Why does anyone pay attention to people that study "economics"?

[dinofelis]

I think Keynesian vs Austrian debate is only because of popularity of young libertarians found on the internet.  Nobody in economics takes Austrian economics seriously.  It's really on the fringe and usually coupled with some heavy political dogma.  It's  the new radical right.

New Keynesian like Krugman represent the liberal left.  Most mainstream economists are probably neoclassical liberals.

In fact, the Austrian school is so close to neoclassical, that it is difficult to see the difference.  There are theorems linking both.  Both have a differently formulated axiomatic system, and I personally prefer the way the Austrians formulate it.  But it is almost but just another view on neoclassical economics.

http://econfaculty.gmu.edu/bcaplan/whyaust.htm

There is for instance a whole bunch on whether utility is cardinal (neoclassical) or ordinal (Austrian), through the notion of subjective preference.  In fact, both are mathematically equivalent.

It is a bit like the Bohm versus the Bohr interpretation of quantum mechanics.

[1714630061]

1714630061

[1714630061]

1714630061

[1714630061]

1714630061

[1714630061]

1714630061

[2112]

In other words, the economic dynamics is not time-invariant, because there is this learning function.  And the states also have learning functions.  if the first time they react to a crisis in a specific way, and that doesn't bring in the hoped-for results (on electoral or economic side), next time that state (which is part of the economic dynamics) will do something else.

Why you omitted time-variant models from the consideration?

So: same situation, different response.  By definition, that cannot be modeled.  Because a mathematical model gives you the same response to the same input.

By whose definition?

Scientific literature is full of hidden-state/unobservable-state and reflective models, "observers" as an algorithmic way of estimating the not-directly-observable state, etc.

I don't know if you are trying to omit most of 20-th century research as a dialectic technique or are you simply unfamiliar with the required math and therefore consider it all false/impossible?
 

Human action, which includes learning, makes that economy is *in principle* not following a mathematical model.  Because it is not time invariant.
And because people learn.

Sounds more like a religious credo than a science. And that is the problem with most Austrian economic literature: they very much resemble theological tracts.

Edit: But in comparison to other religions the Austrian economics is a sort of religion without ethics, (programmatically so, because of the free-choice axiom). This is the reason why many (if not most) consider them morally bankrupt from the start.

[2112]

I believe in Austrian economic, because I believe that the best people that know what the exchange is worth are the two parties themselves. Not a third one located several miles away without any stake in the game.

Spoken like true wife-less, child-less & land-less libertarian. In the real society your personal choices affect the life of your current family, your future family and your neighbors life and land values. Therefore the society has a stake in influencing your decisions.

The general problem with Homo Sapiens is that it is most fertile when young and stupid (or at least when not that smart). Even the ultra-assholey people who would have no problem with killing dumb libertarians (or letting libertarians kill themselves as a result of their stupidity and short-term thinking) will have problem killing children of the libertarians. So the society has a stake in forcing libertarians to send children to school instead of sending them to steal or to beg.

Libertarians are like Roma/Gypsy people in the sense that they can exist as a semi-stable community only at the periphery of the regular society. And the regular society has a stake in periodically culling the worst excesses in those roaming bands.

Anyway, read the "Ethics of Liberty" and please point me to the libertarian way of preventing infanticide or other within-the-family crime.

[2112]

Politicians are forever talking about statistics and the economy so virtually any interview with any politician where they mention anything about the economy should suffice as an example (and they often put forward economic "excuses" as to why climate change is something we can't afford to do anything about now).

But let's get to what riles me most - here is a well known publicly made statement from the current Australian PM (it was made before he was PM but he was leader of the opposition at that point in time).

"Climate change is crap." (Tony Abbot)

It is not surprising that even America (which has dragged its feet on doing anything much to address the issue of climate change) is now finding Australia's complete lack of political will and funding to be of concern (which has been recently been brought up).

Can you distinguish http://en.wikipedia.org/wiki/Economics from http://en.wikipedia.org/wiki/Political_economy ?

Are you really taking your knowledge of economy, statistics and other sciences from the TV and political speeches?

Are you pretending to be member of the uneducated underclass as a way of jump-starting the discussion?

Or do you really behave like the uneducated masses in your real life? I'm really confused. Are you a self-taught programmer or did you study computer science in some formal educational institution?

[dinofelis]

Why you omitted time-variant models from the consideration?

Because by definition, time-variant models are not time invariant, and cannot be found out before the time in question has passed !

Do not confuse time-variant models with time-invariant models which contain pre-known time-varying functions, but which are time-invariant models with a time-variant - but KNOWN - input.

For instance, Newton's equation m x d^2/dt^2 r = F is a time-invariant model.
You can put a time-varying force function in there.  But the model is still time-invariant, although the solution set will not be time invariant of course.  

A time-variant model would be something where today, the model is m x d^2/dt^2 r = F, and tomorrow, it will be, say,
m x d^3/dt^3 r = F^2, and the day after, it will still be something different.

If we were going to write that the derivative order is variable (here, going from second order to third order), then the model:

m x d^n(t) / dt^n(t) {r} = F^(n(t) - 1)

would in itself be a time-invariant model with a time-variant input, here n(t).

Unless we can model human action completely, including *understanding of economics* and *learning of the past*, there is no hope to put it into any time-invariant model, and of course, we don't know any time-varying function that could replace it.

So it is impossible to have a time-invariant model (unless we can model human learning and human inventivity).  It is impossible to have useful time-varying functions which would contain part of the future, because we don't know them.
So in the end we can only have true time-varying models.  Which is what econometrists do, but from the moment they have fitted a past model, it only works for the past (and maybe approximately for the very near future as a kind of short-term extrapolation).

Of course you can *write down* time invariant models.  But they will then contain unknown time-dependent functions, which are the "fitting parameters" which change over time, because they change with human learning and human understanding of economics.

So: same situation, different response.  By definition, that cannot be modeled.  Because a mathematical model gives you the same response to the same input.

By whose definition?

By the definition of a model !  A model is a mathematical transformation of the input to the output.  Same input -> same output.
That output can be a stochastic description.  But if the input is the same, the output should be the same.
If we take into account that the *time* is different, then the output should be the same if the model is time invariant and there are no explicit time-dependent functions.

Scientific literature is full of hidden-state/unobservable-state and reflective models, "observers" as an algorithmic way of estimating the not-directly-observable state, etc.

It is not the fact that it is not observable that is the problem.  It is the fact that its dynamics CHANGES over time.
People will react in a totally different way AFTER having seen the 2008 crash, than before.  People will have different economic opinions after learning about the 1929 crisis than before.  Human action (including state action) was totally different before 1929 than after 1929.  I'm not talking about the crash itself.   I'm talking about the learnings from that crash.  Human action and its dynamics has been altered by that learning.  State reactions have been altered by that learning.  Economic textbooks have been altered.  So people act differently now, than they did before.
And that is only one aspect.

The other aspect is the unpredictability of innovation.  By definition, innovation is unknown before it takes place.

Nobody could foresee the internet in 1930.  The economic effects of the internet were hence unpredictable.  I remember a prediction that the UK would have enough with 4 computers in the sixties :-)

So how can one make economic predictions based upon internet technology when you cannot even predict the technology ?

[twiifm]

If all these algorithms are so good why don't we have stable economies? And theology is probably the best way to view the systems we have today, even the temples are indistinguishable and religion is the only other conceptual thing that's become so engrained in our lives. When we can build models that accurately simulate the workings of the mind we might stand a chance because value is just as much a concept as beauty or taste.

That's like asking why there is still hurricanes even though we are advancing in modelling meteorology.

The models are tools for understanding.  The policies are tools to counteract.

[2112]

If all these algorithms are so good why don't we have stable economies?

I'm just going to guess that you are completely unfamiliar with modeling of chaotic systems.

Here's the simplest to understand example that I know:

http://en.wikipedia.org/wiki/Logistic_map
http://en.wikipedia.org/wiki/Bifurcation_diagram

"These algorithms" give ranges of answers, not single answers. It is up to economic decision maker to choose which portion of the range they are aiming at and which portion they want to avoid.

[dinofelis]

If all these algorithms are so good why don't we have stable economies? And theology is probably the best way to view the systems we have today, even the temples are indistinguishable and religion is the only other conceptual thing that's become so engrained in our lives. When we can build models that accurately simulate the workings of the mind we might stand a chance because value is just as much a concept as beauty or taste.

That's like asking why there is still hurricanes even though we are advancing in modelling meteorology.

No, the comparison doesn't work.  There are perfectly good models for meteorology, but they are chaotic, and so we know that they diverge.  They still allow statistical predictions in fact, and running families of meteorological models for 100 years is exactly what climate modeling does.

What I'm pointing out is something else, and in fact you illustrate it:

The models are tools for understanding.  The policies are tools to counteract.

But the counteraction is EXACTLY PART of the dynamics too !  Policies are a part of "human action".  Policy goals too.  How do you predict and model that ?

Give a climatologist the following task:

- calculate the number of hurricanes in the years 2070 - 2080.
- calculate me the yearly average temperature in North America in 2070-2080.
- calculate me the amount of snowfall in the northern hemisphere in 2070 - 2080.

He will be able to do so, given enough computing power.  His result will be statistical, but with good enough meteorological models, he will in principle be able to calculate you the statistical distribution of hurricanes in 2070 - 2080.  Similar for the other questions.  *in principle* he knows how to do so. Maybe the computing power is not sufficient, maybe the models are slightly off, maybe the parameters in the model are not known accurately enough etc... .but *in principle* he knows how to do so.  In fact, the only external input he needs is the economic one, of land usage, CO2 emissions, and so on !

Now, give an economist the following task:

- calculate the average evolution of unemployment in the northern hemisphere in 2070 - 2080.
- calculate the energy market in 2070 - 2080: what's the price of a KWhr ?
- calculate the average state deficit in 2070 - 2080.

Do you even know in principle with what model to estimate that ?

[2112]

By the definition of a model !  A model is a mathematical transformation of the input to the output.  Same input -> same output.

OK, this is bullshit. You must have slept through some lecture at your school. This is true only if your state vector is zero.

http://en.wikipedia.org/wiki/State-space_representation
http://en.wikipedia.org/wiki/Control_theory

It is hard to argue with the rest of your http://en.wikipedia.org/wiki/Cosmik_Debris if we aren't using the usual mathematical terminology anymore.

Edit: Actually I take back "hard to argue". I should've written "not worth to argue". I re-read you previous reply where you mentioned http://en.wikipedia.org/wiki/Kalman_filter . In any sensible control theory curriculum they discuss things like http://en.wikipedia.org/wiki/Luenberger_observer way earlier, when still explaining LTI (linear, time-invariant) systems.

So the conclusion is that you either:

A) come here to intentionally confuse (a.k.a. troll),
B) studied control theory from the special curriculum for Austrians where they only consider memory-less systems (a.k.a. undereducated or retard).

In any way, not worth further discussion.

[twiifm]

If all these algorithms are so good why don't we have stable economies? And theology is probably the best way to view the systems we have today, even the temples are indistinguishable and religion is the only other conceptual thing that's become so engrained in our lives. When we can build models that accurately simulate the workings of the mind we might stand a chance because value is just as much a concept as beauty or taste.

That's like asking why there is still hurricanes even though we are advancing in modelling meteorology.

No, the comparison doesn't work.  There are perfectly good models for meteorology, but they are chaotic, and so we know that they diverge.  They still allow statistical predictions in fact, and running families of meteorological models for 100 years is exactly what climate modeling does.

What I'm pointing out is something else, and in fact you illustrate it:

The models are tools for understanding.  The policies are tools to counteract.

But the counteraction is EXACTLY PART of the dynamics too !  Policies are a part of "human action".  Policy goals too.  How do you predict and model that ?

Give a climatologist the following task:

- calculate the number of hurricanes in the years 2070 - 2080.
- calculate me the yearly average temperature in North America in 2070-2080.
- calculate me the amount of snowfall in the northern hemisphere in 2070 - 2080.

He will be able to do so, given enough computing power.  His result will be statistical, but with good enough meteorological models, he will in principle be able to calculate you the statistical distribution of hurricanes in 2070 - 2080.  Similar for the other questions.  *in principle* he knows how to do so. Maybe the computing power is not sufficient, maybe the models are slightly off, maybe the parameters in the model are not known accurately enough etc... .but *in principle* he knows how to do so.  In fact, the only external input he needs is the economic one, of land usage, CO2 emissions, and so on !

Now, give an economist the following task:

- calculate the average evolution of unemployment in the northern hemisphere in 2070 - 2080.
- calculate the energy market in 2070 - 2080: what's the price of a KWhr ?
- calculate the average state deficit in 2070 - 2080.

Do you even know in principle with what model to estimate that ?

This is a red herring response.  I didn't claim economics have the forecasting accuracy of meteorology to a specific date.  I don't know anything about meteorolgy to be qualified to make statements about meteorology.  

But if you want to know unemployment figure forecasting you pull past data all variables such as from income, tax rates, interest rates, consumer spending, etc..  Then you model this as a matter of relation.  After you have your model you can run simulations.  E.G what happens to unemployment when there is shock that causes interest to fall or rise.

I don't understand why you think you need to know the reason why interest rise or fall.  If sudden policy raises the rate then you can do simulation in your model.  

[2112]

Nobody could foresee the internet in 1930.  The economic effects of the internet were hence unpredictable.  I remember a prediction that the UK would have enough with 4 computers in the sixties :-)

I kinda like playing bullshit bingo. So lets count who has a computer now:

1) Amazon has EC2 computer
2) Microsoft has Azure computer
3) Google has a BigTable computer
4) Apple too has some computer
5) Rackspace has some generic "cloud" computer

Everyone else has either a tablet or a smartphone.

So the prediction is true, the world has about 5 computers and lots of dumb terminals. Win for economic modeling from the sixties!

[Nicolas Dorier]

I believe in Austrian economic, because I believe that the best people that know what the exchange is worth are the two parties themselves. Not a third one located several miles away without any stake in the game.

Spoken like true wife-less, child-less & land-less libertarian. In the real society your personal choices affect the life of your current family, your future family and your neighbors life and land values. Therefore the society has a stake in influencing your decisions.

The general problem with Homo Sapiens is that it is most fertile when young and stupid (or at least when not that smart). Even the ultra-assholey people who would have no problem with killing dumb libertarians (or letting libertarians kill themselves as a result of their stupidity and short-term thinking) will have problem killing children of the libertarians. So the society has a stake in forcing libertarians to send children to school instead of sending them to steal or to beg.

Libertarians are like Roma/Gypsy people in the sense that they can exist as a semi-stable community only at the periphery of the regular society. And the regular society has a stake in periodically culling the worst excesses in those roaming bands.

Anyway, read the "Ethics of Liberty" and please point me to the libertarian way of preventing infanticide or other within-the-family crime.

Did I talked about libertarianism in this thread ?
What I said is that from Austrian perspective, two persons making a deal knows better than the bureaucrat if it is worth it. If you don't believe you reach Keynes conclusions, and if you believe it, you will reach Mises' one.
I am talking about business concerning two people which don't plane to steal another.

We are talking about economics in this thread, not libertarians.
Libertarian extends this principle beyond business, and this is not the right thread if you want to discuss about it.
But yes, Rothbard have responses about how to deal with crimes in libertarian society. Just it is not the right thread to talk about it.

So stop trolling and comes up with some interesting arguments if you think I mis defined Keynesians or Austrian philosophy. But from what I see in your response, I'd say that I defined very well.

[2112]

Did I talked about libertarianism in this thread ?
What I said is that from Austrian perspective, two persons making a deal knows better than the bureaucrat if it is worth it. If you don't believe you reach Keynes conclusions, and if you believe it, you will reach Mises' one.
I am talking about business concerning two people which don't plane to steal another.

We are talking about economics in this thread, not libertarians.
Libertarian extends this principle beyond business, and this is not the right thread if you want to discuss about it.
But yes, Rothbard have responses about how to deal with crimes in libertarian society. Just it is not the right thread to talk about it.

So stop trolling and comes up with some interesting arguments if you think I mis defined Keynesians or Austrian philosophy. But from what I see in your response, I'd say that I defined very well.

No Nicholas, you didn't talk explicitly about libertarians.

You've just spoken about personal economic decisions like them, especially the ones who are wife-less, child-less & land-less. You are so precariously naive that I don't consider you (plural you, the people in the peculiar social situation like you personally) a worthwhile discussion partner about long-term economic decision making. Because I've seen too many single male programmers/computer technicians with the similar mindset. Either time&life will cure them out of their egotism/egocentrism or they die completely alone and their cadaver is picked up by the municipal services and absolutely nobody shows interest where and when will be the funeral. The employer is finally alerted to their death when those services are trying to collect on the death benefit of the health&life insurance.

I don't remember the exact "death alone" statistics from my employer's human resources department, but the programming department was on top, then technical support, then big gap and then inside sales.

Anyway, since Rothbard was now mentioned twice I'm going to quote a fragment from Chapter 13 "Punishment & Proportionality" from his "Ethics of Liberty".

A problem might arise in the case of murder - since a victim's heirs might prove less than diligent in pursuing the murderer, or be unduly inclined to let the murderer buy his way out of punishment. This problem could be taken care of simply by people stating in their wills that what punishment they should like to inflict on their possible murderers. <I snip the rest, but anyone who was born without the will attached should read that chapter>

This may read like ad hominem towards Nicholas, but it really isn't against him, but the class of people he represents in his thinking and writing.

[dinofelis]

By the definition of a model !  A model is a mathematical transformation of the input to the output.  Same input -> same output.

OK, this is bullshit. You must have slept through some lecture at your school. This is true only if your state vector is zero

I'm not talking about the instantaneous inputs, but about the input time series.

Now, of course, if you consider that the state vector does contain the knowledge from economic happenings from 100 years ago for instance, and that this matters now then you ARE modeling human nature, that is, you *are* modeling the fact that humans learn from the past.  But this is what I claim, is essentially impossible.

What you are saying is that, say, in 1935, of course, the effects of the 1929 crisis of course have an influence on economic data, and that just "putting in the actions of 1935 is silly if you do not take into account the inherited state due to the crisis of 1929".
But that's trivially right of course,  that's not what I'm saying.

I'm saying that the crisis of 1929 prompted Keynesianism, and that that has had an influence on state policies, and hence on economic dynamics even in the 1970-ies and even today.

A good "time invariant" model of the economy would have had in its dynamics, the human action which was Keynesianism as an invention after the 1929 crisis.  That is, the model, built in 1918, say, would have not only to foresee the crisis in 1929, but also the fact that due to that crisis, some economists would have invented Keynesianism, and that that would then influence state behavior (which is a special kind of human action, namely human action of people with state privileges) and that that would have made the states react (their dynamics) differently than before 1929.

I'm claiming that that is essentially impossible.  It would mean that the same model that would have foreseen the invention of Keynesianism after the crisis of 1929 (even before Keynes wrote his stuff) and would have foreseen the reactions of states on the 2008 crisis, all this back in 1918.

Because if you want to make long-term predictions in economy, those are the essential dynamics (state behavior included) you have to be able to model.  And those *change* irrevocably after certain happenings, in ways that are essentially impossible to foresee, because they imply advancement of our economic understanding.

[dinofelis]

This is a red herring response.  I didn't claim economics have the forecasting accuracy of meteorology to a specific date.  I don't know anything about meteorolgy to be qualified to make statements about meteorology.  

But if you want to know unemployment figure forecasting you pull past data all variables such as from income, tax rates, interest rates, consumer spending, etc..  Then you model this as a matter of relation.  After you have your model you can run simulations.  E.G what happens to unemployment when there is shock that causes interest to fall or rise.

I don't understand why you think you need to know the reason why interest rise or fall.  If sudden policy raises the rate then you can do simulation in your model.  

That's what I'm saying.  One is capable to make "models" that are nothing else but grey box extrapolation from the relatively recent past to the very recent future, all else equal.

The point is that "policy" is just a human action of an economic agent, like any other, and that a genuine model can PREDICT that instead of having it as an input.  The decision to raise or lower rates is a human action in the same way (but by state-related agents) as business people take decisions, and as inventors do inventions.  Human action.
Especially in a global economy, where the policy response of one country (to a specific economic situation) will influence the economy in another country.  It is hence part of the dynamics, and I tell you that that is essentially impossible to model because it is based upon human action.

Of course as a physicist (I'm one) you can say that the human brain is a physical system which we can model.  But my point is that you'd have to go down that kind of road for economy (with human action) to become modelable.  And for all practical purposes, we can't.  

If I would have as an input all individual decisions of all individual economic agents in an economy, I could also "forecast" what's going to happen.  But forecasting means to be able to calculate it !

[dinofelis]

Nobody could foresee the internet in 1930.  The economic effects of the internet were hence unpredictable.  I remember a prediction that the UK would have enough with 4 computers in the sixties :-)

I kinda like playing bullshit bingo. So lets count who has a computer now:

1) Amazon has EC2 computer
2) Microsoft has Azure computer
3) Google has a BigTable computer
4) Apple too has some computer
5) Rackspace has some generic "cloud" computer

Everyone else has either a tablet or a smartphone.

So the prediction is true, the world has about 5 computers and lots of dumb terminals. Win for economic modeling from the sixties!

 

[picolo]

People studying economics at the university study keynesian economics so they are not learning real economics.

[dinofelis]

QE is a new invention first attempted by BoJ back in 2001.  There's data we can look at.

BTW, as far as I know, QE is nothing else but a variant on what John Law already invented in the beginning of the 18th century, no ?

[Nicolas Dorier]

Either time&life will cure them out of their egotism/egocentrism or they die completely alone and their cadaver is picked up by the municipal services and absolutely nobody shows interest where and when will be the funeral. The employer is finally alerted to their death when those services are trying to collect on the death benefit of the health&life insurance.

I don't remember the exact "death alone" statistics from my employer's human resources department, but the programming department was on top, then technical support, then big gap and then inside sales.

You are stereotyping and just wrong about that. Not giving something worthwhile to the discussion.
But yes, I am not one of those that think my child should not pay for my retirement, and that parent's duty is to make his child stands by his own feets. It is not egocentrical, it is what independence means and the only way to live as a free individual.
Again, this has nothing to do with austrian economics.

For the keynesian, you can always try to model something and prove that from stats, your model maximize some metrics.
But if we don't agree with the metric you are maximizing, then the reasoning that led you to the conclusion no matter how coherent and mathematical it is, won't make it legitimate.

[dinofelis]

In the real society your personal choices affect the life of your current family, your future family and your neighbors life and land values. Therefore the society has a stake in influencing your decisions.

The problem with that is that "society" is not a conscious being that can take decisions.  It are only individuals that take decisions.  Those individuals just take decisions in their *own* interest, whether they are vested with powers to decide "in the name of the general well-being" or not.  In other words, from the moment that an individual decides "in the name of society", he's *actually* not deciding in the name of society (whatever that might mean), but in his own name, and then just has to lie and manipulate enough to make people believe that he did that "for the general good".
When a politician takes a decision, he does this for his own advantage (electoral, financial,...).  He will make it sound as if it is "according to his views for the general good", but that is never the case.  When a central banker takes a decision, he does that in his own advantage (political nomination, having his favourite politicians win elections, having friends and family members make money from the pre-knowledge, ....) and not for some or other macro-economic reason, although he will pretend to do so.

In other words, giving other individuals *power* to decide "in the name of society" is always going to lead to total corruption, because everybody always takes all of his decisions in the name of his personal advantage. Having other individuals decide over aspects of *your* life from the viewpoint of *their* personal advantage is not something that I would think is positive for anyone, except for the powerful deciders.
The difference between libertarians and collectivists is that libertarians are honest about that, that's all.

The discussion is not whether "a state that would try to achieve the general good" is a good idea, the discussion is whether such a state (made up of individuals) could even exist.  In my opinion, it can't, because individuals always decide first in their own advantage.