Wednesday, February 18, 2009

Challenge: Formulate your issue in one sentence

When I see criticism of economic theories they often seem to go on and on, use lots of “ism” words (positivism, pluralism, autism and worse), and be very vague and abstract and wordy. Which is why I have the following challenge: Take your criticism and formulate the main point in one sentence free from overly abstract and philosophical/methodological jargon.

Here’s my attempt:

  1. Against some types of welfare theory: If you have no empirical evidence concerning what people actually care about and what choice alternatives they actually face, then you do not have sufficient information to establish that they are welfare maximizing.
  2. Against some types of “explanatory theories”: If you claim to explain why something is the case, then all sorts of empirical information are relevant in assessing your driving assumptions and proposed mechanism– not just market data.

I don’t feel these two claims should be controversial. They seem rather obvious to me – yet these two “principles” (if accepted) are sufficient to kill off (I believe) much of the weirder (but accepted) theories in economics. My Ph. D. work on Rational Addiction theory was (in retrospect) basically the attempt to make these two claims and show why they were sufficient to discard many of the claims made in top 5 journals concerning the validity and implications of rational addiction theories:

  • Welfare: As long as you merely observe market choices and develop a “choice model” to accurately reproduce/fit patterns in the market data (whether “stylized facts” or econometrically through structural models) – this does not prove/establish/support the claim that the consumers in question are maximizing their long term welfare. All you’ve proved is that they are acting “as if” they are maximizing some utility function (which you have not shown to be related to their actual welfare) given some choice set (that you have not shown to be the one they are facing). No matter how insanely and self-destructively you act – it will be possible to develop some model that rationalizes this as optimal behavior for some set of preferences and constraints.
  • Explanation: Merely observing economic data and developing a “model” to accurately reproduce/fit patterns in this, is insufficient to establish that your proposed causal mechanism is actually at work in the real world. If your theory, e.g.,  asserts that people have information that does not exist, are solving problems that we know they are not facing, or influencing the world through relationships that we have no reason to believe in – then these are good reasons to discard your proposed mechanism as an explanation.

Is there some non-obvious flaw in my claims that rescues the weirder economic theories? I’m open to that possibility - please let me know.

Also interested in other attempts to formulate one-sentence summaries of

4 comments:

  1. !. "If you have no empirical evidence concerning what people actually care about and what choice alternatives they actually face, then you do not have sufficient information to establish that they are welfare maximizing."

    Clear pattern would emerge when people act to achieve some goals and chaos is when they act randomly. If people are not maximizing some objective, they would have acted randomly. As far as I see the market, I see order instead of chaos and hence agents must have been maximizing.

    2. "If you claim to explain why something is the case, then all sorts of empirical information are relevant in assessing your driving assumptions and proposed mechanism– not just market data."

    I disagree. It all depends upon whether one offers a COMPLETE vs PARTIAL explanation. The RBC people for example say that they explain a major portion, not complete, of the observed variation.

    ReplyDelete
  2. Hi Anonym - thanks for your response.

    On point 1: Do you mean to claim that IF AND ONLY IF people act to achieve long run welfare maximization THEN we will observe patterns? That seems like a strong and unsupported claim. Habits, norms, persistent cognitive biases and decision flaws can all generate patterns as well - and it may be that some of these cases will involve poor choices that fail to be in the acting individual's best interests (even given the information he has).

    On your second point: I explicitly said "driving assumptions and proposed mechanism". In that, I tried to distinguish between different types of assumption and their relevant to the empirical claim that you have described some real world mechanism. To link to a recent post here: If, for instance, your proposed mechanism REQUIRES that people in a game-theoretic situation are able to take into account "that they know that their opponent knows that they know that their opponent knows that..." and so on TO INFINITY, then a relevant objection to your proposed explanation is that empirical evidence suggests few people are able to follow such reasoning more than one level.

    ReplyDelete
  3. Let me expand on point 1 again, I don't think there is a welfare function that people maximize,at least as far as I and the people I know concerned. What we have rather is an objective or goal, you may be excused if your "welfare" notion matches with a "goal", people want to achieve.

    The goal could be securing permanent income over life time, living a life that one values, etc. With such goal, there are people going to school for getting the degrees that helps them meet these goals. One needs to show that our daily activities, like working for salary, traveling to work place, eating, socializing etc have nothing to do with well designed actions to meet some well defined goals before forwarding a grand claim of lack of maximization. It needs to noted that one need only a single counter example to invalidate the idea that people do not maximize[or choose actions to meet a given goal].

    In some cases, meeting these goals might require conformity with norms, habits, cultural values etc. But, again the mere existence of habits, norms, etc can not be as evidence for lack of maximization.

    ReplyDelete
  4. Thanks for a thoughtful reply - but I think it is necessary to take a step back and review what we are discussing.

    My claim relates to how economic theories are used for welfare analysis. I started worrying about this when I came upon rational addiction theories, whose authors claimed that their results establish that real addicts are rationally optimizing their welfare in a forward looking, well-informed manner. In other words: These economists seemingly believe their theories/models can be used as tools that establish whether or not some real world behaviour is optimal.

    To use these theories/models for such a purpose, we need to assume that they correctly relate people's welfare to the determinants of this welfare. There may be various ways to establish this. Two that come to mind:

    1. Provide evidence on the specific goals/interests of people and their actual opportunity sets

    2. Provide evidence that the "utility" of your model can consistently predict the level of some credible empirical proxy of well-being/satisfaction/etc.

    My claim is not that no people ever maximize anything. I am not claiming that all human behavior is self-destructive. What I AM claiming is that we can not assume out of hand that ALL human behaviour at ALL times is well-informed, forward-looking and optimal.

    This means that we need to face the question of WHEN the "choices are well-informed, forward looking and sensible" assumption is suitable, and when it is not. And, to return to my original claim, it means that we need MORE AND DIFFERENT EVIDENCE from what economists typically provide in their current work to know whether their model accurately informs us of real people's welfare under different scenarios.

    ReplyDelete