How economists think differently from other humans

[136or142]

Paul Solman report on PBS Newshour

http://www.pbs.org/newshour/bb/economists-think-differently-humans/

[© tweed]

resident trained economist lambasts the video in 3 2 1 ...

[136or142]

The guy in the interview, Richard Thaler, is a trained economist while Paul Solman has an MBA focusing in economics.

[Ban my account ffs!]

The guy in the interview, Richard Thaler, is a trained economist while Paul Solman has an MBA focusing in economics.

Not only do they think differently than other humans...they cant seem to ever agree with each other either.

Economists would criticize me for being so fascinated by a forest.  And yet you can never really understand an economy without understanding a forest.

[Gustaf]

It's a bit dishonest. Firstly, behavioural economics is not a one-man crusade it's a large field in economics.

Secondly, no economist ever believed everyone was a perfectly rational machine or whatever. The question, which is still an open one, is to what extent these individual idiosyncracies must be taken into account in order for aggregate predictions to work.

[136or142]

It's a bit dishonest. Firstly, behavioural economics is not a one-man crusade it's a large field in economics.

Secondly, no economist ever believed everyone was a perfectly rational machine or whatever. The question, which is still an open one, is to what extent these individual idiosyncracies must be taken into account in order for aggregate predictions to work.

1.Richard Thaler said exactly that.  They said he founded the field, not that he's the only person in it.

2.There were all sorts of economic theory that came out of the 'rational person' hypothesis though that behavioral economists are showing don't stand up to scrutiny. This piece mentions a couple of them, especially the 'sunk cost' one.

[ag]

It's a bit dishonest. Firstly, behavioural economics is not a one-man crusade it's a large field in economics.

Secondly, no economist ever believed everyone was a perfectly rational machine or whatever. The question, which is still an open one, is to what extent these individual idiosyncracies must be taken into account in order for aggregate predictions to work.

1.Richard Thaler said exactly that.  They said he founded the field, not that he's the only person in it.

Even that is an... exaggeration

[ag]

resident trained economist lambasts the video in 3 2 1 ...

I have not seen the video - but, trust me, as a trained economist I do have a fairly nuanced view of these things

[ag]

Economics, in general, is a behavioral science All of economics is. And, trust me, the profession is very much aware of it. For gods' sake, Kahneman got his Nobel back in 2002! Long before he published his popular book

Nevertheless, one thing to be made very clear. Before one says that "rationality is contradicted by..." one has to understand very well what "rationality" means - for an economist. Perhaps, if there is a popular demand here, I could do a little exposition. This is very much my area, actually

[ag]

The guy in the interview, Richard Thaler, is a trained economist while Paul Solman has an MBA focusing in economics.

Not only do they think differently than other humans...they cant seem to ever agree with each other either.

Actually, you are going to be surprised on how many things there is a general agreement.

Anyway, it is unquestionably true that people with some economics training sometimes tend to behave strange. In fact, economists study that as well

[136or142]

The guy in the interview, Richard Thaler, is a trained economist while Paul Solman has an MBA focusing in economics.

Not only do they think differently than other humans...they cant seem to ever agree with each other either.

Actually, you are going to be surprised on how many things there is a general agreement.

Anyway, it is unquestionably true that people with some economics training sometimes tend to behave strange. In fact, economists study that as well

If you google "where economists agree" you'll get slightly over 2 million results and if you 'tor' "where economists agree" you'll get 2 million results and offers to wack any economist who doesn't agree

[Gustaf]

It's a bit dishonest. Firstly, behavioural economics is not a one-man crusade it's a large field in economics.

Secondly, no economist ever believed everyone was a perfectly rational machine or whatever. The question, which is still an open one, is to what extent these individual idiosyncracies must be taken into account in order for aggregate predictions to work.

1.Richard Thaler said exactly that.  They said he founded the field, not that he's the only person in it.

2.There were all sorts of economic theory that came out of the 'rational person' hypothesis though that behavioral economists are showing don't stand up to scrutiny. This piece mentions a couple of them, especially the 'sunk cost' one.

I'm doing a PhD mostly about behavioural econ - I'm well aware.

My point is that there is a difference between the behavioural lab experiments and transferring that to aggregate macro variables.

And as AG points out I think Thaler's importance is being a bit inflated in this video.

[ag]

It's a bit dishonest. Firstly, behavioural economics is not a one-man crusade it's a large field in economics.

Secondly, no economist ever believed everyone was a perfectly rational machine or whatever. The question, which is still an open one, is to what extent these individual idiosyncracies must be taken into account in order for aggregate predictions to work.

1.Richard Thaler said exactly that.  They said he founded the field, not that he's the only person in it.

2.There were all sorts of economic theory that came out of the 'rational person' hypothesis though that behavioral economists are showing don't stand up to scrutiny. This piece mentions a couple of them, especially the 'sunk cost' one.

I'm doing a PhD mostly about behavioural econ - I'm well aware.

Here may go our anonymity Our profession is far too small a town for this to be preserved on any forum less inane than the econjobrumors  

Still, just out of curiosity, are you doing it in Europe or in the US?

[Oakvale]

It's confusing to me why some still think saying HEY PEOPLE AREN'T PURELY RATIONAL! is some kind of devastating blow to the field.

[All Along The Watchtower]

It's confusing to me why some still think saying HEY PEOPLE AREN'T PURELY RATIONAL! is some kind of devastating blow to the field.

We still live in the mid-19th century or something.

[Ban my account ffs!]

It's confusing to me why some still think saying HEY PEOPLE AREN'T PURELY RATIONAL! is some kind of devastating blow to the field.

Jaichind

[Gustaf]

It's a bit dishonest. Firstly, behavioural economics is not a one-man crusade it's a large field in economics.

Secondly, no economist ever believed everyone was a perfectly rational machine or whatever. The question, which is still an open one, is to what extent these individual idiosyncracies must be taken into account in order for aggregate predictions to work.

1.Richard Thaler said exactly that.  They said he founded the field, not that he's the only person in it.

2.There were all sorts of economic theory that came out of the 'rational person' hypothesis though that behavioral economists are showing don't stand up to scrutiny. This piece mentions a couple of them, especially the 'sunk cost' one.

I'm doing a PhD mostly about behavioural econ - I'm well aware.

Here may go our anonymity Our profession is far too small a town for this to be preserved on any forum less inane than the econjobrumors  

Still, just out of curiosity, are you doing it in Europe or in the US?

Haha, I've stayed away from that one, because every time a friend referenced it, it sounded awful.

I'm in Europe, not the US.

[anvi]

Nevertheless, one thing to be made very clear. Before one says that "rationality is contradicted by..." one has to understand very well what "rationality" means - for an economist. Perhaps, if there is a popular demand here, I could do a little exposition. This is very much my area, actually

I'm not sure one person constitutes popular demand.  But I'd be interested in your exposition, ag.

[ag]

Nevertheless, one thing to be made very clear. Before one says that "rationality is contradicted by..." one has to understand very well what "rationality" means - for an economist. Perhaps, if there is a popular demand here, I could do a little exposition. This is very much my area, actually

I'm not sure one person constitutes popular demand.  But I'd be interested in your exposition, ag.

Ok, I will try

[ag]

I will make it into several short posts.

1. First of all, one should start with the explanation of what is it that economists consider worth studying. These days, it is a professional consensus that everything starts with individual behavior. You could argue that, perhaps, aggregating from individual (micro) level to macro phenomena is difficult - and, indeed, it is. But even there economists tend to start from "micro foundations." In this sense, all of economics is, really, a science of individual choice - i.e., a behavioral science. Individual choice is what we try to understand. That - understanding individual choices - is what I will be talking about.

2. So economists start with a record of choices. Think of it as a record of observations by a keeper in the monkey house in your neighboring zoo. "I offered the chimp an apple and an orange and the chimp chose orange".  Now, notice, that if I am trying to explain this behavior I should try to record everything that is relevant for the chimp choice.  But what is this "everything"? Certainly, no record can include every single detail, so the keeper has to choose what to write down. For instance, in the example above I skipped the day of the week: surely, the chimp does not know whether it is Monday or Tuesday. Of course, it COULD happen that the chimp is, in fact aware: on Monday the zoo is closed, so he does not see the visitors, and that affects the choice. If that is the case, I am off to a bad start: I did not record pertinent information. Why? Perhaps I had a wrong theory of behavior (did not realize the chimp knew days of the week).

To sum up:

a) even to record data one needs some theory of what is relevant for behavior.
b) a complete description of a choice-relevant state of the world we shall call a consumption bundle
c) a collection of all such possible states we shall call a consumption space. Note, that the specification of the consumption space is a crucial part of our theory of behavior.

To continue.

[ag]

Ok. Now on to "rationality". When an economist is saying that an individual is "rational" s/he means, basically the following.

1. She means that individuals can make statements of the sort "apple is better than orange" about consumption bundles available (preferences)
2. She means that this individual can, in fact, compare in this way any two bundles out there (preferences are complete).
3. She means that if for some individual apple is better than orange and orange is better than banana this must logically imply that for this individual apple is better than banana (preferences are transitive).
4. Finally, she means that when making choices this individual maximizes his own preference: i.e., chooses the best bundle available according to this preference.

That is it. This is the only meaning of "rationality" that a trained economist will recognize. In a more general setting with time and uncertainty, a few of other assumptions would be a part of a "standard" model, but rationality is just this: maximizing a complete and transitive preference relation.

Now, these four assumptions are highly non-trivial (of which later). But, for the moment, the question for us will be, what are their testable implications: i.e., what do we need to observe to be able to claim that "rationality failed" to explain behavior.

To be continued.

[anvi]

Thanks much, ag.  I am following you so far.  I have a few questions already, but I will wait for your further posts to ask anything.  This is quite helpful for me, so thanks again!

[ag]

It turns out that, assuming we have the right consumption space, the implications of rationality, as defined above, are easy (well...) to characterize.

Essentially, the idea is that if you choose an apple, while orange is also available, you can never choose orange if an apple is present. The reason is obvious: you choosing an apple in the presence of an orange means that you liked the apple more than the orange. Since your preferences are defined independently of the alternatives available, this means that whenever you see an orange and an apple next to each other you will prefer the apple and, hence, not choose the orange. This is what economists call Weak Axiom of Revealed Preference (WARP) and it is an obvious consequence of rational preference maximization.  Equivalently, it can be restated as saying that removing an alternative you were not choosing should not affect your choice.

WARP can be strengthened by also prohibiting cyclical choices: for instance, if you choose an apple over orange and an orange over banana, you should never choose a banana over apple. In a way, WARP itself is prohibiting cycles length two (apple over banana, banana over apple). The stronger version (SARP) prohibits cycles of any length.

It turns out that SARP is, really, the necessary and sufficient restriction on choices for them to be "explainable" with rational preferenes ("rationalizable"). In other words, not only it is an implication of our "rational" model, but also if choices satisfy SARP then one can find a rational preference that would imply exactly such choices. Hence, SARP gives us what we can call an "empirical content" of rationality: if it fails, we can argue that the chooser is not rational. If it holds, however, we cannot reject the hypothesis of rationality.

Note, however, what I remarked at the beginning of this installment: I have assumed that we have the correct consumption space. If we misspecify the consumption space by omitting choice-relevant information (as when we forgot to mention the day of the week talking about that zoo monkey and his apples and oranges) we shall sometimes observe what would seem to be violations of SARP or even WARP: sometimes the monkey chooses apple over orange, other times it chooses orange over apple. Of course, that does not mean the monkey is not rational: it may only mean that the monkey prefers monday apples over monday oranges and tuesday oranges over tuesday apples and we just goofed up in our records, by failing to distinguish these consumption bundles. Thus, an empirical violation of rationality can in this case be "repaired" by simply making the consumption space "finer": SARP holds, once we correctly distinguish between different consumption bundles.

Unfortunately, it turns out that this is not a very interesting observation. The reason for this is that IT CAN ALWAYS BE DONE. If I define each apple as different from every other apple, so that I never see the same apple twice, the axioms of revealed preference have no bite whatsoever.  So, any empirical observation of failure of rationality can be ascribed to simply a misspecification of the consumption space. The reason this is bad is that it implies that, thought of this way, rationality is not a testable theory whatsoever: it can never be shown to be wrong, no matter what observation you have. Of course, that would take us completely outside of the realm of science. Hence, even though it may well be true that the mistake is not in assuming rationality, but in specifying the consumption space, we cannot independently test this.

To sum up:

1. Rationality is only a testable theory when taken together with a stand about what is relevant for choice (specification of the consumption space). Hence, our model is never just rationality: it is rationality together with the statement of what matters for individual choice.

2. With consumption space fixed, SARP (lack of cyclical choices) provides the full empirical content of  rationality.

To be continued.

[ag]

So, should I be continuing, or this is just boring?

[anvi]

Not boring at all, ag.  I've been off Atlas for a while working on some other things, so I'd like it if you continued.

But maybe I'll ask some of my questions now just to confirm my interest and curiosity.  This is about as much of a non-economist talking as there can be, so you have fair warning for the simple-mindedness of any questions I ask.

On a first couple of readings, it sounds as if what is being called "rationality" here boils down to the capacity of a person to have preferences, to articulate them to themselves or others, and to prefer consistently given the consistency of the consumption space.  In other words, "rationality" seems to be "consistency of preference" that is predictable under conditions which are clear to whomever is doing the empirical economic research.  Is this characterization accurate?

If so, then one question that comes to my mind immediately is whether or not this conception of economic rationality can take into account the possibility that a person, even in cases where the consumption space stays the same, can change their preferences?  This change, let's say, would not be relevant to refining our understanding of the consumption space.  Let's say as a child I hate spinach, and as an adult, I love it (which in my case happens to be true); and that change is not relevant to what kinds of spinach have been available to me or the circumstances in which I'm eating a particular meal.  It's just that, when I was a child, I didn't like spinach, and when I'm an adult, I do.  My preference has changed, it's not consistent, and nothing about my consumption space could predict the change.  Would such a change in preference be considered on the view you're describing "irrational?"  And would it be considered "irrational" not because I'm crazy, but merely because nothing about the consumption space could help predict the change? 

I ask this not because I'm just trying to find a snarky exception or a set of them, but because I guess it could have implications for lots of other things.  Calculations of expected value, for instances, may be rendered much more uncertain if in people we are dealing with possible changes in preference which can't be predicted.  I guess there may be other kinds of important implications too.  But I'll leave the question there for now.

Thanks again, ag; all this is helpful.