from the continuation of the bubble. And now the notion that nobody could see it is used as excuses for the powerful to excuse themselves from responsibility from fraud and crime.
Down in the guts of banks, there were both risk modeling quants in the fancy banks, and the traditional "ladies with a bun" in the retail banks who processed the paperwork who saw how much outright fraud and insanity there was. Their jobs were threatened when they attempted to speak up and stop the madness, because the business side executives were making shitloads of shekels on volume.
Few people saw the collapse coming? Really?
All you had to do was turn on some form of broadcast radio after about 1995 and listen for a little while. When the commercial break appeared you heard one mortgage mill after another hawking refis, credit lines, etc. Bad credit? No credit? No problem! Interest only mortgage. Balloon mortgage. Jumbo mortgage!
This went on for years and years.
I saw it coming. If you missed it you're a fool. Maybe we just have a lot of fools.
Even if everyone acted rationally, you would then have the instability which is generated because all of these rational people would then change their behavior based on ... the model. It's unclear, and in my eyes rather unlikely, that a "fixed point" exists where all of these rational people start behaving identically and predictably.
The unpredictability doesn't only come out of irrationality. If you look at game theory, you see that many optimal (i.e., rational) strategies are "mixed" strategies where the rational party necessarily behaves probabilistically, not deterministically.
Models aren't equal to models, and even rough models of chaotic phenomena can be very useful and predictive, if they are the right ones. Read this for some acknowledgement of which brand of economics has been right during the last few years. Here is another account, including some pointers to predictions of the current crisis reaching as far back as 1999. Krugman even has a "model" of how good models get out of fashion.
Economics suffers from the manipulation by political interests, and by the wish of many practitioners to project their moral ideals onto the world. Many economists simply go and try to prove that the world works however they want it to work, and find funding for that from rich supporters. That makes the endeavour biased.
The problem is, that once it's known what's going to happen, the arbitage people will make it happen sooner, and sooner, until it becomes faster to predict again.
It's like if next weeks lottery numbers were printed in a newpaper, everyone buys tickets with those same numbers, so that a $1 ticket wins a one two-millionth share of a million dollars...
I got horrible marks in my University Economics classes. I grasp the basics, but not the various approaches that are espoused by the "experts" in the field.
But it's pretty intuitive to me that allowing the economy of the world to be impacted by the greed of a few is bad for everyone. Especially when those US economic interests are interfering with the policies, government, and economies of foreign nations.
I'll leave it at that, because otherwise I'll get slammed as an anti-American ranter instead of a proud Canadian who is affected by US policy, but unable to impact it.
I do not fail; I succeed at finding out what does not work.
The basic theories of economics slavishly followed by bankers and governments that rule all our lives are based on patently wrong assumptions and simplifications, compounded by bad maths. Professor Steve Keen's book 'Debunking Economics' explains all. You don't know whether to laugh or cry when the stupidities are revealed one by one as the book progresses. We need a major change to the system to escape from it. (Debunking Economics - The Naked Emperor Dethroned, Steve Keen, Zed books Ltd)
I recall an article in the Econonist a few years back that described a time when a macro economist visited his chum, who worked on a trading desk in a large bank. The economnist basically came away saying that there's simply no time or space for elegant theories in anything that went on in that environment. The science of economics was more applicable to fly fishng than high frequency trading. But I think the real issue for economics is that it has historically been very prescriptive - what people should do - rather than descriptive - explaining what people actually do. This was forcefully highlighted by the psychologist Kahneman, who went on to win the nobel prize for economics. That said, I enjoy economics and do feel it offers a lot to the world. I also think the premise that economic theory didn't predict the economic collapse is wrong. My limited reading of economic theory left me with the impression something was going wrong (in particular Shleifer, A (2000). Inefficient Markets: An Introduction to Behavioral Finance). But any prediction using economic theory presumes we have access to the evidence required to make accurate predictions. What we are slowly seeing is that, in their rush for profit, institutions effectively hid their exposure to risk. This duplicity, more than anything, compounded by a healthy dose of herd behaviour, is the reason the approaching precipice was obscured.
In a religion, you just tell people what is the Truth. In science, you try to observe and learn.
The models are self fulfilling prophecies.
The high priests of the Economy tell us the Truth. The lower priests spread the word. And the people believe. Without the belief of the people, the system would instantly collapse. And if reality turns out differently, then they/we just invent a New Truth.
I mean, is it really necessary to give trillions of euros/dollars to banks to bail them out? In which pockets is that money disappearing? The bailouts are presented as "The Only Way"... but nobody actually knows.
From the article:
Wait ... you are saying the growing number on my bath scale isn't because the constant of gravity is growing? :-)
The Tao of math: The numbers you can count are not the real numbers.
The reason for the failure of models is much simpler: economics is not physics!
If you find the perfect model and the perfect set of parameters today and use it, you change reality - thus markets will see that you're making loads of money, react and anticipate your knowledge, which renders it useless so that you'll need a new model that takes your old model plus the new reality into consideration.
However, finance and economics are still useful as they help to understand reality better and improve allocation of ressources and information, although we did not achieve perfect markets yet. If we would want a constantly growing global economy without volatility, we would need perfect information flow so that every agent in the market can anticipate the future. Anyway - until we can predict every earthquake, every storm and even the day of death of every single goat perfectly, we'll have financial volatility - so we gotta learn how to deal with it.
Now how to deal with the fact that our world is dynamic?
We need insurances. The current financial crisis, for instance, hit people who lost their jobs and don't have an unemployment insurance much worse than capital owners who lost a far higher share of their fortunes. For instance, if you had 10 millions in 2007 and now have 2 millions, you probably won't have to change your lifestyle as much as a father of a family who had 10000 dollars on his bank account and now has 8000 but lost his job and can't find a new one.
The perfect solution, however, would be to make labour markets more efficient: companies should be able to reduce worker's salaries easily and workers should be able to buy an insurance against having their salaries reduced. Thus, we shouldn't try to change financial markets, as we can't - we should improve labour market to reduce suffering.
From a more practical point of view, this is how we (an egineering company) dealt with the financial crisis: ;)
When we didn't get new contracts anymore due to investment stops all around the world, we just finished our ongoing projects a little slower and asked our staff to finally take their well-deserved holidays in their holiday accounts, which they accumulated over several years because we've been really busy and workers had to work extra hours as we couldn't find enough qualified engineers.
Now as the crisis is easing, we got the same efficient team we've had before, but have everything in our office tided up and a far more relaxed atmoshpere because our workers finally could do what they never had time for before. So although we certainly didn't benefit from the crisis financially, we benefited hugely in terms of living quality. And of course, when engineers are bored, they start inventing stuff - so we can now offer much better products than before the crisis, which will also pay off
Sigh, misleading thread name which makes basic error of confusing finance with economics. This may be news to some people but they are not the same - you only have to talk to a Financial Quant Analysts for around ten seconds before this becomes immediately apparent (they are basically statisticians)
Anyhow on the main points:
1) Lots of economists were pointing out the increased risks in the global economy prior to the financial crash, they were ignored all called 'doomsayers'. In part because it was in the interest of vested interests to do so, but also because it is now clear that actually some of those interests (e.g. the Banks) really didn't know what they were doing.
2) Nobody predicted exactly when it was going to happen because you can't predict the kind of confluence of several events which trigger that kind of crisis. Contrary to what some financial experts were saying this was not a one in a million event, as ultimately they lost sight of the difference between systematic and non-systematic risk.
3) In part this is because economics handles confidence effects poorly, a lot of models try to pin things down to fundamentals on the assumption that eventually they will act as an 'attractor' and drag the econony back on track.
4) Economics also handles financial markets poorly, partly because of the confidence issue (above) but also because of how wealth and confidence effects effect behaviour in asset markets.
e.g. you buy a house, if everyone buys houses their value starts going up due to scarcity, this makes homeowners more wealthy on paper, so they upgrade to a bigger house or spend their wealth on other things, add in speculation in the housing market and you can see how this cycle can go on and on. But, confidence effects can inflate this well beyond fundamental values which means if something serves to kock that confidence (say a recession) it can go into reverse. So this combination of factors acts like an amplifier in either direction, which is obviously not good for market stabilty, particularly when factors like confidence can shift very rapidly.
The main problem with most commentary on economics is that it comes from pundits or half experts who have an interest in taking a position one way or another (typcially working back from the conclusion they want and choosing their model to get there).
the article suggests that our financial woes are caused by miscalibration of bank and stockmarket software models. I submit that the system itself is flawed. You can't make a working model of a broken system. The idea that if we could find a better software solution for banks all our financial problems would end is absurd.
So now stuff done in an introduction to numerics class is news.
Not much known to the general population, there are problems that cannot be calculated numerically, usually because a change of input magnifies immensely on the output. So an error of 1 on input becomes an error of 10^n (with n in the two digit range and bigger). The issue here is that basically by definition all numerical systems used to calculate in a computer have builtin error sources, and errors do accumulate.
The pain becomes even bigger if you consider that in floating point numerics basic mathematical laws do not apply, e.g. (a + b) + c = a + (b + c) is true in mathematical sense, but is in general untrue for floats. Using fixed point arithmetics, while in theory better, implies a rigorous error analysis (floating point is kind the "automatic" solution to the required analysis for fixed point arithmetics), which is especially in such "fantasy" models hard to do. (For many coefficients the "designer" of the model has no idea what the range of valid values might be. If you do not know the value range, you basically are forced to floating point, implicitly accepting that you have no real idea about the error to be expected, beyond the general knowledge of the used floating point system.)
1) economic model are used to represent a steady state and can't represent well, or even AT ALL, rare outliers event.
2) they are worth shit anyway as the "perpetual growth" model can only fail. You can't have perpetual growth with finite resources.
of Economics tries to take the veneer of science by using a lot of mathematics. But this is not good. With powerful enough mathematics you can make almost any story you please fit your historical data. And there is certainly plenty of motive to do just that. Economics is rarely based on experiment. Granted there might be some psychological experiments that can inform economics, but most economics isn't based on that.
Democracy Now! - your daily, uncensored, corporate-free
The phenomenon this guy has observed is nothing to do with chaos theory, as several posters think, but rather to do with error propagation and model uncertainty. This is an issue whether the model is chaotic or not. His mistake is to think that calibration has to choose a single set of parameters, and then one has to make a single prediction from the model. Statistical methods can take into account many sources of uncertainty, including the range of parameters that could have produced the original data and intrinsic stochasticity in the model. The best way to do this is using Bayesian techniques.
You're still limited by how realistic your model is, and this is likely to be the real problem with economic models. However, Carter's argument (that it's fundamentally impossible to fit a model to itself and then make consistent predictions) is wrong.
http://www.debtdeflation.com/blogs/2011/10/27/george-monbiot-seminar/
comment first, facts later. http://chem.tufts.edu/AnswersInScience/RelativityofWrong.htm
"Or how very few even saw the current economic collapse"
Y'know, there's an entire school of economics that predicted the collapse. And the collapse before it and the ones before that. It's called the Austrian school. But even though they predicted every single damned collapse because they didn't use shiny models and after the mid 90's shiny powerpoints nobody pays any attention to them.
It's not that the model's are wrong, it's that they're based on a theory that has proven time and time again that it doesn't work. Have a read of these two articles and guess which one we're currently using.
http://en.wikipedia.org/wiki/Keynesian_economics
http://en.wikipedia.org/wiki/Austrian_School
And now try to guess which has correctly predicted ALL of the previous economic disasters (and the even larger one looming next year).
The failure of the former model comes from trying to mathematically model human actions themselves which are inherently unpredictable.
But people don't want good models. They want models that predict massive proffits. Doubly so when they're paid on commision and it's someone else's money.
404: sig not found.
"Remember that all models are wrong; the practical question is how wrong do they have to be to not be useful." (George E.P. Box and Norman R. Draper, Empirical Model-Building and Response Surfaces (1987), p. 74)
"One of the most insidious and nefarious properties of scientific models is their tendency to take over, and sometimes supplant, reality." (Erwin Chargaff)
I think that says it all, really.
--Bud
The Keynesian system of economics fails to model reality well because it's seriously flawed. The Austrian school of economics, which is similar to what was used in the United States pre-Keynes and whose economists did accurately predict the current economic crisis, is an entirely different matter. Look at Peter Schiff, Ron Paul, and a number of others -- they have been debunking Keynes for years. They've also been pointing out that the artificial "stimuation" of spending and the idea that your house is an investment rather than a liability or at best an item whose value is controlled by supply and demand and with the retirement of the baby boomers was bound to experience a slump in demand. With the addition of government interference in backing unrealistic loans for those who couldn't afford them, the writing was on the wall. We need to ditch the idea of a "centrally planned" economy and Keynesian economics generally if we want to have any kind of realistic understanding of how markets really work, and before we shoot ourselves in the foot yet again.
NetShadow
Just a clue, the "alternative ways to organize society" always fail when they scale up. Its like people claiming Sweden is a model, but then fail to note that Sweden is about 1/50th the size of the US in economy and even smaller in terms of population.
Or to put it another way... I am the head of my family. I provide food, shelter, education, money, cars, and whatever they want all in exchange for love & obedience (at least from the children).
And it works great, everybody gets what they want, when there is disagreement we work it out over dinner, and when I make a decision for the good of the family, everyone accepts that. I decided what religion we all follow, what personal habits they can have (smoking is not tolerated, not working is not tolerated, etc. with punishments meted out when they don't fit the established rules)
That doesn't scale very well.
Its the same for nations. A small nation with a homogeneous population can be governed much differently than a huge population with wide ranging opinions, backgrounds, beliefs, creeds, and temperaments.
Its why you can't compare China to any other nation. It why you can't compare the U.S. to Finland or Germany.
They didn't use economic science! No way is this stuff due to any real economists input.
The purpose of existence is to make money.
That man knows nothing about economic modeling. His whole story about "calibrating the model" is just pure and utter bullshit - so much it makes my head hurt to read that. Sure, someone trying to model who knows nothing about it might try to force the model to fit the data, but that's not how actual Economists do it - you'd get laughed out of grad school if you tried the things he mentioned in his articles in a research paper. I'm currently finishing up my Masters in Applied Economics and do quite a bit of modeling on a regular basis. There is no "calibration" - merely statistics. You include all the variables you consider relevant and then start whittling out the ones that are statistically insignificant in explaining the variation in your dependent variable. When dealing with forecasting, you never use the full time series data set in your model - you use most of it and then leave part of it for testing the accuracy of your model (so you can compare the forecast values with the actual values). If you've done a good job collecting a large enough data set and including the necessary variables, you'll have some pretty damn good predictions for the first part of your time series. Obviously, like with any type of prediction, the farther into the future you try to forecast, the less accurate you'll be. Hence why you continually gather more data and further refine your model and keep redoing the predictions.
The other wonderfully fallacious thing that he had in his article was not pointing out that Finance is NOT Economics. Financial data, such as the stock market, is VERY hard to predict (and actually due to basic financial theory regarding market efficiency, if the market is efficient then you should NOT be able to predict stock prices) due to the obscene amount of variables involved and the fact that there are decisions made based off emotion and not purely mathematical logic (such as bank runs and panics in the stock market).
"The tree of liberty must be refreshed from time to time with the blood of patriots and tyrants." ~Thomas Jefferson
...and Peter Drucker observed or rather stated the obvious years ago: One can't really compare models in physics with models in economics, though it's tempting. The problem is, the model or a theory that tries to explain the real world beaviour will be applied in the real world which will in turn influence the real world system, which will eventually adapt, rendering the initial observations (that led to the theory in the first place) irrelevant for future explanations. For example: Every theroy we build on which parameters influence inflation will eventually influence economic behaviour by tuning monetary and fiscal policies according to theory. Market participants will eventually accomodate and alter their inflation expectations and in turn economic activity. Compare that to a model in physics: No matter what the explanation we find for the real world system, that theory will not influence the system. It doesn't matter to the system whether we think Newtonian physics or relativity is correct.
So, every economic model or theory will eventually become wrong. Econophysics, statistical mechanics and complex network theory may be the key to unlocking economic science and taking it a step further (from crystal ball gazing)....let me dream....
I feel so sig.
All money is debt; money is created from debt. It is created out of thin air when a loan or credit is taken. It does not exist outright in the money supply until the debtor makes payments plus interest to the creditor. Furthermore, it is impossible to repay all debts in existence because the money for the interest payments does not exist in circulation. Economists throw more debt at debt expecting it to make a difference, when all we get is the inflation and we pay for it with higher prices and our savings lose value. Where is the sense here?
It is only a matter of time before our monetary systems collapse entirely. I believe we are seeing this right now. Let's apply science to our social systems and implement a resource based economy for the good of all mankind! See The Venus Project, I think you slashdotters will love the concept.
I work with groundwater models, and (at least where I work) we do not consider the models to be predictive. Rather we hope they will give us a better idea of what is happing with a particular site. We combine this with "trend" analysis and real world data to get an idea of what's actually going on with a site. We do not believe we know what's going to happen based on the model, but we hope that the model will tell us what direction things are headed so we can prepare for it.
The landlord makes money either because (a) he inherited it and cannot find a better return on investment, or (b) the system is fixed to keep his interest artificially low, whereas the renter does not have the same access to cheap credit and so would have to pay more on a mortgage.
From scarped cliff or quarried stone she cries "A thousand types are gone, I care for nothing, no not one."
Many, many, many people saw the economic collapse.
A newsletter from an economics professor and CNBC financial commentator: ...
Any talking head who tells you that this market is a buying opportunity has his/her head screwed on backwards. The only buys are the kind of value plays that the likes of Buffett are pulling off. That is, it is very much a stock picker’s market.
Recession plus inflation plus a credit crisis plus a softening European economy plus an inflation-plagued Chinese economy plus Russian strong-arming in natural gas plus two leading presidential candidates who are ignoramuses on economics plus a rising long bond in the face of Fed rate cuts does not a bull market make."
http://www.peternavarro.com/2008.02.01_arch.html
"Thursday, February 28, 2008
That is his oldest newsletter but I understand he was telling his economics students to "get out" of the market in fall 2007. Plus he was showing them a whole bunch of historical indicators that were all pointing in the wrong direction.
i) ?????
ii) ????
iii) Free market solves all known problems
iv) Profit
Ugg boots hot outlet has set off a craze all over the globe these days, when the winter come, we have to face the cold conditions outside, the chilly outside has stopped most people’s step, with the ugg boots discount outlet , you can enjoy your playing as you want, the most warm and fashion uggs on clearance has attract most young and adults attention, do you want to own the stylish ugg clearance sale boots ?
instead of working ones. They derive every behavior from one single model. Not even physics can do that.
Since I already posted, I'll just have to "tip my hat" to your succinct and slightly sardonic reply by posting (which means something, since I almost never post "me too" comments).
Hope some of the late-comer moderators mod you up!
Of course they are wrong!
Models are made by mathematicians. Those scientists use a huge pile of data and crunch them using powerful computers. In certain degree, the model must be simplified in order to be "crunchable", even on modern computers. The results are presented to high-level economists, the people who know hot to manipulate other people and money, not data. They don't understand why the results are just the way they are. So, they will either rely on the results, or will reject them. Or even worse, the will insist on further simplification in order to get full awareness of the background process. Relying on results is risky yet it may bring some benefit. Discarding results is even more risky because it imply using old models which are usually less reliable than the old ones. Further simplification is the worst solution - simplified models are generally unreliable, depending on the simplification itself (simplification means a lot of approximations, assumptions and rejections - you have a great number of approximate-assume-reject combinations, especially on large models).
So, many economists and many choices. Some choose high-quality model acceptance and gain benefits - you see them as CEOs of large successful companies in economic collapse. Other guys are desperate managers pointing at wrong models. Of course they are wrong! You can't run the process without comprehension. Even then, it is hard to assume every little perturbation in such a complex system. Some will say that economics is not mathematics. But, the global economy is a dynamical process which means that it IS a complex entity which requires solid mathematical expression. So, understand or die. Obviously, many choose to die.
The models that are so continually wrong are those with an ideological bias. The models which have been consistently correct are ignored because they're out of favor.
Why Bad Writing Easily Gets Published
Why Logical Reasoning is so Hard to Find in Today's News
Read an Entire Article that Provides No Supporting Examples
Why I can't get a job writing articles for a decent newpaper and pay off my student loans (I got tired of capitalising).
Horrible article. I honestly feel sorry for whomever hired this guy to write for them.
That's politics, not economics. You want room 12A, just along the corridor.
Confucius say, "Find worm in apple - bad. Find half a worm - worse."
Because the quantity of data required for global warming models dwarfs that of nearly every other predictive modeling humanity has ever attempted...do you think scientists would dare question the validity of claims of "proof" similar to those of economists? I recognized long ago while using Ansys to model the distortion of a piece of metal weighing 2 pounds under simple loading that numerical computer modeling has a scientific value equivalent to that of toilet paper. That is, it could be printed onto a piece of paper and then used to wipe your ass. But beyond that, any claim made with such a model is worthless for scientific purposes.
We should learn what we need to know about issues, before we decide what we need to feel about them.
is talking far more about valuation models, not economic models. Hence the quote from Wilmott, a financial valuation model expert.
As far as economics models go, standard Keynesian macro has done a fine job forecasting this mess. The issue was not forecasting, it was getting anyone to act on the forecast (we Americans live in a democracy, not a technocracy).
Or, more precisely, many people saw that the mortgage-backed securities fiasco was going to unfold, and that getting back out of that hole after it did was going to be difficult. I mean, seriously, anyone would tell you from the last couple of centuries of financial experience that offering home loans to people with no income, no jobs, and no assets was fine (if stupid) if you did it a couple of times, but not if you started doing it systematically and based the future of your whole company on those loans. Likewise for subprime mortgages. Likewise, anyone with half a financial brain would tell you that, no, although housing would usually go up in price over time, it wasn't a guarantee. It was all financial garbage, sustained for years in a "false economy" based on assets and money that didn't really exist. It was inevitable that it was going to come crashing down hard.
The problem is, financial businesses were too busy making money off the bad loans and other dubious financial instruments derived from them to care about the likelihood that they could destroy the economy, they were busily hiding the risks of things they knew were bogus, and they were still making money by shorting stocks even as the crash itself unfolded. They knew damn well what they were doing was risky, but the money was good and thus they didn't care. The regulators were asleep at the wheel the whole time, were rolling back banking regulations, and thus didn't stop them.
Don't confuse lack of care for lack of foresight.
The problem, of course, is that while these different versions of the model might all match the historical data, they would in general generate different predictions going forward
If these morons knew a little more about data mining techniques and in special, if they weren't so greedy, maybe we all wouldn't be facing this global economic mess right now..
Others have commented on the economics, I can add a little about the modelling.
The problems with the article are numerous - if the original researcher had know more about modelling he wouldn't have thought any of this was worthy of comment.
For a start, in parameterising a model you need to choose a paramterisation which is parsimonius with respect to the amount of data and the level of noise in that data. Otherwise the parameters will be overfit and the results meaningless. A careful researcher will use statistical techniques such as cross validation (leaving out some of the data when determining the parameters to see how well the model predicts the omitted data - statictical equivalent of a double-blind). Similarly sensitivity analysis can tell you about how well your parameters are determined, or alternatively the sensitivity of the parameters to the data can often be determined analytically. Whether the response of the model is linear or non-linear is also relevant.
1. Bet against the economic models.
2. ???
3. Profit!
I'm not a lawyer, but I play one on the Internet. Blog
But that if someone important doesn't like what the model says, then they'll go with the answer of a model they DO like the answer to.
For example, see the economists who saw in their models that the deregulation of banks and their betting on naked swaps would lead to economic catastrophe. These models were right, but ignored by the bankers, their backers and the politicians who were and or have rich friends who liked a different answer.
They analyze data (past events) and try to determine a pattern or discern a formula out of it.
It's like having a historian predict the future of a nation - a psychic with a degree, if you will. And historians are not that, nor do they intend to become one.
Speaking of which...
You haven't seen anything yet. These criminals which run everything now, are not happy with just owning everything thing, they want control. To do this, they will collapse the economy and then call for even _MORE_ centralized control.
When I say criminals I mean:
Lloyd Blankfein
Larry Summers
Ben Bernanke
The Federal Reserve Board
(I would list the owners, but the list is _CLASSIFIED_ by the US Military and State Department....and the Federal Reserve isn't even a government institution!)
They create the crisis, then they propose the solution. When this collapse happens, decentralization will be the only way out.
We must reverse all forms of Globalism and excessive centralization of power, otherwise things will get much much worse.
Finally we see the Catholic Church's real face, as they call for a One World Bank, where only one financial entity controls all the rules of commerce, all buying and selling and trade world wide. Right out of the book of the Apocalypse, Jesus Christ.
If these Federal Reserve/Wall Street who are planning this, successfully reform a new even more centralized financial system, humanity will be plunged into a new dark age.
With a single formed entity that all nations pay to in the world, they will be all but unstoppable. They will have the ability to raise their own private Army, and every nation on the earth will be shackled under a tyranny that has never before been seen.
This level of centralization of power will breed forms of corruption the world must never see or humanity won't survive.
-Hack
Got Geometrodynamics? Awe, too hard to figure out? Too bad.
and made ~30% per year between 09 and 11.
i.e. It is predictable. I'm currently positioning myself for 12 and beyond. Most of the economists in positions of authority just now are Neo Classical or Keynesian, what this means is they are ignorant, arrogant and have a habit of painting radio dials on rocks and worshiping them[1].
I'm not going to tell you how, I need someone to hold the bag. So if you don't want to be the empty bag holder go figure it out yourselves, I will give you a hint though.
Go and find out exactly what money actually is and what growth actually is.
[1] Cargo cult: http://en.wikipedia.org/wiki/Cargo_cult
Deleted
Allow me to correct you - this is what Economics actually is.
-- "So they told me that using the download page to download something was not something they anticipated." - Bill Gates
unless all its activity is used in the act of prediction (the universe predicts itself in real time, but does nothing else). It will be always difficult for humans to openly predict the behavior of humanity. The reason is simple: if you openly predict accurately the future, the margin to profit from that knowledge is so big, that the market will modify the trend, and your prediction will meaningless. The only way to predict the future with a high degree of accuracy, without influencing it with the prediction you must do two things: - do it secretly, in a small group, so that there is no "market" - do not use your knowledge to alter in a big way your prediction. That is, you can profit from your prediction, but not too much compared to the quantity being predicted. A concrete example: I predict rightly that the price of oil will be 200$ in one month. If I make that prediction open, and the market trusts me, the market will immediately adjust to that prediction. Those market forces will render my prediction completely moot. If, on the other hand, I keep that prediction to myself, I can profit from it as long as my profits are not affecting in a big way the price of oil.
Quite a few people who had good savings still lost jobs, burned through their savings and retirement funds and in the end lost their homes anyway.
The idea that only bad or irresponsible people lost their homes in foreclosure is magical thinking. You can be a responsible person and still get wiped out
during a deep recession.
They never take into account asshats.
Every economic model has rational demand and rational supply, both with equal amounts of information and that everyone acts in his/her own best interests all the time, even over the long term. This is bogus, and has always been bogus.
It's the same reason why pure ideology never works. It's the reason why communism failed and it's the reason why Libertarianism fails. Because asshats.
--
BMO
Is the art of examining the impact of human reasoning (or more often emotion; suchs as fear and greed), on the financial world. By using common historical models, a fair amount of predictive element becomes realized, but it cannot be a definitive model. Human behavior to one series of events will not have the same outcome as a very similar series of events 5 years or even 50 years later. But there is a pattern to it in a general sort of way.
The 'economic' systems are NOT mathematic. Money is a 'perceived' value artifact.
Economic systems are roughtly equilivant to an analog volume control operated by a 2 year old, linked to the clock chip of your (overclocked) home computer.
IT makes for interesting performance figures, or can crash your computer.
With apologies to Scott Adams:
And next week we'll have a doctor with a flashlight show you just where economic models come from.
Go ahead, laugh. It's funny because it's true.
The point the author may be missing here is that in a competitive environment such as financial models, the goal is not necessarily to model the market but to make a model better than everyone else's model. Consider sports, where you have a "spread". If I can model a sport and generate a better spread than the one being generated by the Vegas model, I win. If I can't, I lose. Stocks are very much the same thing, where the price of a company represents the "spread" of future earnings and I need a model that is better than everyone else's model in order to win $$$. As Einstein might have said, the quality of your model is relative to what you are comparing it to. If you compare it to perfection (actual data), surely any model will be less than perfect. But compared to OTHER models, it may be excellent. The article doesn't really quantitize what is meant by "really bad predictions" going forward so its hard to judge his findings anyway. If its wrong more than half the time, that's probably a bad model. Was it? Or is he calling 75% "really bad"?
So the guy 'discovered' something that is taught in any introductory numerical methods course. Yippee!
If the perfect description of reality could be written down in an equation but that equation involved hundreds of parameters that could not be exogenously measured but must be 'fit' in some manner - least squares regression, method of moments, maximum likelihood or any number of other techniques for CURVE / MODEL fitting - then determining the model parameters, which reflect the perfect description will be difficult and stability of the parameter estimation will be an issue.
There is nothing novel or even interesting about this 'result'.
There is no spoon.
My first Journal Entry ever, in 8 years! http://slashdot.org/journal/365947/aphelion-scifi-fantasy-horror-poetry-webzine
In the recent PNAS paper http://bit.ly/uI1nxG one can read that prevailing economic models of credit risk assume that price fluctuations form a bell-shaped curve, with very large fluctuations essentially never occurring. But during financial crises, wild fluctuations occur more frequently than these models predict. Authors developed a method to incorporate these fluctuations in their analysis of financial data from 488 publicly traded manufacturing firms for each quarter from 2000–2009. The researchers used multiple types of known calculations to analyze financial data such as the ratio of working capital to total assets, and sales divided by total assets. These data were plugged into multiple ratio calculations to estimate credit risk for the companies. Particular attention was paid to the years 2007–2009, a time of overall financial crisis. According to the authors, the results suggest that even during stock market crashes, the basic dynamics that underlie less volatile periods still govern credit risk. The study revealed that credit risk follows slowly decaying functional form, implying that dangerous credit positions are more likely than is commonly believed. According to the authors, the credit rating approach may help improve the estimation of credit risk, particularly in the event that financial services companies respond slowly to changes in corporate credit quality.
The current financial crisis was manufactured, not a poorly modeled prediction.
"If any question why we died, Tell them because our fathers lied."
Most economic models will continue to be wrong because they are predicated on the lie that an economy can continue to grow exponentially, forever. It's not just the models that are wrong - Since it's hard to find any information relating to business and the economy that does not promote continuous growth as being the number one metric of a successful economy, this tenet of free-market capitalism is engrained in the minds of most people. And yet, it is demonstrable using very simple arithmetic, that it is impossible to achieve.
We're collectively fucked unless that belief can be reversed.
To quote William Melvin Hicks
"Let's get this food, water and population thing sorted out first. Then we can go and explore space"
Bill is right, but a few people with too much money and too much control, have too much to lose...
Ron Paul foresaw the collapse way back in 2003, and he's not even an economist (watch the first 1:30 of this video: http://www.youtube.com/watch?v=n6V8N8Um9Q4 ). Economic models are more or less useless, in terms of attempting to enforce one that theorists believe to be the best. But in terms of predicting where we will end up when one is enforced, that seems to be much much easier.
I know the article claims "Calibration--a standard procedure used by all modelers in all fields, including finance--had rendered a perfect model seriously flawed." but obviously our climate models are simple enough that you can't just calibrate the parameters until it matches the data, right?
If motivated financiers can't get their models correct, why do people believe that climate models are anywhere close to reality?
Because Economists only deal with econermic models!
Really? A system of equations - the model - can be non-unique?
That's not news. That's basic linear algebra.
In other news: Economists are interested in and model things other than the macro-economy!
At 11: Optimal jurisdiction size, and then the application of behavioral parameters to game theory.
Most of today's economists believe in Keynesian central planning, not true free market economics and sound money like the Austrian economists. Students of the Austrian School predicted the economic bubble and subsequent collapse because they very well know the obvious causes. RON PAUL 2012!
Where the article contends that this is because of "calibration" issues, it manages to really confuse the issue.
The fact is that model calibration is usually not a sinecure that can be (blindly) entrusted to a software package (as so many practitioners are fond of doing).
There is a large grain of truth in the suggestion that model calibration fails because various sets of parameters can fit the data on which models are calibrated equally well. In order to reliably calibrate such models, one must calibrate submodels (that describe observable phenomena) in isolation. Next those parameters should be kept constant and the other parameters calibrated. That's how university researchers would (usually) do it.
Practitioners (consultants) are usually under constraints of time and budget and will typically face clients who (a) know absolutely nothing about the models they commission and use, (b) know less than nothing about how models ought to be calibrated (c) believe that any calibration that fits the data is OK and (d) will simply pick the very lowest proposal to calibrate their model because they cannot distinguish between a methodologically sound calibration proposal and a trashy one.
As a result, consultants cannot sell a ' proper' calibration when someone else is offering a quick-and-dirty calibration, and give up trying after a few failed proposals. This in turn ensures that models are often calibrated in zero-knowledge mode (i.e. just fit the data and don't think), and hence are open to large errors in cases where such simplistic approaches are inappropriate.
As so often, the market mechanism will ensure optimum (read minimum) pricing for specified objectives and measurable deliverables at the expense of unspecified or unmeasurable ones (such as methodological soundness of model calibration).
And that has little to do with any inherent weakness of models, but a lot with the inherent flaws in the way models are calibrated and used.
What did you do back in your grade 6 science class?
You ran experiments and you learned these two important lessons:
1, you ran the experiment multiple times (repeatability)
2. you try and very only 1 variable at a time
Neither of these is prevalent in any human system. Whether that is economics, politics, or even sociology. This makes predicting anything very hard. You can have 10 PHDs, and you really have as much insight into the economic system as someone with reason.
The great depression happened once. It happened under certain condition. A certain set of policies were tried. We know that those policies under those condition did not solve the problem.
But we have no way of knowing how another set of policies would have reacted. We have no way of knowing how some policies that applied under the situation of the great depression applies to the situation today. It's a human system, ever variable matters. Technology, family size, urban divide, globalization, the media... It's a billion variable equation.
In something that is more of a science, you can at least attempt answers by running experiments over and over and changing variables... you can test out any answers.
You can't do that in economics, because... we are running live.
This is not to say you shouldn't study economics. You should. But it doesn't provide anywhere near the reliability of science and really shouldn't be used heavily by politicians.
Because all economist mentioned are Keynesian economists. Browse around mises.org. Search for articles in 2003-2007, and it is obvious they saw it comming. Here is one notable austrian economist. You would think politicians would be knocking at his door constantly to help them see. If you claim it was a fluke, here is another much more famous guy that follows austrian economy, that predicted every single recession since 83. Heck, you can also predict the next recession, just spend a few hours reading on mises.org, they have courses for free.
An economist is someone who, if you don't know a phone number, will estimate it for you.
Sorry, but gray text on gray background is making my eyes bleed.
It seems like all of the various "economists" do not live in reality where various policies - like not allowing someone who works more to keep more - have a discouraging impact. I know that in the last year I've actually cut back on how much I earn - why? Simple - the government can't take 50+% of my day if I'm not earning anything which is what they do for every dollar I earn. So if I earn nothing, I get to keep the results of whatever I do that day for myself - even if it is sit on my ass and laugh at the OWS crowd.
and an entire field of economics, econometrics, has been trying to tackle this issue for decades.
People did predict the economic Collapse. Ron Paul predicted it all along, as many others did. Those who "failed to predict" it are simply those economists and politicians who stood to gain financially from the whole thing. The issue isn't that it is impossible to predict what is to happen, but rather, that those who are in positions to control and predict are those who stand to gain in the bubbles that end in economic turbulent times [read the Fed, Wall Street, Washington].
The biggest problem with modeling is with the modelers. They don't ever seem to look at what they're doing with a critical eye. Any time that you have a huge parameter set and a limited data set (economics, weather, science) you can fit anything to a model. It basically reduces to fitting a spline to two data points.
I've watched it happen for years in drug discovery research in the pharmaceutical industry. The problem is that the modelers know how to run their programs and the more they know about how to run their programs, the less they have time to understand about the stuff they're trying to model. But the managers love it cause they don't have time to understand either and they like the quantitative aspect of the whole exercise. They can use the models to show how they're gonna be more productive. In the meantime, the whole world goes to crap.
I believe you are all missing the point. Not since Malachi has anyone predicted the future! If an economist could predict the future, there would be a lot of rich economists around. However, even if they cant predict the future doesn't mean they are not valuable. They can "nudge" the future by looking at the past and what works and implementing it.
You cannot predict, to any reasonable level of accuracy, a binary value of "Sunny / Not Sunny" for anything past 3-5 days.
That's despite the fact that we have centuries/millennia of data, hugely complex models, thousands of amateurs and scientists in the field, that there are huge benefits to even the common man in doing so (let alone things like fisheries, farming, etc.), that the question is simple to define and simple to answer once the date draws nearer, etc.
What on earth makes anyone think that ANY mathematical formulae will reasonably predict the actions of millions of individual, self-managing, inter-connected entities that consider themselves to have free-will, can act impulsively and without reason, and interact in a billion times more complex ways than we can ever model?
Economics and forecasts are NOT about predicting the future. That's stupid and impossible. They are about determining what the most likely outcome is and "hedging your bets" that way. Sometimes (in fact, quite often) that will be wrong because you have insufficient data and are only projecting a "most likely" outcome from all of the statistics. That's why we don't let banks "run on empty" and they have to have some insurance, backup, funds, procedures etc. (even if they *aren't* perfect).
Economical models are not mathematical precise. They contain mathematics. They use mathematics. They rely on mathematics. But their inputs and results are chaotic and random and the best we can do to that sort of data is statistics (i.e. making arbitrary things equal to numbers, then guessing what that means based on those numbers).
There is one single change that will have an immediate affect on all this business.
... all the funds in their collection are making money !!! They tell you ... had you invested your pocket money in 1980 ... you would have been a millionaire.
... had you invested in those funds that started 20 years ago, you would have been a millionaire. That is like saying had you bought the lottery ... because that chart behind them is made up with the winning funds. The loosing funds have been removed from the collection that makes the trend, as they are no longer available to invest in today, so why include them in the trend of the funds they are offering that you invest in !! Off course they keep saying past performance is not indicative of future trends :-) ...
... selling insurance my friend as well as fees and charges on their advice (PS: Fees and charges is the same word, they just invented two words to charge more :-)
... if your investment starts turning south, somebody that works for you might advise you to move your money into a safer investment.
... you couldn't make up such a scam (at least I couldn't) :-) even though I am being burnt by it as we speak :-)
KPIs on financial advisers.
If you go to a financial adviser, and pay him money to advise you where to put your money.
You should be able to choose the financial adviser that has made the most profit for his customers.
Well that information is only known to banks. Why is that ? Because even though you pay financial advisers, they actually work for banks and not for you !
Most financial advisers are actually crap, and don't make money for their customers. When you walk in their office, they show you amazing trend graphs, What they don't tell you is that they started with 1000 funds 20 years ago, and slowly retire the bad performing funds, and introduce new ones as time advances. So amazingly
Now be careful how they word it. They say
Where do they make their money you ask
One last detail
Somebody that works for the bank might rather keep the money there to protect the fund.
Guess who the financial adviser protects ?
I love this stuff
If you laid all of the economists in the world end to end, you still wouldn't reach a conclusion. Duh.
Chewbacon
The Bible is like Wikipedia: written by a bunch of people and verifiable by questionable sources.
Hayek criticizes the "scientistic" approach to economics.
http://www.nobelprize.org/nobel_prizes/economics/laureates/1974/hayek-lecture.html
I am curious though. A model for a complex economic system is perceived as untrustworthy. A model for a complex atmospheric system is perceived as trustworthy. I am of the opinion that our knowledge is limited. I also believe as, Hayek states, that not all truths can be reduced to mathematical models.
Regards,
Jason C. Wells
Several people saw the crash coming (Krugman, Stigliz et. al.) the problem was lots of people were getting paid really well to justify what their bosses were doing to make boatloads of cash. It is called control fraud - look it up. I am ashamed scientific american is trying to blur this crystal clear problem.
Did you ever wonder how and why professional economists often seem to get it wrong in terms of predicting consequences or policies accurately (or even at all)?
No, but I have been wondering how and why they get to keep their job after they get it wrong.
The three laws of thermodynamics:(1) You can't win. (2) You can't break even. (3) You can't even quit.
“An economist is an expert who will know tomorrow why the things he predicted yesterday didn't happen today.” -Laurence J. Peter
The model can be fine for the existing data. But the problem with any model comes in extrapolating to data points beyond your current data set. http://en.wikipedia.org/wiki/Extrapolation#Quality_of_extrapolation It's not a surprise that many models extrapolate poorly.
lots of people predicted the economic meltdown. They just weren't very popular because a) Goldman Sachs et al run this country, and they pocketed trillions (20% of our GDP baby) and b) for anyone that mattered it was safe to ignore the cries of impending doom because there was a Republican in the Whitehouse.
So yeah, broad statical analysis can predict economic behavior. It doesn't mean anyone will do anything about it.
Hi! I make Firefox Plug-ins. Check 'em out @ https://addons.mozilla.org/en-US/firefox/addon/youtube-mp3-podcaster/
False. My mother told me and my sister when we were very little that advertisements were all lies. So we watched TV looked at commercials and said "No, that washing powder is not the best, they are lying!" Etc. I remember that, and it worked although I presumably would not have been influenced anyway.
I have never bought any product because of advertising. I have only once thought in a supermarket "Hey I remember that froma commercial, perhaps I shoud try it". All other times when I want to try something new I just go for no-name stuff and/or something that seems interesting. Not because of advertising.
You mentioned Derren Brown, well, his stuff doesn't work well on me either, I've seen his programmes and almost all of it is clear to me, how he influences. It is scary how easily it is to influence people, but I will give you an example:
He tried influencing people by asking say someone in a say to show something then at the same time asking directions or something, and he had for example an expensive watch (IIRC) in his hands, then said 'It is ok' (or similar). The seller assumed the situation was ok, Derren Brown walked out of the store with the expensive watch. Later the guy in the store realised something was wrong.
The interesting thing is, I know that this doesn't work because when someone comes round to buy something I always have this feeling to be really careful and not get distracted. And no, I had never seen Brown nor anything like that before. Also, in the same episode he showed a hot dog salesman, whom he could NOT influence, he wouldn't have this "it's ok" as "Payment was made". I think a hotdog salesman will have seen the bullshit people try so much, that it's impossible to fool him.
So I think that when people get more aware of themselves and the way they are being manipulated, the less they can be manipulated.
For myself, any purchases of devices are made by going to websites to compare specs, prices, experiences of others, and of course my own wishlist of features. I am aware of how people are, thus that for example negative experiences are by nature more prominent because people are disappointed, posititive experiences usually contain little information... Advertising certainly does not work on me...
Parents should do as my mother did, and I'm sure advertising would have to change. That would be nice...
by case and shiller. go look it up, read it, and toss this out.
modeling people is difficult if not impossible, and they dont always listen when economists DO know whats going on
PS: I don't reply to ACs.
The original analysis isn't as relevant as it seems to be. There are major aspects of economic policy that don't depend upon our ability to do a real model. E.g. the current crisis is due to too much debt, both private and public. (That's oversimplifying. Lots of things went wrong.) It's pretty obvious that something bad would happen. No model may have predicted what and when. But there are still some basic principles that guaranteed that we would get in trouble, and there are basic principles that tell us some things about our current situation, although not the specifics of what is going to happen when. These are more like conservation laws in physics than specific models. The fact that energy is conserved tells you that if you dump energy into a system it will go somewhere. You need to know more about the physics to predict exactly where and how. But a lot of economic policy can be done without specific economic models.
When we have a housing market with a total value of $13 Trillion, and a Credit Default Swap market based solely on that housing market, that has a total value of $35 Trillion, and people are still making models that say we're financially sound, the problem is obvious.
They are the ones holding the $22 Trillion of unbacked debt and they're trying to fool eveyone else into buying it off of them before the whole shit storm blows up.
-Rick
"Most people in the U.S. wouldn't know they live in a tyrannical state if it walked up and grabbed their junk." - MyFirs
If all economists were laid end to end, they would not reach a conclusion.
George Bernard Shaw
Back in 2003, in response to a Bush effort to tighten accountability over Fannie Mae and Freddy Mac, Barney Frank said they "are not facing any kind of financial crisis" and complained that people "exaggerate these problems."
So even Bush knew a problem was looming. If even he could see it, then a lot of people must have been able to see it.
I'm not sure that GM is the company that I would trust with any prediction of future needs or supply. Math doesn't seem to be among their stronger skill sets.
So this guys model is not stable in the sense that it's own data can not be used to calibrate it. The natural conclusion is that one should run this test on their own models to determine how robust they are to their calibration procedure. If you can't produce the same model by calibrating it with it's own output, you've got a serious problem. This guy assumes that all models suffer from this, but I'm not convinced by his single data point.
My friends and I are not economists. We predicted consequences in 2004 of policy shifts accurately. Not seeing the economic collapse on the horizon was caused by politicians, C*Os, and economists with their heads up their ass. Most politicians, C*Os, economists ... delusionally love what they see from their perspective and provide their shit to US, EU .... Also, we knew that the Rove-Chaney-Bush plan for private retirement accounts to replace social-security was a scam to pump money into the economy, prevent the pending collapse, and distribute more money into WallStreet, Banks ....
Most US companies (C*Os) pillaged and destroyed the corporate employee retirement funds and then put US on the hook for providing those corporate employees retirements.
Anyway, most C*Os and politicians are criminals or idiots, and almost all walked away with our money and no punishment. We should take C*Os and politicians money, retirements, homes, cars, health coverage, college funds/tuition ... and pay down their fair share of the national debt, because We The People will still be paying the bulk of the national debt bill.
Unaccountable leaders are masters, and unrepresented people are slaves. How do US and EU fare?
Doctors haven't cured cancer
Physicists haven't made a fusion reactor
Brain Surgeons can' t operate on Alzheimers
Rocket Scientists can't fly us to Mars and Back
News at 11
I'll see your hokum and raise you a boondoggle.
A lot of people are holding forth on why economic models are wrong, but few comments are related to the actual subject of the article. (By the way, unless I missed something, the article itself is very vague on what work is done. I think it may be referring to this Jonathan Carter, and the research findings may be related to this 2005 paper.
The article is about the following situation: you have a model (statistical model, computer simulation, etc.) that you want to use for prediction. It has some "knobs" (parameters) that you can twiddle to change its output; this is necessary because the settings of these knobs are often unknown a-priori. So people "tune" or "fit" or "calibrate" the model to observed data to determine the parameter settings in order to make predictions.
A problem occurs if there are many different "knob settings" that cause the model to behave similarly on past observed data. Statisticians call this an "identifiability problem" (since you can't hope to identify the true value of the parameters from the observed data. Ecologists call it "equifinality", since there are many equally good ways to reach the same final outcome. And engineers call it "multimodality", where the fit of the model has many local minima. (Or you could get a whole "ridge" in parameter space that is equally good everywhere along the ridge crest.)
In such circumstances, you can't determine the true values of the parameters very well, even if the model is perfect. This isn't about imperfections in the numerical model, or in the mathematical theory. It's an inherent consequence of the relationship some models have with the data.
This also is not a consequence of imprecise data. There is always some uncertainty about model parameters given noisy data, so you'll never determine the true value of parameters exactly. But this isn't what it means to be non-identifiable.
An example of the real problem of non-identifiability: suppose your model is y = (A+B) * x + error. It's pretty clear that if you measure y and x, all you can hope to determine is the linear combination A+B, and not A or B individually, even if you have perfect data. (That is, unless you have some additional source of information to constrain their values other than y and x.)
The above is a case of perfect non-identifiability. Other models are just "nearly" non-identifiable (e.g., they have "almost flat" ridges in parameter space). Then you can identify the parameters eventually, but only with unusually good data, or multiple data constraints. As an example of the latter, you could observe one quantity that constrains the parameters to a ridge in parameter space, and another quantity that constrains the parameters to a perpendicular ridge, and the intersection of the ridges is well constrained. (Think of an "X" shape, or something like this figure, except the ellipses are stretched into ridges extending across the whole parameter space).
Non-identifiability is sometimes a problem for prediction, and sometimes not. The issue is that different parameter values can be consistent with the same data. If this relationship also holds into the future, then it may not matter: you might not know what the true value of a parameter is, but if all the allowed parameter settings lead to the same predictions, maybe you don't care if you get the parameters themselves wrong.
However, the relationship may not hold into the future: parameter settings that give similar predictions for historical data may lead to very different predictions for the future. This is the real problem, and it can't necessarily be solved with better data if the model is truly non-identifiable. Then you have to simply prepare for the wide range of possible outcomes.
What the article doesn't make clear is that not all models have this problem.
Ah. Not a business developer from before 2000 I see.
Tell me, what was the way 6-digit numeric dates were made sortable before datestamps were common?
Now - what affect did Y2K have on that?
How is the problem fixed? Especially as most systems that did not have dynamic file descriptions?
And that's just one.
Game theory always does account for things like that, primarily because the behavior you're describing is not irrational. The very fact that you are predicting that "he gets ahead" is what makes it rational.
Same for your "when they zig, you zag" idea: I have never heard of anyone using game theory that doesn't account for (and in fact, predict) that sort of behavior.
If you want to come up with an example where game theory doesn't work, you're going to have to try a few thousand times harder than that.
The reason game theory tends to disappoint, is that peoples' intuitive hunches for the payoffs of certain actions don't match the theory, but those hunches are what they act upon -- and that in turn changes all the payoffs, sometimes toward causing the hunches to becomes true (!) and sometimes toward causing the hunches to be more false. And that itself can be analyzed and predicted, but only if you just happen to know what other people's hunches are going to be -- and that is never predictable.
Game theory is about finding optimum equilibriums for behavior; it can never tell you what people believe.
BTW, back onto GP's subject.. a few months ago I went on an AdamCurtis-athon with some high expectations. It was a letdown, and not nearly as serious a criticism of the targets as I had hoped, especially since I just assumed some of them (e.g. the neo-cons) would be shooting fish in a barrel. I won't say watching all his docs is a waste of time -- it's not -- but don't get your hopes up. You'll find some good anecdotes, carefully selected interesting trivia, and great quotes like the one about economists and psychopaths .. but that's all.
As copyright owner of this comment, I authorize everyone to defeat any technological measure which limits access to it.
Comment removed based on user account deletion
"Essentially, all models are wrong, but some are useful."
-George E.P. Box
A crisis like this doesn't occur in just a year or two. It has to grow for a while.
The prime mover of the collapse was the longer-term buildup of bad debt due to subprime mortgages. Right smack in the middle of the subprime mess were Fannie Mae and Freddie Mac, underwriting about a quarter of all subprime mortgages.
At university, boring lectures were relieved by staring into the back of the prototype Phillips Machine, face to the wall because it was then unrestored. WikiP shows it was/is a fluid-dynamic macroeconomic model, only needs filling with coloured water. Something reminded me of my Northumberland G-Granny's meaningless rhyme "I fell into a bucket of eggs / and all the yaller ran down me legs". The best in that room went into banking - now they know too.
"Math is hard."
The cake is a lie.
No brain, no pain.
Does no one understand the Heisenberg uncertainty principle?
No system can be observed, or modeled, without affecting the system itself. In short, knowledge of the model invalidates the model.
Consider stock trading strategies. Presume that there exists an optimum strategy. A trader implements this strategy and is successful. Other traders, desiring to be successful, attempt to discern and emulate this strategy. When enough traders start to adopt the model, the model fails. In mathematical terms, there are high order terms that become increasingly significant as time goes on, until the model breaks down and becomes chaotic.
This effect is seen daily, in weather forecasting. At the end of last century, megabucks were thrown at weather modeling, trying to develop the ability to have long term, specific forecasts. These efforts failed in their original form, but resulted in a deeper understanding of complex systems and in well developed chaos theory. The end result is the knowledge that we can forecast weather pretty well tomorrow, not so well three days from now, and only in the most vague general way for next week. No model can work long term.
When economists, and budget offices, and (omg) poiliticians start talking about "over ten years" one has to laugh. The economy can't be forecast next month, and trying to extrapolate a trend over ten years is complete and utter handwaving.
Don't take life too seriously; it isn't permanent.
He has been maintaining a blog for over a decade, go read is predictions... Hint: They came true.
The real issue is that the media will put anyone on the air under the title of 'economist’ and give them all equal weight.
So you have a Nobel prize winning Dr. of economics being compared to some self proclaimed 'economist'.
It's sad. Now uncommon, I see this in science 'debates' as well.
I'm not going to get int a discussion about Paul Krugman and politics. All I am saying here is read his blog and look at the predictions.
Then read his book. THEN form an opinion.
The Kruger Dunning explains most post on
"There is no spoon"
True, but there are five interwoven economies that can all acquire spoons in different ways: :-)
"Five Interwoven Economies: Subsistence, Gift, Exchange, Planned, and Theft"
http://www.youtube.com/watch?v=4vK-M_e0JoY
"This video presents a simplified education model about socioeconomics and technological change. It discusses five interwoven economies (subsistence, gift, exchange, planned, and theft) and how the balance will shift with cultural changes and technological changes. It suggests that things like a basic income, better planning, improved subsistence, and an expanded gift economy can compensate in part for an exchange economy that is having problems."
The text for the presentation is here: http://www.pdfernhout.net/media/FiveInterwovenEconomies.pdf
A 21st century issue: the irony of technologies of abundance in the hands of those still thinking in terms of scarcity.
http://i.imgur.com/tPdL9.jpg
Also:
http://krugman.blogs.nytimes.com/
Go back 11 years and start reading that blog.
The Kruger Dunning explains most post on
And, once again, someone rediscovers the joy of an ill-posed mathematical model. And a geologist at that, who really ought to know better since all seismic imaging models are inherently ill-posed (and have their own body of specialized techniques for regularizing the inversion problems).
-JS
What about venture capital? They don't get a dime until IPO or buyout, and yet them putting money in startups so that they could grow is considered a valid investment, not a speculation.
Anything that you buy expecting/encouraging it to grow in value could be considered an investment. Risk is a different matter. You could argue there is higher risk that growth will never happen than the risk that you won't get any dividends/interest- but there are risks in both situations.
--Coder
There seems to be general confusion between general economic modelling and quantitative models for pricing derivatives etc. This is not least helped by the TFA which also seems to confuse the two. Quant pricing models are not predictive models, they are tools for pricing and risk-managing derivatives. As some famous quant once put it, all pricing models are just fancy interpolation schemes.
A pricing model starts from the position that you can't predict the future with certainty. Instead, the idea of the model is to describe the joint probability distribution of the relevant variables underlying the derivative. These are things like short-term interest rates, stock prices etc., and generally not macro variables like GDP or jobless rates.
The probability distribution produced by the model doesn't even have to match the real world probabilities of the events, it just has to match the probabilities which are implied by the market prices of instruments which can be used to hedge the relevant risks, the so-called "risk-neutral" distribution and its variants. Given enough reference prices, a complete picture of the risk-neutral distribution can be built up. The issue of calibration comes in because in practice, there aren't enough reference prices to build up the complete picture so you have to start making assumptions, i.e. specifying functional a form with some free parameters, and fitting those parameters to what you can observe. But this is still not about prediction, it's about getting a full description of "what the market thinks now about all possible futures".
Ross Perot predicted the banking meltdown long before it happened. As memory serves, his biggest problem was that he was off on just how soon it would take place. Just a bit of historical trivia. No, I'm not a Ross Perot worshiper...but the man certainly called a number of things correctly...or certainly more correctly that the many of the powers that be...
The point of the article is not that economic models are flawed (don't represent reality), but that models can give wrong predictions even if they're perfect (accurately represent reality), due to unavoidable uncertainty in their inputs. I go into more detail in this comment.
See subject.
"There are monetary units and there are measurable physical units of various economic goods and of many--but not of all-services bought and sold. But the exchange ratios which we have to deal with are permanently fluctuating. There is nothing constant and invariable in them. They defy any attempt to measure them. They are not facts in the sense in which a physicist calls the establishment of the weight of a quantity of copper a fact. They are historical events, expressive of what happened once at a definite instant and under definite circumstances. The same numerical exchange ratio may appear again, but it is by no means certain whether this will really happen and, if it happens, the question is open whether this identical result was the outcome of preservation of the same circumstances or of a return to them rather than the outcome of the interplay of a very different constellation of price-determining factors. Numbers applied by acting man in economic calculation do not refer to quantities measured but the exchange ratios as they are expected--on the basis of understanding--to be realized on the markets of the future to which alone all acting is directed and which alone counts for acting man.
We are not dealing at this point of our investigation with the problem of a "quantitative science of economics," but with the analysis of the mental processes performed by acting man in applying quantitative distinctions when planning conduct. As action is always directed toward influencing a future state of affairs, economic calculation always deals with the future. As far as it takes past events and exchange ratios of the past into consideration, it does so only for the sake of an arrangement of future action."
There are those that forecasted 2008 economic collapse but those in responsible positions, i.e. Alan Greenspan and Suze Orman, chose to ignore and continue to dupe ordinary people to make themselves more wealthier. Real crime is these people are doing the same crap again and again.
mfwright@batnet.com
The problem is that there's no fish, and all the conjuring tricks in the world can change that, and can't fool the punters forever.
-- The Grand Teddy Bear has Spoken: "Windows 8 Source Code Available NOW! more disgusting than your pr..."
The article deals with the reality of physical systems and their models. Who could doubt it, as it is well-known that there are an infinite number of equations that will match any given set of data points to any desired accuracy.
The reality of human attempts to control economic and social systems is even worse, as the standard story doesn't deal with the flaws in the regulatory model.
We use the law to build static institutions in the middle of rapidly evolving, extremely complex systems. The static institutions immediately occupy themselves with fostering the careers of their employees, and tolerate any amount of negative or perverse results of their institution's actions, so long as their institution is doing well.
Money buys power, so of course regulatory institutions become incestuous with their regulatees. Power attracts money, so legislators get their cut of the regulatees political budget to vote against regulations (many of which were proposed for precisely this reason) and to vote for regulations that the regulatees want.
The evolving system evolves, conditions change so the laws and regulations are not quite right. Everyone covers that up, it might affect their career. Iterate until the crisis is too large to cover up.
At which point, the political system repeats the Progressive mantra "Only government action can fix this failure of markets. Elect the right people, give them the power, and the problem will be fixed."
Iterate 100 years, that is how we got here, entering a world-wide Very Greatest Depression. Brought to you by the best and the brightest, educated at the finest institutions on the planet, Progressives who have the same uniform mind-set and have dominated the planets political, educational and social institutions for at least 60 years, group-thinking themselves into disaster.
Poor people have been seriously screwed, effectively enslaved. The rest of us are not in much better condition.
...that it would be adequate to staunch the bleeding, but not enough to jump start a recovery. At the time, the US economy was shedding jobs like crazy, since the stimulus the shedding stopped and there were some modest employment gains. One can argue whether that's due to the stimulus til the cows come home, but it's pretty much what the Krugster predicted.
Never let a lack of data get in the way of a good rant.
I work in higher ed. Three years ago we were cutting to the bone, now it's raining money and we're building like there's no tomorrow. When the loan bubble bursts it's going to be very ugly.
Never let a lack of data get in the way of a good rant.
Ignore facts, expertise, education; do what FEELS GOOD to you personally. This anti-intellectual attitude isn't the only reason people are more foolish today than in the past (including the ironic lip service payed to education.) The consumer culture we've built to extremes since WW2, raises us upon following our thoughtless and emotional impulses. It doesn't feel good to hear things that are unpleasant.
With heavy personalization, we are taking this to another extreme where one is automatically censored from even seeing something that doesn't feel good; it has and will continue to even change how people interact as they become more sensitive to unpleasant things -- since they grow up not being exposed to them as often.
Democracy Now! - uncensored, anti-establishment news
Take a look at the Freakanomics and Super Freakanomics books. They talk all about how irrational folks are with financial decisions.
Sure the Austrian school has predicted plenty of crashes that happened but they've also predicted a lot more crashes that haven't happened. If you want to look at their hit/miss ratio, I'd imagine that their statistics are piss poor.
I don't buy the premise of the article at all. If wildly different parameters for your model can satisfy the data, then you don't have enough data. That is all (though it is occasionally a legit problem in economics).
The real problem (as has been mentioned here already) is that most economists are heavily biased in favor of certain outcomes. Most economist jobs exist solely for the purpose of advocating particular views or policies. They end being wrong not because they use the wrong model or insufficient data, but because they reach a predetermined conclusion that has no real basis other than that someone paid them for the effort.
The only person to successfuly predict an economy was Hari Seldon.
I claimed, truthfully, that they were a big part. Two companies, 1/4 of the subprimes, leveraged for more than the five top investment banks combined.
The subprimes wouldn't have even been there if not for government interference from the likes of Frank. In fact, he resisted regulation specifically because he wanted more subprimes issued ("affordable housing").
The change in regulation brought it to a head, but it only brought to a head a situation that had been building for years.
I've always said that trying to put a fixed equation to something as random as human nature should have your ass tossed into a mental institute for life.
If you think you can predict and perfectly model human behavior, not across one person but across the entire population, and build an economy around it, you're fucking insane.
Still waiting on Serviscope_minor to wake up to fucking reality and realize that Jessica Price isn't going to fuck him.
O wait, no I don't.
"Economic models are always wrong", geez. So, when interest rates fell people didn't take out more loans? When unemployment is high, overall demand doesn't tend to drop? When cellphone service is provided by just a few carriers prices don't rise?
*Some* models are very sensitive to their parameterizations. And yeah, they'll be really tricky. Lots of economic models are really, basically, correct.
If there were a mathematical model which could predict market fluctuations, people would incorporate that model's decisions into their investment choices, driving up the price of things now which the model predicts would rise later.
Any such model, if it were accurate, would be very useful to investors - and its usefulness to investors would preclude its accuracy.
It's a bit of a Godel problem, ultimately.
DRM: Terminator crops for your mind!
Don't let it put you off his earlier documentaries.
The article is really about models of complex systems more than it is about economics.
Economist: A person who will gladly tell you tomorrow why the prediction they made yesterday did not come true today.
As a slow learner, taking 33 years to go from my HS diploma to a BS degree, I got to see some of the changes in Economics first hand
My first Macro Economics class was concerned with STUDYING the market to see how it was affected by consumers and take advantage of their NEEDS, WANTS, and DESIRES to produce products to fulfill their expectations... thus providing products, and services that people would exchange their hard earned money for and make a profit for anyone serving the market successfully...
By the time I graduated (Business Management) it had changed to an "Applied Science" focused on CHANGING consumer behavior and INFLUENCING society... the problem being it only influences that portion of society that doesn't understand economics...
The rules of the game as I was taught early on would have predicted (and did from the mouths of those still using them) all of the economic woes we've seen...
The Social Tool that "Modern Economics" theories are based on don't really work quite the way the books claim... kinda like when the butcher puts his thumb on the scale and says "this is for Timmy’s College Fund" and the customers don't appreciate the "contribution" they're making
In this case, the dude had a physical based model that is known to describe the phenomenon. Simplifying this model to fit the data is a bit weird in this situation, though it might be a fun suggestion that physicists should provide models that can be calibrated with different number of data points. Oh, if you have only 10 points, use Kepler's law, between 10 and 100 you can use Newton, more than one 100, use Einstein.
Everyone assumes that prices are set fairly and that's how the models work, and they just ignore that these huge financial market players are fucking everything up with these arbitrage deals.
Blah, blah, blah. Forrest and trees, people!
The question is not how to predict bubbles, but how to *eliminate* them! C. H. Douglas had this all figured out a long time ago. Read!
Social Credit would solve everything...