Is Economics Logical or Empirical? (and why should you care?) Warning: Long

[Oakman]

Richard A. Abbott wrote:

Because of the chaotic nature (read dynamics) of economics as well as weather and countless other disciplines, it is devilishly difficult to formulate a precise prediction for nothing more than a few days ahead. Anything greater than that, irrespective of historical records, is just guessing

I am translating that as you agreeing with me that Economics is not an empirical science. Would that be fair?

Richard A. Abbott wrote:

yet if you must attempt to predict from chaos, the "Lyapunov Exponent of Noisy Nonlinear Systems" tests are perhaps (perhaps not) most relevant.

In all honesty, I suspect a crystal ball or a deck of tarot cards would be as relevant when predicting from chaos. "Chaos" may be a good explanation for why it's impossible to know what circumstances and systems produce what outcome. It is not an excuse for talking as if you can until after the fact. (Not the weatherman's fault.)

Jon Smith & Wesson: The original point and click interface

[Oakman]

Fisticuffs wrote:

Is that the weather isn't reading your predictions.

When I first saw a game of jai lai with people wagering on it as if it was a horserace, someone told me a wise man once said that the difference between the two was that the horses couldn't read the tote board.

Jon Smith & Wesson: The original point and click interface

[Oakman]

Fisticuffs wrote:

I disagree that one could consider logic and mathematics a science of any sort.

With all due respect, that pretty much means we don't have anything to discuss, right? Most of what I wrote was based on the assumption of there being two, rigorous, forms of scientific thinking. If you say that one of them is witch doctery, I suspect there is no argument available to me to change your mind.

Fisticuffs wrote:

Nobody would argue that Vioxx wasn't studied empirically, and yet once we started using it, the number of cardiovascular events increased. Which only means our understanding of COX inhibitors and the cardiovascular system is incomplete.

It also is an argument against having testing done only by a group with a vested interest in what the outcome is. However, the two cases ( level 5 hurricane in existence and bearing down on NYC / the long-term effects of Vioxx ) are not comparable. As Ben pointed out, there's no onus in not predicting exactly when dealing with complex systems - but when you get the results completely, tragically, and utterly wrong then it is time to find a different predictor.

Jon Smith & Wesson: The original point and click interface

[Lost User]

Essentially, yes and yes. This damned butterfly has a lot to answer for!

[Chris Austin]

My personal opinion is that economics works at a certain boundary conditions. Basic micro economic principles like supply & demand and price elasticity are pretty well proven. But, I think macro economics is on far less solid ground. It is hard for me to believe that any predictive model based on true and tested natural laws could have not seen the problems when an entire globe moves to a fiat currency. Supposedly, there is some economist out there who studied business cycles that really hit the nail on the head with a prediction about this. His analysis included some very nice charts and predictions. I really wish I remembered his name.

Sovereign ingredient for a happy marriage: Pay cash or do without. Interest charges not only eat up a household budget; awareness of debt eats up domestic felicity. --Lazarus Long Avoid the crowd. Do your own thinking independently. Be the chess player, not the chess piece. --?

[Rob Graham]

Oakman wrote:

So, to answer your question directly, I think they're faking it. Ironically, they are being guided, in part, by the wisdom of the same people whose lack of foresight contributed to the country and the world being sucker-punched.

Hence the joke icon on my post. If not faking it, at least betting the farm on their unproven/untested theories in hopes it will prove correct when the experiments results are known.

Oakman wrote:

A few, mostly Austrian School economists (and Ron Paul who follows their lead) were predicting the collapse, or so I am told.

Yep. I was lucky enough to listen, end put my 401k money on the sidelines right after the Countrywide collapse in late 2007 (which convinced me they were right). May have been the best investment move I've ever made.

[Rob Graham]

It may not be possible to predict exact outcomes for economics, but like the weather, it should be possible to recognize when the chaotic system has been pushed over an instability boundary, and most likely outcomes are not of the desirable kind. When most .Com stocks had P/E rations that meant they would have to have unrealistic returns for 50 years to justify their pricing, one could reasonably predict (and some did) the .COM bomb. In this case housing valuations (and mortgage backed securities) probably showed the same thing. Beck showed a chart on a recent show that claimed to show the average house value in constant $ was around 100K, except for just before the bust, when it rose to nearly triple that, something it had not done in the previous century (some dips and peaks before, but no previous deviations of more than 30%). That might have served as a "boundary warning" had someone noticed before the collapse, especially considering the massive multiplier effect of the Credit Swaps that everyone was in love with.

[John Carson]

Economics works by constructing simplified representations of reality known as "models", which usually take the form of systems of equations. Often these models are used simply to give qualitative insights (e.g., putting a tax on something will tend to raise its price and lead to decreased purchases of it). Sometimes the parameters in the models are empirically estimated and used for the purposes of quantitative prediction. All of these models are "wrong" in the sense of being over-simplified, but the hope is that they capture the most important elements of a situation and thus give a good "handle" on it. How simplified models are depends in part on how big a piece of the economy the model is meant to represent. A model that looks at a single firm will typically incorporate more complexity in firm behaviour than a model of the whole economy. Even at the level of a single firm, some models may have a very rich description of labor relations but an exceeding primitive representation of financing. Other models do the reverse. The reason for this is simply one of analytical tractability. The more realistic a model and the larger the proportion of the economy that it covers, the harder it is to analyse behaviour within the model. It is also the case that much of the data about the economy is privately held in a decentralised way. Thus the more detailed a model, the harder it is to get the data needed to estimate its parameters. Models that look specifically at the financial sector may incorporate aspects of regulatory behaviour and consider a range of financial securities (mortgages, shares, derivatives and all that). Models of the economy as a whole will typically represent the financial sector in skeletal form and many of the issues at the centre of current debates (behaviour of ratings agencies, transparency of debt instruments, incentives to risky behaviour) may not be modelled at all (banks just borrow and lend and everyone is assumed to understand what is going on). For decades at a time, it is possible to get away with this, but occasionally an aspect omitted from a model proves to be crucial. Model construction is an ongoing enterprise. At any given time, there is a lot of behaviour that noone has ever gotten around to trying to model and there is some behaviour that is extremely difficult to model accurately (the modelling of human reasoning, creativity etc.). For an analogy that illustrates the difficulties, consider attempts by biologists to predict species populations. We know that biologists

[Lost User]

John Carson wrote:

Models of the economy as a whole will typically represent the financial sector in skeletal form

To flesh-out this skeleton will be subjected to dynamic forces that may or may not be known. The flesh could be thin or obese and may vary over time. The governments of this world have been throwing money, almost confetti like, trying to assess the size of this flesh. My questions are : (1) What are the general economic rules of thumb that says "we now know how large or small this flashed skeleton is" (2) When do we know that enough medicine has been applied so that the flesh becomes "normalized". and (3) How much of this will be guesswork?

[John Carson]

Richard A. Abbott wrote:

To flesh-out this skeleton will be subjected to dynamic forces that may or may not be known. The flesh could be thin or obese and may vary over time. The governments of this world have been throwing money, almost confetti like, trying to assess the size of this flesh.

My reference to "skeletal" was meant to convey that the representation of the financial sector is very simplified in most models of the economy as a whole. It was not a matter of the size or profitability or health of the financial sector.

John Carson

[Oakman]

John Carson wrote:

Economics works by constructing simplified representations of reality known as "models", which usually take the form of systems of equations.

It is the rigorous approach that you define that makes it obvious to me, at least, that Economics is a science.

John Carson wrote:

All of these models are "wrong" in the sense of being over-simplified, but the hope is that they capture the most important elements of a situation and thus give a good "handle" on it.

And this is why I think of it as as logical science rather than empirical one. As I referenced, had observation proved that Einstein had gotten a "good handle on it" but not been able to precisely predict the displacement of light by gravity, his theory would have had to have been discarded.

John Carson wrote:

We know that biologists cannot reliably predict the future course of evolution

I may be wrong, but isn't that the purview of evolutionary scientists, or possibly naturalists (like Darwin)? It did cross my mind while I was puzzling through the writing of the OP, that both the study of Evolution, and of Climatology, are indeed also logical sciences, not empirical ones. Perhaps that's why their discussion generates more heat than light. Please believe that I am not trying to show any disrespect any of the above fields of study. Geometry - to put forth my own example - is extremely challenging and once we advance beyond Euclidean Geometry, I begin to hear clearly, but understand poorly. Yet Geometry depends on its theorems, definitions, and postulates to prove the validity of any of its solutions. Change those axioms and the solutions change, too. So rather than saying that Economics is like Evolutionary Science, I'd rather say it is like Geometry. A very useful tool in the real world, but one where one must take care to know and understand the postulates that underlie it. Otherwise, you can end up in a segment of the universe where perspective is truth and parallel lines meet in the distance. ;)

Jon Smith & Wesson: The original point and click interface

[John Carson]

Oakman wrote:

And this is why I think of it as as logical science rather than empirical one. As I referenced, had observation proved that Einstein had gotten a "good handle on it" but not been able to precisely predict the displacement of light by gravity, his theory would have had to have been discarded.

I don't think this is normal usage. As I would understand it, all empirical sciences (including physics) make use of mathematical models and all make testable predictions. However, some empirical sciences have more success and stricter standards of success than others. Einstein's work would not have been discarded if it predicted inaccurately until something that could predict better came along. Even when something better does come along, the old theory may not be wholly discarded. Newton's work is still used even though Einstein's work showed that it provided merely a good approximation under certain conditions. Certainly, a robust failure on the part of any theory of physics to accurately predict something tells us that the theory is imperfect. It doesn't tell us if the theory merely needs modification and refinement or if a completely new way of looking at the problem is required. In fields like economics, perfect quantitative prediction is more or less never achieved. Nevertheless theories are discarded if they predict badly enough, so empirical results do matter.

Oakman wrote:

I may be wrong, but isn't that the purview of evolutionary scientists, or possibly naturalists (like Darwin)?

Evolutionary science is a multi-disciplinary field covering biology, geology, nuclear physics (for dating) and various other fields. But predicting the future course of evolution (as opposed to figuring out the past course of evolution) would mainly fall to biologists. Naturalists are biologists.

Oakman wrote:

So rather than saying that Economics is like Evolutionary Science, I'd rather say it is like Geometry. A very useful tool in the real world, but one where one must take care to know and understand the postulates that underlie it. Otherwise, you can end up in a segment of the universe where perspective is truth and parallel lines meet in the distance.

There are empirical tests of economics and not of geometry, so I think there is a fundamental difference. However, your remarks touch on issues that attract some controversy i

[73Zeppelin]

Hi Jon. I missed this because I was travelling for many, many hours on a train through the snowy high Alps, dreaming of living in a ski chalet with a roaring fireplace doing nothing but skiing and relaxing.... Anyways, to make it short, I am both a physicist and a financial economist. The problem with economics is that, while it is empirical, the results are not reproducible. You cannot conduct and repeat an experiment in economics. You have but a single data sample: history. This means that it is hard to put an exact definition of what a "bubble" is, unless it is placed in a historical context. This makes identifying current bubbles extremely difficult to some extent. The laws of physics are quite stable, economic laws and conditions change with time. So macroeconomic theory has to try to predict the future from past historical trends - the problem is that the past is not a great predictor of the future. Mathematically, most economic quantities are what are called martingales - that is, the best predictor of future values is the current value. That's obviously bad, since human behaviour and unexpected occurrances (Taleb calls these "Black Swans") can easily ruin things. In fact, most stochastic processes used in modeling economic time series have similar properties. Another is the Markov property. What the mathematics is telling us is that we can't predict the future with any accuracy - which is how the world should be. But that's bad for economic forecasting and that's why the physicist you reference looks down upon the economist. So, while Richards may tend to look down on economics as a science, having worked in both fields, I would say the economist operates at a distinct disadvantage. There are no nice economic laws like there are physical laws. Economics does not operate in accordance with finely-tuned natural constants and laws, economics has inherent elements of uncertainty, risk and unpredictable human behaviour to it. As much as Richards claims he can predict down to 10 decimal places, I'd like to see him pinpoint both the position and momentum of an electron. Then we'll sit and talk about prediction. So while economics may be an empirical science, methodologically it is fundamentally different than physics and it's really quite absurd compare the two.

[Oakman]

73Zeppelin wrote:

dreaming of living in a ski chalet with a roaring fireplace doing nothing but skiing and relaxing....

Sounds like a trip worth taking. Are you back in Switzerland now? For good?

73Zeppelin wrote:

The problem with economics is that, while it is empirical, the results are not reproducible.

An important point. One I thought might be true, but was afraid that if I tried to make it, you or John or Robert would point out that I was guessing again and unaware of the work don't by Carleton Phleblemeister when. . .

73Zeppelin wrote:

The laws of physics are quite stable, economic laws and conditions change with time.

It almost seems that there is such a thing as quantum economics. We weren't in a recession until someone name Schroedinger at Countrywide opened the box so we could see inside.

73Zeppelin wrote:

So macroeconomic theory has to try to predict the future from past historical trends

Something we all do in theory, although it occurred to me way back in the '80's that we were not learning from the past as we removed one after another of the regulations put in place after 1929. I was told back then and as recently as a couple of years ago that 1929 held no real value as a lesson any longer. . . Just as a matter of interest are there any, besides Gresham's Law, axioms that are widely accepted by every school of economics? For that matter, am I right in thinking that Gresham's Law is widely accepted? (I suppose that since it's all fiat money, his Law is both proved and no longer valid.)

73Zeppelin wrote:

Another is the Markov property.

We are starting to reach beyond the limits of my knowledge, but, if I understand correctly, it should be a waste of time to claim that any human directed process is Markovian.

73Zeppelin wrote:

I'd like to see him pinpoint both the position and momentum of an electron.

I wanted to ask him about quantum mechanics, but that really wasn't his point - nor do I think it was to put economists as a whole down - just those who claim that their prescriptions are always accurate - and maybe those who are willing to risk a lot of other people's wealth without their consent on that accuracy.

[Lost User]

Sorry John, I read "skeletal" as "skeleton". I recognize skeletal as a system in itself and part of an overall larger system much like as you would find in a human body.

[Oakman]

John Carson wrote:

all empirical sciences (including physics) make use of mathematical models and all make testable predictions. However, some empirical sciences have more success and stricter standards of success than others

The last sentence, if you are correct, begs the question: When are models and predictions given such a low standard of success to achieve that they become useless? Fat_Boy would have told us that the prevailing climatological theories have reached that point; Others with more couth than Ilion are able to argue a similar point regarding evolution. And therein lies the rub, once standards are lowered, then all results, even ones that could meet a much higher standard, are suspect and open to attack.

John Carson wrote:

I have never agreed with this since models can never be comprehensively tested in advance of their use.

A close friend of mine was project chief when the Columbia Pictures conglomerate changed their internal accounting software over to a totally new system. When it came time, they ran both systems, side by side for over a year, until they were satisfied that given the same data, both systems produced the same results. It would seem that before a set of economic theories is offered to the public, it should be run for awhile without fanfare, to see whether it's predictions prove accurate. It may be that there are such tests going on (by definition, I wouldn't hear of them) but it appears to this observer that many economists do not meet this level of testing before publication. Which is what leads me to conclude that they are not empirical scientists, even if I accept your assurance that economics itself is or at least can be. Side note: You are frighteningly bright and well-informed. I am sorry that we locked horns for awhile and strongly prefer the exchanges we are having now. (Even if I am not sure we agree any more than we did before ;) ) I hasten to add that I have never learned much worth knowing from someone who alweays agreed with me, and a lot from those who always didn't.

Jon Smith & Wesson: The original point and click interface

modified on Thursday, March 12, 2009 12:58 PM

[73Zeppelin]

Oakman wrote:

Sounds like a trip worth taking. Are you back in Switzerland now? For good?

No. I went back to consult with a coauthor on a paper. I don't think I will be going back to Switzerland in the immediate future. My job possibility is in an EU country.

Oakman wrote:

An important point. One I thought might be true, but was afraid that if I tried to make it, you or John or Robert would point out that I was guessing again and unaware of the work don't by Carleton Phleblemeister when. . .

Well, it's really more basic than that. Stock prices only have one trajectory - the one that happened. It's rather pointless to simulate various price histories of, say, Google. The price history of Google is the price history of Google that we can download from Yahoo finance. So, while you can simulate stock prices, the only important series of stock prices are the historical ones. It's like history, the only history that we can study is the history that happened. Alternative histories may be interesting, but they are of limited value since they are fictional.

Oakman wrote:

It almost seems that there is such a thing as quantum economics. We weren't in a recession until someone name Schroedinger at Countrywide opened the box so we could see inside.

I think the problem is that the growth of financial products exceeded the regulatory framework. Innovation is great if it's not destructive (e.g. Schumpeter's "creative destruction[^]").

Oakman wrote:

Something we all do in theory, although it occurred to me way back in the '80's that we were not learning from the past as we removed one after another of the regulations put in place after 1929. I was told back then and as recently as a couple of years ago that 1929 held no real value as a lesson any longer. . . Just as a matter of interest are there any, besides Gresham's Law, axioms that are widely accepted by every school of economics? For that matter, am I right in thinking that Gresham's Law is widely accepted? (I suppose that since it's all fiat money, his Law is both proved and no longer valid.)

I work in an industry where predicting the future equates to

[RichardM1]

You missed a whole section of science. There are, as you put it, hard sciences, and 'soft' sciences. You forgot the really squishy sciences. Economics is not a soft science, it is a squishy one. It is really the application of statistics and logic to psychology, which is even squishier. Economics is couched in 'firm' terms, but at its base is attempting to define how people react, and learning how people reacts changes how you react, which changes how people react. The result is that when people learn how the market reacts, they change the way it reacts, and how its reaction changes has to do with the 'mood' of the people who make up the market.

Silver member by constant and unflinching longevity.

[73Zeppelin]

Would you say the price of an option or financial instrument should depend on the "mood" of the crowd, then?

[John Carson]

Oakman wrote:

The last sentence, if you are correct, begs the question: When are models and predictions given such a low standard of success to achieve that they become useless? Fat_Boy would have told us that the prevailing climatological theories have reached that point; Others with more couth than Ilion are able to argue a similar point regarding evolution. And therein lies the rub, once standards are lowered, then all results, even ones that could meet a much higher standard, are suspect and open to attack.

I think the important point is that there is a competition, a battle for hearts and minds and for professional prestige. The standards are set by what people are able to actually achieve and tend to evolve upwards as a discipline matures. But it is certainly true that when a discipline is in a state where lots of things aren't well understood and standards reflect this, then that tends to have unfortunate effects across the board.

Oakman wrote:

It would seem that before a set of economic theories is offered to the public, it should be run for awhile without fanfare, to see whether it's predictions prove accurate. It may be that there are such tests going on (by definition, I wouldn't hear of them) but it appears to this observer that many economists do not meet this level of testing before publication. Which is what leads me to conclude that they are not empirical scientists, even if I accept your assurance that economics itself is or at least can be.

Two points. 1. As Zepp noted at another point in this thread: "You cannot conduct and repeat an experiment in economics. You have but a single data sample: history." Empirical models in economics are estimated from historical data, but that doesn't guarantee that they will fit future data as well as they fit past data. 2. Much economic writing is purely theoretical. It is, as you earlier suggested, like geometry. Some theoretical economists will resolutely refuse to draw any policy conclusions from their work. They see themselves as simply clarifying the logic of an argument. "If X and Y are true, what does that mean for Z?" Other economists are more willing to draw policy conclusions from theoretical models that have been subjected to little if any testing. Generally this would be justified on the basis that the assumptions in the model have proved serviceable in other contexts and/or are empirically supported. Reliance on theoreti