Thursday, October 26, 2006

Econ101: Preference

Hi again. It’s time now for the third installment of our Econ101 series. After introducing some key concepts needed to speak the 'baby' language of economics, let’s now turn to a more structured and systematic approach. That is, we’re going to follow a text book structure, without having to religiously adopt its verbal and mathematical presentation.1 Yes, we’re going to do it the Cafe-way (and that may as well mean irregular schedule!). Lean back and enjoy your coffee.

When we analyze an individual behavior (in making decision, or more accurately in choosing between available options, given his constraints), we need to make some assumptions with regards to his preference. The most important assumption is that the guy is rational.

What do we mean by rational in this context? We mean his preference relation is complete and transitive. Complete means you can describe the relation between any two goods that he is considering. So, if the guy is considering apple, orange, and banana, you have to be able to say whether he prefers apple to orange. Also, you have to be able to tell his preference over apple and banana, as well as banana and orange. The good thing is, telling that he likes banana as much as apple is a valid statement – we say he is indifferent between banana and apple.2

Transitive means consistent in choice ordering. If our guy prefers apple to orange and orange to banana, he should prefer apple to banana. Yes, this assumption is strong: I know a friend who likes Manchester United more than Liverpool and prefers Liverpool to AC Milan, but he loves AC Milan more than MU. That’s fine, but for now, let’s assume away such intransitivity. Don’t worry; we will relax the transitivity assumption sometime later. (That soccer freak friend of mine; he ain't crazy, he's just irrational, as we 'have to' call him).

How do we conveniently talk about preference? By assigning numbers to the preference order. In our example, the preference order of the guy is: apple-orange-banana (in decreasing order of importance). Now let’s assign some numbers. Yes, we’re assuming that we somehow can measure satisfaction. Suppose the satisfaction experienced by the guy if he consumes an apple is 10. Then, the corresponding number of an orange should be less than 10. Say 7. How about a banana? Yes, it should be less than 7. Say 5. We say, for the guy, the utility of apple, orange, and banana are 10, 7, and 5, respectively. Can we change the numbers? Yes, we can. But mind the order! So, if you like you can use 1,000-700-5, or 356,464-100-0.3. But combination like 3-5-1 or 7-4-10 is not allowed, given the guy’s preference. You see, utility function is an ordinal concept, not cardinal. That is, all that matters is the order, not the number itself. So, if we can use simple numbers as long as we keep the order, why make it complicated?

Stay tuned.


1 The text I’m referring to is Mas-Colell, Winston and Green. This book is one of the most elaborate modern microeconomics text. However, it is designed for graduate course. In one of its strongest part i.e. general equilibrium analysis, it uses differential topology, so you might want to consult some graduate math texts. Many times, students find it useful to combine this text with the more compact, Varian. If you want a good text for undergraduate level, we recommend Mankiw.

2 Seriously, guys, this is just an illustration. I really don't care if you happen to like orange more than apple :-)


  1. I think the assumption that people are rational is the biggest weakness of economics. Without it economics wouldn't really function, with it human society wouldn't.

  2. Obviously rationality is also economics biggest strength...

  3. if i had too much apples, i don't mind to have an orange. having too much of something sometimes alters my preference.

  4. John,
    quoting Milton Friedman, you don't have to agree that people are rational. In fact, the assumption needs not be correct. But see whether the assumption makes the theory predicts human behavior quite well...

    And, according to David Friedman (Milton's son), the reason we need the rationality assumption is to give a reason why people do something. Otherwise, we'll say that people eats orange or throw it away just for the sake of eating or throwing away.

  5. BTW guys,
    anyone who prefers MU than any clubs is not only irrational. he/she is crazy...

  6. I know this is a simple-minded question, but I'm wondering how you guys would answer it

    where are these preferences coming from?

  7. Anymatters, that's called diminishing marginal utility. We'll get to there. But please read this. We always talk at the margin. Sorry, your 'too much apples' is irrelevant.

    Roby, you mean the information about his preference? From observation, or by asking him directly. Please see this again. We get information either via revealed preference (RP) or stated preference (SP).

  8. I get the general basics (really, i'm not that stupid and i am very interested in this subject, but without trying to distract too much, i want to pick up where John there left us).

    a.p.: given that the assumptions aren't required to be correct, i can make three gazillions statement to be right. Any statement will be true if we can custom order the assumptions, no?
    In most behavioral economics, the theories really explain certain observations pretty well, that much is right. But does this predict (as opposed to transcribing passing observation)?
    To be useful at all as predictions, the model really falls back for assumptions, and creating specific (and/or incorrect) assumptions to simplify dangerously distort the output when the model is used for other things.

    To me, this is the biggest problem with economy. As much as economists would like to claim that they could predict things, the fact remains, economists are still at work these days trying to explain too many things.

    which rebuts the entire premise that economics really predict things: if it were really that predictable, your job would've been done some years ago and we'll have a perfectly uniform and functional market already.

    of course i understand that your issue at the moment here isn't really rationality: per se. Maybe you will offer that in a separate lesson? we'll throw in morality, too and i can have a jolly good exercise for my aging braincells.

    PS: guys, try to be more rational in posting the next lesson, otherwise we'll forget all these by the time you get to the numbery bits.

  9. re: i mean rational in scheduling. i am not at all suggesting your lack of irrationality.

    anyone who custom made assumptions to arrive at a designated hypothesis must be very rational indeed.

  10. no - i mean how come, say, i have the preference for apples not oranges? why do i prefer indonesian food?

    in short, it's not about how observers know the preferences of individual, but about how the individuals themselves acquire preferences in the first place.

  11. Well economists seem to assume rationailty even in situations where outcomes don't match reality (when they see this, generally they construct some extravagant and not so plausible model which gives them the "right" result -- not v scientific!).

    Look at the iterated prisoner's dilemma for example, rationality fails miserably for finite games. This occurs for many very basic games.

    And as most of us live finitely - QED.

    There are other ways of modelling human decision making -- e.g. evolutionary equilibria. Nnot all that practical, but they successfully model fundamental facets of human behaviour which rationality cannot.

  12. Well economists seem to assume rationailty even in situations where outcomes don't match reality (when they see this, generally they construct some extravagant and not so plausible model which gives them the "right" result -- not v scientific!).

    Look at the iterated prisoner's dilemma for example, rationality fails miserably for finite games. This occurs for many very basic games.

    And as most of us live finitely - QED.

    There are other ways of modelling human decision making -- e.g. evolutionary equilibria. Nnot all that practical, but they successfully model fundamental facets of human behaviour which rationality cannot.

  13. rationality is a fine assumption, especially from the utilitarian point of view, it simplifies a lot of things.
    it is also a useful assumption for political and economical purposes.

    the problem is only from the scientific point of view: the rationality assumption does not make a deep connection to reality.

    ps. i know that not all economists agree with milton friedman in this case (including some heavy weights too).

  14. Heavy!

    If an early post allegedly explaining "the basics" get such comments, can't imagine what will come out of lessons "intermediate" or "advanced topics".

    John and T/S
    Think of economists as behavioral statisticians. Sure, at different times, on different issues, different individuals may not act rationally. Most economists are realistic enough to recognize this.

    Economists, however, also believe that in most cases, for a specific decision, at a specific time, on aggregate, most people (or more than not) behave rationally. I think it's quite a reasonable assumption with good enough predictive power to describe many aggregate phenomena.

    Obviously, there are some cases where "outcomes do not match reality [as defined by the rationality assumption]". So? Does it mean it's high time to drop the rationality assumption?

    Of course not: We shouldn't throw away the baby along with the bath water. After all, it did quite a good job in explaining a lot of social phenomena.

    This merely shows that economics -- like any other natural and social sciences -- is far from complete. Behavioral economics (also the evolutionary equilibria approach, which, by the way, is being absorbed into economics through game theory) among others, is part of the effort to move towards refining or complementing economics's rationality assumption.

    Economics can't answer your question about the origin of preferences. I think you should ask evolutionary biologists for that.

  15. arya - that questions is one of the reasons i dropped out from the econ program (i just realized i have that mas-collel book, we used that book when i took the micro class here years ago). i think the answer to that question is less related to the brain than to the social structure. but that's another issue.

    and one more thing: i think one of the problem with the rationality assumption is related to the explanation for aggregated outcome. thomas schelling put this clearly in his micromotive and macrobehavior book : inference of individual preferences from collective outcome is tenuous.

    ps: i believe ujang has something to say. he should have put his comment here, not on my friendster :D

  16. This comment has been removed by a blog administrator.

  17. i'll pick it up another time. sorry guys to be too troublesome a student :D

  18. Arya: Actually I sometimes wonder about the relevance (and possibly the plausibility) of the basics (Debreu etc.) being taught to most econ undergrads.

    I read an interesting piece by JEAN PAUL SCHMETZ [] I must admit I don't know about his background, but it made me stop to think (which is good i suppose :)).

  19. John, Roby, and T/S. Thanks for the elaboration on the rationality assumption. Arya's comment I think is what I would have responded with (thanks, Arya). But let me add some points.

    Roby, Arya's right. I can't answer your question. Sorry. (But if it were an exam, I would have written this on my paper for a partial credit: preference comes from the fact that choices are constrained)

    T/S, thanks for you thoughtful comment. I guess we would have to address your concern (and others) in a separate posting. If time permits -- ah, forgive my escape -- that would focus on the bounded rationality literature. We're aware that in the current frontier of economics, this issue of how strict we are on the rationality assumption is high on the list. Many textbooks have sporadically touched on this.

    John, as you're into game theory I would suggest you to search the works by Ariel Rubinstein on this rationality issue. Sorry can't provide you the links now.

  20. well,i'm still rather disappointed with the answer to be honest. two things:

    1. ap statement that the assumptions need not be correct. i think that's really ridiculous. of course you can claim just about anything, but they will be as much random rambling as they are theories. Faulty assumptions lead to fantasies, not theories.

    if you claim that the assumption doesn't have to be correct, why did we bother picking on various gov't statements etc, on this very space for the last few months (or a year, yes :D). afterall, they have their own assumptions to make for bold prediction, and well, they don't have to be correct, right?

    2. the assumption that people are rational, but yes, that can go into its own post.

  21. and well, roby, preferences come from needs. very simple.

  22. Rationality isn't actually a bad assumption for many applications. It is a simplification for sure, but also can lead to very clean results (which don't deviate too much from "reality") -- something which bounded rationality doesn't generally give you...

    I believe many economists (and their students) are brainwashed into thinking that rationality is the only way to go. I have seen many finite game models where rationality seems to complicate things more than anything else. (Experiment: see who is more successful at playing the prisoner's dilemma game, a group of economics or non-economics students :))

    Another plus for rationality is that economists don't bother spending too much time on philosophising on the inner workings of the mind -- something a psychologist friend really loved the idea of.

    It's just a matter of using the right tool for the job!

    Maybe I just need to get back to college, great post BTW :)

  23. T/S: given that the assumptions aren't required to be correct, i can make three gazillions statement to be right.

    AP: you may, but necessarily that you are able to -- in the sense of making your building blocks of assumptions consistent with your theory and what it predicts.

    take the extreme situation. let's say that the the neoclassical assumptions that people are rational (complete, transitive and non-satiated) is incorrect. nevertheless, the theory predicts that when the price of a good increases, compared with other goods and everything else constant, people will demand for less. true, it may not be correct all the time. but it predicts human behavior very well. not perfectly, but still very well.

    now, let's say we want to assume that people are altruistic, not selfish, and many other 'ideal' assumptions. we can give thousands arguments why it is so. but can we make a good model of human behavior based on those assumptions?

    but then, we can reverse the question. if the theory predicts human behavior very well, how true must the assumptions be?

  24. I had this very discussion with an economist after complaining about this post and they offered the same argument.

    But really, the difference between theorizing and storytelling is the very fundamental fact that theories have assumptions to support the statement and therefore its conclusion. a made up assumption and a theory based on made up assumption is as much fiction as it is a theory - which is why i raised the issue.

    I've no problems with the theories explaining a phenomenon. Explanation (eg. to describe the mechanics of something) is perfectly acceptable, yes. Economics is indeed very useful to understand the mechanics of things.

    It is the predictive qualities of these that i am questioning (again particularly with the dismissal of correct assumptions).

    As i mentioned in the email, i find the theories to be acceptable and reasonable when it deals with the basics (eg. basic needs), as you well pointed out, it works in the neoclassical sense.

    but soon as you take the limitation (of basic needs) out and you apply the same stuff in a bigger pond with no obvious limit, then the predictive qualities fall apart.

    This is very much so because the assumptions NEED to be correct. In a bigger environment (and thus bigger models), it is harder to provide for reliable facts (and therefore to have the assumptions).

    Again, my issues there are with the need of the assumptions to be correct (and how it affects the predictive quality of the theory).

    In a neoclassical sense as you mentioned, if you measure apple and oranges according to their needs and therefore instead of having preference, you have the qualities to which consuming the said items fulfills certain needs, then we will have no issues about assuming the rationalities of the decision. It simply illustrate the degree of which certain items satisfy certain needs.

    then we can go full circle and respond to Roby why Preference comes from Need.

  25. i don't think the theory has any predictive power for human behavior at all. if we have the ability to predict then we should be filthy rich already by predicting stock prices.

    we don't even have a complete explanation yet for some *theoretical* problems such as the ultimatum game; not to mention predicting real cases (i can easily give a large number of cases where rational assumption can fail).

    treespotter - your explanation is ok. however, it is still problematic. first, i don't think it can explain (in non-trivial way) the variation in preferences. i need to eat, and i prefer pempek. so you still need to explain why pempek, not ketoprak. of course you can say you need pempek, but i don't really know what that means.
    so it's not clear how need translates to preferences

    on the other hand, the need explanation has a stint of functionality argument which is problematic because of its post-factum nature; you can explain anything away by need, so it's not really helpful.

    i tend to aggree with aco, preference arises from constraint. or, in my lingo, structure; although structure can enable *and* constraint.

    a.p:"if the theory predicts human behavior very well, how true must the assumptions be?"

    try to look at the papers about social outcomes in physics journals. their models fit the data well (not predicting anything though); but most of them are silly models (arise from silly assumptions). the thing is that the mapping from assumption/mechanism to outcome is many to one.

    anyway, two general comments:
    maybe it's not really fair to the cafe to criticize fundamental of economics. it is econ 101 after all.
    shouldn't we just accept all assumptions in 101?

    the second thing is, i always become very nervous whenever anyone offers one relatively simple mechanism (e.g., market, rationality, preferences, networks, culture, ideology, biology, psychology) for real social phenomena. if it were only for intellectual exercise, then it is perfectly fine. we, scientists, build models to simplify things. But, of course, models are not substitute for reality (assuming we are all realists, no postmo crowds here).

  26. All,
    I suggest you look at this link pointed out by John Orford. That, I think, is as far a claim as any economist can make.

    Are pragmatists part of the post-mo crowd? ;-) A pragmatist would ask: What is "reality" anyway? ;-) ;-)

    Also, we should take up this topic on the origin of preferences in your blog (BTW, which one is it? It seems, you keep changing it). Yeah, you're right, I should have added "sociologists" on people to ask on it. ;-)

  27. how can you say it's irrelevant. there still contains a preference issue cos we still need to choose between two or three things. i'm talking changes here. the diminishing marginal utility concerns for one thing. in reality, there are choices. the diminishing marginal utility may change our preference. i think, econometrically, there might be sort of change in the preference function, considering the variable of the diminishing marginal utility.

  28. I would not submit myself to the one-million dollar question on whether an assumption should be realistic. Friedman, the old one, has a point here,but oftentimes, alas, we forget to empirically examine how good the predictive power is.

    Instead I subscribe myself to more pragmatic opinion that the assumption facilitates the gradual understanding on the issue. You can then restate the assumption, once you feel it is deemed irrelevant. It is more a learning tools, indeed.

  29. no offence to pick this up at Roby's only we need to register at multiply and that's so blah! (unless he's another place not on multiply/friendster, etc).

    can't you guys do it here instead once we're thru with the babies?

    or maybe i'll pick it up :D

  30. Anymatters, you're right, preference can change, but we are analysing situations one at a time. And that time is that at the margin. Meaning, we're interested in the situation where you are about to spend your last dollar to either apple, orange, or banana, regardless of whether you once had different preference prior to this situation. Put it another way, we're interested in the value of an additional apple to your consumption, given what you decide on orange and/or banana.

  31. thanks i understand that.

  32. i don't have your email roby, so will reply here...

    1) "inference of individual preferences from collective outcome is tenuous."

    I agree. But think of modeling one person like modelling the motion of one planet - easy. Modelling a crowd is difficult because of feedback etc -- in much the same way that modelling planets in our solar system analytically is impossible (it's a non-linear dynamic problem).

    2) "if we have the ability to predict then we should be filthy rich already by predicting stock prices."

    many people have found non-randomness in financial markets and have made themselves filthy rich. the problem is that one someone finds non-randomness it is sooner or later found by others and the markets are random again. i have a book by an MIT prof that found a few instances, name escapes me now - interesting stuff tho!!

    Economics is /the/ hard science. When predicting things like interest rates it's similar to predicting weather with data that's a month or two old (takes time to collect). Essentially it's a system with non-linear dynamics which is very sensitive to initial conditions (i.e. chaotic).

    If anything it's surprising how /good/ economic predictions can be. Ask a meteorologist what the weather will be like next quarter with data from last month ;)

    3) rationality.

    humans play the ultimatum game differently from rational agents - because rational agents are exceedingly smart.

    it's always nice to have /smart/ people in mind when creating a model - because they'll find loopholes and oversights in it for sure (economists are generally pessimists).

    i imagine humans would figure out the rational way of playing the ultimatum game soon enough, if they had large enough incentive to!

    4) "that questions is one of the reasons i dropped out from the econ program"

    that's a real pity :(

    feel free to ask more questions:

  33. that should be ;)

  34. john - you raised interesting points that can lead to potentially fascinating discussions. thanks. i'll think about it and write it up (arya, i'm going to set up a new blog :D)

    ps1: maybe you guys want to take a look at this.
    i don't necessarily agree with him, but i met him once and he's quite a character.

    ps2: i'm thinking to do modeling for living so any comments are appreciated. so if you prefer email please convey your thoughts forcefully to:

  35. john - "But think of modeling one person like modelling the motion of one planet - easy." just wondering where you get this idea from? i'm a psychologist, and i think my field would be finished if your case is true.

    roby - preferences come from the brain, from the interaction of complex neural networks yet to be discovered. put simply, from neuroscience. structures or constraints are fine, but bear in mind that it is people who often 'choose' their structure and constraints, so the social explanation is indeed a tautology.

  36. Dear all,
    To avoid being overtly OOT, I have just re-started the discussion on the origin of preference (taken out of the comments here) to my blog here.

  37. ROBY: modelling for living?? You're modelling a life? or you're gonna enter modelling for a living (ala kiera knightley)??

    Good luck either way!! :p

  38. ace: thanks! btw, keira is a respected actress, modeling is only her side job; does she do modeling really? zealots will curse me, but i think she's really wonderful in pride and prejudice. i can't really comprehend why girls love Mr. Darcy so much, can you ace?