by Mario Rizzo
Behavioral economists who like to indulge in normative pronouncements have decided that quasi-hyperbolic discounting violates rationality. In other words, suppose a person decides today that he will give up the hamburgers he loves beginning in 2010 (because of the high fat content). But then when 2010 arrives he reverses his decision and continues to eat them. To stress the point, let’s suppose that he repeats this preference-reversal one or two more times during 2010.
The poor fellow is, in addition to all his other troubles, violating standard economic rationality.
There are many things one can say about this. Glen Whitman and I have made more technical criticisms in “The Knowledge Problem of New Paternalism” and in a previous blog posting.
Here I want to point out that this form of analysis misses something basic about human decisionmaking and the pursuit of happiness. Read what Immanuel Kant had to say in 1785:
“There is one end…which we may presuppose as actual in all rational beings… This purpose is happiness. … But it is a misfortune that the concept of happiness is so indefinite that, although each person wishes to attain it, he can never definitely and self-consistently state what it is that he really wishes and wills. …[I]t is impossible for even a most clear-sighted and most capable but finite being to form here a definite concept of that which he really wills. If he wills riches, how much anxiety, envy, and intrigues might he not thereby draw up his shoulders! If he wills much knowledge and vision, perhaps it might become only an eye that-much-sharper to show him as more dreadful the evils which are now hidden from him and which are yet unavoidable; or it might be to burden his desires…with even more needs!
…[H]appiness is an ideal not of reason but of imagination, depending only on empirical grounds which one would expect in vain [strictly] to determine an action through which the totality of consequences – which in fact is infinite – could be achieved.” (Foundations of the Metaphysics of Morals, Section II, §§ 415 ff.)
The upshot: Don’t expect the most “rational” person to clearly understand and consistently pursue what will make him happy.
This is another one of those examples where a simple concept of rationality – perhaps useful instrument in understanding market phenomena or some aspects of decisionmaking – is transformed into a normative ideal. The economist quite reasonably simplifies the world to understand it. But then he should not criticize the real individual for being more complex than the neoclassical model of rationality allows.
ThinkMarkets 
Here is where I think the Austrians stand head and shoulders above current high-brow opinion. For Mises, rationality is simply applying the appropriate means to an end,given an individual’s state of knowledge at the time.
It means nothing more.
Herbert Simon made the same observations about the problems with neoclassical rationality. But (like Mises, apparently) he didn’t give up on the idea that, at least taken in the aggregate, we’re still optimizing, albeit imperfectly.
But optimizing what? Kant’s critique of happiness is far more serious than the behavioralists, Simon, or Mises. One has to start somewhere in modeling, and any constrained optimization must have an end (in Mises sense).
In this regard, it seems that the neoclassical notion of utility might still be useful, although Samuelson’s proposal of revealed preference got things off on the wrong foot leading, as it does, to this concept of rationality that is not realized by any actual agent.
But what else, other than revealed preference, could be used to operationalize utility? Some have little imagination for this question.
My own suggestion is that we ought to be measuring the frequency distributions of activities — i.e., how often does a particular agent engage in a particular unit of activity per unit time. For example, how many calories per day, how many hours of work, of sleep, &c.
From such frequency distributions, one could construct a model of consumer behavior in the aggregate, a model not restricted to being consistent in preferences over time. Should we call the function underlying such frequency distributions utility? I don’t care. The point is that it relieves the constraints of operationalizing through revealed preferences.
As a consistency check, one could observe whether the same model of preferences comes out when these frequency distributions are stable over same-width time windows — i.e., in the limit of ergodicity.
“The economist quite reasonably simplifies the world to understand it. But then he should not criticize the real individual for being more complex than the neoclassical model of rationality allows.”
This reminds me of what Pietro said over at AE:
“as an engineer, it appears to me that the approach is just like: design a rocket assuming there is no atmosphere, launch the rocket, blame the air.” – Pietro
Why do we insist on this “modeling” of human behavior? I mean, not in the everyday commonsense of it (in the way “Austrians” often have) that individuals have ends, they implicitly have a rank order of the specific options they are facing at a moment of choice, and decide which means might be useful or appropriate, and decide on whether in a particular concrete trade-off situation a specific goal is or is not worth the cost of what actually must be foregone.
I mean it in the hyper-rational sense of a global and fully consistent (through time) “rationality.”
This Neo-classical conception reminds me of the madman of whom G.K. Chesterton speaks in his book, “Orthodoxy.” The madman, Chesterton says, is the one “who has lost everything except his reason…. He is not hampered by a sense of humor or by charity, or by the dumb uncertainties of experience. The madman’s explanation of a thing is always complete, and often in a purely rational sense satisfactory.” The madman has a “most sinister quality” of “connecting of one thing with another in a map more elaborate than a maze.”
The fact is, people are, well, inconsistent; they change their minds, they modify their views of things are they learn and experience the events of their life. They are emotional. They are guided by material and “ideological” goals and values (and they often get them mixed up and rationalize one in terms of the other). And they weigh the present and the future differently, including differently from when that future actually comes (i.e., “talk is cheap” in terms of deciding or saying what you’ll do when tomorrow really comes).
So why don’t we just accept all this and live with it? Use our general conceptual framework about ends and means, (marginal) costs and (marginal) benefits, trade-offs, etc., as a wide, but very qualitative “grid” upon which to give degrees of intelligibility to human doings?
Let’s stop being Chesterton’s “madman” who is looking for a reason for everything, an ulterior “utility function,” hidden somewhere in people’s behavior, which if we could only find it, would give us the “open sesame” to know and predict everything about what people are doing and will do.
Searching for the answer to the unpredictable, we forget that much in life, including our own behavior and the “why” behind it is, well, just unpredictable.
Trying to make the unpredictable, predictable will drive you mad, as Chesterton suggested.
Richard Ebeling
I’m being cited. I’m famous! 🙂
Thanks.
I was discussing a point similar to Ebeling’s with a friend of mine, yesterday. He insisted on how to model human behaviour realistically.
My knowledge of the history of economic thought is limited, but I tried to give him what I thought was a meaningful background.
The problem is about the relation between psychology and economics in neoclassical and misesian economics, and about the role of falsifiable theories in economic method.
Mises conception of rationality was subjective: whatever the agent does, he can’t be irrational, because his actions can always be understood with an ends/means analysis.
Neoclassical rationality is, on the other hand, the perfect adequatio rei et intellectus.
Misesian rationality lacked operational content and forced no falsifiable restriction on behaviour. It was a metaphysical diktat about interpreting men as men and not as automata.
Both constructions may be a response to the same problem: the mystery of human intelligence, cognition and creativity.
Without isolating economics from psychology, no economic theory can be done without first having a full theory of the mind. The two solutions were formal, but in two opposite meanings. Formalism in neoclassical economics means “minds don’t contrain economic dynamics”, and in Mises means “the logic of action is independent on cognitive processes”. Both types of formalism can be easily pushed to the point of irrelevance, but are answers to a real problem.
While neoclassical economics in its pure form assumes the problem away, the misesian solution isolates a domain for economic theorizing that is independent, but possibly complementary, to any development in psychological theories.
So, why models? Neoclassical economists insisted on fully articulated models. They needed a way to reduce men to a formal equation. The first attempt was to neglect the scarcity of cognitive powers. Probably in the future limited cognition will become part of the neoclassical toolbox.
A problem I see is that there is one only way to be objectively rational, but infinite ways not to be. A model is required to obey the operationalist and falsificationist view of method. Maybe after the billionth model of irrationality, this view of method will be abandoned.
Just a quick point, but surely we need to contextualise Kant’s remarks a bit?
His argument in the Groundwork is that human desire is fallible, and attempts by human beings to achieve their “happiness” frequently turn out to be incoherent and self-defeating. It’s a stop-gap on the way to arguing that what humans “ought” to do is follow the moral imperatives revealed by rational reflection, incumbent upon all rational beings.
But what really does the work in Kant’s thinking is to – effectively by fiat – deny the Humean argument that moral vice and virtue consist essentially of pleasure and pain, and that therefore emotion and subjectivity are the bedrock of moral valuation. Kant excludes this, postulating instead that any presence of sentiment (or in the usual translation, “inclination”) effectively infects a moral action and renders it devalued in terms of moral worth.
So the “human beings pursue desires and aren’t fully rational” stuff in Kant is being deployed to lament the fact human beings are not perfectly rational beings, and hence condemned to never achieving morally pure deeds, lives or actions (he explicity states that a truly moral action may never have been done in the history of the world).
Now it may be true that in understanding the way people are buffetted about by their passions and desires, Kant has a better picture of humanity than behavioural economists. But he translates that into a most severe view of the foundations of ethics. And we might want to pause before bestowing praise on him from that directions.
If I may add one more point.
Who are these “new paternalists,” who (implicitly at least) presume to above the rest of us?
One pictures these “social engineers” of human behavior in their white laboratory coats with their pocket protectors filled with colored pens, holding a clip board upon which they are recording all their dispassionate observations of those other, poor, imperfect, and irrationally inconsistent human beings.
Give me a break. As if THEY never broke a New Year’s resolution to stop smoking or keep their garage clean; or never swore to stop drinking so much at parties, as they dangled at the edge of a toilet bowl later that night, and then found themselves in the same situation after the next “wild” night out; or never gave in to the momentary desire to eat that rich dessert, and telling themselves AGAIN, that they’ll get on that diet tomorrow.
Who monitors the proper choices for and the actions of the paternalistic planners?
These people seem to presume that they are somehow outside and above the human circumstance they are evaluating in the context of their constructed notion of “rational” choice.
Am I the only one that finds this a bit arrogant on their part?
Richard Ebeling
No to Richard’s last question.
Perhaps because he was influenced by Kant, I don’t recall Mises ever writing about optimizing something or even much about happiness. In Human Action, he talked about removing a sense of uneasiness.
That is a purely formal sense of motivation, and admits of the widest possible set of motivations and goals, form the loftiest to the most mudane. It seems to me to say more is to move from economics to psychology.
Paul,
I am being myopic about Kant. I know the full context of what he is saying takes us in a direction not to everyone’s taste (nor mine). However, his observations about what the pursuit of happiness entails are useful.
Jerry,
I think the difference between the Austrian concept of “uneasiness” and the neoclassical assumption of maximizing behavior is oversold and mostly nominalistic. I had the patience, during this summer, of reading Rothbard’s gargantuan treatise “Man, Economy and State”. To my surprise, in the section “10. Action as an Exchange”, p. 72, I read the following passage: “[…] at the inauguration of any action, the actor will believe that this course of action will, among other alternatives, maximize his psychic income or psychic revenue, i.e., attain the greatest height on his value scale”. So, what’s the radical difference between the Austrian view and the current neoclassical approach of maximizing individual expected utility, given a context of informative limitations and possible future shocks to preferences and external economic data due to pervasive uncertainty?