"Economics makes my Brain hurt" - Opinions?

[Epanechnikov]

"My failure led me to think about economists, as opposed to economics, and they’re much easier to figure out. This is how it works. An economist starts with a few axioms, ones that bear a vague similarity to a small part of the human condition under restricted situations and in an idealized world. (You get my drift here?) From those axioms follows a theorem. More often than not this will be a theorem based upon rational behaviour. That theorem gets a name. And that’s the point I identify as being the problem: The jargonizing of complex ideas based upon irrelevant assumptions into an easily used and abused building block on which to build the edifice of nonsense that is modern economics.

 

 

 

Small assumption by small assumption, the economist builds up his theories into useless gibberish. By acceptance of each step he is able to kid himself he is making progress. And that’s why I struggle with economics. It is not mathematics where, barring mistakes, each step is true and indisputable and therefore you can accept it, even forget it, and move on. And others can do the same, using everyone else’s results without question. This you cannot do in a soft science. I’ve mentioned this in another blog, beware of anyone talking about ‘results’ in finance or economics, it says more about them and their perception of the world than it does about the subject."

 

 

 

 

Paul Wilmott's Blog: Economics Makes My Brain Hurt

[Gecko]

hmmm, it is up to the users to construct 2nd order explanations to bring it back into the real world then :)

 

it is always possible to construct different explanations by adding new assumptions, so if u find what you see unrealistic, then add your own assumptions ^_^

[One on One]

I pretty much agree with him completely, which is why I am interested in economic policy and theory not so much. Theories are cool though.

[pevdoki1]

The post he links to is even better:

 

Paul Wilmott's Blog: Results and Ideas: Two classical putdowns

 

I don't really have time now (micro prelim on Wednesday :(), but in short I agree with him. After a year of macro, I don't find "add more assumptions" to be a very good argument. What you start with in your models is preference relations that at least lead to utility representations, forward-looking agents etc. Complicate your assumptions all you want, but what can you buy in the real world with dynamic programming?

[murni]

The species Wilmott describes comprises maybe 5% of current economists, maybe even less. Has he never heard of empirical economists ?

[murni]

The post he links to is even better:

 

Paul Wilmott's Blog: Results and Ideas: Two classical putdowns

 

I don't really have time now (micro prelim on Wednesday :(), but in short I agree with him. After a year of macro, I don't find "add more assumptions" to be a very good argument. What you start with in your models is preference relations that at least lead to utility representations, forward-looking agents etc. Complicate your assumptions all you want, but what can you buy in the real world with dynamic programming?

 

First: typical macro models answer questions you never asked (why are real interest rates (short term) between 1 and 5% and not 500 percent, ...) and second: explaining the macroeconomy is a tough job, does anybody have anything better ?

And third: do not expect to learn much about modern macroeconomic analysis in a macro first year sequence. You learn some building blocks and some techniques, but typically not up-to-date stuff.

 

But I do not like DSGE, no ponzi, infinite horizon, representative agent models as well, I think one can gain more from empirical research these days.

[thesparky]

I'm glad behavioural and experimental economists exist.

[greeneggs]

It is not mathematics where, barring mistakes, each step is true and indisputable and therefore you can accept it, even forget it, and move on.

 

Even I know this isn't true after reading a little bit about mathematical history and taking a few phd level math courses. Economics actually isn't so different from math. You have a problem or a question that you're wondering about, you start making assumptions and seeing what conclusions you can draw from those assumptions - and then you go back and look at the assumptions more. Math has the axiom of choice, which people use more because so many powerful results depend on it rather than they actually beleive in it. Is this really so different than how economists make assumptions?

 

I haven't taken graduate macro, but I have done some stuff with computer algorithms and dynamic programming is very very applicable to the real world. It is used all the time by everything from mapquest to firms deciding optimal manufacturing processes.

 

Economics is all about giving you tools to gain insight into complicated problems - but all anybody seems to want is the answers, they don't want to understand what's going on or why. And so they complain that economics can't answer unanswerable questions.

 

Also I would point out that most of the assumptions made for empirical analysis fail to hold in reality with real data sets as well; it's not just theory where assumptions fail to hold and mistakes are made.

[Epanechnikov]

"Economics is all about giving you tools to gain insight into complicated problems - but all anybody seems to want is the answers, they don't want to understand what's going on or why. And so they complain that economics can't answer unanswerable questions."

 

That is a very good point.

[ThJefferson]

Since my senior year in college (07-08), I have dealt with the "problem" in economics described in the thread, i.e. building "the edifice of nonsense" from wrong assumptions about human behavior. Let me say up front that I think empirical economics is on the right track with finding better ways to handle data via more sophisticated statistical procedures and/or better data sets. My main objection to economics, on the other hand, comes from economic theory.

 

It infuriates me when I read a paper that assumes something that is not even empirically observed, but something that makes the math convenient. Say, for a simple example, that we assume people maximize their discounted utility when they make a decision over time periods. Dynamic programming or calculus of variations tells us how to solve for the person's decisions. This is a wrong assumption about human behavior. Why? Who can honestly say that they behave like that? Who sits down and solves the complicated math problem? I have a degree in math and cannot solve most dynamic programming problems by hand. The theorist might respond that people don't know that they act like that but they in fact do, or they act close to it. Why don't theoretical studies site empirical evidence defending their assumptions then?? The theorist might respond saying that it is too difficult to attain the required empirical evidence. If this is the case, then we still have a problem, in that we assume something that is not empirically justified.

 

What I would like to see in the future of economics are ALL models and results based on empirical evidence. This would lead to high emphasis on experimentation and observation, which is what every science has (or should have).

[Golden Rule]

What I would like to see in the future of economics are ALL models and results based on empirical evidence. This would lead to high emphasis on experimentation and observation, which is what every science has (or should have).

Have you ever spent a few days taking an "unrealistic" model and try to make it more consistent with empirical evidence? Here are three possible outcomes:

(1) the model becomes intractable

(2) the more realistic assumptions do not yield any new interesting results, they just make everything much more complicated (or the new results aren't particularly interesting or important and are not worth the complications)

(3) you've actually advanced the field. Congrats. Go submit to a good journal.

 

But (1) or (2) are probably much more common results for researchers.

 

When you make assumptions in a theoretic model, there are tradeoffs between simplicity and realism. A good economic model is probably going to have a good mix between the two.

 

Krugman has a nice essay on this here. The big reason why he has a solo Nobel Prize is because monopolistic competition was just as simple an assumption as was necessary to generate insights about trade and economic geography that were interesting at the time.

[Golden Rule]

Why don't theoretical studies site empirical evidence defending their assumptions then??

Actually, I've seen this quite a bit. Not always though. Simplicity and tractability is one reason. Another is that certain assumptions may yield results that have been well-studied in past literature.

 

The theorist might respond saying that it is too difficult to attain the required empirical evidence. If this is the case, then we still have a problem, in that we assume something that is not empirically justified.

By this logic, economists cannot attempt to offer any insight to problems for which we don't have perfect data, and we should never do anything. Unfortunately, in the real world, people have to actually make decisions without perfect data.

[ThJefferson]

Actually, I've seen this quite a bit. Not always though. Simplicity and tractability is one reason. Another is that certain assumptions may yield results that have been well-studied in past literature.

 

By this logic, economists cannot attempt to offer any insight to problems for which we don't have perfect data, and we should never do anything. Unfortunately, in the real world, people have to actually make decisions without perfect data.

 

To your first point. I hope to see more what you have seen. And I do agree that simplicity and tractability are important in economics.

 

To your second point. I didn't say anything about perfect data. Come to think of it, does any empirical study use perfect data? All I am saying is that some data is better than no data.

 

And if there is no data and the data cannot be accumulated, then the theorist should (and many might already) make an attempt to come up with an experimental design for validating/rejecting their results. Think of the alternative. A theorist comes up with a way that people decide to save money. For some reason the theory cannot be tested. Is it wise for public policy to use the theory to make judgments?

 

On a slightly different topic: I'd like to propose a fun challenge to the test magic community. Can anyone find a mathematical theorem in economics describing the way people make decisions that has been "verified" by empirical evidence? And when I say verified, I mean verified in the sense that Albert Einstein's theory of relativity has been verified by empirical observation. I am sure some are out there that I have not yet read; it would greatly help me ease my angst toward economic theory. If you find one, please send a link to the empirical paper that verifies the theorem.

[JohntheExistentialist]

I found the following quote from Krugman's article astoundingly fitting to this discussion:

 

In particular, I have no sympathy for those people who criticize the unrealistic simplifications of model-builders, and imagine that they achieve greater sophistication by avoiding stating their assumptions clearly. The point is to realize that economic models are metaphors, not truth. By all means express your thoughts in models, as pretty as possible (more on that below). But always remember that you may have gotten the metaphor wrong, and that someone else with a different metaphor may be seeing something that you are missing.

 

It first addresses the criticism from the Willmott/Taleb tribe. Personally I enjoyed the two Taleb's books and was, fortunately, forced to scrutinize about my economic training. However, the main bullets, I believe, should have been aiming at those economists who think they got the truth from their models, not the study of Economics itself. Milton Friedman had also commented that one shouldn't dismiss an economic model simply by its assumptions, which are defined to be unrealistic but if appropriately put can help us to gain deep insight.

 

On the other hand, we should always be conscious about the limitation of economic model and be especially cautious when it comes to policy implication. After all, it is just a simplified version of reality. This is particularly sobering for some arrogant and/or overly optimistic economists (like the Robert Merton described by Taleb) who thought they had discovered an objective, rigorous, scientific truth. As Krugman has put it, economic models are just metaphors. Anyone who claims he had grasped a comprehensive modeling of human/market behavior should be injected a dose of skepticism and self-reflection.

[Antichron]

And if there is no data and the data cannot be accumulated, then the theorist should (and many might already) make an attempt to come up with an experimental design for validating/rejecting their results. Think of the alternative. A theorist comes up with a way that people decide to save money. For some reason the theory cannot be tested. Is it wise for public policy to use the theory to make judgments?

 

Just some random thoughts:

 

Though my views on the way economics is practiced are undoubtedly Cambridged-biased, I think there may be some confusion in this thread about the role of economic theory in our understanding of economic behavior as well as some confusion about the allocation of tasks within the discipline.

 

I can't speak for how macroeconomics is done (simply my own ignorance), but my understanding of how micro works is that there are essentially four types of microeconomists: pure theorists, applied theorists, empirical economists, and policy people (and many economists wear multiple hats). Pure theorists typically generate (and characterize properties of) tools and frameworks. Applied theorists use these tools to construct and analyze models that explain real-world phenomena (that have typically been documented elsewhere) when existing models prove to be insufficient. A careful applied theorist should think long and hard about any assumptions that go into his/her model, and a strong goal should be parsimony (reducing the model to its key elements and nothing more) and tractability so that others can embed the model in a more realistic context. Empirical economists are the ones who take a model and adapt it to a particular setting (accounting for the key elements of the relevant institutions), generating empirical predictions which they then test (econometrically). (Experimental people often test the primitives of a model more directly.) Policy people are the ones who have a connection with the real world and understand well the institutional detail in place in a particular setting and, in suggesting policy, should be aware of the relevant empirical work supporting whatever claims they may make.

 

Additionally, there is a substantial amount of interaction in the opposite direction as well. People with closer connections with the real world are the ones who document facts that shape the direction of applied theory. This often necessitates the development of theoretical tools for handling situations not usefully analyzed with the current set of tools.

 

Obviously a model that is not falsifiable has no place in any of this. The same need not necessarily be true about a framework that itself is not falsifiable (e.g. game theory), since the models generated using such a framework should have empirical predictions.

 

To make a long story short, it isn't the people that axiomatically derive preferences that are the ones advising the government about what its tax policy should look like. It is not to say that they are not the ones writing op-eds and getting quoted in newspapers, though.

[Golden Rule]

To your second point. I didn't say anything about perfect data. Come to think of it, does any empirical study use perfect data? All I am saying is that some data is better than no data.

Ok, I was too vague -- by "perfect data" I meant any data that can do a full test of the theory in question, i.e. test every prediction of the theory. Most empirical papers will tend to test the most important aspects of the theory, and won't be able to fully identify all the parameters of the model.

A theorist comes up with a way that people decide to save money. For some reason the theory cannot be tested. Is it wise for public policy to use the theory to make judgments?

Ok, but consider this -- even if there is empirical evidence that a theory is valid, you still can raise questions about the external validity of the prior empirics. To say that the past study is relevant for the current policy decision, you're always going to need to make some untestable assumptions before making the decision.

 

That seems kind of hopeless, and it is to some extent, but I think the tools of economics -- whether empirical or theoretical -- when used appropriately can help us make better decisions. Though yes, I agree it's possible for overconfidence in a theory to lead to worse results than total ignorance. But it's overconfidence is the problem here, not the theorizing (people could just as well be overconfident about a view of the world with no scientific basis).

 

Can anyone find a mathematical theorem in economics describing the way people make decisions that has been "verified" by empirical evidence? And when I say verified, I mean verified in the sense that Albert Einstein's theory of relativity has been verified by empirical observation.

I'm confused by this request. Most empirical papers test economic models that involve some assumptions about decision-making. If you're thinking of direct tests of individual decision-making, look to the neuroeconomics literature.

[murni]

It infuriates me when I read a paper that assumes something that is not even empirically observed, but something that makes the math convenient. Say, for a simple example, that we assume people maximize their discounted utility when they make a decision over time periods. Dynamic programming or calculus of variations tells us how to solve for the person's decisions. This is a wrong assumption about human behavior. Why? Who can honestly say that they behave like that? Who sits down and solves the complicated math problem? .

 

The world is full of idiots and nearly noone can solve complicated dynamic programming problems. However, people act as if the would solve such problems, concave utility functions are a VERY realistic assumption, starting from that, it is most convinient to do some dyn prog. But that should be the starting point of the analysis (and every serious researcher treats it like that). Dyn prog is not wrong, people smooth their consumption, but it is just a rough description of reality, but the best tool out I can imagine to rationalize peoples behavior. Again: what is better ? It is not ok just to state: this is unrealistic, dont do it. If you tackle a very difficult problem, you should not reject every solution that is not perfect. As long as nobody comes up with something better, dyn prog ist the best thing I can imagine. As starting point of analysis, however.

[murni]

 

I'm confused by this request. Most empirical papers test economic models that involve some assumptions about decision-making. If you're thinking of direct tests of individual decision-making, look to the neuroeconomics literature.

Pleas keep in mind that (to the best of my knowledge) EVERY empirical study relies on UNTESTABLE assumptions.

[LouisBD]

A bit off topic:

 

The starting point of this debate was something from Paul Wilmott's blog. While the debate is interesting, I would like to point out that he (Paul Wilmott) is a 'very, very, very' controversial figure in the math fin community (just check out one of the many fora eg. nuclearphynance and you'll get my drift). I do not want to say anything about the quality of his ideas, but I know that citing him can be a bit risky (in the sense that he raises strong emotions).

[murni]

A bit off topic:

 

The starting point of this debate was something from Paul Wilmott's blog. While the debate is interesting, I would like to point out that he (Paul Wilmott) is a 'very, very, very' controversial figure in the math fin community (just check out one of the many fora eg. nuclearphynance and you'll get my drift). I do not want to say anything about the quality of his ideas, but I know that citing him can be a bit risky (in the sense that he raises strong emotions).

 

could you elaborate on that ? It would take ours to understand his status by browsing through the web.

[Epanechnikov]

realestate

 

 

DON'T VISIT THE ABOVE SITE.

 

IT IS AN ATTACK SITE!!!!

[ThJefferson]

I found the following quote from Krugman's article astoundingly fitting to this discussion:

 

In particular, I have no sympathy for those people who criticize the unrealistic simplifications of model-builders, and imagine that they achieve greater sophistication by avoiding stating their assumptions clearly. The point is to realize that economic models are metaphors, not truth. By all means express your thoughts in models, as pretty as possible (more on that below). But always remember that you may have gotten the metaphor wrong, and that someone else with a different metaphor may be seeing something that you are missing.

 

It first addresses the criticism from the Willmott/Taleb tribe. Personally I enjoyed the two Taleb's books and was, fortunately, forced to scrutinize about my economic training. However, the main bullets, I believe, should have been aiming at those economists who think they got the truth from their models, not the study of Economics itself. Milton Friedman had also commented that one shouldn't dismiss an economic model simply by its assumptions, which are defined to be unrealistic but if appropriately put can help us to gain deep insight.

 

On the other hand, we should always be conscious about the limitation of economic model and be especially cautious when it comes to policy implication. After all, it is just a simplified version of reality. This is particularly sobering for some arrogant and/or overly optimistic economists (like the Robert Merton described by Taleb) who thought they had discovered an objective, rigorous, scientific truth. As Krugman has put it, economic models are just metaphors. Anyone who claims he had grasped a comprehensive modeling of human/market behavior should be injected a dose of skepticism and self-reflection.

 

Krugman makes a good point about realizing economic models are metaphors and not "truth." I especially find his last point helpful.

 

I don't fully agree with Friedman's comment. If a model with an unrealistic assumption helps us gain deep insight, we must remain skeptical of that insight because it is based on an unrealistic assumption. On the other hand, there is a range of realism that assumptions can take on. The less realistic an assumption, the more skeptical we should be.

[ThJefferson]

The world is full of idiots and nearly noone can solve complicated dynamic programming problems. However, people act as if the would solve such problems, concave utility functions are a VERY realistic assumption, starting from that, it is most convinient to do some dyn prog. But that should be the starting point of the analysis (and every serious researcher treats it like that). Dyn prog is not wrong, people smooth their consumption, but it is just a rough description of reality, but the best tool out I can imagine to rationalize peoples behavior. Again: what is better ? It is not ok just to state: this is unrealistic, dont do it. If you tackle a very difficult problem, you should not reject every solution that is not perfect. As long as nobody comes up with something better, dyn prog ist the best thing I can imagine. As starting point of analysis, however.

 

I never said dynamic programming is wrong. I think it is a great mathematical tool that has a lot of application in many sciences. It, however, is not always used best when dealing with an economic problem. And because people are complex beings, a lot of the "simple" assumptions used in dynamic programming problems make the drawn conclusions wrong and/or useless.

 

I agree with what pevdoki1 said earlier:

"After a year of macro, I don't find "add more assumptions" to be a very good argument. What you start with in your models is preference relations that at least lead to utility representations, forward-looking agents etc. Complicate your assumptions all you want, but what can you buy in the real world with dynamic programming [with respect to human behavior]?"

 

You say that "It is not ok just to state: this is unrealistic, dont do it." I agree with you to an extent, in that not much research could be done if every economist acted like this. And so yes, we as scientists need to judge when an "unrealistic" assumption is justifiable in a model. For example, if it makes the model extremely simple, then it might be worthwhile to put it in there to see what happens.

 

After reading through some of the other responses, it seems like a model should be rated in three categories: Simplicity, tractability, and realism, where realism is based on how realistic the assumptions are. I think there should be some minimum rating in each category that a model must have to be considered worthwhile.

 

For example, take a physicist who is estimating the trajectory of a fired cannon ball. Suppose he assumes that the force of gravity is twice as much as its true value. This is a poor assumption and he will get a wrong answer. Suppose, however, that the physicist gets his paper published in a good journal because his results are simple and tractable. People read his work and continue to use his assumption and it becomes common practice. In my opinion, the physicist was wasting his own research effort as well as others research effort because the assumption used is wildly incorrect and yields incorrect answers. Most researchers don't have time to consider altering assumptions that have been used time and time again and, as a result, take those assumptions for granted. In an ideal world, every researcher should take a lot of time to challange every assumption that has been used in past work. Unfortunately, scientists have deadlines and they must "publish or perish." I feel that this phenomenon is a central criticism of current economic theory.

 

At the same time, it might just be that I am misinformed with the education I have received (undergrad degree and the first year of phd econ). I want to be in the camp among the TM'ers that are confident about the way economic thoery and application is performed. You say that "people act as if the would solve such problems" as you refer to people maximizing their utility over periods. Could you direct me to a few studies that have documented this?

[Epanechnikov]

"people act as if the would solve such problems"

 

I can definitely say that this is an unrealistic assumption. See for example Hyperbolic discounting.

 

How can an individual solve his utility maximization problem if he prefers smaller amounts of utility (or cash) now rather than bigger amounts at a later point of time?

[Antichron]

I can definitely say that this is an unrealistic assumption. See for example Hyperbolic discounting.

 

How can an individual solve his utility maximization problem if he prefers smaller amounts of utility (or cash) now rather than bigger amounts at a later point of time?

 

The way economics progresses is exactly by (1) pointing out deficiencies in current ways of thinking, (2) showing how it is inconsistent with empirical evidence (i.e. that people make time inconsistent choices), and (3) proposing an alternative model that captures this (i.e. Phelps-Pollak, 1968) in a way that makes additional testable predictions (i.e. that there is scope for commitment mechanisms (Laibson, 1997) and we should see them in the data). I noticed that when I was taking undergraduate economics courses, there was a lot of (1) and (2) going on, but very little (3), and from the comments in this thread, it appears that I wasn't alone. However, I think the difference between scientific progress and scientific regress is exactly (3).