Preferences and Incentives

Our voluntary service he requires,
Not our necessitated, such with him
Finds no acceptance, nor can find, for how
Can hearts, not free, be tri’d whether they serve
Willing or no, who will but what they must
By Destiny, and can no other choose?

Paradise Lost : V : 529-34.

Why is it so important that obedience be voluntary and not necessitated?  After all, if the point is obedience, either will do.  The problem with necessitated obedience seems to be that it is fragile.

Voluntary obedience means obedience that isn’t compelled by external circumstances.  I clean your home because I love cleaning homes or love you or think it’s my duty.  You’ll stop me cleaning your home when you kill me.  This contrasts with necessitated obedience.  If you take all my money and then offer me a buck to clean your home then I’ll do it because otherwise I’ll die.  But if you don’t create a situation in which I have no alternative but to clean your home, I won’t do it.

Voluntary obedience is choice in the sense of preference alteration.  Necessitated obedience is choice in the face of static preferences.  Voluntary obedience is when you make your own preferences, necessitated obedience is when your preferences make you.

Basic economics can’t really distinguish between voluntary and necessitated obedience because it tends to treat preferences as static.  So it only admits of the existence of necessitated obedience.  In economics, the only way you’ll clean my house is if you’re forced to choose between doing that and starving.

Let’s look at another example.  Suppose you have the following game.

A's payoff / B's payoff B Cooperates B Betrays
A COOPERATES 5 / 5 0 / 8
A BETRAYS 8 / 0 4 / 4

Cooperation is efficient in the sense that the sum of the payoffs to parties A and B is the greatest when both try to cooperate. The trouble is that neither will want to cooperate.  For each, the payoff of betrayal exceeds the payoff of cooperation both when the other cooperates (5 > 4) and when the other does not (4 > 0).

Economics typically solves this problem through redistribution.  Suppose the penalty for betrayal is that you must pay 4 to the other party.  Now the game looks like this.

A's payoff / B's payoff B Cooperates B Betrays
A COOPERATES 5 / 5 4 / 4
A BETRAYS 4 / 4 4 / 4

Both parties have a chance to end up better off if they cooperate, so now they will both cooperate.

Both parties behave well, but they do so only because redistribution compels them to behave well.  If the redistribution that compels good behavior is removed, then they will regress back to their old noncooperative behavior.  In this sense their good behavior is necessitated and therefore fragile.  It is a world in which each party bides her time, waiting for the moment when the system will break down and betrayal can again produce a windfall.

Suppose that instead of redistributing payoffs, you were to change the preferences of the parties such that betrayal ceased to have any value to the parties.  They value 8 in betrayal as 0 and 4 in betrayal as 0 too.

Now the game looks like this.

A's payoff / B's payoff B Cooperates B Betrays
A COOPERATES 5 / 5 0 / 0
A BETRAYS 0 / 0 0 / 0

Now both stand to gain only through cooperation, so both will cooperate.  The difference here is that no amount of redistribution can compel the parties to prefer betrayal.  They are no longer mercenaries, blowing with the prevailing wind.

If your goal is to create an enduring commitment to cooperation, you want this second scenario, because it is immune to changes in the power of government to create felicitous incentives.  You want to change preferences, not incentives.  Only in this way can you ensure that Abdiel

Among innumerable false, unmov’d,
Unshak’n, unseduc’d, unterrifi’d
His Loyalty he kept, his Love, his Zeal;
Nor number, nor example with him wrought
To swerve from truth, or change his constant mind
Though single.

V : 898-903

For the economist, changing the betrayal payoffs to 0 makes the problem trivial, because the problem, for economists, is always how to achieve efficiency in the face of static preferences.  Rewriting preferences to make things work is cheating.  But of course rewriting preferences in real life is not as easy as zeroing betrayal payoffs in a matrix.  It was the major project of the pre-modern world; and remains a major project of the modern business world (though more in service of getting consumers to tolerate betrayal than of procuring universal cooperation).

It is no wonder that voluntary fidelity was elevated to divine command in the pre-modern world, where government was too weak to  redistribute wealth.   Where incentives could not be manipulated reliably, cooperation leading to an improvement in the fortunes of the group as a whole could be achieved only by manipulating preferences.

. . . unjust thou say’st
Flatly unjust, to bind with Laws the free,
And equal over equals to let Reign,
One over all with unsucceeded power.
. . .
Yet by experience taught we know how good,
And of our good, and of our dignity
How provident he is, how far from thought
To make us less, bent rather to exalt,
Our happy state under one Head more near

V : 818-21, 826-31

In many older cultures, the aggressive anticipation of needs is the signature of commitment, loyalty, and friendship.  You find that friends expend a great deal of energy giving you help you didn’t ask or pay for.  This behavior expresses felicitous preferences.  Your friends are showing you that they want to do these things, not that they must do them.