« An update on the concentration of income in Canada | Main | Ban the penny! And the nickel, while we're at it »


Feed You can follow this conversation by subscribing to the comment feed for this post.

I would be interested in seeing you and others (sumner, beckworth, woolsey etc.) respond to this Mark Thoma post http://economistsview.typepad.com/economistsview/2010/05/whos-really-standing-in-the-way-of-progress-on-macroeconomic-theory.html. It seems to me that this is the new, Standard View as you have discussed, and it would be nice to hear again why you think this view is wrong.

On a related note. I thought the findings about "reference prices" were interesting http://www.voxeu.org/index.php?q=node/3611 . I think Thoma talked a little about this a while back.

jsalvati: there are 3 things going on in that post by Mark Thoma: a minor blogosphere war; the general question of what's wrong with macro; the more specific question of whether monetary and/or fiscal policy can increase aggregate demand.

Was it the third question you wanted addressed? If so, the most important component of my views on this question is found in my old post: http://worthwhile.typepad.com/worthwhile_canadian_initi/2009/11/interest-rate-targeting-as-a-social-construction.html
But, as I confessed, that was the weirdest post I have ever written, and likely to be met with incredulity. The Standard view has created the very social reality of the trap from which it can see no escape, at least via monetary policy.

On the "reference prices": Yes, I read that when it came out. Fascinating. But also depressing. Because it seems to mean that any simple "menu cost" story of sticky prices won't work. They change prices a lot, but put them down, then back up again, *to exactly where they were originally*. It sounds very plausible. But it totally contradicts any simple model of price adjustment we might be able to come up with.

And those reference prices are just another social reality, made of the same sort of stuff as e.g. the 49th parallel as the US/Canada border. It's not a real physical fact; it's something we made up, and agreed on. Only, unlike the US/Canada border, which we really can assume fixed (I hope?), those reference prices do adjust, occasionally. It's hard enough to model a fixed Schelling focal point. But modelling a Schelling focal point that sometimes moves is going to be very hard.

Very good post. I never thought the problem on this perspective. The asset markets are much more volatile than the goods markets. Can we explain this without any reference to institutional rigidity? If yes, must we think in slowing the asset market volatility?
I´m thinking in Cooper´s "The Origin of the Crises" were he says that the monetary policy must counteract to excessive movement in asset prices, whatever the effect on the real economy. Hummm! I´m not sure.

Thanks Luis! (And I'm especially flattered by your quote from me on your blog - I sound so much better in Spanish!)

I haven't read Cooper. Russ(?) Cooper?

There is an argument that the central bank ought to tighten monetary policy if asset prices rise *above their fundamental values* (i.e. to deflate a bubble), even if this causes inflation to go below target, and even if it causes a recession. I'm not really in favour of that view. It's a bit like tightening monetary policy because there's inflation in Vancouver, even if it means putting the country as a whole into deflationary recession. Monetary policy can't do everything.

My view is that monetary policy should try to ensure that those prices that won't change quickly (the "sticky" ones), don't *want* to change. So trying to keep the CPI, or core CPI, growing slowly and steadily seems the best policy. But it would be easier to implement if we could actually observe directly where those sticky prices *want* to go.

My view is that many of the prices measured by CPI are stickier to downward moves than upward ones. I think (this is all just theory) that competition drives prices down more slowly than exogenous factors like monetary inflation or commodity price shocks are able to force them up.

If I have a downward move in some commodity that is important to my business, my first instinct is going to be to leave everything as it was. My customers are used to the price they are paying, and I'm probably unsure how long that downward trend will last. In the meantime, I can profit off the difference. All my competitors are very likely to do the same at first. Only after the trend becomes fairly predictable will we start battling each other over price. On the other hand, if there is a sudden uptick in the price of that commodity, I'll feel fairly safe raising prices to compensate. I know it is going to affect all of my competitors similarly, and the odds are that I (and they) need to protect their margins.

By the view, they work like a kind of geared ratchet, where it can move in either direction but the gear ratios are different.

Hi Nick.

While I don’t disagree that some prices (such as those covered by the CPI) may be sticky, I wonder if there might be an alternative explanation for the recent combination of CPI stability and asset price volatility, specifically, that: a) investors do not have confidence in the recovery, b) they believe that the rise in asset prices since March 2009 has had a significant monetary component, c) they are thus especially alert to events that may cause a spike in money demand, d) in order to preserve their capital, they would ideally immediately respond to any such event by increasing their own demand for money, and e) when such an event occurs, they satisfy that increase in demand for money entirely by selling assets.

Regardless of recent circumstances, I suppose that larger short-term swings in demand between money and financial assets are more likely than those between money and constituents of the CPI, simply because money demanded for portfolio purposes can be a good short- and medium-term substitute for financial assets, whereas the same cannot be said for everything covered by the CPI. So perhaps some of the observed “stickiness” could in fact be stability on the demand side rather than impediments to price changes on the supply side.

As an aside, if one believed that prices covered by the CPI were sticky, would one have to conclude that movements in the CPI were a poor indicator of excess money supply and that asset prices or quantity changes were potentially better indicators? And, if one believes that the purpose of monetary policy is ultimately to maintain some sort of desired relationship between the supply of and demand for money, would one also not have to conclude that movements in the CPI were a poor guide to/target for monetary policy?

Nick Rowe: I actually meant the second part about whether there is something fundamentally wrong with monetary models so that they must be either fixed or abandoned. Thoma seems to talk mostly about New Keynesian and I don't know enough about monetary theory to say how different New Keynesian theory is from you and Sumner talk about (not sure what the best descriptive name is: Modern Monetary Theory? money-demand based models?). Thoma seems to think that current monetary theories miss something fundamental about macroeconomic fluctuations and you seem to think that current monetary theories capture the fundamental nature of recessions.

As asset prices discount future returns, you would expect them to be more volatile than CPI. In general, the volatility would increase with the duration, so if the duration is zero, that is by definition the asset with the least price volatility.

jsalvati, when economists can correctly describe the difference between price inflating with currency and price inflating with currency denominated debt along with correctly describing what a liquidity trap actually is, you might (and I emphasize might) get somewhere.

Cuadernodearenacom.blogspot.com said: "I´m thinking in Cooper´s "The Origin of the Crises" were he says that the monetary policy must counteract to excessive movement in asset prices, whatever the effect on the real economy. Hummm! I´m not sure."

I prefer this "The origins of the economic crisis"


"In the past, the dilemma of capitalism was that the firms had to keep real wages growing in line with productivity to ensure that the consumptions goods produced were sold. But in the recent period, capital has found a new way to accomplish this which allowed them to suppress real wages growth and pocket increasing shares of the national income produced as profits. Along the way, this munificence also manifested as the ridiculous executive pay deals that we have read about constantly over the last decade or so.

The trick was found in the rise of “financial engineering” which pushed ever increasing debt onto the household sector. The capitalists found that they could sustain purchasing power and receive a bonus along the way in the form of interest payments. This seemed to be a much better strategy than paying higher real wages. The household sector, already squeezed for liquidity by the move to build increasing federal surpluses were enticed by the lower interest rates and the vehement marketing strategies of the financial engineers. The financial planning industry fell prey to the urgency of capital to push as much debt as possible to as many people as possible to ensure the “profit gap” grew and the output was sold. And greed got the better of the industry as they sought to broaden the debt base. Riskier loans were created and eventually the relationship between capacity to pay and the size of the loan was stretched beyond any reasonable limit. This is the origins of the sub-prime crisis."

That brings me back to a question from earlier that no one seemed to answer.

If low interest rates keep producing asset bubbles and higher interest rates produce price deflation, what is the problem?

Nick said: "There is an argument that the central bank ought to tighten monetary policy if asset prices rise *above their fundamental values* (i.e. to deflate a bubble), even if this causes inflation to go below target, and even if it causes a recession. I'm not really in favour of that view. It's a bit like tightening monetary policy because there's inflation in Vancouver, even if it means putting the country as a whole into deflationary recession. Monetary policy can't do everything."

Really?!?!? It seems to me that bernanke, greenspan, and every other worthless central banker think there is not a problem in the world that can't be solved with more currency denominated debt from low interest rates as long as wages don't rise and head for the chinese level.

Nick, glad to read you, I enjoy it a lot. Your entries on the euro and the ECB have been very clever to me.
I agree with your skepticism about the MP and asset prices.
Spain is in a complicated situation. Thanks to the euro, we have an enormous private and public external debt. Thanks to the euro, we have the dubious privilege of unnemployment rate of 20%, the highest in the Western world.
But the hard financial links between our countries (because the euro) make very risky to break it.

"Thanks to the euro, we have an enormous private and public external debt"

The euro forced you to borrow a lot of money? The fact that being in the eurozone lowered your borrowing costs is a benefit to you. Nobody forced you to squander that benefit on bad investments and outright waste.


If the CPI is an asset price, it is of extremely short "duration". That is, the price of the asset does not incorporate information about future CPI. The predictability of near-term CPI simply means its near-term volatility (what you call "stickiness") is low, and the best indicator of this short-term volatility is the cost of an option on a three month T-bill.

In contrast, other assets are long duration. They incorporate all known information about the future path of the CPI. Gold is ultimately the longest duration asset, and since its supply is more or less fixed, one could hypothesize that it is the asset most sensitive to the future value of 1/CPI. Gold, and other commodities, have been signaling since about 2002 that the Global War on Deflation will ultimately produce a less sticky CPI. By ultimately, I mean within a time period and magnitude sufficiently large and small (respectively) to influence the present value calculation. This is why I cringe when people explain that gold is not a "hedge" against inflation as it sometimes rises when inflation is falling: they miss the fact that gold is discounting all future inflation.

So, the fact that the assets you mention fluctuate more than the CPI is mostly an indication of their duration, and not the "stickiness" of the CPI.

Two points.

1. Most consumer goods do not have secondary markets like assets do.

2. The stability of the money markets is determined by the interest rate not reflected in the CPI which determines the purchasing power of money (facility).What is truly the cost of money?

ThomasL; Yes, that idea has been "in the air" in economics for some time, though it's not universally accepted: that prices rise more quickly than they fall. I think there's something to that idea. But I don't find your explanation fully convincing. If I re-write your argument the other way round (replace "rise" with "fall" etc.) it seems to work just as well.

David: If there were exactly the right covariances in all the shocks to the demands and supplies of assets and goods, then the equilibrium CPI would stay constant despite those shocks. But that would seem to be a very big coincidence. (Over the longer term, like 2 years or more, an inflation-targeting central bank, like the Bank of Canada, does indeed try to adjust the money supply so that that exact coincidence does arise, but this takes some skill, plus luck, to get it right). And over the last few days, I would find it very unlikely that this exact correlation would happen. When you see all the exchange rates change so much so quickly, it would be a real fluke if equilibrium CPIs didn't want to change too.

I find your second point really interesting. We want to target something we don't observe, namely the equilibrium level of the prices that are sticky (i.e. where they would be if they weren't sticky). So we are forced to target a second-best. It is indeed possible that some flexible asset price might be the second best. It could never be ideal, because there would always be real shocks that would change the equilibrium price of that asset relative to the CPI. But, it might still be better than the actual CPI, which is sticky, and hence adjusts only slowly to the equilibrium CPI, which is the thing you want to target. Central banks finesse this issue by targeting the forecast of the CPI. But even if their forecast is exactly right, the forecast CPI might not be an ideal proxy for the current equilibrium CPI.

I'm going to mull this one over.

jsalvati: "Modern Monetary Theory" means something quite different. I share some views with MMT, but differ on others. In some ways I'm a New Keynesian, like Mark Thoma, in others more monetarist. Both Scott and I are rather idiosyncratic. I vaguely remember writing a "What's wrong with macro?" post a year or two back. I think I said the main thing is that it needs to integrate finance into macro. Basically the same as what others are saying. But I'm not sure how useful it is to talk about what needs doing in general terms. The sort of model you need depends on the question you're addressing. Plus, it's better just to try to do things, rather than make a list of what needs done. You never really know what needs done in advance of doing it. You might find that fixing one problem solves a lot of others, or creates a lot of others, as a by-product.

RSJ: "As asset prices discount future returns, you would expect them to be more volatile than CPI. In general, the volatility would increase with the duration, so if the duration is zero, that is by definition the asset with the least price volatility."

I'm going to take some time with this, since it raises a number of important issues.

1. Basically, I see your point. Other things equal, the price of a long-lived asset will be more volatile than a short-lived asset. If the price of an asset is the Net Present Value of the future stream of returns, then: the longer asset has more years of returns that might vary; plus, mathematically, a change in the interest rate will have a bigger effect on the NPV of a longer asset.


2. Other things might not be equal. Fresh strawberries might fluctuate in price more than houses.

3. This is the main point you are missing. *All prices are the price of one thing in terms of another thing*.

We have to look at both things when asking what determines that price. The price of strawberries in terms of peaches depends (supply side) on the strawberry harvest and the peach harvest, and (demand side) the pleasure people get from eating strawberries and from eating peaches. That price depends on the *relative* supply of strawberries vs peaches, and on the *relative* demand for strawberries vs peaches.

Now, the dollar price of strawberries is the price of strawberries in terms of money. In partial equilibrium analysis of the price of strawberries we talk about the demand and supply of strawberries, but ignore the demand and supply of money. Strictly, we can't do this. But it's an OK simplification for partial equilibrium analysis. But when we talk about the CPI, that simplification won't work at all. We are doing general equilibrium analysis. We can't ignore the demand and supply of money. The equilibrium CPI is as much determined by the demand and supply of money as it is by the demand and supply of the goods in the CPI. In fact, those are sort of mirrors of each other.

If you start in equilibrium, and permanently increase the supply of money, other things equal, it is the CPI that must eventually adjust to restore equilibrium. So in that sense it is correct to think of the CPI (or 1/CPI) as the price of money. And money is a durable asset. If the value of a durable asset is volatile, then the equilibrium CPI should be volatile too. *Even if every good in the CPI were a non-durable good*.

Too much Fed: Please stay on topic.
"Really?!?!? It seems to me that bernanke, greenspan, and every other worthless central banker think there is not a problem in the world that can't be solved with more currency denominated debt from low interest rates as long as wages don't rise and head for the chinese level."

That statement is totally false.

Luis and Adam: That is one of the things that has puzzled me about the Eurozone. When someone starts to borrow too much for future solvency, you would expect to see *both* the borrower and the lender (or at least one of them) start to progressively tighten up, in some way. This didn't seem to happen in the Eurozone. Why? Some people (Simon Johnson, for example) I have read say this is a structural feature of the Eurozone. But I have never understood their explanation. I just don't get it.

Money has a store and a purchasing power value. The first depends on the time value(interest rate), the second depends on the inverse of the price level. Goods with duration have secondary markets while goods without duration have only primary markets whose price given asymmetry is set by the seller. Thus the price level is relatively fixed. On the other hand the time value effect of storage fluctuates in secondary markets subject to duration. These are the hints in my previous comment.


"That is one of the things that has puzzled me about the Eurozone. When someone starts to borrow too much for future solvency, you would expect to see *both* the borrower and the lender (or at least one of them) start to progressively tighten up, in some way."

I'll forward another un-provable, tin-foil hat theory, which I will name: Denial and the Indomitable Belief in Deus ex Machina.

DIBDEM (for short) is where seeds of doubt in a system begin to take hold, but only so far as realizing that if one even allows their mind to go any further there will be no stopping.

In context, if one even allows themself to question the solvency of any one of these agents, they may just realize that the debt level and [most importantly] trajectory of every major nation on every side of the Atlantic, Pacific, and in between is unrealistic. That is, that the entirety of modern* sovereign debt finance is a con, and that we have bought in far too far in to escape with a cushy landing.

People that recognize neither half are in pure denial. People that recognize the first half, but not the second, are the believers in deus ex machina. It will all be fixed "somehow." "No one acts till the point of crisis, but then we'll act decisively," etc...

The idea behind DIBDEM is that it might be _cheaper_ (both emotionally and in dollar terms) to keep feeding the system despite any doubt, because the doubts are too terrible to contemplate. In fact, "cheaper" may not even have much meaning if the doubts prove right, since, what is anything one does now going to matter if they come true. If the doubts are wrong, then the debt will be good. If they are right, then what will any of it matter. So may as well play along one more day...

*Post-war, but particularly post-Bretton Woods.

Why the euro has boosted the credit in the peripheral countries? I do not know exactly, but the fact is that foreign credit in Spain has risen astronomically, while the risk premium differential between countries disappeared. I asked to a friend in an investment bank, I was told that the markets expected the ECB would never allow a default in one member country. Obviously, there was a housing bubble (as in US), but I think that without the euro, the spreads between countries had been higher. To this it must be added that for years, the euro interest rate was appropriate to the weakness of Germany, but very low for Spain and other countries (lower than CPI inflation).
Now, the situation is complicated: Germany does not want the ECB to give more credit facilities, and so the ECB is sterilizing purchases of Spanish and other countriesdebt(as accorded by euro Governments), but I think the risk is clearly deflation, at least in these countries. In any case, if the risk is divergent (deflation in the south and inflation in the north) the euro is not an optimum currency area.
Does it make sense to buy debt and then sterilized injected funds? the first alters the relative price of debts, the latter threatens the banking, given the signs of illiquidity. Good luck with the puzzle! Perssonaly. Good luck with the puzzle.
Personally I think Trichet is outweighed by the problema. Nick, remember what you said about a Central Bank with 16 Minister of Finance...

There are several things going on with inflation:

0. Stickyness. Inconvenient or not, many retail prices are inevitably sticky because they incur significant costs to change. Also, there is artificial stickyness introduced when inflation data is compiled: some judgment is always required, and in practice this is notoriously biased toward preserving the trend.

1. Genuine low volatility. So much of the input into a typical CPI basket is labor, and labor prices are not volatiles. (And yes, that implies positive feedback between CPI and labor.)

2. Wide bid-asks. It is normal for such prices to exhibit less measurable volatility than narrowly priced assets. Here you are comparing the prices for (baskets of) items with 300-400% markups against FX with maybe a 5bp range.

3. Selection bias. Not withstanding the above, you have over-generalized. Had you chosen a different period, or the price of a different money, you could have found more volatility.

4. Wrong variable. Many (most?) asset prices exhibit higher short- than long-term volatility. Your concern with CPI as an anomaly to this, and the wording "wanting to change", suggests you are thinking in terms of a situation like a currency peg, where the interfering hand of government dams pressure until it becomes irresistible, causing catastrophic change. There may be something to that, but in fact central banks target the *rate* of change rather than the price level (as your buddy Scott Sumner likes to complain.) Inflation rates in fact exhibit both significant volatility (even after seasonal adjustment) and appreciable mean-reversion. (And yes, I understand that the price of money is the variable you care about.)

=> Why the euro has boosted the credit in the peripheral countries?

Did your investment banker friend mention that banks could post peripheral sovereign bonds as collateral at the ECB? And that the haircuts set by the ECB did not fully reflect differences in collateral value?

In effect, the ECB was giving away a cheapest-to-deliver option to European banks; to exercise, they had to lend money to the periphery. This free money is at the root of the large exposures owned by French and German banks.

Nick Rowe,
In my previous comment we can generalize for all products to have a store value and a purchasing power value. The price for this good is equal to its purchasing power (price/price level) plus the yield service discounted by the time value plus the price change as reflected by the duration effect induced by the time value change.If it has no yield and duration is zero, it is a typical immediately consumed product without any secondary market to trade. Again, the purchasing power value is set by the seller given asymmetry of advantages between buyers/investors and sellers/issuers and is relatively fixed. A demonstration of the asymmetry issue I posted it in the Game Theory post of your associate this weekend.

Philip, yes, sure, you're right. The hair cuts were decided by the calification of S&P, Moody´s and Fitch. this criterium has been suspended to face the crisis. But the origin of the problem has been PRIVATE indebtedness. The public debt has been a consequence of the crisis (as say Rogoff & Reinhard). At the end the asimetry between the countries has increased.


Yes, I was only making the point that you would expect asset prices to be more volatile than CPI. I was not addressing the causes of CPI volatility -- I have no clear model to explain the fluctuations cited in the voxeu paper, but the "inertial" price theory seems to correspond well to the notion of markup pricing, which makes sense to me.

"If you start in equilibrium, and permanently increase the supply of money, other things equal, it is the CPI that must eventually adjust to restore equilibrium. "

I find these types of statements frustrating. What do you mean by "money" -- M0, M1, M2, MZM? I think MZM is probably what you mean -- but is it? Japan has certainly increased the monetary base, and yet household deposits have not increased, and prices have continued to fall. They have been playing this game for 2 decades now, so hopefully someone has noticed that you can increase the supply of money massively without rising prices. It all depends on what you are increasing and how. As long as salaries remain low, the mere fact that there are fewer bonds and more currency can't cause the price of strawberries to go up.

Moreover, what do you mean by "permanently increase the supply"? The supply of money is partially endogenous and partially set by policy -- in fact lending standards can be more important to influencing the "supply" of money than changes in the fed funds -- and FFR is just a change in the marginal *cost* of money, and not need correspond to a change in the quantity of deposits.

It may or may not. If I can earn a return of 3%, then setting my marginal costs of borrowing to be 2% will make me take out the loan, increasing MZM. On the other hand, if I can only earn a return of 1%, then 2% is still too high. So when trying to pass from the marginal cost of money to an increase in the amount of money, you need to pass through the channel of borrower expectations, which are volatile. The borrower's own expectations may stymie your attempts to increase or decrease MZM -- assuming MZM is even what you are talking about. On the other hand, if I am a business, and you offer to buy my strawberries, then this would not only increase the supply of money, but it would also increase my return expectations -- if the commitment to buy strawberries was credible. So there is a big difference between buying strawberries and buying bonds, i.e. between subtracting one liquid asset and replacing it with another, and subtracting one good and replacing it with money.

As you point out, it's about relative prices. If the prices of bonds are rising, that means people want to hold more bonds as well as more money. If, simultaneous to that, the prices of goods are falling, then doesn't that tell you that people want fewer goods and more money? In that case, buying a bond with money is not going to increase the stock of what people want. It would be more effective to buy what people don't want to buy (goods) and replace that with what people want (bonds or money). That would be more effective at price stabilization, IMO. But perhaps that is ruled out from consideration because the economic models rule out the possibility of people not wanting to buy real goods? Who knows, but it shouldn't be ruled out.

Anyways, I don't mean to rant, but I think these conversations would be more productive if we all:

1. specified what we meant by "money" -- MZM, M1, all government liabilities, etc?
2. specified which operations we are going to use to "add money": replace bonds with money, or replace goods with money, or replace labor with money -- as the effects will be different.
3. Instead of deciding a priori what people want to hold, we would look at which prices are falling and which are rising to determine what people want to hold.

Anyways, good discussion.

Food for thought.

There are many studies that show empirically that changes of the federal funds rate or equivalent in other monetary systems alter the risk taking activity of banks inversely altering credit lines given borrower demand. Similarly for bank regulation which has a positive effect upon risk taking.

"But the origin of the problem has been PRIVATE indebtedness. The public debt has been a consequence of the crisis (as say Rogoff & Reinhard)."

I agree with both these statements. Nevertheless, in my opinion easier terms for public borrowing translated into easier terms for private borrowing long before Spain, say, encountered any fiscal problems. I see three reasons for this: direct subsidies by governments using borrowed money; lowering cost of private borrowing since spreads over local sovereigns did not widen as the cost of sovereign borrowing fell; displacement as improved ability for government to borrow abroad freed local money for private investment.

Phil, right in your three reasons. Greece is a country were ther is no private sector; Spain is not the same, but the level of intervention of the central and regional governments is quite high.
Yes, all that is true. But I suppose that all these reason would have had to rise the risk premia between countries. I suppose the Euro has prevent this.

Luis H Arroyo,

What do you mean that Greece has no private sector? What do you know of Greece?

Phil, what do you mean by "improved ability for government to borrow abroad freed local money for private investment."?

Where was this money previously trapped, and why couldn't it get out?

On a completely different note, "inertia" means that it takes energy to change your moment. In the case of prices, it certainly makes sense that there are costs to changing prices, and this can be put under the rubric of menu costs, provided that the definition of menu is expanded.

Whenever a project is undertaken, the decision makers need to know the future costs and revenues up front in order to make a decision. If there is a lot of uncertainty as to future prices, then they will demand a premium, and as a result, a producer that can provide stable prices will be at an advantage over a producer that cannot -- stable prices have a value.

For the same reason, if households had more uncertainty as to their incomes, they would spend less and save more, and they would make fewer long term commitments. So stable wages have value. Those suppliers who change prices more often than their competitors will lose customers, just as those employers who do not provide a stable wage will need to pay higher wages in order to attract the same people.

And in that case, you can imagine a situation in which the economy will organize itself in such a way as to create sticky prices, by means of long term contracts, hedging, and electing to increase or decrease output rather than prices whenever possible.

RSJ: In the 1970's there was an attempt to explain sticky wages on very similar lines to what you describe. Azariades, Bailey, and Gordon, did models in which the worker was risk-averse, the firm risk neutral, and so sticky wages were part of an insurance contract that could be mutually beneficial to form and worker. It looked very promising initially, but on closer inspection had a fatal flaw. I doubt anybody remembers this stuff now. The flaw was that if you had contracts specifying future wages, you would also want contracts specifying future employment as well. And as a first order approximation (except for the wealth effects of the insurance payouts on labour supply) the employment would be exactly the same as if wages were perfectly flexible. As Barro noted, what this means is that wages and prices might *appear* to be totally sticky, while in reality employment and sales might be determined *as if* wages and prices were perfectly flexible -- at the efficient levels of employment and output.

In short, it was yet one more dead-end in the search for the theoretical underpinnings of the assumption of sticky prices.

On another note: there appears to be "intertia" in the price *level*, but also intertia in the *rate of inflation*. The standard Calvo model "explains" price level intertia, but not inflation inertia. That's one of the biggest empirical flaws in the standard New Keynesian models.

My own guess, by the way, is that it was at precisely this point that Barro abandoned Keynesian macro (where he had done some really good work) and became a Classical RBC theorist.

"The flaw was that if you had contracts specifying future wages, you would also want contracts specifying future employment as well. And as a first order approximation (except for the wealth effects of the insurance payouts on labour supply) the employment would be exactly the same as if wages were perfectly flexible."

That's interesting, Nick -- are you saying that prices would be flexible, or hours worked? Or is there an implicit assumption of labor market clearing (e.g. flexible prices) that allows you to equate one with the other? Do you have a reference?

In related news, Steve Keen reminds us that:

* 89-95% of firms report constant or falling marginal costs
* 85% of goods and services in the non-farm sector are sold to "regular customers" with whom the firm has a relationship, and 70% of gross sales is business-to-business, not business-to-customer.
* "one-quarter of output is sold under contracts that fix nominal prices for a nontrivial period of time. And it appears that discounts from contract prices are rare. Roughly another 60 percent of output is covered by Okun-style implicit contracts which slow down price adjustments."

How to reconcile a situation in which each individual firm faces a declining total cost curve, yet the economy as a whole can clearly overheat?

RSJ: "How to reconcile a situation in which each individual firm faces a declining total cost curve, yet the economy as a whole can clearly overheat?"

You mean "declining *marginal* cost curve".

That one I can answer. As soon as you start thinking of macroeconomics with imperfect competition (which was the one great key feature of the New Keynesian revolution), it becomes a lot clearer.

Let me answer in math first. Define the individual firm's total cost function as C(y,Y), where y is the individual firm's output, and Y is average output at all the other firms. Marginal cost is Cy (dC/dy). The slope of the individual firm's marginal cost curve is Cyy. This tells us whether an individual firm "overheats" (wants to raise prices) if the individual firm expands output, holding other firms' output constant. But if we are looking at the macroeconomic level, at what happens to an individual firm's MC when *all* firms expand output, we need to look at (Cyy + CyY). That second term is the upward shift in the individual firm's MC curve when other firms expand output.

Put it another way: for partial equilibrium micro, we want to look at the partial derivative of the marginal cost function; but for macro, we need to look at the total derivative of the marginal cost function. The individual firm can have a flat or even downward-sloping MC curve, but the economy as a whole (assume symmetry, so y=Y) can have an upward-sloping MC curve.

In words, an individual firm can expand output alone and have a negligible affect on employment, wages (and other input prices). But when all firms together expand output, employment increases, unemployment falls, and (depending on your model of the labour market) wages get pushed up either because you are moving up along an upward-sloping aggregate labour supply curve, or workers' bargaining power is stronger, or it's harder for firms to find suitable workers so try to bribe them away from other firms, or whatever.

Here's my old post

(and the previous post linked in that post).

RSJ: Here's a (pdf) link to the Barro paper: http://www.google.ca/url?sa=t&source=web&ct=res&cd=4&ved=0CCUQFjAD&url=http%3A%2F%2Fisites.harvard.edu%2Ffs%2Fdocs%2Ficb.topic500592.files%2Fbarro%2520long-term%2520contracting.pdf&rct=j&q=Barro+Azariadis+bailey+Gordon&ei=vs0DTKKnIIP78Aa057HEDQ&usg=AFQjCNHqW79qpyV9_BNRjDqDfvIanL2AIw

Intuitively, ignoring externalities, the intersection of the labour demand and supply curves determines the equilibrium wage and employment, and also the efficient level of employment. {W*, N*}.

Suppose, that for whatever reason (like risk aversion), firm and worker do a forward contract, that specifies W and N in advance, at {W',N'}.

First off, the firm and worker would agree on a contract that makes N' state-contingent, so that N' in state s is where the firm's marginal benefit equals the worker's marginal cost of labour. Because this maximises the total surplus to the firm+worker.

Second, even if they didn't make N' state-contingent, the firm and worker would still have the incentive to make side-deals (at a different W) on the spot market after the state s has been revealed, if MB(N') is not equal to MC(N'). And these side deals would ensure that N=N*.

In other words, the standard Keynesian analysis assumes there are unexploited gains from trade. $100 bills on the floor that are not picked up.

So, sticky wages may be just a facade, says Barro.

Nick @10:50

Yes, clearly when all firms increase output, the aggregate marginal cost behaves differently that the micro MC curve. I think aggregation issues are at the heart of all of this.

But you haven't provided an explanation, but an assertion that the aggregate behavior is different, by introducing, ex-nihilo the aggregate functions C(y,Y) -- there is no micro foundation or reason why this emergent phenomena, in fact, emerges.

Now, at first, it seems obvious -- you have fixed resources that are in greater demand (e.g. labor, oil), etc. But on the other hand this seems to contradict the other assumptions of substitutability of one factor with another. And if firms can produce output more cheaply on a per-unit basis, then who wins out? The shortage of the exogenously set factors or the increasing returns of the endogenous factors? It's not clear, a priori, which would win out.

In the post you cite, that assertion is buried in the introduction of the aggregate production function (at the end of the post), which has diminishing returns. If you picked a "representative" production function, then it would have increasing returns. In either case, you are not deriving the form of the aggregate production function from observations about actual firm characteristics combined with rational behavior, but are imposing it from above -- there are no micro-foundations here.

And the Barro paper has a similar flaw. All of the mathematics is used to verify that the introduction of the contracts can lead to a situation in which labor demand differs from labor supplied. The math ends there! Then, in section 3, philosophy takes over:

Barro *assumes* that forward looking agents would not enter into a situation in which there are, in aggregate, dead-weight losses, so they would form side contracts to ensure that full employment was reached. There are no micro-foundations that explain why this would happen -- other than an assertion that it must happen:

"Consider the situation from the ex ante position where the contract is set up
between a firm and a worker before the value of the monetary shock, v, is
realized. The two parties to the contract know that that v-shock will sometimes
be positive and sometimes negative. If they agree on a fixed value of w
and use an employment rule of the type proposed by Gray and Fischer, they will be
adopting a contract that imposes dead-weight losses for both positive and
negative monetary disturbances.

In the absence of any ‘transaction costs’ that
would, for example, inhibit the possible contractual arrangements from making
side payments between firms and workers, it is *apparent* that the employment
rule would be selected in order to maximize the total pie possessed by the two
parties. "

In other words, he uses a model to show that the forward contracts can cause a deviation from full employment, then
determines, without micro-foundations, that this would result in dead-weight losses, and finally rules
this out on grounds that the results are not pareto-optimal, when the whole point of the original observation is that it can cause these types of losses to occur.

And from this he concludes that rigid prices, rather than full employment, is a "facade":

"The principal contribution of the contracting
approach to short-run macro-analysis may turn out to be its implication that
some frequently discussed aspects of labor markets are a facade with respect to
employment fluctuations. Tn this category one can list sticky wages, layoffs versus
quits, and the failure of real wages to move countercyclically."

But he doesn't provide any micro-model that justifies *why* each individual employee would enter into side contracts that nullify the insurance value of the wage contract, nor does he supply a micro-foundation as to why the dead-weight losses would occur -- the source of the dead-weight loss is another ex-nihilo log linear production function, which itself has no micro-foundations.


When physicists deal with complex systems, they do this with statistical mechanics, not by looking at a "representative molecule", and it's understood that interacting systems, each subject to certain forces, have emergent behavior that is not characteristic of the micro-system. In general, most of the degrees of freedom vanish, and you are left with a set of conjugate variables that control the overall system behavior: pressure, volume, temperature, etc. These macro variables are different from the micro variables, and are often inapplicable at the micro level (e.g. a single molecule does not have pressure, temperature, or volume).

The macro-variables are *derived* as legendre transforms of micro-variables over the statistical distribution of the entire system, and the macro potential function is *derived* from the micro-behavior as well. What would make me happy is if the aggregate production function or similar was derived from plausible micro-foundations, and then you would accept whatever conclusions that reached. This is better than assuming that aggregated behavior can never lead to any dead-weight losses, or that aggregate behavior must have diminishing returns, and then ruling out unemployment or other dead-weight losses on teleological grounds.

Or at least, if you do that, then you should not pretend that your model has micro-foundations.


Thanks for the Barro reference. And I agree with the basic conclusions of the previous post -- e.g. there *must* be some C(y,Y) and MR(y,Y) functions that behave roughly as you describe. But I don't think that this model has micro-foundations, and as a result, I think there may be other hidden effects that are not yet understood. The macro-model is, in the language of physics, a phenomenological model, but I think its plausible.

RSJ: On the C(y,Y) function. If you simplify enough (one factor, representative agent/worker, etc.) it's easy enough to provide *a* microfoundation. But doing so is not very general. I prefer to leave it open as to what the labour market looks like. Just wanted to make the general point that Cyy is not the same as Cyy+CyY.

On Barro: there are two ways to interpret "Why should rational agents leave dollar bills on the table?" Either as a rhetorical question, or as a genuine question, that puts the onus of explanation back on whoever says there is inefficient unemployment.


" If you simplify enough (one factor, representative agent/worker, etc.) it's easy enough to provide *a* microfoundation."

The source of the micro-foundation should be the observed production characteristics of actual firms. If your micro-foundation does not share those characteristics, then the onus is back on you to produce a rationale as to why this doesn't violate the legitimacy of the model's outcome.

But you are arguing that the onus is on the person arguing that we are not in a panglossian world. The onus should be on the person making irrational assumptions when formulating their model.

In a single firm model, the "representative" firm would have increasing returns to scale and falling marginal costs. That it doesn't or can't, should be interpreted as a ruling out of the one-firm approach, just as the inability of a single molecule to exhibit "temperature" should be used to rule out the "representative molecule" approach to modeling ideal gasses.

We should have moved beyond this approach a long time ago, rather that suffering the Friedman argument, in which, on the one hand, micro-foundations are critical while also arguing that the accuracy of the actual micro-assumptions are irrelevant. Don't you agree this is a bit silly?

"There are two ways to interpret "Why should rational agents leave dollar bills on the table?" ..puts the onus of explanation back on whoever says there is inefficient unemployment."

No, my point was a bit more subtle than that.

Disputation 1: Is there really a dead-weight loss over the long run?

By assuming a log linear production function that is just a function of labor, Barro is assuming that output is independent of the price level or price volatility. Only with that assumption can he argue that dollar bills are left on the table.

But this is begging the question. I.e. if Intel spends 3 Billion dollars to build a chip fab, and if prices for chips fall or labor costs increase so that after the fab is built, it becomes unprofitable, then someone in the economy just lost 3 billion dollars, and wasted an enormous amount of real goods. In that case, fewer capital investments will occur, and the economy will be less productive as price uncertainty increases. There is evidence that price volatility slows output.

So you need to weigh the long run dead-weight loss of price rigidities with the long run dead weight loss of flexible prices. I doubt any industrial economy would be able to survive flexible prices.

Disputation 2: Use of representative firms assumes away externalities

A dead-weight loss for the economy is not necessarily a dead-weight loss for any individual worker or firm. Only when you assume that there is a single worker and a single firm do you internalize all of those externalities and observed behavior becomes irrational. Indeed, in the U.S. there is a ton of idle labor, cars sitting in parking lots that cannot be sold -- it *looks* irrational!

Nevertheless each agent *is* acting rationally, because the dealer has signed contracts that prevent him from selling the cars at firesale prices, and these contracts are in his long run best interest. And it is in the interests of firms to not have their long term auto prices lowered if they believe the economy will recover, etc. But the aggregate behavior is that for a period of time, there are people without cars that want to buy, and cars stockpiled in the port of long beach with nowhere to go.

And this is exactly what the Keynesians are arguing does happen, via the sticky price mechanisms as well as other mechanisms. And in that case, you cannot "refute" this argument as being irrational by assuming that there is only a single worker and firm -- that the person buying the car is also the owner of the car company.

In statistical mechanics, there is a 'representative molecule'. It's just the molecule of the gas in question. Boltzmann figured out there is a distribution of molecular speeds (or momenta, if you prefer) and the rest sorta falls out from there. Not sure that there is an analogous distribution for people. At least I hope there isn't.


the point is that the dynamics of the ensemble are not representative of the dynamics of each constituent part. A "molecule" has no pressure, temperature, or volume. The variables that determines the aggregate behavior doesn't even appear as a variable at the level of each individual optimizing agent.

Now in some sense this is true of economics as well -- the aggregate production function is nothing like actual observed production functions, which have declining marginal costs. That would be fine except that economics claims to have "micro-foundations". It does not -- it just makes up the aggregate dynamics using a hodge-podge of fallacies of composition combined with simple intuition that may or may not apply. I.e. the aggregate variables are imposed from above, not determined from looking at the micro behavior.

True micro-foundations would take plausible models of large numbers of actual households and actual firms interacting with each other, would capture the emergent behavior of such an ensemble, and would encode *that* in its dynamics. Now, on the one hand, you can say that this is too hard -- the math is too complicated, but I don't think it is. In general, as the number of actors increases, the math becomes simpler, not more complex. Degrees of freedom start to vanish and new degrees of freedom emerge. All I am saying is that we need a rigorous mechanism to determine what those degrees of freedom actually are, rather than postulating what they are.

Let me give a simple example:

When a business sells a 10 year bond into the credit market, the household buying the bond is not actually foregoing consumption for 10 years. The household may not be foregoing consumption at all -- it may short some other bond and use the funds to buy the first bond, or it may only hold the bond for 6 months, and not for 10 years.

As soon as assets can be resold, then the aggregate dynamics is different from the simple model of me lending you money for 10 years. When the IOU becomes liquid, I no longer give up any consumption opportunities by holding your bond, because the bond is a store of value that can be unlocked at any time. All you can say for certain is that I did not consume on the day I bought the bond. But in that case is the opportunity cost equal to 1 day of missed consumption (e.g. basically zero) or of 10 years of missed consumption -- all of a sudden, you don't know and the standard euler equation is not applicable -- that is an example of an emergent effect. But if you only have a single household, then it makes no sense to re-sell assets, and you lose all of the dynamics of households interacting with each other.

There are many such emergent effects when you look at a large ensemble of actors interacting within institutional frameworks such as credit markets and banks -- hopefully none of this is controversial.

I don't think the math is too complex or difficult -- not more complex the current approach; I hope that we will make more progress in understanding things like depressions and asset bubbles if we truly did have micro-foundations and a rigorous basis for legitimately determining the aggregate dynamics.

RSJ: I'm not an economist, but what you say makes sense. It seems to me the agent based approaches hold some promise:


The comments to this entry are closed.

Search this site

  • Google

Blog powered by Typepad