Tuesday, November 1, 2011

Keynesian Economics

I'm trying something a little different here. Think of this as a response to one of the commenters on my last post. This is a very preliminary version of a paper I'm writing. This is my attempt to understand Keynesian economics. The departure point is this post.

I essentially took Farmer's idea, simplified it so that I could understand what he is doing, and then expanded on it, in part by incorporating monetary exchange. I have some ideas about what it all means, but have not quite settled on how I want to write it or what else I might want to include.

This is what I think is the basic Keynesian idea, with no funny business involved. There's suboptimality, and fiscal and monetary policy can correct it. Then you have to ask whether you buy the story or not. Let me know what you think.


  1. I don’t think you will ever manage to put Keynes into a nutshell but I’m certainly going to spend a bit of time on that paper, which looks very interesting. There’s no point in commenting on it until I think I understand it (which may take a while) but I will comment on this summary of what Keynes was on about: “There is a pricing problem that private sector agents are somehow unable to solve on their own. But a benevolent, smart, and well-informed government is able to step in and solve that problem.”

    I think that’s exactly right (so far). Let A be the market capitalization of the economy’s financial assets. Let W be some index of input costs (mostly wages). What Keynes is saying is that markets will not find the optimal value of A/W without assistance. Mass unemployment is a clear indication that A/W is too low, but don’t try telling a welder that the value of Exxon Mobil, expressed in welders’ labour-hours, is too low and therefore he must take a pay-cut. It’s not his fault that the stock market has the heebie-jeebies. Instead, find a way to convince investors to get out of cash and into real capital while their money is still worth something.

    That’s my Keynes in a Nutshell, but I’m under no illusions that I can wrap it up in a package worthy of a reputable journal.

    But the main thing I want to say is thanks for the pointer to the paper.

  2. On reflection, A should be the market capitalization of the economy’s risky assets. As Brad DeLong frequently points out, the fundamental problem is the (notional) excess demand for safe assets.

  3. Steve: from a quick reading, I think your model should work. But it seems to me to be unnecessarily complex.

    Consider an alternative. A representative agent divides his time between 3 activities: producing; selling (searching for a buyer); and buying (searching for a seller).

    There is a tabu against consuming your own output.

    The output must be consumed at the point of production, and agents are anonymous. So they can't barter and can't use credit and must use money.

    Then use Roger's gambit to make the price of output in terms of money indeterminate, within a range.

    If the stock of money is too low, relative to the price level, the representative agent spends most of his time searching for a buyer.

    Here's my sketch (apologies for my bad math). It's what I thought up after reading your old post on Roger.


    Yep, there is no monetary or fiscal authority. But I reckon it works. And it's very simple.

  4. I would give a different interpretation in your section 4 discussion though. Think of the government as a commonly observed sunspot, that allows agents to coordinate their activities. If agents cannot communicate with each other, but the government can communicate with all, the government can do things that private agents can't. More simply still, the government can act as a focal point in a coordination game, like the conductor of an orchestra, or a cox on a racing 8, calling out a beat. In a Keynesian underemployment equilibrium, each agent wants to increase the speed, but at the same time wants to go at the same speed as everyone else.

  5. "Private sector agents in our model are
    incapable of deciding among themselves on terms of exchange that yield socially
    optimal outcomes... However,
    the government is well-informed, and has sufficient policy tools that it can effectively manipulate the terms of exchange to bring about efficient outcomes"

    Chickens aside, is there any reason to believe private sector agents would wish to make "socially optimal" choices when personally optimal choices would be more in keeping with their desires? The differences between individually personally optimal and more generally socially optimal choices lies at the heart of the tragedy of the commons - wherein wiser collective rule making (government) is the solution.

  6. wherein wiser collective rule making (government) is the solution.

    Yes when the govt includes Barney Frank and Chris Dodd and sundry other hacks, this is not likely.
    You sir suffer from the presuppositions of Harvey Road, to quote Nobelist Buchanan.
    Markets fail and so do governments. The problem is to design institutions to minimize the fallout from the fact that men are no angels.

  7. Nick,

    This seemed pretty simple to me. You even get closed form solutions in some cases.


    Yes, the idea in Keynesian economics seems to be that the individually optimal actions (setting the terms of exchange) are somehow not socially optimal. But that's usually expressed as "setting these wages and prices is really hard," or something like that.

  8. Steve: OK. My definition of "simple" is your "really simple".

    Put it another way. You have a labour market, and an output market. Why not just collapse the two into one market for haircuts? As far as I can see, your having two markets rather than one does not play any essential role in your model. You could simplify it, and lose nothing.

    But to answer the main question in your post: yes, I do *think* your model is capturing the keynesian perspective (at least an important part of it). And, basically I like it. At least I think I would like it, if it were simpler, so I could see what's going on more clearly.

    The whole Keynesian "recessions are a shortage of AD" thing can and should be re-interpreted as "in recessions, it takes more time for sellers to search for buyers holding money, and less time for buyers holding money to search for sellers".

    And (I think) that's what your model does.

  9. why having consumer in the model? it seems you need that just to justify monetary policy. otherwise money is neutral

  10. Nick,

    With search there are not really "markets" in the usual sense, though there are objects that I call "labor market tightness" and "goods market tightness" which are just the ratios of different types of individuals searching. It's not really 2 markets, as the three types of people are coming together, producing something, and then distributing the surplus from exchange. The "shortage of AD" is a situation where prices are high and the number of consumers searching is low relative to producers and workers.

  11. anonymous,

    If you mean that I having the consumer in the model is part of the device for getting monetary exchange in the model, then that is correct. However, you might imagine setups with just workers and producers where you need monetary exchange. For example, suppose a worker and producer have to meet to produce output, but the output is something the worker does not want to consume, so the producer has to pay the worker in money. That would work also.

  12. Stephen,

    In your model, producers and workers bargain over shares of what they can produce together, and problems arise if they can’t agree on the optimal division of the surplus (the Hosios rule). In Keynes’s model, producers and workers bargain over nominal wages the real value of which will be determined later by the prices of the things workers buy. Producers assume the risk involved in incurring costs today without even probabilistic foreknowledge of future sales revenue. For Keynes, problems arise when producers lose their nerve.

    In your model, workers can improve their chances of finding a match (a job) by reducing their claims on output. In Keynes’s model, when workers reduce their claims, prices fall, and it’s not clear that the combination of lower wages and lower prices increases the likelihood that workers and producers will find a match.

    I can’t really tell whether your model captures Keynes’s view, or whether you care, but I appreciate your sharing your paper.

  13. You seem to have a pretty firm view of what Keynes's model was. When I read Keynes, I'm not quite sure what he has in mind. Can you direct me to where in the General Theory you find this?

  14. Stephen,

    On bargaining in terms of the money wage rather than in terms of the real wage, see the General Theory, ch. 2, pp. 10-13, where Keynes criticizes the “classical theory” for assuming that workers bargain over the real wage.

    In this same section, Keynes assumes that P = MC and argues that when wages (roughly MC) fall, prices fall too, so that the real wage remains (roughly) unchanged. He concludes ch. 2 by arguing that workers may not have the power to reduce their real wages by revising their money wage bargains with firms because workers have no control over prices.

    In ch. 19, Keynes allows that more labor would be employed if real wages fell, provided that demand (prospective sales revenue) remains unchanged. But, of course, he questions the latter assumption, asking whether a reduction in money wages will be accompanied by the same level of demand as prevailed before the cut in wages (p. 259).

    The idea that producers assume the risk involved in incurring costs today without having even probabilistic foreknowledge of future sales revenue is most clearly expressed in Keynes’s 1937 QJE article.

  15. Stephen: go back to Peter Diamond's coconut paper.

  16. Anonimo: what matters in the coconut paper is the ad hoc increasing return matching function, which may not be a key idea of Keynes.

  17. The substance of the argument is that demand matters for supply. Diamond implemented this idea using a technology unknown to Keynes, that is all.

  18. If the matching function is constant or decreasing return like MP, there is unique equilibrium and hence no role for "demand management".

    hmm, I think I can't follow how increasing return of matching function captures the substance that "demand matters for supply".

  19. Shouldn't this be the default model for macro ? Coordination failure is the norm for economies. Williamson and the DSGE methodology take market clearing as axiomatic. This seems bizarre - why assume this happens? Surely coordination needs to be modeled not assumed. We are still in the pre-Newtonian phase in macro.

  20. Market clearing is not the key, you can incorporate search and matching in any RBC/new Keynesian/new monetarist models as in the coconut/ MP/ Prescott-Lucas. But it seems that doesn't give you much extra insight here.

  21. Anyone who ever bothered looking at the reality of policy making and how economies react to demand shocks will have a hard time denying that a macroeconomic model where some concept of aggregate demand is operational is worth less than the ink used to print it :)

  22. In other words, if one needs increasing returns for "demand management", then it just says that the case of decreasing or constant returns on matching are empirically irrelevant, just as irrelevant as a macroeconomic model where monetary contractions do not have real effects.

  23. "O" Anonimo assumes the fact he purports to prove. "Demand shock" is a meaningless term in general equilibrium -- pray tell, what shocks don't shift the "supply curve" via labor demand/supply effects? Yeah, didn't think so.

  24. anonymous says "you can incorporate search and matching". But doesn't that just shift the black box one notch further down? Don't those guys assume some kind of matching function ? Where is the real "machinery of commerce" underlying that matching function?
    It seems that there are a lot of questions not being answered right now by macro models because we don't fully understand how markets work at the micro level.

  25. hey Anon 8:05, here is Anon 108 339 and 808 (so inhuman...). I think we can go deeper any time , as people have been proposing several micro-foundation for the matching function. Don't be lazy, "google scholar" the literature.

    The point is, since even relaxing market clearing condition gives us not much extra mileage, then why do we bother to go further and deeper along this direction? There are million of things we can look into. Focus on the important one!

  26. The confusion of AD and AS is very clear in the coconut model. Guys don't go out for shopping because they expect less supply of the good they want, which is because producers expect less demand for this. How can one say that it is demand but not supply that matters definitely? There is no much meaning to distinguish whether egg or chicken comes first here. It is just a coordination failure.

    the same Anon above