Melberg, Hans O., Social engineering and rational choice theory

_{[Note for bibliographic reference: Melberg, Hans O. (1996), Social engineering and
rational choice theory, http://www.oocities.org/hmelberg/papers/960509.htm]

[Comment: This observation assumes some familiarity with game theory. I also have to
acknowledge that the main ideas in this observation are now my own, but from various works
by Jon Elster.]

Social engineering and rational choice theory

by Hans O. Melberg

The starting point for this observation is the thought that a change in the environment
(rules, institutions, incentives) will cause individuals to change behaviour - in a
predictable way - as they adapt to the new system. An example of this is the belief of
many economists that freeing prices and introducing market institutions (private property,
bankruptcy laws) automatically will make people adapt their behaviour so that the economy
Soviet Union becomes more like the West (more efficient). It is this belief which I
believe is closely connected to a rational choice conception of man. If people do not
behave rationally, then a change in the environment will not lead to predictable changes
in peoples' behaviour.

The driving motor behind rational choice theory is maximization. If people choose their
actions in order to maximize a function (such as their utility function), then it is
relatively easy to predict how they will behave under a new system with a new set of
incentives. The problem is that rational choice theory is both incomplete and inadequate.
A theory is incomplete when it fails to tell us what we should do. It is inadequate when
people behave according to the predictions of the theory (Elster (1993), p. 181). Thus, my
critique of rational choice theory is not only that people sometimes behave irrationally,
it is also the fact that sometimes it is impossible to behave rationally because rational
choice theory cannot prescribe an action.

That people sometimes are irrational is beyond doubt. Numerous experiments, and everyday
experience, gives evidence that we are weak-willed, imperfectly calculating, and norm
inspired creatures. The question is, of course, to what extent we behave irrationally
compared to rational behaviour. However, in this observation I only want to give one
example of norm-guided behaviour which created (creates) problems for the economists'
social engineering experiment in Russia and Eastern Europe.

Judy Batt in an article entitled "The "Politics of Economic Transition. (in Developments
in Eastern European Politics edited by Stephen White, Judy Batt and Paul Lewis -
London, 1993), mentions that cultural attitudes may prevent a reform from having the
expected result (p. 223). One example of this could be that Russian enterprise directors
view themselves more as "fathers" of their labourforce than profit maximizers on
behalf of their owners. This attitude may be traced back to the old Russian commune, in
which the head of the village was responsible for all the villagers. However it may also
be traced back to the times under communism in which the enterprise director was
responsible for sending their labourforce on holidays, for their housing, for social
benefits etc.. Hence, traditionally the role of an Russian enterprise director has been
very different from the more aggressive profit maximizers of the West. In which way is
this difference relevant today?

Today the enterprise directors face a new environment - they are supposed to maximize
profit by selling their own products in the free market. One would then expect that they
do what seems rational - to cut costs as much as possible - to become more effective.
However, empirically we have witnessed that enterprise directors are unwilling to make
cuts in their labour force in order to become more effective. Instead they send people on
paid (and unpaid) holidays, they reduce the working week, they continue to produce goods
even when they cannot sell their products. One must then ask why they did not behave as
predicted. One answer could then be that they are motivated by the old norms - by their
old culture of protecting the social benefits of the workers in their factory.

One may ask how this is possible since one would expect firms behaving in this way to go
bankrupt. The answer is that in an economy where many enterprise directors are motivated
in the way explained above, they can avoid the "hard budget constraint" by
giving each other credit (or simply by resorting to bartering). This inter-enterprise
credit has increasingly become how factory directors keep their factory in business even
when it is not really profitable. In addition it was possible to get substantial credits
from the central bank. In this way culture has overcome the "hard budget
constraint".

One might argue that the it is only a question of time before the enterprise directors are
forced to behave rationally. Thus, in the long-run the experiment in social engineering
(the big bang!) is going to be successful, as predicted. This may be true, but I have yet
to see a good theory of how long time the transition is going to take. Maybe, as Keynes
once said, the long run is so far away that we are all going to be "dead." when
the experiment becomes successful.

Although the proponents of social engineering might rescue their argument by arguing that
people will become rational in the long run, they cannot use this argument if we can prove
that in some situations it is impossible to behave rationally. Following Elster I shall
discuss two reasons why it sometimes is impossible to behave rationally: Strategic
uncertainty and brute uncertainty.

First there is the problem of forming expectations in strategic situations. The problem is
this: What I want to do depends on what I think you will do and vice versa. In game theory
these situations are solved by backward induction. But the backward induction technique
rests on a fallacy. In short backward induction is a fancy name for the following chain of
thought "If I do A, he will do B and we end up in X. If I don't do B, he will do C
and we'll end up in situation Y. Because the situation in Y is worse than X, I'll do
A". The fallacy is to assume that the other person will do C if you do not do A. The
reason is that if you do not do A the other person can no longer believe that you are
rational, hence the best action for the other person need no longer be C. But, knowing
this it may become rational to appear irrational. Then again - none of the players know if
the other player really is mad/irrational or is simply faking. because of these problems
backward induction may fail. And, if backward induction fails, there is no rational
solution to the game.

(One possible solution could be to argue that people assume it is more likely that the
player is faking, than really being irrational, since they fear the humiliation of being
outsmarted.)

The second source of indeterminacy, is brute uncertainty. In many situations we simply do
not know enough to make rational decisions. For example we can never know the optimal
level of information about how much information we should collect before we make a
decisions (because this leads to an endless chain and because the marginal value of
information is not constantly decreasing). We do not know enough about the future to make
rational investment decisions (although we pretend). In these, and many more, situations
we simply do not know enough to make a rational decision - We might as well use a random
decision-maker (although this option is often prevented by our (irrational) addiction to
reason - the need to always justify our choices by reasons even when this is impossible).

Now, the reader may ask, how is this related to social engineering. Social engineering
relies on being able to predict the outcomes of a new system. This ability rests on the
assumption that people are rational maximizers. In this short article I have showed how it
sometimes is impossible to be rational. This leaves room for norm-guided and other
non-rational (as opposed to irrational) behaviour. We do not know how this kind of
behaviour will change in response to a changed environment. This in turn means that social
engineering even in theory cannot be predictable. The question is then to what degree
predictable social engineering is possible. This leads us to ask to what extent behaviour
is, and can be, guided by rational considerations as opposed to irrational and
non-rational motivations. Since I do not yet feel confident enough to give an answer to
this question, I shall have wait until a later occasion.

Sources

The main source of this article is Jon Elster's article "Some Unresolved Problems in
the Theory of Rational Behaviour" (Acta Sociologica (1993) 36:179-190). The
main ideas and examples in this observation are from this article and not my own.

Two other sources are:

J. Elster (1979), Ulysses and the Sirens (Chapter 3), Cambridge: Cambridge University
Press

J. Elster (1996), Doing our level best, The Times Literary Supplement, 29. March
1996, pp. 12-13.

[Note for bibliographic reference: Melberg, Hans O. (1996), Social engineering and
rational choice theory, http://www.oocities.org/hmelberg/papers/960509.htm]}