Thursday, June 07, 2012

Economic theory that actually works


Economic theorists have worked hard to lower expectations for their field. The standard line (originally due, I think, to Keynes?) is that, unlike a science like biology that can be applied to the real world, economic theory merely gives people a way to think about complex issues in an intellectually disciplined, rigorous manner. One subtext is that economic theory need not be subjected to the rigors of the scientific method; that theory need not match data, and that the criterion for judging a theory a success should basically be that other economists think the theory is cool. Another implication is that the study of economic theory need not ever be justified by the present or future creation of useful technologies. In other words, it doesn't have to be right or useful, only interesting.

In a recent interview, game theorist Ariel Rubinstein says this of his own field:
Is game theory useful in a concrete sense or not? Game theory is an area of economics that has enjoyed fantastic public relations...I think it’s a very tempting idea for people, that they can take something simple and apply it to situations that are very complicated, like the economic crisis or nuclear deterrence. But this is an illusion...I don’t respect the claims that [game theory] has direct applications... 
Game theory is about a collection of fables. Are fables useful or not? In some sense, you can say that they are useful, because good fables can give you some new insight into the world and allow you to think about a situation differently. But fables are not useful in the sense of giving you advice about what to do tomorrow, or how to reach an agreement between the West and Iran. The same is true about game theory... 
In general, I would say there were too many claims made by game theoreticians about its relevance. Every book of game theory starts with “Game theory is very relevant to everything that you can imagine, and probably many things that you can’t imagine.” In my opinion that’s just a marketing device... 
[We do game theory] because it is interesting...I believe that intellectual thinking – philosophy or logic or game theory – is very useful in the cultural sense. It’s part of the culture, it’s a part of our perpetual attempt to understand ourselves better and understand the way that we think. What I’m opposing is the approach that says, in a practical situation, “OK, there are some very clever game theoreticians in the world, let’s ask them what to do.” I have not seen, in all my life, a single example where a game theorist could give advice, based on the theory, which was more useful than that of the layman... 
I categorically cannot see any case where game theory could be helpful.
Naturally, this opens the economics field to the obvious criticism of "Well, fine and good, but then why do we pay econ theory professors so much more than literature professors?" (That, as you might expect, is not a question that many economic theorists are eager to bring up in conversation.)

But here - surprise! - is where I come to game theory's defense. Rubenstein seems to be talking about the national defense applications of game theory, which were much-ballyhooed in the Cold War but (I agree with Rubinstein) were pretty heavily oversold. Yet, although I am not a game theorist, I believe I know of some very useful real-world technological applications of game theory that have nothing to do with nuclear deterrence.

I'm talking about auction theory. Auction theory describes a situation where the sequence of moves is known, and the payoffs are (close to) constant - i.e., just the kind of situation where game theory allows concrete predictions. And lo and behold, auction theory is used to power the internet's most successful advertising technology: Google's AdWords. Everyone except Google has had trouble making money off Web advertising (look at Facebook), but with the help of auction theory, Google has built one of America's most profitable technology businesses.

That's real technology based on economic theory. It's really just one subset of mechanism design, which (people tell me) is the hottest sub-field of game theory right now. Mechanism is deep, pure economics - it's the science of incentives. And because there are real situations in which incentives are observable and well-defined, you can use mechanism design to predict and control the outcomes of certain kinds of transactions and interactions. That's applied economic technology, just as surely as your smartphone is applied computer science technology.

So why do we pay all these theorists all this money? Are a few auction applications really such a big deal? To this I respond: There may be big payoffs in the future. If economists can use theory to understand the business cycle, then this may lead to technologies that allow companies to anticipate and prepare for downturns, as well as - even more importantly! - government policy technologies that allow the government to prevent or effectively fight recessions. That would be a huge prize. In fact, many people believed (and some still believe) that the development of monetarist/New Keynesian/demand-based macro theory in the mid to late 20th Century has dramatically improved central banks' ability to fight recessions. If so, then that is a very important economic technology. But maybe we can do a lot better than that.

One obvious question raised by this post is: Why is there essentially no game theory in macroeconomics?...

Update: Speak of the devil. I'm now watching Gunter Strobl of the NY Fed UNC Chapel Hill present a model that uses game theory (with adverse selection and moral hazard) to explain liquidity cycles. Not sure how well the model works yet, but glad to see people doing this sort of thing...

Update 2: I am now watching Marcus Opp of Berkeley add I/O-style game theory to a macro model to show how oligopolistic competition could amplify business cycles. Love it. Not sure why Opp doesn't follow Rotemberg & Woodford (1992) and use AD shocks (he uses RBC-style productivity shocks). I do like that the model uses inelastic labor supply.

49 comments:

  1. Given the current economic circumstances I think it may be a bit premature to declare that monetary policy has "dramatically improved." I also remain skeptical that recessions can ever be prevented. That being said, I think this post raises some very interesting questions for economics that deserve further attention.

    ReplyDelete
    Replies
    1. Anonymous11:13 AM

      Bubbles and bursts result from a mismatch between real and financial assets brought about through leverage. Leverage is the basis of financial profits.

      Get rid of the leverage, problem solved.

      Delete
    2. Getting rid of leverage would reduce the volatility greatly but I still don't think it would eliminate recessions. Allowing some leverage can also be beneficial, the problem arises when leverage/debt is subsidized or guaranteed.

      Delete
  2. labor market auction. when thats figured out, someone will make a killing.

    ReplyDelete
  3. Hey Noah - great blog! I'm curious where you stand on the upcoming cliff. Are you for or against letting the tax cuts expire?

    ReplyDelete
  4. Here's my latest post, discussing something Lord John Maynard Keynes wrote:

    http://socialmacro.blogspot.com/2012/06/taking-minor-exception-to-lord-keynes.html

    ReplyDelete
  5. Noah: "One obvious question raised by this post is: Why is there essentially no game theory in macroeconomics?... "

    Read Krugman - it's clear by now that the macro elites have figured that being right-wing wh*res pays better and has no downside. So, in the face of the data, they wh*re.

    ReplyDelete
  6. > Google has built one of America's most profitable technology businesses.

    That can't be right, because Google pays hardly any tax, especially internationally.

    Or is Google profitable *because* it avoids paying taxes - games the system, in other words?

    ReplyDelete
  7. Anonymous12:41 AM

    rotemberg-woodford (1992, JPE)....?

    ReplyDelete
  8. Anonymous1:22 AM

    I thought game theory+macro was a field pretty much taking off. Like what this guy does: https://sites.google.com/site/wfuchs/research

    Or the new dynamic public finance of Kocherlakota.

    (Not to start a fight, but Barry needs to read someone other than Krugman to know what is going on in macro).

    ReplyDelete
    Replies
    1. Moar links! MOAR!!

      Delete
    2. My macro teachers are telling me the same thing... here are a few of the articles we read...

      currency crises - http://www.jstor.org/stable/116850
      bank runs - http://www.aeaweb.org/articles.php?doi=10.1257/aer.99.4.1588
      dynamic public finance - http://onlinelibrary.wiley.com/doi/10.1111/1468-0262.00255/abstract

      (not sure why these are not showing up as links)

      Delete
    3. "(Not to start a fight, but Barry needs to read someone other than Krugman to know what is going on in macro)."

      My comment was a little unclear; what I meant was that the elites in Macro had worked out the game, and figured that truth just didn't pay.

      Now, I am unclear about the details of macro guys, so please inform me as to elite macro guys who have Krugman's record of truth, and who've publicly challenged the existing elites.

      Delete
    4. MC - Yes, I've seen some of the papers on currency crises (and I should probably read the ones on bank runs). The dynamic public finance I believe in a lot less, though it's fine as far as it goes.

      Anyway, what's really important is that these things get labeled "macro theory". As of now, people still mostly think of macro theory as being models that involve stochastic fluctuations around long-term trends; models of specific events that have macro ramifications seem to be considered of secondary interest by macro folks.

      Delete
    5. "Anyway, what's really important is that these things get labeled "macro theory". As of now, people still mostly think of macro theory as being models that involve stochastic fluctuations around long-term trends; models of specific events that have macro ramifications seem to be considered of secondary interest by macro folks."

      What's funny to me as a statistician is that 'models that involve stochastic fluctuations around long-term trends' sounds like the sort of stuff that a statistician who didn't know economics would do, if he/she decided to 'do' economics (i.e., naive about the subject matter).

      Delete
    6. Dynamic public finance lives in a universe very far away from our own...

      Delete
  9. Anonymous2:59 AM

    "Why is there essentially no game theory in macroeconomics?" You're kidding right?

    ReplyDelete
    Replies
    1. You may consider that rational expectations is an attempt at inserting game theory into macroeconomics.
      But a more convincing game is the one described by Bagehot, and recently unearthed by Brad De Long
      http://delong.typepad.com/20120411-russell-sage-delong-paper.pdf

      Delete
  10. Game theory is based on individual decisions. Macro is split between trying to sum up millions of micro (game theory relevant) decisions, and trying to influence those decisions thru policy, either monetary/ FED, or fiscal.

    Those trying to influence policy then use macro, selectively, to push their preferred policy, usually with less than full honesty.

    Finally, here's Grey's Law: any economic "law" which can be used to make money, will lead to changes in behavior to make money, but such changes will invalidate the "law".

    And of course, implicit in Rubenstein, his game theory results can't be used to make money, so seem less, if not irrelevant -- but they won't be invalidated.

    ReplyDelete
    Replies
    1. "Finally, here's Grey's Law: any economic "law" which can be used to make money, will lead to changes in behavior to make money, but such changes will invalidate the "law".

      And of course, implicit in Rubenstein, his game theory results can't be used to make money, so seem less, if not irrelevant -- but they won't be invalidated."

      Don't those 'laws' depend on an efficient and rational market?

      Delete
  11. Anonymous6:43 AM

    One application for game theory that is hugely important and changed the world is the application of Game Theory on war.
    For example on missiles, and automatic fighters (drones etc), those are real applications. Don't forget Nash was crazy but his game theory stuff was very important to the military.

    ReplyDelete
    Replies
    1. Actually, this is exactly what Rubinstein spends a lot of time arguing against in his article.

      Delete
    2. Seconding this - is there anything *successful* about the West in the Cold War which wouldn't have been clear before game theory?

      Delete
    3. @Noah: RBC + game theory have gone together since at least the early 80s. In addition, there's a world of difference between the older *strategic* style of game theory used in the cold war (basically making up numbers for prisoner dilemma set-ups) and what today's tacticians use game theory for.

      @Barry: The transition from Strategic Deterrence to brinksmanship,Dead Hand/Cobalt Bomb-type weapons, and Domino Theory were all events/strategies employed during the cold war that were explicitly modeled as games by RAND. Not sure how to measure success. But Dead Hand was never used, Domino Theory was basically true*, and the number of military conflicts the US was directly involved during the period of strategic deterrence was small relative to later decades. The RAND modelers weren't far off.

      *It wasn't true in fact because Communism was never a sustainable economic system. Every state that became Communist was actually a small draw on the Soviet Union's resources. But in terms of a model that predicts which countries are going to get attacked next and where the opponent would be the strongest, it doesn't get better than domino theory.

      Delete
  12. Anonymous8:42 AM

    @ Grey

    Your law should be restated: any economic law which can be used to make money, will lead to changes in behavior such that excess returns on applying that law fall to zero. Given that economic theory has productive applications, it earns its equilibrium contribution to society. I otherwise agree, especially with your emphasis on aggregation.

    ReplyDelete
  13. To this I respond: There may be big payoffs in the future. If economists can use theory to understand the business cycle, then this may lead to technologies that allow companies to anticipate and prepare for downturns, as well as - even more importantly! - government policy technologies that allow the government to prevent or effectively fight recessions. That would be a huge prize.

    Given that economists seem willing -- happy even -- to embrace the notion that the scientific method can't and shouldn't apply to their so-called discipline, I don't see how they're ever going to get to such a place. Y'all mostly (yourself excepted) seem to be using economics to (thinly) veil political ideology in a gloss of science.

    ReplyDelete
  14. Economics is still largely in the same stage as pre-Newtonian science. It was Keynes himself who said "Newton was not the first of the age of reason, he was the last of the magicians." In fact, when Newton's papers on alchemy surfaced at a Southeby's auction, it was Keynes who purchased many of them to add to his already extensive collection of Newton 's writings on the alchemy and the philosophers stone. (Newton Project Manuscript list at: http://www.newtonproject.sussex.ac.uk/prism.php?id=82&cat=Alchemy

    You might want to look up this article by Julian Reiss, A Plea for (Good) Simulations: Nudging Economics Toward and Experimental Science at:http://somewhatlogically.com/?p=489 Simulations have been used to provide workable analysis of highly complex situations where the intractability of equations describing non-linear phenomena prevent practical answers.

    Too many practitioners of modern economics seems to be doing nothing more than justifying clearly non-provable dogma driven by self interests, though some of its practitioners seem to have been quite good at helping Wall Street spin dross into gold. And the belief that exponential growth will solve all problems is not only absurd but dangerous. For a funnily serious look at this issue, see http://physics.ucsd.edu/do-the-math/2011/07/galactic-scale-energy/ where physicist Tom Murphy calculates that at current growth rates, we will consume all of the energy in the galaxy. He ties these to historical benchmarks, and the time scales should give even the most dogmatic economist pause to think.

    As far as usefulness, keep in mind that Newton's work in alchemy stood him in good stead when he was placed in charge of the Royal Mint, and in an epic battle against a master counterfeiter, where his exploits were worthy of Sherlock Holmes. This is well documented in a good read in Newton and the Counterfeiter by Levenson http://www.guardian.co.uk/books/2009/aug/16/newton-counterfeiter-thomas-levenson

    There are lots of interesting problems in economics that lend themselves to game theory, control systems theory, statistical mechanics, econophysics and analog dynamic simulation (I'm doing some work in this area...very much inspired by Phillips work in the 50's (see notes in this blog post http://somewhatlogically.com/?p=598

    Properly done, with an open mind, economics might even become scientific and much more fun, and perhaps more useful.


    JRHulls
    somewhatlogically.com


    JRHulls

    ReplyDelete
    Replies
    1. OOPS....wrong link on Reiss paper....http://www.jreiss.org/papers/S%26G_42(2)_2011.pdf

      Delete
  15. Anonymous11:15 AM

    I would suggest that the problem is intractability at large scales (i.e. the macro-level). Like real business cycle and agent based models, game theory based models have yet to produce reliable forecasts of future periods.

    ReplyDelete
  16. Anonymous11:18 AM

    I would suggest the problem is intractability. Like RBC and ABM models, game theory based models have yet to produce reliable forecasts or predictions at large scales (i.e. the macro-level).

    ReplyDelete
  17. Anonymous11:19 AM

    Sorry about the double post, I got a page error the first time and it appears like the comment still went through

    ReplyDelete
  18. Anonymous11:32 AM

    There used to be a time when Christianity dominated Europe. Theology was considered a science, and the learned spent a great deal of time debating obscure questions that had real consequences for the peons who were just trying to make a living. Theologians had chairs in the university, were well paid and respected. The peons went to church to face the back of a priest hardly more educated than they were to hear the truths of the science in Latin. It all must have sounded very important, since nobody could understand a word of it, or why it was relevant. To keep the lesser educated entertained, stained glass with colorful images were installed in churches.

    Over time, the Enlightenment happened and exposed theologians for the quacks and magical thinkers that they are.

    Today we have economic theorists, who consider what they do to be scientific. They spend a great deal of time debating obscure questions that have real consequences for the peons who are trying to earn a living. They talk in a rarefied language called math that escapes the peons. Economic theorists have chairs in the university, are well paid and respected. The peons listen to pundits who are barely more educated than they are deliver chewed-down versions of the grand theories of the day. It sounds very important, because nobody understands what is being said, or why it is relevant. To keep the lesser educated entertained, pundits make heavy use of colorful charts and graphs.

    The free thinkers watch and endure with skepticism, and hope for another Enlightenment.

    ReplyDelete
    Replies
    1. Anonymous11:53 AM

      Who are these free thinkers? The people who dont understand the math or economic models? The sound like the ignorant, particularly because they are waiting for another enlightenment rather than working through the models to produce it.

      Delete
    2. Anonymous12:13 PM

      I wrote this post, didn't I? Don't presume your way is the only way to challenge the orthodoxy of the day.

      Delete
    3. Today we have economic theorists, who consider what they do to be scientific.

      Yeah, well, see original post. Auction theory works really well. It's real. So what do you have to say to that, eh? ;)

      Delete
    4. Anonymous1:35 AM

      You are all just a bunch of humorless pedants.

      Delete
  19. Anonymous12:06 PM

    They are the ignorant indeed, and are counting on you, Wise One, to do all the hard work for them.

    Oh, Kingdom come!! : )

    ReplyDelete
  20. I think the truth is that economics is in a highly decadent state. But then one could say the same thing about English and any number of other disciplines outside the hard sciences. (History itself is not so much decadent as decayed. There's nothing left. ) Higher education is in desperate need of reform, but it's hard to see where it will come from.

    The only thing I can think of is new schools. Oxford and Cambridge were in a highly decadent state in the 18th century. But then along came the University of Edinburgh. Centi-millionaires, who are a dime a dozen, like to endow fancy buildings on college campuses. Often with their names engraved on them. What a bore!

    If you really want to make your mark endow a whole liberal arts college, with low tuition, high standards, a conservative curriculum, and student bodies that represent the full ethnic and geographical diversity of America. The field is wide open. It only takes a billion dollars or so. We've got billionaires. What are they waiting for?

    ReplyDelete
  21. Anonymous12:17 PM

    I wasn't saying I was the wise one, believe me I am not. I was just trying to say that skepticism without reason or understanding is as ridiculous as blind faith in some old Latin sayings.

    ReplyDelete
  22. "Why is there essentially no game theory in macroeconomics?..."

    Do you mean theoretically or concretely?

    Because while there may be none theoretically, concretely I think we are currently witnessing a two player game between the FOMC and the Congress where the Nash equilibrium results in a lack of aggregate demand stimulus. The payoffs appear to be unilaterally optimal to the players involved but the results are wholly unsatisfactory for the American people.

    ReplyDelete
  23. This comment has been removed by the author.

    ReplyDelete
  24. "Why is there essentially no game theory in macroeconomics?..."

    I think for the same reason Isaac Asimov's psychohistory would only work if the subject(s) have no knowledge about it's existence.

    ReplyDelete
  25. Anonymous8:55 PM

    macro is like weather prediction, micro is like heat transfer along a beam

    ReplyDelete
  26. This is my take on what Ariel Rubinstein (btw this isn’t anything new from him, he’s been critical of economist acting as sages since like forever) is saying:

    Economist study social (as opposed to natural) phenomena. Game theory are 'fables' created by economists born out of the observation that people act strategically. The fact that economists know the theory behind strategic phenomena doesn’t make them the best people to go to learn how to be good e.g. chess players, soccer coaches, military/political/whatever strategists. In other words you don’t need to know game theory to be a good whatever 'player'. It might help, but people have been using 'game theory' since forever without knowing it’s game theory.

    ReplyDelete
    Replies
    1. i.e. best theorists doesn’t necessarily overlap with best practitioners.

      Delete
  27. dilbert dogbert11:16 AM

    "Sylvia Nasar was a reporter for the New York Times when she covered the success of the [telecommunications] spectrum auctions in 1994. The auction was described – in my opinion wrongly – by the popular press and by some game theoreticians as the glorious success of the field of game theory, in terms of making it applicable. But in any case, the success was in contrast to the misery of one of its important contributors, John Nash."
    Just read this.

    ReplyDelete
  28. Lulz4l1f39:26 PM

    "The third weakness that I find in economics is what I call physics envy. And of course, that term has been borrowed from....one of the world’s great idiots, Sigmund Freud. But he was very popular in his time, and the concept got a wide vogue.

    One of the worst examples of what physics envy did to economics was cause adaptation and hard-form efficient market theory. And then when you logically derived consequences from this wrong theory, you would get conclusions such as: it can never be correct for any corporation to buy its own stock. Because the price by definition is totally efficient, there could never be any advantage. And they taught this theory to some partner at McKinsey when he was at some school of business that had adopted this crazy line of reasoning from economics, and the partner became a paid consultant for the Washington Post. And Washington Post stock was selling at a fifth of what an orangutan could figure was the plain value per share by just counting up the values and dividing. But he so believed what he'd been taught in graduate school that he told the Washington Post they shouldn't buy their own stock. Well, fortunately, they put Warren Buffett on the Board, and he convinced them to buy back more than half of the outstanding stock, which enriched the remaining shareholders by much more than a billion dollars. So, there was at least one instance of a place that quickly killed a wrong academic theory." - Charlie Munger

    ReplyDelete
  29. "That's real technology based on economic theory. It's really just one subset of mechanism design, which (people tell me) is the hottest sub-field of game theory right now. Mechanism is deep, pure economics - it's the science of incentives. And because there are real situations in which incentives are observable and well-defined, you can use mechanism design to predict and control the outcomes of certain kinds of transactions and interactions. That's applied economic technology, just as surely as your smartphone is applied computer science technology."

    This made me smile, mechanism design is algorithmic game theory, and the best results (including a lot of the reason why Google uses it) are coming from theoretical computer science. Funny that you claim this for economics. To me it looks like both auctions and smartphones are applied computer science technology.

    ReplyDelete