No, I’m not linking to somebody’s silly article on this topic, so that we can all make fun of him. For a few years now I’ve thought there were a few uncomfortable problems with subjective value theory (in the Austrian tradition, the only one with which I’m familiar), and Stephan Kinsella’s recent concerns about profit (on a private email list) prodded me to post this. (I’m not saying subjective value theory is wrong, just that I think there are a few holes that need to be plugged...)
Ex Ante vs. Ex Post Profit/Loss
Rothbard says that people always maximize utility ex ante, but that ex post they might realize they made a mistake and didn’t choose the most highly valued option. But I always teach my micro class (following the LSE lectures on cost) that cost is always an anticipation, and hence is never “realized.” E.g. if a woman chooses a suitor to marry, she thinks he will be the best choice. She may discover to her surprise that he becomes an abusive alcoholic. Still, strictly speaking she can’t know that she made a mistake, because her choice is irrevocable. She will never know how happy she might have been had she chosen someone else (or stayed single).
Diminishing Marginal Utility
Consider the following value scale, adapted from Rothbard:
- 1st egg
- 1st piece of toast
- 2nd egg
- 3rd egg
- 2nd piece of toast
This ordinal ranking is consistent with DMU. Now my question: What the heck does it even mean to say that you rank the 1st unit of a good higher than the 2nd unit? If I say I rank X over Y, that means, if faced with a choice, I pick X over Y. But it makes no sense to say I would rather have the nth unit of a good rather than the (n+1)th unit; if you have (n+1) units, you necessarily have the nth too.
I know, I know; you’ll respond that this just means that the end to which the nth unit is devoted is more important than the end to which the (n+1)th unit is devoted. But let’s switch out of successive units to see that the fishiness doesn’t come along for the ride:
- Shovel
- Sunglasses
Someone with the above ranking could explain by saying, “I value the removal of snow from my driveway as more urgent than keeping the sun out of my eyes, and so that’s why I rank those items in this fashion. If faced with a choice between shovel and sunglasses, I pick the shovel.” Nothing fishy here. So I still think the fishiness above is due to the fact that we’re trying to apply a “law of DMU” that might not make any sense if we don’t believe in cardinal utility. (Note that—I think—the law of DMU is neither necessary nor sufficient for the Law of Demand.)
Value Scales
More generally, what sense are we to make of a long ordinal ranking? Mises says that the scale of values shouldn’t be given too much meaning; action is what’s fundamental, and the analyst constructs a scale of values to understand choices.
So what does something like this mean?
- Good X
- Good Y
- Good Z
If someone can choose between X, Y, and Z, and picks X, then we conclude that he ranks X higher than Y, and X higher than Z. But how do we know the ranking between Y and Z? You could say, “Oh, well if we hypothetically removed X, then the person would choose Y.” But now aren’t we starting to sound dangerously neoclassical?
