What you chose to prefer is a constraint you place on your own behavior. One reason we may act consistently with reference to our chosen preferences is because our behavior become predictable to others and thus they repose greater trust that their transactions with us will have the desired outcome.
It is foolish to distinguish preferences from constraints we place on ourselves. Rather we may speak of our preferences as arising from such self-imposed constraints.
Sen, naturally, takes the opposite view.
PREFERENCE AND SELF-IMPOSED CONSTRAINTS In the discussion so far, the influence of the process of choice, and in particular of the menu, has been considered interchangeably (i) through the preference ranking (incorporating concerns about choice acts within the preference ranking), and (ii) through self-imposed choice constraints, excluding some options from "permissible" conduct (we leaned towards this latter way in the formulation of the "fruit-passing game").
The fruit-passing game was mis-specified. The best outcome is to take command of a scarce resource and gain credit for disposing of it in some optimal manner. The process of choice does not influence the menu considered as the set of feasible outcomes. No doubt, in a restaurant, if we dither and dither and speak at great length about how what we want is not on the menu, the maitre de might say 'look, I'll tell the chef to prepare the dish as you stipulate'. But in that case the restaurant menu wasn't really fully specified. It is not the case that your preferences caused the restaurant to suddenly be able to make something it could not otherwise do.
As for 'permissible conduct'- that is a constraint on the feasible set. True, the restaurant kitchen could prepare you a dish from a freshly slaughtered baby. But the police would arrest all concerned. The restaurant would be closed down.
They are not, of course, formally equivalent,
Nonsense is formally equivalent to nonsense. Ex falso quod libet applies.
and it is useful to consider how they may relate. We must also examine the nature of self-imposed constraints as parts of "norms" of behavior or "rules" of choice.
Why? If 'norms' or 'rules' exist, they are not really 'self-imposed'. In so far as we speak of ourselves as following rules or norms, it is in connection with what we believe ought to constrain everybody else. There is nothing individual or idiosyncratic about them- unless we are speaking facetiously.
The practice of enjoining rules of conduct that go beyond the pursuit of specified goals has a long tradition.
But rules of conduct have specified goals- e.g. gaining a reputation for rectitude.
As Adam Smith (1790) had noted, our behavioral choices often reflect "general rules" that "actions" of a particular sort "are to be avoided" (p. 159).
This is because an 'impartial observer' inside us has drawn the conclusion from long experience of society that violating those 'general rules' will have adverse consequences for our reputation, our position in society, and thus our own flourishing.
To represent this formally, we can consider a different structure from choosing a maximal element, according to a comprehensive preference ranking (incorporating inter alia the importance of choice acts), from the given feasible set S (allowed by externally given constraints).
We could minimize regret. That would be the sensible course because Knightian uncertainty obtains and thus we don't know all possible states of the world and thus the maximal is inaccessible to us.
Instead, the person may first restrict the choice options further by taking a "permissible" subset K(S), reflecting self-imposed constraints,
The result is exactly the same as ruling out everything in the set S which conflicts with our preferences- e.g. all the non-veg options, anything containing garlic, etc, etc. Indeed, our preference may be to have the same thing we always have. If the restaurant can't provide it today we may go to another restaurant or decide to skip the meal. This is purely a matter of one's own tastes and preferences. We may speak of it as arising from a self-imposed constraint, but we could equally speak of all our preferences as arising from our own limitations which, in some sense, we choose to take as binding upon us.
and then seek the maximal elements M(K(S), R) in K(S).
In which case no 'different structure' obtains. We are still choosing a maximal element.
The "permissibility function" K identifies the permissible subset K(S) of each option set (or menu) S.
No it does not. The law may have such a function. What is permissible is not as sharply defined as what is legal. No mathematical function can describe what is inchoate and only very vaguely defined. We can no more choose the 'maximal' element in what is permissible than we can choose to always do the best possible thing with the result that we soon become the richest and most powerful person on the planet. It may be that an infinite intelligence can compute what that trajectory should be. But it is inaccessible to us because the 'time class' of that computation greatly exceeds the life-span of the universe.
How different an approach is the use of such a permissibility function in comparison with incorporating our concerns fully in the preference ranking itself?
The difference is that 'permissibility' is a partition on the set S of feasible outcomes.
The formal features of the difference can be more readily disposed of than its substantive relevance. Consider a person with a preference R over the universal set X;
The universal set is unknowable. No functional relation on it is accessible to us.
I am taking this R to be menu-independent, but the argument to be presented would hold a fortiori if the preference were menu-dependent.
The menu is unknowable. It is foolish to speak of independence or otherwise in this connection.
When it comes to choosing from a specified menu S (determined only by externally-given limits, but no self-imposed constraints), the person aims at identifying the maximal elements M(S, R) of S with respect to R. The effect of a self-imposed constraint that specifies a permissible set K(S) to which she deliberately confines her selection is to make her pick a maximal element, according to R, of K(S) rather than of S: (6.1) C(S) = M(K(S), R). Can the route of self-imposed choice constraint be represented as maximization with an as if preference relation R *? The answer to this question turns on the issue of menu dependence, as the following results immediately establish (for proofs, see the Appendix). THEOREM 6.1: For any permissibility function K and any S, there exists an "as if" preference Rs such that: (6.2) M(S, Rs) = M(K(S), R) = C(S).
The theoretical existence of a theory of everything does not mean I can choose to know that theory. If I claim to do so, by reason of my choice, people will laugh at me. I may then say 'a permissibility function' exists such that it is not permissible for you to know that I'm a genius. It is only permissible for you to laugh at me.' You may well say 'you have an 'as if' preference for eating only dog turds. That whatever it is that you dine on, it is as if you are dining on dog turds'.
This sort of back and forth represents the entire epistemic content of Sen-tentious Economics.
A morally exacting choice constraint can lead to an outcome that the person does not, in any sense, "desire," but which simply mimics the effect of his self-restraining constraint.
Preferences are a constraint on choices. No mimicry is involved.
To illustrate, there has been a good deal of discussion recently on the alleged tendency of many Japanese workers to work extraordinarily hard, and the idea of "karoshi" (death through overwork) has been discussed in that context (see, for example, Morishima (1995)). The tendency to do one's "duty" to the point of severely damaging one's health (whether or not leading literally to "death") is easier to explain as the consequence of adhering to a deontological obligation rather than as an outcome that is actually "preferred" by the hapless worker.
This is foolish. Soldiers, in doing their duty, get severely wounded or killed. We don't say they preferred to get killed. We say they preferred to do their duty even though this meant risking death. If you 'work yourself to death', it may be that you overestimated your stamina or recuperative capacity.
Japan took the 'samurai spirit' from the army into the work-place. What Morishima was commenting on had nothing to do with the theory of consumption. Like other Japanese Marxists, he believed that Japan's 'damaged modernity' and 'dual' economic structure had deep cultural roots. Thus, he was fond of telling his students, when the first Japanese scholars to gather Confucian texts were on the ship back to Japan, they decided to translate 'benevolence' as 'obedience'.
Social psychology can be important here.
Because we are talking of the psychology of a given Society- does it encourage its people to be happy-go-lucky? Does it try to make them thrifty and hard working?
The as if preference works well enough formally,
They are utterly useless. They do no work whatsoever formally or otherwise.
but the sociology of the phenomenon calls for something more than the establishment of formal equivalences.
There is no formal equivalence. Sociology requires a lot of descriptive work and has a wholly independent analytical structure.
This issue is close to Adam Smith's general point that many behavior regularities can be explained better by understanding people's attitude to actions, rather than their valuation of final outcomes.
Nonsense! Smith said that people's attitudes are affected by their empirical observations and these attitudes determine whether they will act prudently and thus gain good outcomes.
Similarly, Immanuel Kant gave a central position in social ethics
No. His 'Anthropology' was quite different from his critique of practical reason.
to a class of restrictions on actions,
No. He had a positive view of the Law. It was 'command' simply.
which formed a part of what he saw as the "categorical imperative," as elucidated by the following remark in the Groundwork: "There is...but one categorical imperative, namely this: Act only on that maxim whereby thou canst at the same time will that it should become a universal law" (translated by Abbott (1889, p. 38)).
Which is cool if, like Kant, you believe pure practical reason delivers an assurance of the immortality of the soul. Thus if you are killed because of your virtuous action, you get eternity in paradise. That's a good deal from the consequentialist point of view.
The form of the imperative, which is crucial to Kant's reasoning, is the need to impose on oneself some constraint on how one can act.
Because your soul is immortal whereas your body will give out in 90 or a 100 years.
While the focus of Smith's and Kant's reasoning is normative rather than descriptive,
Both did plenty of descriptive work- or rather relied on the descriptive work of others. Smith's reasoning is empirical and 'associationalist'. Kant was more of a Pietist upholding the immortality of the Soul.
the two are closely linked in their analyses, since both understood actual behavior to be partly based on norms.
Smith derived norms from the actual behavior of successful people who, like him, had risen in Society and who enjoyed a high reputation for prudence and probity.
Their behavioral analysis included seeing the process of actual choice through K(S), and not just through an "everything considered" grand preference ranking Rs.
Neither believed choice was based on an unknowable set of feasible outcomes or that mere mortals could construct 'everything considered' grand preference rankings. Smith was saying 'emulate the guys getting ahead thanks to their prudence and perceived probity.' Kant was saying that just as Pure Reason could justify credence of Newton's Absolute Space and Time so too could Practical reason justify faith in the immortality of the soul. We know Kant was wrong. Absolute Space and Time don't exist. Even Godel's proof of God was shown to be wrong by a computer.
We live in a world of Knightian Uncertainty. That is why evolution is a true theory whereas creationism or occassionalism is not scientific or useful. The Church itself declares Faith to be founded upon a mystery which neither pure nor practical reason can elucidate. Economists need to admit that 'utility maximization' is just a limit case, where uncertainty is zero, of 'regret minimization'.
Sen, oblivious, of the growing literature on Hannan consistency, comes to the opposite conclusion
In this paper I have tried to examine the role of the choice act in maximizing behavior, which has to be distinguished from maximization without volitional choice by a maximizer, as, for example, in standard models of physics (Section 1).
It is not possible to do so. A mathematical theory has no way of distinguishing whether it is being applied to elementary particle or cognitively complex humans.
The process of choice can be an important concern
which is why 'behavioral economics' was finding useful applications outside the classroom whereas Sen-tentious shite was favored by useless UN bureaucrats
and so can be the necessity of choice
If the thing is 'necessary', it is not free and thus not really choice at all. It is something coerced. We don't think 'Sophie's Choice' was any reflection of Sophie as opposed to the beastliness of Nazi cunts.
even when the alternatives are not fully ordered and the conflicting considerations not fully resolved (Section 5). The analysis shows how the maximizing framework can adequately accommodate both issues, once its axiomatic structure is correspondingly adjusted
So, Sen thinks shit will stop being shit if its 'axiomatic structure' is correspondingly adjusted such that shit is defined as tasty and nutritious food which did not come out any being's rear end.
Some of the findings can be briefly identified. First, one aspect of volitional choice is the possibility that choice acts may have to be undertaken with substantial incompleteness in judgements (arising from instrumental or valuational reasons). While this is problematic for the framework of classical optimization standardly used in economics, there is no great difficulty in systematically accommodating such incompleteness in a framework of maximizing behavior and to study its regularity properties as distinct from those of optimization (Section 5).
We now know that this wasn't true. The Arrow Debreu framework can't accommodate regret minimization. We need to go Bayesian or go home.
Exploration of the relationship between maximization and optimization (characterized in Theorems 5.1-5.6) shows exactly how they relate and where the gaps are.
No. Maximization is not economic. There may be a maximally good choice of food, clothing, employment etc. Optimization is economic. It involves trade offs and occurs under Knightian Uncertainty.
The difference between maximizing and optimizing can be formally closed,
by waving a wand and saying 'abracadabra'!
in one direction (from maximization to optimization, not vice versa), through an "as if" preference,
which is like saying 'dog turds are as delicious and nutritious as truffles & caviar provided we act as if this were so'.
but a substantive interpretative difference remains even here. The directional asymmetry lends further support (in addition to the larger reach of maximization) to the case for taking maximization to be the mainstay for rational choice functions.
In other words, Sen was only pretending to find fault with Arrow-Debreu. The truth is his own shitty 'Capabilities' was built on that nonsense. This is not to say that Scientists can't improve 'capabilities'- e.g. figuring out a way for athletes to run faster or fat cunts like me to fill our bellies with tasty treats which don't cause diabetes or heart disease or whatever the fuck it is which will put an end to my swinish existence.
Simon's formulation of "satisficing" behavior,
was useful in the context of competition policy- the monopolist may prefer 'a quiet life' to maximizing profits- as well as nuclear strategy- it is enough to be able to take out a few cities belonging to the enemy rather than just blow up the entire world.
connected with his important idea of bounded rationality, can be accommodated within a general maximizing framework, eliminating the tension between satisficing and maximizing (but the tension with optimization remains, except in terms of the formal device of an "as if" preference).
This was always obvious. However, the future lay with Bayesian methods and machine learning and so forth.
Second, the process of choice-and in particular the act of choice-can make substantial difference to what is chosen.
But that process is itself decomposable into choices just as an act can be decomposed into choices between mental states.
While the differences can take various complex and subtle forms (Sections 2-4 and 6),
No. The differences can take stupid and shitty forms if the guy making the distinctions is a Sen-tentious cretin.
there is a particular necessity to take note of (i) chooser dependence,
Choice does depend on the chooser just as a fart depends on the farter. This no particular necessity to take note of the bleeding obvious.
and (ii) menu dependence, of preference, even judged from a particular person's perspective.
There is no 'menu dependence'. However, if you are as stupid as Sen, you might be too stupid to work out what is actually on the menu. Indeed, you may not be able to distinguish the restaurant from the side-walk. You may also thing a lump of dog shit which you find on the pavement is actually the best dish on the menu of the posh restaurant nearby. But the fact that you get down on all fours to feast on that dog-shit and then leave a large pile of banknotes to pay for your repast, does not mean you dined well from the menu of the restaurant in question.
The parametric preference relation RjV' of person i can reasonably rank the same elements x and y differently depending on who (j) is making the choice (in particular whether it is the person i herself: i =j), and the menu S from which the choice of x or y is being considered (Section 3).
No. This is a stupid, not a reasonable, way to proceed. If you starve to death while standing beside a free buffet of delicious food just because you wanted Beyonce to come and fill up your plate and to feed with her own dainty hands, we consider you a cretin. It is another matter that you could fill a couple of plates and take them to a beautiful lady who, in consideration of the sprained wrist you claim to have, may be persuaded to express her gratitude by popping a samosa into your slavering mouth. Well when I say 'you could'- I don't mean you. I mean a handsome dude wearing a gold rolex watch around his sizable dick.
This is analytically important for understanding the nature of rational choice and maximizing behavior (it militates, in particular, against many widely-used "consistency conditions" that ignore these parametric variations).
It militates for eating dog shit outside the restaurant and then paying through the nose for that privilege.
It is also practically important in explaining a variety of behavioral regularities in economic, political, and social affairs-from variations in work discipline and in economic corruption to the operation of social norms and of voting behavior (Sections 2-4 and 6).
Nonsense! Choice may be delegated where information asymmetry obtain or transaction costs are high. But that has nothing to do with preferences.
Third, it is necessary to distinguish between menu-independence of preferences and menu-independence of choice functions,
Rubbish! Preferences give rise to choice functions within the 'budget set'.
since there is, in general, no one-to-one correspondence between preference relations and choice functions.
Yes there is. The choice function is a preference relation on the budget set. However, both may be unknowable and thus it is foolish to pretend an 'existence proof' means the thing is accessible or computable.
While menu-independence of preference entails menu-independence of the generated choice function, menu-independence of a choice function need not entail menu-independence of the preference that generated that choice function, as shown by Theorem 3.1.
So what? If preferences were menu dependent, the choice function would be too.
The connection between binariness and menu independence can also be identified,
because either the axiom of choice or something like countable choice is assumed
and it is in fact convenient to see binariness of choice as a condition of menu independence (Section 3).
only if preferences are binary
Fourth, the role of the choice act can be particularly significant in decisions made on behalf of others-a feature of economic policy-making on which Ragnar Frisch himself had put much emphasis.
Frisch used thought experiments to put demand theory on an independent footing from utility theory. What Sen is talking about is something different- viz. agent-principal hazard- which must be tackled by incentive compatible mechanism design- i.e. the incentives facing the agent must elicit behavior advantageous to the principal.
The presence of fiduciary responsibility calls for some reformulation of the standard axioms of choice theory because of the role of the choice acts.
No. Only mechanism design matters. It is obvious that if the agent can safely get rich by screwing over his principal, they no fucking 'fiduciary responsibility' will obtain.
This also has implications for the formulation of games and strategic concerns, as the "fruit-passing game" illustrates (Section 4).
It illustrates nothing except Sen's stupidity. Take command over a scarce resource and then dispose of it as you think best. Don't starve to death while standing next to a free buffet because you'd prefer Beyonce to come and feed you with her own dainty hands.
The role of behavioral norms in general, and of the common knowledge of norms in particular, can be quite important for understanding strategic actions (including "strategic nobility") and the corresponding game outcomes.
No. A stupid action is explained by stupidity not some supposed norm that the stupid cunt claims to be upholding. On the other hand, it is common knowledge that Beyonce would like nothing better than to feed me with her own dainty hands.
Finally, the accountability and obligation to others may take the form of self-imposed choice constraints (as formulated by Immanuel Kant and Adam Smith) rather than being incorporated within reflective preferences in the binary form.
Either such 'accountability' and 'obligation' is voluntarily chosen or we are not talking of the free action of a rational agent at all. Economics is not concerned with actions taken under duress.
This is not a major technical gulf, unless we insist on preferences being menu-independent (as is standardly assumed in traditional theory of preference and choice).
A list of commands is not a menu. It is stuff we must do to avoid being beaten to death. Sen doesn't get this. If 'menu-dependence' is admitted, then preferences are irrelevant. Suppose, as you read through the menu you come across the following item 'egg salad- patrons are advised that failure to order this item and no other will lead to immediate decapitation.' You then notice that the maitre de is holding an executioner's axe. There are photographs on the walls depicting the severed heads of former customers. You decide to order the egg salad. Why? What we actually have is a command, backed up by a credible threat of decapitation, not a choice menu.
The operation of self-imposed choice constraints can be readily represented through devised "as if' binary preferences in a menu-dependent format (Theorem 6.1),
I suppose some cretin might say 'you have an 'as if' preference for egg salad' but this isn't genuinely the case. We have a preference for not getting our fucking heads chopped off.
but not in general through menu-independent "as if`' preferences (Theorem 6.2).
Nonsense! It doesn't matter whether you read ' egg salad- patrons are advised that failure to order this item and no other will lead to immediate decapitation' on the menu or if some fellow diner whispers it in your ear. Your preference not to be decapitated is 'menu independent'.
However, irrespective of formal representabilty, the tangible differences made by the use of choice constraints can be materially important for the psychology of choice
Twenty years later we can be sure that wasn't the case. There's no pop-psychology book or self-improvement TED talk which tells you that 'menu dependence' can improve your life. Mindfulness, sure. Transcendental Meditation- maybe. But Sen-tentious shite about Menus?- fuck off!
No comments:
Post a Comment