Wednesday 8 December 2010

The other DeLong quibble

I know Brad DeLong outranks me by... ok, I'm starting to think about problems when dividing by zero. But when I teach micro theory, I draw a really sharp line between market failures (defined relative to the first welfare theorem) and things that folks might just not like about market outcomes. So if we think it's not fair that the result of markets is that some people are poor, that doesn't get to count as a market failure. It's a failure to achieve certain distributional outcomes that may be viewed as desirable by some; it can be a failure to maximize particular social welfare functions other than the default one implied by efficiency. But it's no more a market failure than that Eric isn't a billionaire (as much as Eric would like to be one). If we care about "some people might be poor" and want to solve that, that's when we bring in the second welfare theorem and suggest that we try to fix that problem with redistribution, if it's a problem. But it's not a market failure; it's just a (potentially) undesirable outcome of markets. If we're on the contract curve, it can't be a market failure. And "some people are poor" doesn't push us off the contract curve; the on-diagonal corners of the Edgeworth box are just as Pareto efficient as every other point on the curve.

What's the first problem that DeLong points to in his second big lesson for Econ 1 students?
First, the market will go wrong if the wealth distribution is wrong. The market judges value by willingness to pay, and the rich are much more willing to pay them the poor, and those without wealth or income have no willingness to pay at all. If your wealth and income are zero, then the market literally does not care whether you live or die--it is of no interest to it at all.
Well, it might go wrong relative to Brad's SWF that worries a lot about poor people. But what makes Brad's "not wrong" point on the contract curve any better than mine?* You have to point to something outside of economics to make the case for one point over another, barring the point that emerges as outcome of voluntary trade from folks' initial endowments so long as those endowments are also the result of prior voluntary trade. That point on the contract curve gets slight privilege as moves away from there in the real world will encounter leaky bucket problems if we can't run the transfers of endowments without incentive effects.

I'm not saying that we shouldn't care about the poor, or that there might not be good reasons for wanting redistribution to get to a different point on the contract curve (of a likely somewhat smaller Edgeworth Box). But I'd not say that markets have "gone wrong" if we wind up at some point on the contract curve that isn't my favorite one.

*For my favorite point on the Edgeworth Box, draw it as follows. In the bottom left hand corner, write "Eric". In the top right hand corner write "Everyone Else". Label the horizontal and vertical axes as being two composite commodities that Eric views as goods: horizontal can be "All the good things that begin with the letter B", vertical can be "All the good things that do not begin with the letter B". Draw a contract curve stretching from bottom left to top right. Draw some indifference curves too, if you like, noting what the labeling of the axes implies about the shapes of the curves. Write in a nice title for the box too. Then, put a great big dot on the top right hand corner that says "Markets haven't gone wrong". For all other points on the contract curve, write "Markets went kinda wrong". Put a small dot somewhere close to the bottom left hand corner, on the contract curve, labeled "Brad thinks markets went right (but they didn't). For all points off the contract curve, write "Markets went really wrong - Pareto inferior solutions".


  1. My hunch would be to say that the market "fails" when it doesn't do something it could do, but doesn't "fail" when it doesn't do something we have no reason to expect it to do. Like distribute wealth ideally, or make the Ottawa senators win the Stanley Cup.

    We don't say a car fails when it won't cross the ocean.

    But my hunch eventually fails too, of course. Since it depends on how much we expect the market to do.

  2. @Nick: And we also don't say an engine fails when it doesn't get 100% transformation of energy into useful work. Engineering efficiency is a blackboard ideal that's useful as a benchmark, but nobody's crazy enough to condemn engineers or to demand government-built engines because of our failure to achieve engineering efficiency...

  3. Continuing with your engine analogy, I think the problem for some people is that they expect *all* engines to meet a minimum efficiency threshold. You are maximizing performance for the population of engines, but there are specific engines that are definitively suboptimal.