Friedman on the gift giving puzzle

David Friedman writes,

Economists, especially those familiar with Gary Becker's analysis of altruism, find the practice of giving gifts puzzling for two different reasons.

If, as a Becker altruist, I take your utility as one of the things I value, the obvious way to increase it is by giving you money and letting you spend it. While there are exceptions, in most other contexts we assume that each individual knows what is in his interest better than others do. Gifts, however, are usually things, not cash.

Becker altruism implies, roughly speaking, that from the altruist's point of view there is an optimum division of the combined income of altruist and beneficiary between the two, a division that maximizes the altruist's utility. If the beneficiary already has more than his share, the result is a corner solution--no transfer. If he has less, the altruist transfers money to him until that division is reached--that being the point at which an additional dollar of transfer costs the altruist as much in lost utility from his own reduced consumption as it gains him in increased utility from the beneficiary's consumption.

One implication of this is that, when transfers occur, they should be large. How likely, after all, is it that my beneficiary's share of our combined income, adding up to many tens of thousands of dollars, will be precisely five dollars less than the optimum? Yet gifts are usually small.

He then puts forward a possible solution to these puzzles.