Friday, February 4, 2011

Bedtime Stories

Stories are helpful in economics. Particularly in class, a simple story can get the essence of an idea across, and then you can go on to develop the full-blown analysis that puts some rigorous structure on the idea. One story I like is this one, which comes from Joseph Ostroy (UCLA). There are two people, George and Martha, who agree that they will eat dinner at one or the other's home each Saturday night. They want to share the burden of the cooking equally, but they have very bad memories, and can never remember who last cooked dinner. However, they have a stone. George and Martha have good enough memories that they each can remember where the stone is stored in their respective homes. On Saturday night, each looks in that spot. If the stone is there, they take it over to the other person's house and eat dinner. If the stone is not there, they cook dinner and wait for the other to show up. The person who brought the stone leaves it behind, and the other stores it.

This is a money parable, of course, and it captures the essence of the most important idea that has taken root in monetary economics in the last 30 years: Money is Memory. People use money in exchange because of information frictions - essentially limited recordkeeping, or limited memory.

Paul Krugman has been writing about stories recently, and he too finds these useful. Here's one of his:
Real business cycle theory says that economic fluctuations are the result of technological shocks, amplified by intertemporal labor substitution. My version: think of a farmer who faces sunny and rainy days. On rainy days his labor won’t be as productive as on sunny days; this effect on his output is amplified by his rational decision to stay in bed on rainy days and work extra hard when the sun shines. I think this gets at the essence of the concept; it also makes you wonder, is this really, really what you think happens in recessions?
His RBC story is quite standard. The one I tell to students is pretty close to it. Again, it captures the essence of the idea. Of course, Krugman uses the story to make fun of the idea, but aside from the issue of whether the RBC model helps us understand recessions, the story is quite helpful. At the minimum it can help a student understand intertemporal substitution, which is at the root of much of dynamic macroeconomics.

I have a Keynesian story. There are two farmers, George and Martha (again). George grows peanuts, Martha grows cranberries. George eats only cranberries, and Martha eats only peanuts. These are weird peanuts and cranberries which sprout from the ground every morning, and have to be picked or they rot. Each evening, George and Martha meet at Martha's house, drink beer, and play cards. They get quite drunk, which is what it takes for them to settle down and negotiate the price at which they will trade in the morning. George's peanuts must be harvested before Martha's cranberries each morning. On a typical day, George gets out of bed, harvests his peanuts, and goes to Martha's house. When George arrives, Martha harvests her cranberries, and they trade peanuts for cranberries at the price they negotiated the night before. However, there are days when George wakes up, and it's not a day when he is so enthusiastic about eating his cranberries. The price of cranberries is too high, so he rolls over in bed and goes back to sleep. Martha makes up, and George has not arrived. Martha knows what has happened. She could call George on the phone and renegotiate the price, but she can't bear to do it, and goes back to bed. The situation is bleak. Both George and Martha express a desire to work, but there is no demand. However, there is a government, which is monitoring the situation. The government agent gets George out of bed, and puts them to work digging holes. In exchange for the hole-digging, George is given some bonds which are claims to tomorrow's peanuts. The government plans to tax George in peanuts on the next day, in order to deliver the payoff to the bondholder. However, George does not see through this. His circumstances are now changed. He goes to Martha's, trades bonds and cranberries for peanuts, and everyone is happy.

Does my Keynesian story or Krugman's RBC story help us understand the recent recession? Krugman's story has the advantage of being simple. RBC is a piece of cake relative to Keynesian economics, if you do each properly. But to argue that the recent recession resulted from a productivity shock is not helpful. Is my Keynesian story helpful? No, I'm afraid it does not cut it either.

24 comments:

  1. Hi David,

    Another satisfied customer. Always glad to serve.

    ReplyDelete
  2. Nice :) However, there is a problem with the stone - there is implied 'obligation' to cook a dinner if one obtains (or does-not-have) a stone. Nothing like that exists with money - money is somehow different. Money is memory - but it is not the stone-obligation. Is there a way to convert remove the obligation from the stone and still retain the story? I don't think so.

    As for the Keynesian story - I don't get it. There seems to be problem with 'communication'. So George digs the hole, gets a bond for peanuts. What he does with the bond tomorrow? He goes to Martha, sells her the bond for peanuts, obtains the cranberries. What shall Martha do with the 'bond for peanuts'? Not sure - unless you make a slave from George, there's nobody who would provide her with peanuts. Did it solve the 'communication problem'? I'm not sure, now we are comparing digging holes vs. cranberries... and unless you are arguing 'phone cost' vs. 'personal contact'... even if it somehow did work, you have probably destroyed quite a lot of the price information in the meantime. I don't even get the story, let alone to try to fit it to current situation :/

    andy

    ReplyDelete
  3. Andy,

    1. Stone money: No, it's not an obligation. This is all voluntary exchange. Martha accepts the stone in exchange because she believes that George will in turn accept the stone in exchange, etc., and the belief is self-fulfilling. Of course, there is another equilibrium where the whole thing unravels due to the belief that the stone will not be accepted in the future - also self-fulfilling.

    2. George takes the bond today, and exchanges it (along with some peanuts) for cranberries with Martha. Martha takes the bonds, because she likes peanuts, and the bonds are claims to tomorrow's peanuts. The story has all the elements of a Keynesian sticky price model - the market failure, the multiplier, etc. If you are struggling, that's good. Keynesian economics is not obvious. It's hard.

    ReplyDelete
  4. Sorry, but your story is far more complicated than the Keynesian model itself. I think Krugman's babysitting story works better; you really need some kind of money in a Keynesian world, otherwise it gets silly.

    In any case, I disagree with Krugman that we can judge a theory by whether it's mechanisms seem realistic. Milton Friedman taught us that what matters is the theories predictions, e.g. RBC theory predicts there will be irregular fluctuations in employment as the economy is subject to random shocks. That prediction seems to have been borne out by events.

    ReplyDelete
  5. 1. Part of my point is that it needs to be complicated. You should talk to Mike Woodford about the necessity of money for the story.
    2. Yes, I agree about the realism. We have a phenomenon we are trying to understand, and we include the bare minimum of details in the model to deal with understanding the phenomenon and making the model useful. The beauty of RBC is that it is simple, clean, and easy to understand. It does pretty well in replicating key comovements in the data. But you can go only so far with it. I think financial factors are important, and Prescott thinks the financial stuff is just a sideshow.

    ReplyDelete
  6. Part of my point is that it needs to be complicated.

    Right, but not more complicated than the model itself, surely? I mean, part of the problem is that, whereas the RBC model is a bottom-up approach that lends itself to such simple stories, the Keynesian model is designed to take a broad brush, aggregate approach. So trying to explain the Keynesian model at the level of Robinson Crusoe (plus Friday) simply obscures the key ideas. Having said that, I think Krugman's babysitting story works quite well, and also shows the flaws in that approach.

    You should talk to Mike Woodford about the necessity of money for the story.

    Well, I'd be happy to, although I thought you were discussing the traditional Keynesian model. But I regard the presence of money as implicit in Woodford type models. How else do you control interest rates other than by allowing the money supply to adjust accordingly?

    I think financial factors are important, and Prescott thinks the financial stuff is just a sideshow.

    What did you make of the recent Ohenian paper that showed problems in the financial/capital side were an unlikely cause of the recsession? He found distortions in the labour side to be much more significant, pointing to a policy explaination as a more lIkely cause.

    ReplyDelete
  7. "the Keynesian model is designed to take a broad brush, aggregate approach."

    Do you mean the approach is generally just sloppy?

    "So trying to explain the Keynesian model at the level of Robinson Crusoe (plus Friday) simply obscures the key ideas."

    For me, it actually highlights the key ideas. Maybe we think in different ways.

    "I think Krugman's babysitting story works quite well, and also shows the flaws in that approach."

    Krugman's story does not help me. I read it and think: "Someone set up a poor allocation mechanism here. They could do much better with a credit system. You just need a bookkeeper."

    "How else do you control interest rates other than by allowing the money supply to adjust accordingly?"

    That was exactly my reaction when I saw my first New Keynesian model. Take a look at Woodford's book, i.e. the "cashless economy." Prices are quoted in units of money, and the central bank is moving the nominal interest rate around, but in equilibrium no one holds any money.

    Ohanian: Yes, he calculates various wedges. The fact that there is a "labor wedge" does not mean that's got something to do with the labor market. It could come from anywhere. Indeed, a financial friction could show up in that wedge.

    ReplyDelete
  8. "1. Stone money: No, it's not an obligation. This is all voluntary exchange. Martha accepts the stone in exchange because she believes that George will in turn accept the stone in exchange, etc., and the belief is self-fulfilling. "

    I beg to disagree; exactly the same situation would arise if they made a non-enforcable agreement; you have fixed the 'price' of the stone. Money as such does not have fixed price and does not have such implied obligation. The fallacy here is that the story is argumenting from what you want it to be (I want $1 to represent some universal unit of value; I want the stone to represent the dinner) and from this you try to create a model in which 1 dinner is the unit of value and a stone 'can', obviously, represent it. However in real world there is no 'unit of value', so you cannot generalize such story to money.

    The article you linked to has the same problem - suppose I had perfect information you provided me in the past with 3 lemons, 2 dinners and 1 car wash. Now you want me to provide you with a new refrigerator. How does this information help me? I have no idea. They propose some 'strategy' with overlapping generations, however this strategy is quite strange.

    The same with Krugman story - he fixes the price and then tells the story. You could call it 'Keynesian-fixed-price-story' - if the problem was solved by changing the price. But the story doesn't allow for non-fixed-price, it would make story absurd. In reality we can change the price. The point is that the story cannot be generalized to the real world. (at least not without a long explanation how you would do it)

    "Martha takes the bonds, because she likes peanuts, and the bonds are claims to tomorrow's peanuts. The story has all the elements of a Keynesian sticky price model"

    I guess the state bankruptcy as it doesn't have the peanuts to fulfill Martha's bond is the element of the Keynesian model as well? I just didn't figure out how did the state help in such situation. You mean spending and not helping is also the feature of Keynesian model? I would understand it that way... :)

    andy

    ReplyDelete
  9. Do you mean the approach is generally just sloppy?

    No, I mean it's based on certain assumptions about how an aggregate economy works. Specifically, that aggregate relationships are stable (essentially, that the Lucas critique doesn't hold) and that the idea of a Walrasian concept of general equilibrium doesn't apply in the real world (or at least that it represents a special case). Now, 70 years on we might think those assumptions are wrong, but it doesn't follow that models built up from the decisions of individual agents necessarily lead to more accurate models of the aggregate economy. When I want to build a bridge, say, I use Newtonian physics; I don't find it necessary to analyze the bridge at the atomic level using quantum mechanics.

    Krugman's story does not help me. I read it and think: "Someone set up a poor allocation mechanism here. They could do much better with a credit system. You just need a bookkeeper."

    Yes, but I think it was based on a true story, which shows how people can set up inefficient systems that seem OK at first but lead to problems later on. Plus, it doesn't have the unfortunate feature of your story that recessions occur because people can't be bothered to get out of bed. The key point I think Krugman was trying to get across is that everyone in his story behaves quite reasonably from an individual perspective but on aggregate they lose out.

    The fact that there is a "labor wedge" does not mean that's got something to do with the labor market. It could come from anywhere. Indeed, a financial friction could show up in that wedge.

    I guess. But is it plausible that such a friction would have negligible effects on the capital market?

    ReplyDelete
  10. "it doesn't follow that models built up from the decisions of individual agents necessarily lead to more accurate models of the aggregate economy."

    Then you're not on the same page with Woodford and friends. He takes great pains to argue that his models are built up from individual optimizing behavior and are immune to the Lucas Critique, which he appears to think is important. You forget the history. People once thought of Phillips curves, money demand functions, etc. as "stable aggregate relations," but they are not, and we can write down models with individual optimizing agents to explain why they are not.

    "everyone in his story behaves quite reasonably from an individual perspective but on aggregate they lose out."

    Yes, same here.

    "is it plausible that such a friction would have negligible effects on the capital market?"

    Yes.

    ReplyDelete
  11. Then you're not on the same page with Woodford and friends

    Well, I thought your story was about the old Keynesian model. In a New Keynesian world there's Ricardian Equivalence, so George wouldn't spend his bond on cranberries. The New Keynesian approach better lends itself to such stories, but you would need some sort of credit market and an explanation for how the government changes the interest rate.

    Yes, same here.

    I don't think your characters behave reasonably at all. Why get drunk before negotiating? Why not negotiate the next day? Why does George not realize he will have to repay the government?

    ReplyDelete
  12. "In a New Keynesian world there's Ricardian Equivalence, so George wouldn't spend his bond on cranberries."

    "I don't think your characters behave reasonably at all. Why get drunk before negotiating? Why not negotiate the next day? Why does George not realize he will have to repay the government?"

    Excellent. You've got it. To make it a New Keynesian model, the intervention is different. The government agent sees that the price is wrong and changes it.

    ReplyDelete
  13. Regarding labor wedges and financial frictions, Chari, Kehoe and McGrattan (Ecta 2007) show that in many environments (Bernanke-Gilchrist-Gertler, for example) financial frictions show up as labor wedges in the prototype RBC model used for business cycle accounting.

    In Arellano, Bai and Kehoe (working paper), financial frictions plus time-varying uncertainty about firm-specific TFP shocks give rise to a time-varying labor wedge that qualitatively matches what happened in 2007-2009.

    Basically, a large labor wedge isn't a good reason to rule out financial frictions as the cause of the crisis.

    ReplyDelete
  14. Joe,

    Thanks for the references. In principle, it seems a financial friction (and even the term "financial friction" may be too vague) could manifest itself as various wedges if we're thinking in terms of some baseline frictionless model.

    ReplyDelete
  15. Stephen,

    You wrote that the idea of money as memory is "the most important idea that has taken root in monetary economics in the last 30 years".

    I am familiar with both Kocherlakota's paper as well as the corresponding literature on money and I am wondering why it is that you believe that this is the most important idea in the last 30 years. Bear in mind I am not trying to be antagonistic, but rather am trying to understand your perspective. For example, I think that this idea is important, but I much prefer Peter Howitt's paper "Beyond Search: Fiat Money in Organized Exchange." (Admittedly, this is because I much prefer the work of Ross Starr in the area of the microeconomics of money and Howitt explicitly follows that line of research.) In other words, I am asking two questions:

    1. Is it Kocherlakota's paper that you find to be the most important contribution or the idea of money as memory itself? From the context, it is not clear.

    2. Why do you believe this is one of the most important contribution to monetary economics in the last 30 years?

    Again, bear in mind, that I am not being antagonistic, but rather interested in your perspective.

    ReplyDelete
  16. Josh,

    Kocherlakota's paper is just some examples, that make the idea clear. You can find the idea earlier in some of Townsend's work as well. I think when some people (including some of the original practitioners of the approach) looked at early search models of money (Kiyotaki and Wright's 89 paper for example) it seemed that the role for money was coming from spatial separation. It is hard to find the people I would really like to trade with, and money helps me overcome that difficulty. It's clear the double coincidence problem is necessary for the monetary exchange, but if you thank about what happens in any monetary model when you allow for perfect memory, i.e. perfect recordkeeping, then you can see how critical the information structure is. In general, you can support very nice equilibria (in fact no worse than with monetary exchange) without money, in an arrangement that looks like a credit system, as long as memory is perfect. Getting a role for money requires that you shut down memory. In Peter Howitt's paper (available at http://www.econ.brown.edu/fac/peter_howitt/publication/beyond.pdf), he assumes a particular transactions technology. You have to pay a fixed cost to set up a trading post on which two objects are exchanged, and people are limited in how many trading posts they can visit each period. As you say, it's related to Starr's work. The insight you get from this is that money increases welfare, in part by allowing us to economize on transactions costs (in this case, setting up the trading posts). The limitation of the approach is that it is hard to think about credit, banks, etc., which is important to thinking about central banking, for example. In some of the models I write down recently I take some of Howitt's criticisms of search models to heart. The exchange frictions are not about finding the person with the good you want; it's all about what assets you can use in exchange, and whether you can use credit or not.

    ReplyDelete
  17. Stephen,

    Thanks for your response.

    I suppose that the reason I prefer Howitt's approach is because spatial separation generates an incentive for trading posts (or firms) to develop. In other words, I see both money and the organization of firms as a response to the double coincidence of wants problem. I would also point out that the reason I like these trading post models is because the introduction of non-convex transaction costs can explain why money exists even when there exists a double coincidence of wants (a grocery clerk is paid in money and not groceries, despite the obvious consumption thereof).

    You are correct, however, that this approach is often intractable for analyzing credit and banks -- at least at present.

    ReplyDelete
  18. Josh,

    I may be wrong, but I tend to think that, in the US economy, means of payment does not have a lot to do with industrial organization. The key elements are the information technology and theft (in various forms). There is something to the Howitt/Starr approach, in that it is another way to think about specialization and the role of monetary exchange. Of course, it all depends on what problem you are trying to address.

    ReplyDelete
  19. Stephen,

    Sorry, I think I was unclear with what I was saying. In a world in which people are spatially separated from trade, it is my view that we will not only see the emergence of money, but also the emergence of firms. For example, if I am selling shoes one way to minimize search costs is to set up a trading post where I sell my shoes. Thus, I am not arguing that means of payment are important for industrial organization, but that the emergence of firms (or trading posts) coincides with spatial separation.

    You are correct, however, that the Howitt/Starr-type models do not yet provide the means to analyze some larger questions.

    ReplyDelete
  20. Here is a simpler description of what Keynes was getting at:

    You and I live in an economy which produces just one good: delicious root beer floats. Each float has two ingredients: 12 ounces of root beer, and a scoop of vanilla ice cream. (It is considered perverse and unseemly in our country to consume either ingredient alone; the root beer float is the only socially acceptable meal.)

    Due to draconian state education budget cuts, I have acquired no skills except for producing root beer. You can do nothing except make ice cream. In fact, neither of us can even speak or write down our production plans each day.

    Every morning, I attempt to guess your mood. If I think you look energetic and full of animal spirits, I produce many barrels of root beer, anticipating a huge output of ice cream. You do the same thing in reverse. Any excess root beer or ice cream gets mailed to Ed Prescott. If either of us gets into a lethargic mood--or anticipates lethargy in the other worker--the economy produces little or no root beer floats, and we become pale and weak.

    This is a situation I might have described as "low aggregate demand"...before I learned on this blog that it is a naive, unscientific and defunct term.

    ReplyDelete
  21. 'This is a situation I might have described as "low aggregate demand"...before I learned on this blog that it is a naive, unscientific and defunct term.'

    No, it is simply an equilibrium outcome with low output and low consumption. Aggregate demand is just an unhelpful term -- the animal spirit shock altered both the production function (labor effort dropped) and the consumption function (wealth fell). So maybe it is "low aggregate supply" instead?

    Please return to your undergraduate professor and demand your money back -- you were instructed poorly.

    Go Steve!

    ReplyDelete
  22. There is room for a good Austrian economics critique of the money as memory literature.

    who wrote and updates the massive spreadsheet?

    the money as memory idea is sound but the countless Hayekian social interactions and trial and error experiments that form the unit of account to allow Misesian economic calculation cannot be done with a truly massive spreadsheet in a large society of constant change.

    money is better than memory.

    Memories are overtaken by events, the new situation must be learned, there must be rewards for learning quickly.

    Economic memories need a cardinal number that emerged after testing in market competition to add and subtract different memories and forecast the future.

    the unit of account is not manner from heaven.

    ReplyDelete