Tuesday, October 16, 2012

A Nobel for economics that really works


Out here in the blogosphere, it is common to hear things like the following:

1. "Economics doesn't work; it has no practical applications."

2. "Economic will never discover any stable scientific laws, because human behavior changes."

3. "Economics shouldn't use math, because math can't describe human behavior."

4. "Economics is not a science."

I have some sympathy for these viewpoints. But economics is a very broad discipline, and I think there are many cases in which these criticisms couldn't be more wrong. There are cases in which economics works, in which it does discover "laws", and in which difficult math is absolutely essential. For example, consider the theories that won the Economics Nobel Prize this week.

The prize, given to Lloyd Shapley (who, by the way, spends his summers at Stony Brook) and to Alvin Roth (who was recently hired by Stanford despite being commonly cited as working for Harvard), was awarded for the invention of Matching Theory. Matching Theory is basically an algorithm - a mathematical technology - for finding optimal matches between pairs or groups of people. It incorporates human preferences, optimization, and strategic behavior, so it is economics. Alex Tabarrok gives a great introduction to the theory in this blog post.

I would like to point out some things about Matching Theory:

1. It is testable, tested, and correct for a very broad class of situations. There are many situations in which the theory makes predictions about when matching will be stable. As Tabarrok points out, these predictions have been confirmed in a number of real-world situations (and, I am sure, in controlled experiments as well).

2. It is practically applicable. Implementing the algorithms designed by Al Roth has resulted in much improved availability of kidney transplants.

3. The theory uses a lot of math. It does not rely on verbal characterizations of human behavior, but on hard quantitative predictions derived from non-trivial mathematics. Without that math, the theory would be useless.

In other words, Matching Theory is what most scientists would call science. Nor, I believe, is it the only such example. Critics of the economics profession should realize this. It's an important fact that not all fields of economics - and not all techniques and theories and schools of thought within each field - are created equal, in terms of their testability, real-world applicability, and appropriate use of mathematics.

As for the Nobel, I see this decision as increasing the credibility of the prize itself. The econ Nobel traditionally lies somewhere in between the Peace Prize - which everyone knows is a big joke - and the Physics, Chemistry, and Medicine Prizes, which have managed to maintain very high levels of credibility. But the Economics Prize has always seemed to alternate between testable, applicable, "science-y" sorts of economics (think James Heckman's selection models, Daniel Kahneman's experiments, William Vickrey's auction theory, or the 2007 prize for mechanism design) and less testable, more "storytelling" kinds of econ (I am sure you know which ones I'm talking about). It is no coincidence that the former tend to be prizes for microeconomics and the latter tend to go to macroeconomists.

My guess - and this is just a wild guess - is that in the years since the Crisis of 2008, the enormous wave of criticism directed at the econ profession has not been lost on the Nobel Prize committee. The only macroeconomists selected for the prize in the past few years have been Chris Sims and Thomas Sargent - two hardcore empiricists whose work serves to illuminate the data limitations and huge error margins faced by macroeconomics. It is my sincere hope that the prize will continue to move in the direction of the science prizes, toward testable, applicable theories and credible empirical results.

And in the meantime, my heartiest congratulations to Lloyd Shapley and Al Roth, who richly deserved the prize.

Update: Mark Thoma is thinking along similar lines.

24 comments:

  1. On economic science. I think most of it is much more empirical than Roth's work. Economists doing obvious experiments or assessing natural experiments.

    Obviously I don't think a Nobel should go to typical work -- it should go to excellent work and developing a brilliant hypothesis which fits the data is better than testing an old hypothesis (or bit of conventional wisdom) and finding it fits, or doesn't.

    Also you know, Muhammad Yunis (PhD in Economics) could have gotten the economics prize not the peace prize (as Norman Borlaug (mr green revolution) could have gotten biology not peace). Yunis discovered an amazing stylized fact (almost a law if you will) that poor people repay tiny loans even if there is no cost effective way to enforce collection. Later work has demonstrated that it is possible for micro lenders to mess up (basically if there are at least two and they compete for maximum loan volume). Finding conditions under which micro lending works and in which it doesn't are equally interesting as science. I propose a joint economics nobel for empirical work in micro finance with one for Yunis (showing how to do it) and one for someone who showed how not to do it (plus Esther Duflo cause she works hard and gets fascinating results).

    ReplyDelete
  2. Feeling trollish....

    "There are cases in which economics works, in which it does discover "laws", and in which difficult math is absolutely essential."

    What do you mean by math? Thinking quantitatively? What is (are) the borderline (s) between math and nonmath? How do we differentiate between when we should use qualitative and quantitative reasoning?

    ReplyDelete
    Replies
    1. That's not trollish, that's a good question!

      Delete
    2. Socrates was the ultimate troll (IRL!), and he asked more good questions than I ever will...

      Delete
    3. Actually, many mathematical results are qualitative, not quantitative.

      As for when it is appropriate, the Einstein criterion still seems the best: "Make things as simple as possible, but not simpler.”

      Delete
    4. Wonks Anonymous —

      I follow the principle, but I really dislike Shapley's definition.

      What if John never proposed to Mary because he was too shy, for example? What if he did propose to her and she wanted to marry him, but her parents pressured her into marrying someone else?

      As such I would modify Shapley's definition: mathematics is any abstract, linear argument carried out with precision. Those two qualifications make an awful lot of difference.

      That said, I convinced Noah yesterday to write about this, so I am interested to see what he has to say.

      Delete
    5. Yep, I'm going to write a post about this. Basically, I think of "math" as meaning different things in different situations, but in the case of economics, I essentially just equate it with "very precise language".

      Delete
  3. "so it is economics"

    Sounds like you want to convince yourself and others. Is it really economics?

    ReplyDelete
    Replies
    1. Anonymous7:58 AM

      Nope! Matching models are *puts on sunglasses* all chemistry.

      Delete
    2. Even Prof. Shapley was surprised by the honour. “I consider myself a mathematician and the award is for economics,” he told the Associated Press on Monday. “I never, never in my life took a course in economics.”

      Yes sunglasses.

      Delete
    3. @Anonymous: YEAHHHHHHHHHHHHHHHHHHHHHHHHHH!

      Delete
    4. Sounds like you want to convince yourself and others. Is it really economics?

      Yes.

      Delete
    5. Phil Koop7:32 PM

      "Is it really economics?"

      Exactly. Once you find something testable, practically applicable that uses lots of math, you also find that it's only given a Nobel for economics because there are no Nobel prizes for math or computer science or cognitive psychology etc etc.

      Delete
    6. Let us see, an algorithm that matches buyers and sellers in markets without a Walrasian auctioneer. I wonder what that has to do with economics. Right!

      I also think philosophers should claim Adam Smith, mathematicians should claim John Nash, James Heckman, and others, historians should claim Robert Lucas (his B.A. was in history), and so on. I wonder if economics even exists as a field. Hmmm...

      Delete
    7. I also think philosophers should claim Adam Smith, mathematicians should claim John Nash, James Heckman, and others, historians should claim Robert Lucas (his B.A. was in history), and so on. I wonder if economics even exists as a field. Hmmm...

      And even if all those other fields did claim those people, it wouldn't mean economics doesn't exist as a distinct field! Every theoretical physicist could be claimed by math departments, and every experimental physicist by engineering departments. There's always discipline overlap.

      Delete
  4. JM Pinder8:13 AM

    "The only macroeconomists selected for the prize in the past few years have been Chris Sims and Thomas Sargent - two hardcore empiricists whose work serves to illuminate the data limitations and huge error margins faced by macroeconomics."

    What about Peter Diamond, Dale Mortensen and Christopher Pissarides (the 2010 prize) for their work on search/matching models of the labour market. This is microfounded theoretical macro..

    ReplyDelete
    Replies
    1. Andrew1:52 PM

      Shhh. Don't ruin the narrative.

      Delete
    2. I don't think it ruins the narrative. The Diamond/Mortensen/Pissarides model gives good results for certain (micro) features of the labor market, but when you try to combine it with business-cycle theory - usually just by slapping on some RBC-style technology shocks - it gives counterfactual results, as found by Robert Shimer. So I think if the model didn't give those good micro results, it wouldn't have won the prize.

      Delete
  5. I have read that matching theory can produce outcomes where no alternative will make the results better for all the participants.

    Isn't that a Pareto-optimal outcome?

    Is this indeed a centralized process that can produce Pareto-optimal outcomes?

    If so, that would account for a lot of the disparagement from conservatives/libertarians/market-fetishists.

    ReplyDelete
    Replies
    1. Wonks Anonymous1:21 PM

      Maybe I'm reading the wrong blogs, but I got the impression from MR, EconLog and Greg Mankiw that the prize was richly deserved.

      Delete
    2. Yeah, I haven't seen anyone disparage the prize...got links??

      Delete
    3. If you want to find libertarians, you'll find more at libertarian blogs. The Tabarrok article (which is excellent, and I don't usually compliment libertarians) has at least two disparagers.

      But what I really want to know is if Pareto-optimality is an accurate description. It looked like it to me.

      And yet another question: what is the real synonymy for "Matching Theory"? It seems to go under a variety of names, and it is difficult to locate through wikipedia.

      Delete
  6. Anonymous7:58 PM

    You are calling engineering & architecture "science".

    That is all.

    ReplyDelete