The peanut theory of recessions

I've got three targets in this post: economists who say that recessions are caused by real wages being too high; economists who say that recessions are caused by real interest rates being too high; and economists who say that recessions can't be caused by an excess demand for money, because we can always go to the bond market/ATM and get some more. That's a lot of economists. Most economists, in fact.

Assume a monetary exchange economy. You can't buy or sell anything except by giving or accepting money in return. Every single market is a money market -- where one of the other goods is traded for money. Money is both medium of exchange and medium of account, because all prices are measured in money. People hold stocks of money. When an individual buys something his stock of money falls, and the seller's stock of money rises.

Start in full equilibrium. Every single market has demand equals supply. Now assume all prices are fixed. Every single price for every single good, including all inputs like labour, and all asset like bonds, is fixed. Now halve the stock supply of money. (Or double the stock demand, it doesn't matter). Every individual: tries (and fails) to sell more goods; tries (and succeeds) to buy less goods; because he is trying (and failing) to increase his stock of money. The result is a recession. Absolutely standard monetary disequilibrium story.

Now, let's add one tiny twist to that monetary disequilibrium story. Let's make one tiny exception to the assumption that all prices are fixed. Assume the price of peanuts is perfectly flexible. So the market for peanuts always clears. It's only a tiny twist, and a tiny exception, because by assumption peanuts are a tiny and unimportant part of the economy. If some disease destroyed all the peanuts, or some new method of peanut farming doubled the supply of peanuts, nothing much else would change. Peanuts are, well, peanuts, in the big macroeconomic scheme of things. By assumption.

What happens to the price of peanuts when the money supply halves, and all other prices stay fixed? I don't know. It depends. You might think it would halve, because the halving of the stock of money should mean that the equilibrium price level halves. But a moment's reflection is enough to say that that answer is very probably wrong. The recession had a purely nominal cause, by assumption, but the recession itself is a real event. Real income is lower, and there's excess supply in other markets, so people cannot sell stuff, like labour. That real shock would change the demand and supply for peanuts as a function of the relative price of peanuts. If peanuts were a luxury good, their demand would fall now that people are poorer. If peanuts were an inferior good, that people would only eat when they are unemployed and desperate, their demand would rise. Then there's the supply side of the peanut market as well. The fixed prices of all the inputs needed to produce peanuts may prevent the nominal price of peanuts halving. Alternatively, a flood of unemployed workers growing peanuts in their backyards to earn money, because you can always sell peanuts, might cause the price of peanuts to more than halve.

The price of peanuts will almost certainly change. My guess is that it will probably fall. But I don't know that for certain. It will depend on a lot of things. It might depend on how deep the recession is, and how long it is expected to last. Or I could even assume that all prices adjust slowly, but that the price of peanuts adjusts more quickly or more slowly than the prices of all the other goods. But it doesn't matter. Because peanuts don't matter. By assumption.

We know that the recession in this economy is caused by an excess demand for money, and that the fall (or rise) in the price of peanuts is an almost irrelevant side-effect that we can safely ignore. We know this because we built the model economy and we assumed that peanuts are almost irrelevant. But the people living in the economy do not know this. What they see is a fall (or rise) in the price of peanuts, accompanied by a recession.

Someone living in that economy, who didn't understand monetary economics, would be very tempted to say that the fall in the price of peanuts is what caused the recession. Microeconomic theorists would be very tempted to say that the recession is caused by real/relative prices being wrong. The price of peanuts is too low relative to other goods, or the price of other goods is too high relative to peanuts. That's why the peanut market clears but other goods are in excess supply. And if a monetary disequilibrium theorist suggested that the recession was caused by an excess demand for money, he would be met with the objection that people could always get more money by selling peanuts. Which is true, but irrelevant.

Now let's add a second tiny twist to the story. Assume that the central bank conducts monetary policy by buying and selling peanuts. In fact, every month it sets a new target price for peanuts, at which it will buy and sell unlimited amounts of peanuts. But further assume that the stock of central bank money is tiny relative to the stock of peanuts, so even though the bank's peanut market operations have a significant effect on the supply of money, they have only a tiny effect on the supply and demand for peanuts.

Now all the macroeconomists join the microeconomists, and the common people, in focusing on the price of peanuts. "The price of peanuts is set by the central bank, not by production and consumption of peanuts". "The price of money is the inverse of the price of peanuts". "Monetary policy is the price of peanuts, rather than the stock of money, which is endogenous and demand-determined". "There cannot be an excess demand for money, because the central bank is willing to sell an unlimited amount of money". "There is a natural equilibrium (real) price of peanuts, and if the bank sets the price of peanuts below that natural price, there will be a recession and deflation".

All these things the macroeconomists are saying are true. But they are also all irrelevant. Because, by assumption, peanuts are a tiny irrelevant part of the economy, and almost exactly the same things would happen if the central bank did not implement monetary policy in the peanut market, and if the price of peanuts were fixed like all the other goods. Plus, since the recession will affect the demand and supply of peanuts in many unknown ways, the partial equilibrium relative price of peanuts may either rise or fall in a recession, and so you can't even use the price of peanuts as an indicator of the tightness or looseness of monetary policy.

"The recession was caused by an increased demand for peanuts and a consequent rise in the natural equilibrium real price of peanuts, which the bank failed to match by increasing the actual price of peanuts sufficiently". "Perhaps the government should step in and start growing peanuts to reduce the natural price of peanuts, since monetary policy isn't working". This would be the diagnosis and one suggested cure.

Now, it is true, peanuts are a tiny and almost irrelevant part of this economy by assumption. In the real economy, bonds aren't. But so what. And central banks are a tiny part of the bond market.

Comments

[David Andolfatto]

Nick, can you point to the evidence you use to convince yourself that nominal price rigidities are quantitatively important in terms of explaining the recent recession and weak recovery dynamic we have experienced in the U.S.?

[Greg Ransom]

Hey, this is _my_ thought experiment challenge to the monetary equilibruim position:

"And if a monetary disequilibrium theorist suggested that the recession was caused by an excess demand for money, he would be met with the objection that people could always get more money by selling peanuts."

Only I throw in the fact that people are selling and buying most everything most every day -- i.e they are giving up the chance to hold more money constantly throughout every day.

I've never suggested this is an air tight counter-example, but it does deserve a better parlay than disequil theoriest typically offer up.

[reason]

Greg Ransom
"Only I throw in the fact that people are selling and buying most everything most every day -- i.e they are giving up the chance to hold more money constantly throughout every day."

No they are not. (I take it you actually mean to say that "... SOME people are selling and buying almost everything different good and service ..." - if I take your sentence literally it is clearly empirically wrong.

[reason]

David Andoldatto
I suggest you ask yourself if any particular price governs the demand for money - or expectations of general prices. If the latter - then could this adjust more slowly than actual prices?

[Bill Woolsey]

Great analogy.

My problem with the analysis is that we need to suppose that money is not only the medium of exchange and the medium of account but also a snack food.

(Like other financial assets, money can be accumulated to save. Like Peanuts, money is a snack.)

[Nick Rowe]

David:

I had all my defences trained in one direction, then you come at me from a totally unexpected different direction! Let me try to swivel my guns around.

And let me change your question into "why do you (me, Nick) believe that nominal price rigidities are quantitatively important?"

1. Because it is so very easy to forecast the CPI or nominal wages with very precise accuracy at about a 1 month or 1 year horizon compared to forecasting (say) stock prices or exchange rates or the prices of homogenous commodities sold in competitive organised markets. At much longer horizons this isn't true (unless of course the central bank is targeting inflation). Prices look sticky.

2. Because recessions are times when it gets much harder to sell stuff and easier to buy stuff, except that stuff whose prices are hard to forecast. So it looks like those prices that look sticky are causing excess supply in recessions.

3. Because if you add 1 and 2 together, you get, errr, 3, in that the short side rule predicts quantity sold will fall if the price moves above the equilibrium, so you get a recession.

4. Because I don't find any alternative theories convincing.

But, maybe 100 years from now, when we are all very wise, and understand all the myriad "frictions" that determine P and Q in real world markets we will look back on the assumption of sticky prices and Q=min{Qd,Qs} and say "well, that is not really what's going on at all, but it might be acceptable as a crude approximate high-level abstraction to represent underlying reality". (Just as a CIA constraint might be a crude approximate abstraction for all the stuff Steve Williamson talks about about explaining why people use money.)

2.

[Nick Rowe]

Greg: "Only I throw in the fact that people are selling and buying most everything most every day -- i.e they are giving up the chance to hold more money constantly throughout every day."

Assume there's only a 10% drop in Ms, hold V and P constant for simplicity, and so there's a 10% drop in Y, and 10% excess supply. 90% of the people may be selling everything they want to sell, and 10% can't sell anything. Or maybe 100% of the people are selling 90% of what they want to sell. Or maybe 100% of the people are selling 100% of what they want to sell but only 90% of the time. But that's still a very big recession.

Bill: thanks. Yes, to put it another way, peanuts are a good that gets produced and consumed. It's not the ideal analogy. Except for the metaphorical name. Peanuts are peanuts.

[Scott Sumner]

Nick, I don't think you actually critiqued the real wage transmission mechanism, as you simply assumed real wages were fixed, and then showed we could get a recession. But all real wage proponents concede that. If both wages and prices are fixed, you also get a recession. I agree.

I believe wages are stickier than prices, and are the dominant problem--a claim this post doesn't address. As a side issue I prefer to use "relative wages" not real wages. Relative wages are nominal wages over per capita NGDP. They did rise in your example, although of course that doesn't prove causation.

I'd like to take a shot at David's question:

I see three factors that always point us in the direction of wage/price stickiness when we have recessions with sharp falls in NGDP growth:

1. We have a mountain of evidence that prices, and even more so wages, are sticky. For wages, we also have a mountain of evidence that there is money illusion around 0% wage increases. Krugman has an excellent column showing that last week.

2. We have a mountain of evidence that when central banks take actions that suddenly change the path of NGDP, there are powerful real effects in the same direction. This evidence goes way back into history. Indeed the evidence is strongest in historical cases, when monetary authorities did wild and crazy things. It's harder to find in modern inflation targeting settings.

3. We have models that show how nominal shocks can cause real effects if there is wage/price stickiness, and WE HAVE NO PLAUSIBLE MODELS SHOWING HOW NOMINAL SHOCKS CAN HAVE IMPORTANT REAL EFFECTS WITHOUT WAGE/PRICE STICKINESS.

No one of these reasons are convincing by themselves. Indeed no two are entirely convincing. But all three are.

[Mike Moffatt]

RE: Sticky prices. I'd love to get your (Nick, Scott, et. al.) thoughts on the Bils-Klenow paper on sticky prices.

[Nick Rowe]

Scott:

1. I'm still uncomfortable with NGDP. If NGDP changes, I always want to decompose it into a real change and a nominal change, because real and nominal variables are qualitatively distinct, in my mind, and their effects are usually quantitatively distinct too. Reasoning in terms of NGDP may be OK as a rough approximation, but no more. Plus, W/NGPD per capita is a very weird beast. It would take me some time to figure out what units it's measured in, and how it ought to behave, in equilibrium, in the face of various shocks to labour supply, demand, productivity, whatever.

2. To a first approximation, it doesn't really matter if wages are stickier than prices, or prices stickier than wages. You get roughly the same drop in output and employment either way (as long as the MPL curve (or MRPL curve under imperfect competition) is relatively elastic, which I think it is). Assume the MPL curve is horizontal (diminishing returns to labour don't happen). A fall in AD causes exactly the same fall in Y and L whether we have P fixed W flex, or W fixed P flex. In the former case we have excess supply in the output market, and the labour market clears. In the latter case we have excess supply in the labour market, and the output market clears. But exactly the same recession, and drops in output and employment.

[Andy Harless]

I think Bill's "snack food" objection goes pretty deep. I would argue that the critical feature of money here is not its role as a medium of exchange but its role as a store of value. In your parable, you've stacked the deck by making money the only store of value. If you were to introduce into your model another store of value that's not a medium of exchange (let's say, bonds), a quasi-bondist could say that money's role as a medium of exchange is trivial: people are actually hoarding bonds, not money, and they only use money as a convenience while converting something else into bonds, or as a substitute for bonds when bonds get so expensive that money becomes an equally good store of value. I think this point of view is supported by the observation that, when nominal interest rates go to zero, central banks are able to push massive amounts of base money without having a material effect on the quantity of currency in circulation. When money becomes competitive as a store of value, marginal increases in the quantity of money are no longer used as a medium of exchange.

[Nick Rowe]

Actually Scott, I sort of see why you prefer W/NGDP per capita. Sometimes I think in terms of W/M as being the best definition of real wages. But if you think M is a poor measure of AD (which it is), then I could see why you prefer W/NGDP.

[Nick Rowe]

Mike: was that the paper where they found that stores kept changing prices frequently, but always from $9.99 to $7.99, then back up to $9.99 again?

My reactions: only too plausible, but at the same time profoundly depressing, because it means that aggregate prices are still sticky, but our hope of ever having a simple theory of stickiness in terms of adjustment costs etc. look very slim.

Andy: I may not have fully understood/captured Bill's objection. But there is nothing in my model that says that money is the only store of value. Peanuts are a store of value too. And I did say that asset prices in my model, including bond prices, are sticky too.

[Mike Moffatt]

This is the paper: Some Evidence on the Importance of Sticky
Prices (PDF). Abstract:

We examine the frequency of price changes for 350 categories of goods and services covering about 70 percent of consumer spending, on the basis of unpublished data from the Bureau of Labor Statistics for 1995–97. In comparison with previous studies, we find much more frequent price changes, with half of prices lasting less than 4.3 months. Even excluding temporary price cuts (sales), we find that half of prices last 5.5 months or less. We also find that the frequency of price changes differs dramatically across goods. Compared to the predictions of popular sticky-price models, actual inflation rates are far more volatile and transient for sticky-price goods.

I think it is the one you mean ($9.99-$7.99-$9.99), which they call a 'temporary price cut'.

[Lee Kelly]

I think my story about apples and oranges from a week or so ago was better.

With a fixed the price of apples, a close substitute for oranges, one can resolve shortages or surpluses of apples by manipulating the price of oranges. All else being equal, reducing the supply of oranges makes people demand more apples, and increasing the supply of oranges causes people to demand fewer apples. Even though the price of apples is fixed, there is some price for oranges which indirectly equilibrates the supply and demand for apples. After a while, people might begin to think that phenomena caused by surpluses or shortages of apples are actually caused by the price of oranges being "wrong."

[Andy Harless]

Peanuts aren't a very good store of value. They're bulky and perishable.

[reason]

Nick,
I don't quite understand this:
"Actually Scott, I sort of see why you prefer W/NGDP per capita."

W is a price and NGDP is an amount - this ratio is hugely biased by population growth. Do you actually mean total money wages divided by total nominal income (i.e. the wage share of income)? Or do you actually mean W divided by per capita NGDP (Not quite the same thing as the wage share of income since working hours can change).

[David Andolfatto]

Nick:

Sorry for side-tracking you here. But I am teaching this "excess demand for money" hypothesis to my students right now and I was curious to know the best way to defend the sticky price hypothesis. Let me comment on your points.

[1] Wage and prices indices are relatively forecastable. I never thought of characterizing stickiness in this manner. Let me think about this. (And can you point me to papers that forecast these objects?)
[2] Because recessions are times when it gets much harder to sell stuff and easier to buy stuff, except that stuff whose prices are hard to forecast. Again, can you point to the empirical studies that support this claim; or is it just something you "see" out there? (And if so, please elaborate.)

Scott:

I am not convinced by the "mountains" of evidence you suggest exist in your point [1]. My views on that literature are summarized here: http://andolfatto.blogspot.com/2010/07/sticky-price-hypothesis-critique.html
As for your point [2], this sounds promising. Except that I am not interested in history per se; rather, I asked whether these forces are still quantitatively important today--i.e., over the last 3 years. Still, can you point me to a post of yours that elaborates on this? And I disagree with your point [3]. Flexible price models with heterogeneous portfolios can generate real effects from nominal shocks (though, I presume you might argue that these are not plausible).

One more thing: most of the discussion appears to be in terms of sticky goods and labor prices. What about nominal debt contracts? Oh, and hurry up with your answers...I lecture next week...lol.

[Scott Sumner]

David, I just read the link you provided, and am a bit confused. It seems to me that all the information provided in the link supports my assertions that wages and prices are sticky. Perhaps you misunderstood my point 1. I was not claiming that there is a mountain of evidence supporting the sort of wage and price stickiness that would necessarily causes nominal shocks to have real effects. There are different types of individual price stickiness, and some types do not imply stickiness for the overall price level. (However Nick cited separate evidence that the overall price level is sticky.) In any case, I view wage stickiness as the more likely culprit. So it seem we agree that wages and prices are adjusted infrequently, unlike commodities for instance. My only point was that is a necessary condition for NK-type models.

The other two points are just as important, and without them I'd be the first to agree that the data we have on wage and price stickiness does not, by itself, create a prima facie presumption in favor of sticky wage/price business cycle models. For instance, we all know that RGDP and NGDP are highly correlated over the cycle. By itself that doesn't prove anything. But I think we also have strong evidence that those NGDP fluctuations that can be linked to monetary policy shocks also tend to have real effects.

I'll defer to your expertise on non-sticky wage/price models where nominal shocks have important real effects--I've haven't heard much about them, so I assumed people had trouble coming up with plausible models. That may be incorrect.

Nick, I'm puzzled by your comment on NGDP. It's a 100% unadulterated nominal variable, just like P. Also recall that P is not some sort of feature of the real world, but rather is the creation of government bureaucrats, using entirely unscientific procedures. That's why so many distinguished economists can disagree strongly over Tyler Cowen's TGS idea. They have wildly different views of RGDP growth since 1973, which means they have wildly different views on inflation. There is no "right way" to measure RGDP, as economist don't even agree (at all) on what's it's supposed to be measuring (stuff, utility, or some combination.)

One other point, You said buying and selling peanuts with the objective of influencing their price doesn't matter, and I agree. But if the central bank were to buy and sell something else, with the intention of influencing peanut prices (both nominal and real) it would matter a lot. Here's an analogy. If the US government had supplied more quarters in 2008-09, it would not have helped, but if the Fed had conducted monetary policy in such a way that the quantity if quarters in circulation rose at the normal pace in 2008-09, it would have prevented the (nominal) recession. That's because coins are used for transactions, which decline during recessions. Thus the flow supply of new coins is low in recession years. Of course the flow supply is demand determined at any give monetary base. This probably makes no sense, but I once did a post on it.

[Lee Kelly]

In one of Yeager's essays there is an excellent metaphor for price stickiness. I can't remember if it was original with Yeager or not, but it goes: for years nobody could understand how the giraffe could pump enough blood up its long neck for its brain to function, but only an utter fool would therefore conclude that giraffe's do not have long necks. In other words, all economists know that prices are sticky, it's just that not all of them know it in what they consider methodologically reputable.

Yeager's own explanation of price stickiness with an excess demand for money is that there is a who-goes-first problem, because networks of prices are interdependent.

Ordinarily, competing entrepreneurs lower prices because it will increase their profits. There is a clear incentive. But when there is an excess demand for money, whoever lowers their prices first is likely to suffer losses in the short run. This is because the entrepreneurs inputs are, by assumption, still set at the old level of prices. For example, he buys labour and raw materials at the old high level of prices, but sells his goods on the market at the new low level of prices .In that case, it is far from obvious that lowering his prices first will be good for business. The entrepreneur may sell more units, but at what profit margin? And by the time the long run arrives, the entrepreneur may already be out of business.

Everyone is in the same boat waiting for everyone else to make a move.

Money has no market of its own. It is traded on all markets. It has no price of its own. Whatever "price" it has is an average across many markets and cannot change smoothly. Individual businessmen can't tell the difference between a fall in demand for their particular product and a general fall in demand brought about by an excess demand for money -- they react to both the same.

[Determinant]

Whenever you get storage elements in a model, you will get "stickiness". If we just consider the flow of money in a transaction and then add in a storage element which is proportional to the change in the flow, mathematically you get a differential equation. When you solve for your dependent variable, you will find that you cannot change the flow instantaneously.

To change prices instantaneously mathematically implies that there is infinite amount money in order to maintain general equilibrium. But we do not have unlimited money, we do not live in an economy with infinite money, so price cannot change instantaneously.

To be clear, a storage element like a bank account is mathematically just like a capacitor; a price is just like a resistor. Google introductory Electrical or Heat Engineering Theory and the differential equations are all right there.

Speaking of which, I see a paucity of differential equations in any model that gets thrown around in economics and whenever you deal with flows and storage your get a differential equation.

[David Andolfatto]

Scott,

The point of my "critique" of the sticky-price-hypothesis was, in a nutshell, that price dynamics are largely indeterminate in enduring relationships (the typical firm-worker relationship, in particular). Consequently, even if (say) wages appear to be sticky, this is unimportant, because the wage path is not "allocative" in a wage bargain.

My only point was that is a necessary condition for NK-type models.

And my point is that the NK model, like so much of neoclassical analysis, assumes that labor is traded on anonymous spot markets. Some way is needed to hamper the operation of these markets (from a theoretical perspective). And the natural candidate that people like to pick is price stickiness. Why? Because prices appear "sticky" in the data. This makes no sense to me. I'd rather drop the assumption of anonymous spot markets.

In any case, my question to all of you is this: Look out there...in the real world, I mean. The US GDP path appears to be below linear trend...for years now. Can you tell me where the sticky nominal prices are that are hampering the recovery (and that led to the recession). Point me to something concrete.

[K]

David: Krugman recently did a post which included a pretty striking graph of the distribution of wage changes showing extreme downward stickiness at the zero boundary. That, in my opinion, is the only kind of stickiness you'd ever need to draw a straightforward link between deflation and unemployment/AD shortfall. If you've ever been anywhere near employment contract negotiations, total intransigence against a wage reduction is a fact though people will accept nominal wage freezes for years, even in the face of inflation. Employment contracts often specify that the employee will be paid "no less than...". Maybe nominal wage cuts are insulting.

Determinant: There is nothing you can get in a continuous time model that you can't also get in a finite difference setting. That said, lots of macro papers are continuous time. And your point about infinite money supply makes no sense. You don't even need money at all (as an asset) to have prices (sticky or not). Are you arguing that fx or stock prices are sticky because of a finite money supply?

[Scott Sumner]

David, I like to use NGDP as my nominal benchmark (although I understand why others don't.) But let's suppose the financial crisis caused M*V to fall 8% below trend. Actually, we don't have to assume this, it did fall 8% below trend, but I want to assume it was an exogenous nominal shock. Let's also suppose the sticky wage/price model is wrong, and that nominal shocks don't have real effects via wage/price stickiness. In that case what would have to happen for the economy to stay at full employment? That's easy, both wages and prices would have to rise 8% below trend (i.e. fall.) To put it mildly, that's not going to happen. So in that case it seems to me that unemployment is almost inevitable.

How would I contest this argument? I'd say the Fed was targeting core inflation, and RGDP fell for some reason unrelated to nominal shocks, and because they slightly missed their core inflation target, NGDP fell 8% below trend. In this counterargument (against my real position) I'd claim that the fall in RGDP, combined with the Fed's reaction function, caused the fall in NGDP. And RGDP fell for various "real" reasons.

I don't think it's easy to distinguish between these two theories in the postwar data, although it's not impossible. If I didn't have the pre-war data, I'd be far less convinced that nominal shocks matter (remember, it's the pre-war data that convinced Lucas.) But it seems to me that in the pre-war data we do see monetary policy causing NGDP changes, and then see those changes have real effects. So when I see NGDP fall at the fastest rate since 1938 (in 2009) and nominal wage growth slow far less than 8%, I'm not at all surprised that we see high unemployment. We'd see that even if there had been no banking crisis at all, and NGDP fell 8% below trend between mid-2008 and mid-2009 for some other reason, like currency hoarding by Colombian drug dealers. It's a necessary and sufficent condition for a severe recession, unless wages are really responsive. And they don't seem to be.

I am less convinced I understand the slow recovery. Some days I think the extension of UI from 26 weeks to 99 weeks has made us have a more "European" labor market. Other days I focus on the odd discontinuity of wage increases at zero percent--which implies a lot of money illusion at low inflation rates. Krugman had a post on that last week. I also have a recent post where I suggest demand and supply shocks get "entangled" and reinforce each other.

[Patrick]

"Look out there...in the real world, I mean. ... Can you tell me where the sticky nominal prices"

Here's an anecdote, FWIW:

I work for a Very Big Company that makes some electronic widget - you can think of them as little computers (I can't tell you exactly what it is because then you'd know who I'm talking about and I'll get fired). The widgets are made of lots parts. We stock the parts. Some of the parts are made in very advanced factories - some of the most advanced ever devised by humans. To get a better price on these expensive parts, we buy them in advance on contract. The the widgets are built, and we try to sell them. The price-point for a given widget is set, in part, based on these contracts, which obviously don't change much. They are sticky.

Then we print marketing literature, create catalogs, the finance and accounting people create spreadsheets, the IT staff and an army of consultants feed prices into SAP and whatever other gawdawful enterprise software monstrosities they use, we train salespeople and tell them the range of prices they are expected/allowed to charge, etc ... Changing prices is expensive - you have to redo all that work and undo lots of training and policies. Again, this makes price changes costly, so we don't like to do it - they are sticky.

Last, but certainly not least, we can't change prices willy-nilly because it will piss off our customers. In some jurisdictions and situations it might even be construed as illegal (i.e. Robinson–Patman Act) - though IANAL. In any case, despite our best efforts to keep pricing information secret, the truth will out. If customer A hears from customer B that he paid 10% one week later, we are in for a world of hurt. So prices are sticky.

[David Andolfatto]

K: Yes, I am aware that the distribution of nominal wage changes (among continuing workers) is frequently skewed in this manner (there is considerable heterogeneity across countries as well). My experience in the construction sector during a severe recession (early 80s) leads me to question to what extent this sort of data captures the full degree of wage flexibility in the labour market (workers also adjust labour effort along various dimensions which are de facto real wage changes not captured in the data). Moreover, this type of "snapshot" data is irrelevant from the perspective of surplus division in ongoing relationships. The wage path is largely indeterminate in relationships. In labour market search models, the relevant degree of wage flexibility is in terms of new hires. I have read papers where the degree of wage flexibility among new hires is much greater than in ongoing relationships. In short, Krugman's piece is uninformative and misguided.

Scott: The manner in which you would have contested your own argument was precisely what I had in mind. Thanks for this--I must make an effort to read your blog on a more regular basis.

Patrick: Thanks for your anecdotal evidence. Yes, I believe you when you say the price point is set in part based on contract. But in a contractual relationship, there are other dimensions of adjustment, no? One might want to work a little harder to satisfy a long-time customer at the same price--is this not a de facto price cut (unmeasured in the data)? During a boom, do you tend to spread your resources thinner (per customer) if price is inflexible? I am just wondering here. It's just not as obvious to me that rigidities in the terms-of-trade are responsible for year-long (or longer) recessions. But maybe they are! (I just don't know.)

[Determinant]

K:

Determinant: There is nothing you can get in a continuous time model that you can't also get in a finite difference setting. That said, lots of macro papers are continuous time. And your point about infinite money supply makes no sense. You don't even need money at all (as an asset) to have prices (sticky or not). Are you arguing that fx or stock prices are sticky because of a finite money supply?

I'm not justifying prices, I'm justifying

sticky

prices, that is prices that cannot change instantaneously.

Anytime you have storage and a proportional storage element that responds to a differential, instantaneous change implies that money is infinite. For instance with a capacity the voltage cannot change instantaneously, that implies infinite electrical charge which implies infinite energy. There is no such things as infinite energy.

Empirically there is no such thing as infinite money, no economy has ever behaved like that. No infinite money means that prices cannot change instantaneously to anything they want and still clear. By money I mean a unit of exchange. Coconuts work just as well, so long as they are not in infinite, limitless supply.

fx and stock prices are a canard: read Alan Greenspan's autobiography where he tells about his introduction to a currency desk at a big New York Bank. They moneymaker is not the price of currency, it's the spread between the bid and ask which is much more sticky. That spread is what makes the desk money.

[`]

David: "this type of "snapshot" data is irrelevant from the perspective of surplus division in ongoing relationships. "

I think that's only true in a healthy industry. Nominal wage hikes often only take place in the context of contract renegotiation, which is much less likely to be instigated by the employee where competition from unemployed workers is relevant. But, I agree that the Krugman piece was oversimplified.

[K]

That last comment was me.

[Patrick]

David: Depends on incentives. Say we are in a boom. If we know the long time customer is locked-in (the stuff is deployed and removing it is impossible - or at least too costly to be worth considering), then we may actually work less hard for them and concentrate on getting all the new business we can since that's where the bonuses are. In a recession, we might work harder for an existing customer because their is nothing else to do and it's important to look busy to keep your job. And it's better for our reputation - which can help get what little new business their is. Mostly I'd say it depends the individual situations, the people involved, and the compensation incentives. Not sure their is a hard and fast rule.

[Greg Ransom]

Isn't the "excess demand" for money complicated by.the fact that there are all sorts of close substitutes for money -- the easily traded store of value function can be carried out by various securities, gold, etc.

As much of the problem is the collapse of near monies and shaddow money, isn't it?

What does this reality do the the monetary disequilibrium story?

The money function that can't (usually) be carried out by money substitutes is e tax payment function.

Does this make the tax burden during a bust turn into a special generator of the "excess demand" for money disequilibrium process.

[Nick Rowe]

David: the reason I believe what I believe about sticky prices/wages (I don't really distinguish the two) is more casual observation and reading newspapers than any published papers. I come back from a long weekend out of touch with civilization, and the monthly CPI is almost exactly where I thought it would be. Exchange rates and stock prices have changed more (relative to forecast) in 3 days than the CPI did in one month.

http://worthwhile.typepad.com/worthwhile_canadian_initi/2010/05/the-cpi-and-other-asset-prices.html

There's survey data on ease of hiring, and ease of meeting an increase in demand, that seems to be countercyclical.

Catherine Rampels graph I link to here, for example:

http://worthwhile.typepad.com/worthwhile_canadian_initi/2010/09/poor-sales-minus-quality-of-labour.html

And Bank of Canada surveys.

But mostly I'm just reading the newspapers, and looking out the window. It all seems to fit the story "It's easy to buy, and hard to sell, goods and labour, in a recession".

But there isn't much hard data on how easy/hard it is to buy/sell stuff. I bewail that absence of data here: http://worthwhile.typepad.com/worthwhile_canadian_initi/2010/09/p-data-q-data-and-l-data.html

Actually, that last post goes right to the point that's relevant here.

(If you were closer, I would love to come in and give a guest lecture on this! good luck with your class!.)

[Nick Rowe]

Greg: "As much of the problem is the collapse of near monies and shaddow money, isn't it?"

I think so. Probably. The collapse of close substitutes for money would increase the demand for money, causing the LM curve to shift left.

I never put much emphasis on taxes affecting the demand for money. I pay taxes once a year from my chequing account. If there were no taxes, I would be buying something else from my chequing account. I've always followed Menger rather than Knapp on taxes, chartalism, the state and money.

[`]

Nick: "Exchange rates and stock prices have changed more (relative to forecast) in 3 days than the CPI did in one month."

OK, but how do you distinguish between "sticky" and "less volatile?" I tried to look at lagged autocorrelation of changes the core CPI last night. It's strongly positively autoregressive in the long run (i.e. sticky which will suppress measured volatility), but not so easy to tell in the past 30 years - it's too seasonal. I'll post it here if I can get anywhere tonight.

[anon]

Determinant: This is slightly offtopic, but the econo-physical analogy I'm familiar with relates price to 'voltage' (effort), flow of product to 'current' (flow) and inventory to 'charge' (displacement).[1]

The problem with this analogy is that it does not result in rational behavior: for instance, the damped harmonic oscillator corresponds with a continuous-time version of the 'cobweb model' from economics. Far from involving price stickiness, the cobweb model results in prices overshooting their equilibrium values! Nevertheless, it would be interesting to draw from optimal control theory or similar engineering fields in order to achieve better modeling of disequilibrium processes.

[1] see e.g.: Karnopp, Margolis & Rosenberg; System Dynamics: a Unified Approach [1990] for an explanation of related physical models

[Scott Sumner]

David, A quick note on wage flexibility. The amount of wage flexibility among new hires required for full employment is much greater if existing worker wages are not flexible. And that's if you ignore efficiency wage considerations. Consider the following example. Unless I am mistaken, a skilled auto worker from Michigan can't walk up to a non-union plant in the south, and announce, "fire that guy, I'll work for half his wage." That's not how most firms operate (although I don't doubt that some do.) But even those that do, face greatly diminished demand for their product, because of the wage stickiness of other firms.

[Determinant]

Anon:

Works for me. There's nothing wrong with overshoot if the system can damp the signal and converge to a final result.

Part of my point is that in Engineering we always have to take note of sub-optimal results and models that produce degenerate behaviour. It happens a lot. ISTM that economists prefer to armwave those problems away rather than dealing with them and the result is that the economics debate is poorer for it.

[David Andolfatto]

Patrick: Very interesting, don't you think? When one starts to look at all the different ways in which people in relationships start adjusting their practices in response to changes in economic circumstances, then the de facto real price of goods and services are changing, even if their stipulated nominal levels do not. The econometrician, or Nick Rowe looking at newspapers, sees that nominal prices do not adjust a lot. But maybe they don't have to when resources are allocated in relationship mechanisms (as opposed to the anonymous spot market).

Nick: You do not distinguish between sticky nominal wages and prices? I think it matters, does it not? For example, if both are sticky, then the real wage is sticky. As for your "easy to buy, difficult to sell goods in recession" idea, I agree with it. I might note, however, that this property falls naturally out of search models that feature thin-market and congestion externalities--it has nothing to do with sticky prices. I think you keep referencing sticky prices because your mind has been seduced by the vision of Marshall's scissors. Come to the dark (search) side, Nick! ;)

[Greg Ransom]

The role of the collapse of shadow mony in the recession is argued by Sweezey and Lantz of Credit Suisse:

http://hayekcenter.org/?p=2954

[Nick Rowe]

David:

"You do not distinguish between sticky nominal wages and prices? I think it matters, does it not?"

Not very much, for understanding the business cycle. The model I always have at the back of my mind is a yeoman farmer economy of worker-firms. Or self-employed hairdressers. The price of a haircut is the wage of a hairdresser. If the hairdressers banded together into firms, so that an owner of the salon employed them to cut hair, nothing much would change. If labour were a much smaller share of total inputs, and if the prices of those other inputs were either much more or much less flexible than the wage, then it would matter more.

"Come to the dark (search) side, Nick! ;)"

I like the search side. I don't think it's dark.

1. If I want to model explicitly the idea that buying/selling is easy/hard, I need some sort of search model. But, as I argued in a previous post, we can see Q=min{Qd,Qs} as the limiting case of a search model, where if Qd exceeds Qs it is infinitely costly for an extra buyer to find and extra seller, and vice versa if Qs exceeds Qd. Marshall's scissors are just the limiting case of a search model. I've even got a great picture to show this!
http://worthwhile.typepad.com/worthwhile_canadian_initi/2011/04/excess-supply-excess-demand-and-search.html

2. I have always been hopeful that one day a search model would be able to explain the appearance of sticky prices. So that prices would *look* sticky, even if, at some deep underlying level, they really weren't. But so far none has fulfilled my hope.

"As for your "easy to buy, difficult to sell goods in recession" idea, I agree with it. I might note, however, that this property falls naturally out of search models that feature thin-market and congestion externalities--it has nothing to do with sticky prices."

Can you explain this more, please. My understanding is that in a search model of the Diamond type, where you get 2 (or more) equilibria: in the low level equilibrium it's hard for both buyers and sellers to find a partner (which explains why trade is low); and in the high level equilibrium it's easy for both buyers and sellers to find a partner (which explains why trade is high).

Did you mean that case? Or are you talking about a model with a real shock to one side of the market that shifts the unique search equilibrium? If a shock means it's suddenly easy for a buyer to find a seller, and hard for a seller to find a buyer, bargaining power shifts to buyers, so the price should instantly fall, until bargaining power is once again equalised.

[Nick Rowe]

I have always wanted to define the relevant concept of "real wages" as W/M. Scott wants to define it as W/(per capita NGDP). Both of us are trying to get at the same sort of idea. (Bot measures are inadequate, though I think Scott's is less inadequate than mine).

Just an observation on this: remember Keynes' GT? He defined the money supply in wage units. In other words, he defined the real money supply as M/W. Which is exactly the inverse of my definition of real wages, W/M. And Keynes said that if W always adjusted to ensure full employment, this would be "monetary policy by the trades unions" (not the exact words). We are all talking about the same thing.

[David Andolfatto]

Nick,

Nice post; thanks for the reference!

On your point 2, I'm not sure what you are talking about. I presume you mean sticky nominal prices. Sticky real prices are easy to generate in a search model, since the time path of the real wage in a wage bargain is (under some conditions) indeterminate (i.e., there are many different ways to split the present value of match surplus). And now if we a rationale for money in our model, pay workers in $, I'm not sure why this should change anything. A given split of the match surplus may be implemented even with a nominal wage path that displayed "steps" over the duration of the match. At least, this is what I think...I'm not sure that I can point you to any papers that actually do this!

On your third point, I guess I meant search models with or without multiple equilibria. I think what you say is true under some commonly employed bargaining solutions (like the Nash Bargain). But Hall (and Shimer, I think) have recently written papers that depart from standard bargaining solutions (solutions that entail much more real wage rigidity, in particular). I see now that Hall has an even more recent paper (the one cited by Krugman)...haven't read it yet.

[David Andolfatto]

And Keynes said that if W always adjusted to ensure full employment, this would be "monetary policy by the trades unions" (not the exact words).

This could not have been part of this General Theory (though it may have been part of the opening chapters of the GT, where he invokes sticky wages as a simplification). I say this because he later explains how flexible wages are likely to exacerbate coordination failure. Or am I wrong?

[Nick Rowe]

David: On Keynes, and the GT:

page 267 Chapter 19 "Changes in money Wages":

"If, indeed, labour were always in a position to take action (and were it to do so), whenever there is less than full employment, to reduce its money [he means money wage] demands by concerted action to whatever point was required to make money so abundant relative to the wage-unit [he means increase M/W] that the rate of interest should fall to a level compatible with full employment, we should, in effect, have monetary management by the trade unions, aimed at full employment, instead of by the banking system.

Nevertheless while a flexible wage policy and a flexible money policy come, analytically, to the same thing, inasmuch as they are alternative ways of changing the quantity of money in wage-units [M/W], in other respects there is, of course, a world of difference between them."

He then lists 4 reasons why they are different:
1. Individual workers can't coordinate an economy-wide reduction in W
2. Fixed wages are fairer, since some other nominal incomes are fixed
3. deflation increases the real burden of the debt
4. Falling wages and prices might cause expected deflation, and raise real interest rates and reduce AD.

[Nick Rowe]

David: on indeterminacy of splitting the match surplus.

OK. I have read some of Roger Farmer's (another great UWO PhD!) models like his. Arthur Okun's 1970's "toll model" (imagine that buyer and seller each had to pay a toll before they could enter the market) was an early progenitor, IIRC. They create a continuum of equilibria -- a thick vertical Phillips Curve with a range of natural rates. Here's my own very crude but simple version:

http://worthwhile.typepad.com/worthwhile_canadian_initi/2011/04/a-monetarist-search-model-of-keynesian-unemployment.html

I didn't realise that this was the sort of search model your were talking about.

Yes. If there's a zone of indeterminacy for real wages (as in these models), then you can have sticky real wages, or sticky nominal wages, or sticky relative wages, or... almost anything, really. If something is a Schelling focal point for wages, then that is how wages get set (within those bounds, of course).

These are like "infinitesimally small menu cost" models. And the menu can be written in any language (real, nominal, whatever) you like.

I'm in two minds about that sort of search model. I see them as providing an underlying theory of sticky wages and/or prices. And i believe in focal points (the whole of society is pretty much just one big focal point). I sort of like them. But I can't really bring myself to believe in them. Even if the Nash bargaining solution isn't exactly right, the equilibrium ought nevertheless be some sort of convex combination of the two threat points. Relative bargaining power ought to matter somehow in determining the shares of the surplus. If one side becomes less afraid and the other more afraid of walking away from the deal, that ought to affect the deal. In repeated games I can see the focal point created by past plays should matter. But when a new worker/employer make a match, the two don't have a past history.

Nevertheless, that particular sort of search model is not something i would ignore. There's gold in those models somewhere. But I think it's a very Keynesian sort of gold.

[Scott Sumner]

A couple random observations. One piece of evidence for sticky wages is that it seems like recessions are worse when the government deliberately slows the downward adjustment of wages during deflation (1929-31) than when they don't slow the downward adjustment of wages during deflation. (1920-22.)

Despite all my huffing and puffing about sticky wages, I'm actually not sure it matters whether wages are sticky, or if they are sticky whether they are a transmission mechanism for monetary policy. The strong evidence we have isn't for sticky wage models, it's for nominal shocks have real effects models. And the implication of those models is that one should get rid of nominal shocks. I suppose wage stickiness might matter in choosing which nominal aggregate to target, but that's a much less important issue than the issues David is discussing.

[Richard H. Serlin]

Macro is not a strong point of mine, but I liked this post, and here are some comments:

“And if a monetary disequilibrium theorist suggested that the recession was caused by an excess demand for money, he would be met with the objection that people could always get more money by selling peanuts. Which is true, but irrelevant.”

It seems relevant, just too small an issue. Only peanut producers and owners could do this, just a small percentage of the economy (which you elsewhere say).

"Now, it is true, peanuts are a tiny and almost irrelevant part of this economy by assumption. In the real economy, bonds aren't. But so what"

Bonds serve a special role. The lower the rates, the more people and firms borrow, and so the greater the velocity of money, and the greater the economic activity. From MV = PY, for constant P, you can increase Y just as much with an increase in V as you can with an increase in M.

"1. Because it is so very easy to forecast the CPI or nominal wages with very precise accuracy at about a 1 month or 1 year horizon compared to forecasting (say) stock prices or exchange rates or the prices of homogenous commodities sold in competitive organised markets. At much longer horizons this isn't true (unless of course the central bank is targeting inflation). Prices look sticky."

Great point. You are good (Do you have much knowledge of Stephen Williamson's models and macro. Non-specialists could really use someone to explain his points and positions more clearly to them. Trying to riddle him out, with limited success, has become kind of a hobby for me). There's obvious strong rigidity against lowering nominal wages for reasons that should be obvious to anyone who doesn't consider humans to be robots. Also, could you imagine the response of your restaurant customers if you changed your food prices the way stock prices change!

[Richard H. Serlin]

Also, you could pull out a lot of strong empirics in reply to Andolfato. I remember seeing some really strong studies showing resistance to nominal wage cuts, one was a Canadian study that showed a huge number of wage freezes and very few wage cuts.

[Nick Rowe]

Richard: thanks!

"It seems relevant, just too small an issue. Only peanut producers and owners could do this, just a small percentage of the economy (which you elsewhere say)."

Most of us have a jar of peanut butter, or some peanuts, in the house! Even if everyone could produce peanuts, if the total market for peanuts were very small, it still would make very little difference. Suppose wages for dishwashers were perfectly flexible. "There can't be an excess demand for money/excess supply of labour. Anyone who is short of money and can't earn any can always get a job washing dishes. No, the problem is an insufficient demand for dishwashers, so the natural wage of dishwashers is too low".

Yes, peanuts and bonds are rather different. What I was hoping was that this post would force people to think more closely about what is the relevant different between peanuts and bonds that would allow them to hold the "bond" (saving-investment) theory of recessions while acknowledging that the peanut theory is ludicrous. Your comment is the only one that tries to do this. Ultimately, what you are saying is that bonds are a substitute for money and peanuts aren't.

I don't have a strong knowledge of Steve Williamson's macro. Steve wants to model the role of money/liquidity formally. He thinks it matters that not all goods are perfectly liquid, and you can't understand money, finance, and macro by assuming they are. I applaud. Steve doesn't like the assumption of sticky prices, precisely because it is just an assumption, and not modelled formally. That's where we part company.

I think I've got that right.

[Greg Ransom]

Scott, this is the Hayek position:

"And the implication of those models is that one should get rid of nominal shocks."

Here's the deal. This works better prospectivel by setting a NGDP target in advance, not so much retrospectively after an episode of misdirected labor and capital during an artificial boom leading to an unstoppable re-coordination involving an unavoirdable drop in shadow money values & liquidity -- and resources and labor which must be redirected to new uses at re-calculated values.

Just dumping money & credit willy-nilly into the hands of a slice of the big banks, the mega-wealthy, and the politically favored, etc. can't produce the unavoidable re-calculation, can't substitute for re-calculation/re-coordination, and can't substitute in a fully equiilibrating way for the collapsed of shadow money stocks & liquidity.

[Greg Ransom]

Hayek's repeatly told line is that the time to do something about the recession/bust is _before_ you've created the artificial, unsustainable malinvestment boom setting up an un-avoidable post boom recalcullation.

[reason]

Greg,
so you think the way to avoid satellites being damaged by sunspots is to move to another solar system?

[Greg Ransom]

reason -- as a troll with no point or insight, you could at least be a funny troll.

But no such luck.