What would count as a satisfactory explanation of sticky prices?

Richard Serlin, in a comment on a Stephen Williamson post, explains why he thinks prices are sticky. What Richard says sounds plausible. And it's based on his real-world experience. He may very well be right. But it doesn't constitute a satisfactory explanation of sticky prices, in my eyes.

I'm really just using Richard as an example to illustrate my views on what would count as a satisfactory explanation of sticky prices. His is a good example for my purposes, just because it's simple, and plausible, and it's accessible. I hope Richard doesn't mind my picking on him to make a point.

This is also an example for me to explain why economists want theories, and what "theory" means to an economist.

To oversimplify, Richard says that firms have sticky prices because their customers want sticky prices, and firms want to give their customers what they want.

What's wrong with that?

Suppose Richard is 100% right. Why don't I think it counts as an economic explanation? What more do I need? (To be fair to Richard, I have oversimplified his explanation, and he does start to go into these things).

This is what more I need:

1. I want some sort of explanation of why customers want sticky prices. Why, for example, would customers prefer prices that move in discrete jumps, like a staircase, rather than smoothly, like the banister? (Only assume the banister is at the same average height as the stairs, because I can't think of a better metaphor).

Why do I want an explanation of why customers want sticky prices? Why can't I just be satisfied with the statement that they do? I don't ask why customers want to consume apples.

Partly it's just that sticky prices, unlike apples, strike me as a peculiar sort of thing to want. So if I did understand why customers like sticky prices, it would make the assumption that they did like sticky prices more plausible. But mostly, it's because if I had an explanation of why customers like sticky prices, it would help me answer the following questions.

2. I want to know what exactly is the type of stickiness that customers like? What does "stickiness" mean, precisely? There are lots of different models of sticky prices that give different sorts of stickiness, and they can have very different macroeconomic implications, both for the data they predict, and for the macroeconomic policies they would recommend.

There can be models with fixed costs of changing prices, which don't depend on how big a change is made. Or quadratic costs, which do. (The first has prices jumping in steps; the second has prices moving smoothly.) Or models where firms can change prices every d periods. Or where firms have a 1/d probability of being allowed to change price every period. (That apparently trivial difference alone makes the difference between whether there is inflation inertia or just price level inertia in the Phillips Curve.) Or models where firms' price changes are bunched at the same time, or staggered. Or "s,S" models where firms change prices whenever they get too far away from the optimum. (Which can sometimes make average prices perfectly flexible even if each individual price is sticky.) And so on.

If I could understand why customers like sticky prices, that would help me understand precisely what "sticky" means in this context. And different meanings of stickiness would have very different consequences.

Knowing those consequences would help us test that theory, by seeing if the predicted consequences match the data. And let us predict the effects of different policies, and which policies would make people better off.

3. I want to know why some prices seem to be stickier than others, and why some prices don't seem to be sticky at all. If I understood why customers liked sticky prices, this should help me understand these differences in price stickiness. Perhaps customers for crude oil are different from customers for groceries, and don't care about price stickiness. An explanation of why some customers in some circumstances like price stickiness could help me test the theory. It could also have policy implications, if a change in some policy would make prices more sticky or less sticky, that's something I need to know when recommending policy.

4. This is the most important bit. I want to know why customers like sticky prices because this will help me understand whether and how their preference for sticky prices will affect their demand for goods. We can't just assume that customers have a preference for sticky prices, and at the same time model their demands for goods by maximising U(X,Y) subject to the budget constraint, ignoring the fact that we have just assumed their preferences don't look as simple as that. Or, at least, it might be dangerous to do that. The predictions of a model of the demand for goods which ignored a preference for sticky prices might look the same as a model which included it, but it might not. We don't know, until we know a bit more about that preference for sticky prices.

And look, there's no way it can be exactly the same, if we also assume that firms want to cater to their customers' preference for sticky prices. Firms presumably don't just cater to their customers' whims for the sake of it. They cater to their customers' whims because they believe they will lose customers and sales and profits if they don't. Which means the customers' preference for sticky prices must have some sort of effect on their demand functions, if Richard's theory is correct.

It's that last point, point 4, that is at the heart of what David Andolfatto, for example, means by saying that sticky price macroeconomics is "incoherent". If we had a good explanation of sticky prices, for why customers prefer sticky prices, and how their demands for goods reflect that preference, we would have a better and a consistent theory of what determines quantities with sticky prices, and which monetary policies could improve people's well-being. As it is we have a theory of the effects of monetary policy under sticky prices that might be inconsistent with our assumption that prices are sticky.

So that's what I think would count as a satisfactory explanation of sticky prices. And why a satisfactory explanation matters. And why sticky price macroeconomists should keep looking for an explanation, as they have been. Though that shouldn't stop us from doing the best we can with what we've got. Unsatisfactory though it is, it's better than ignoring the evidence for sticky prices.

Comments

[Mike Sproul]

Nick:

Another explanation of why prices are sticky:
1) The central bank issues more money.
2) The central bank's assets rise in step with the issue of money, so there is no change in money's value. Where the bank previously might have had assets worth 100 oz of silver backing 100 paper dollars, the central bank now has assets worth 200 oz. backing 200 paper dollars.

Quantity theorists are then shocked to find that prices did not respond to the increase in the money supply, and they proceed to throw out the entire supply and demand apparatus in favor of Keynesian fallacies.

[Adam P]

but there is a perfectly coherent theory explaining why customers prefer sticky prices, it reduces search costs.

If I know that the prices at the various supermakets that I could potentially go to are basically the same as they were last week then I don't have to go to all of them each and every week to check which is cheaper for the things I want to buy.

[Nick Rowe]

Mike: That is a theory of why the equilibrium price level doesn't change when you change the money supply. It's not a theory of why prices only adjust slowly towards the equilibrium. And you need a theory of slow price adjustment (or something) if you are going to explain why monetary disequilibria can exist and cause changes in the amount of excess supplies or demands for goods.

Adam: IIRC, that theory has been floating around since the 1970's. Did Okun talk about it in Prices and Quantities? I think there's something to it, but I don't know of anyone who has formalised it. I think the problem is this: prices must adjust eventually. Why does it reduce search costs to have infrequent big changes rather than frequent small changes? But I think there's something to it, in at least some contexts. The catalogue model, for example. You publish your prices in an annual (or seasonal) catalogue, because this saves potential customers from needing to visit the store to discover your prices. And you have an implicit contract not to change prices between catalogues (except under extreme conditions, maybe).

In fact, when I said in the post that we don't have a satisfactory explanation of sticky prices, that isn't really true. We have many explanations, that are at least partly satisfactory. And I think they are more satisfactory, all in all, than the models that don't have sticky prices (RBC models).

[Adam P]

Yes, it's certainly been around a long time. I think I have seen attempts to formalize it but that is tricky, as are all attempts to formalize models with differential information. I think some of Sims' irrational inattention stuff tried to model it.

But really the issue is not so much big changes vs more small changes in the transition from one basically stable equilibrium to another. What consumers most certainly won't want is prices that are fluctuating like stock (or oil) prices to clear markets on that particular day.

Instead firms absorb variation in the market clearing price into inventories and/or margins (for example Forex fluctuations have slow, incomplete pass through).

Also, for perishables like bread prices do move to clear the market, if the bread is a day old you can buy it a substantial discount. We are all just used to that.

[jsalvati]

: Adam P
Remember the study that came out a while back about how grocery prices changed a lot but then came back to a "reference price". To me that seems like pretty convincing evidence that the stickiness comes from supplier side issue.

[Adam P]

Yep, I remember it. Actually this is the sort of study I had in mind, to me the extreme stability of the reference prices and the fact that prices kept going back there is perfectly consistent with the idea that consumers want to have a good idea what general prices are without having to do a lot of searching every time they shop.

[Adam P]

BTW, correction to my comment from before, the Sims stuff was about rational inattention.

[Blikktheterrible]

I'm not dismissing the validity of the concerns expressed in this post, but I think it's easy to get carried away asking "why?" I think it's possible to have a perfectly good economic theory that doesn't touch the issue of why things happen. I also think that demanding an explanation of why things are happening in the world could set the bar too high for us to ever reach.

As I see it, there are several different angles from which we could assess a theory's value. Here are a few of them that I see being relevant, listed in order of relative value:

1. predictive power (What will happen?)
2. consistency with past/current observations (What is happening?)
3. explanatory power (Why does this happen?)
4. internal consistency (Does this make sense?)

So, I would find a theory that had predictive power to be automatically preferable to a theory that did not, even if the latter was superior in all other attributes. The reason that preference isn't terribly meaningful is that very few economic theories have been shown to have any predictive power. Thus, we settle for traits #2-#4. There is significant overlap between the 4 criteria. A theory must have a certain degree of internal consistency in order to be comprehensible enough to meet any of the first three criteria.

It seems that you're looking for #3, which I clearly don't see as the most desirable trait in a theory, though I certainly consider it significant. My assessment could be off and I would be happy to hear people's feedback on it.

[K]

What is more important - prices that are sticky or prices that are not volatile?  Awhile ago I remember Nick comparing FX rates and groceries.  To me, the big difference between those two is volatility rather than stickiness.  Groceries are not volatile because their cost is principally driven by wages which are not volatile if inflation expectations are well anchored.  In my imagination (which is not at all well read on this matter) this is how it works: prices that are not volatile are interesting because 1) since they are driven by wages they are a window into the feedback of inflation into wage demands; and 2) since they are not volatile one can observe the drift rate with a higher degree of confidence.  But is there something particular about stickiness (as opposed to low volatility) that makes it really important?

[Nick Rowe]

Adam: I could see your explanation making sense if there were fixed costs of absorbing new information. (Like that interesting paper that another commenter just linked to today, and now I've forgotten where. And like Sims, as you say.) It's cheaper to process information on a small number of large jumps than a large number of small jumps. Yes, that does make sense. Hmmm.

Blik: In this post I concentrated on your 3 and 4. To my mind, if you satisfy 3 and 4, it helps check whether a theory satisfies your 1 and 2. If you get a complete deep and internally consistent explanation, it has a lot more predictions and retrodictions that can be tested against the data.

To look at it another way: first we want 3 and 4. When we've got them, we've got a theory. Next we see if the theory stands up against the data (1 and 2), or if it stands up better than other competing theories, that also satisfy 3 and 4.

But your way, of listing 1,2,3,4 as separate desiderata, makes sense too.

K: "But is there something particular about stickiness (as opposed to low volatility) that makes it really important?"

Good question. To economists, "Yes!" is the answer.

Sticky price macroeconomists look at the world like this: there is an underlying equilibrium price P* that we don't directly observe. It's the price that equilibrates demand and supply for example (if we assume competitive markets). It's where the price would be if prices were perfectly flexible. It's a theoretical construct. Then there's the actual price P that we actually observe. And it's nearly always different from the equilibrium price, even if it's equal to P* on average, and even if it's always moving towards P*, but because P* itself is always moving, P never gets to equal P* except on very rare occasions.

If prices are sticky, then P will generally not equal P*. And this gap between P and P* causes a lot of problems, like excess supply and demand. Unemployment and shortages, etc.

It's not the variance of P that's the problem. It's the fact that P is not equal to P*.

And the job of monetary policy is to change the money supply, which affects P*, in such a way that P* doesn't move, so that P stays as close to P* as possible, even though P adjusts only slowly towards P*.

But why doesn't P always adjust instantly to P*? That's the puzzle.

[Lord]

Grocery prices are somewhat flexible in produce though there are noticeable price points, perhaps to simplify calculations and comparisons and make advertising easier and more prominent. Brand product pricing is controlled by the manufacture who prefer sticky prices as part of brand image. They pick a sales price as a standard discount and then alter the timing and frequency to smooth production and keep inventories low but prevent out of stock conditions. Another party could buy when on sale and sell at regular prices but this would require inventory costs and management that the manufacturer doesn't have to bear making it difficult to do profitably. Product ads in grocery circulars are paid for by the manufacturer. Manufacturers of competing products often play sales tag with each other to try to scoop each others sales. They like particular price points so much they will often alter packaging and sizes to maintain a price point or to ease the transition to another price point. I imagine some do as well in adjusting portion size to keep nutritional information in convenient sizes as well. There are certainly fewer manufacturers than individual stores which is why this is so evident.

[Jon]

Maybe what customers want is to avoid volatility and the price-stickiness follows as an accident. Basically what that would mean is that risk aversion leads to people preferring contracts to the spot market. Then there is stickiness per se but not because people want sticky prices.

[Sam]

Anecdote from my local small liquor store (in Alberta):

Q: Do you change your prices when taxes go up?

A: No. When I get a shipment, I bring it in and tag everything with its retail price. It would be too much work to go back and reprice existing inventory just because AGLC prices/taxes/whatever changed.

[Richard H. Serlin]

Ok, I'm pressed for time, and hope to elaborate further later, but quickly here you go:

Why do customers prefer sticky prices for many things, and workers sticky (in a certain way) wages?

1) Really it depends on the product; it's not that simple. But raising prices can sometimes really anger consumers, so as a businessperson, especially with certain products, you don't want to cut the price if you think soon after you will have to raise it back up.

2) For wages you really get a clear strong effect. Workers hate a cut in their nominal wage, and a lot more than the same real cut from a wage freeze with inflation. The empirical evidence is really strong here. Nominal wages seem to be the really inflexible thing, especially down. You can have sales to temporarily lower a lot of prices without too much problem, but try cutting a worker's nominal wage and see how he feels. Think about how it might affect morale and loyalty and all the ways he can hurt you, or do a worse job, without you really seeing or knowing for a long time, if ever, that he did it. And think of the costs of turnover, recruitment, and training for the many complicated jobs in today's world.

Ok, so you ask why.

And here I think is the problem a lot of economists have: They think it's not an explanation if it doesn't involve not breaking the assumptions that all people are completely rational, care only about maximizing a simple utility function that only considers current consumption with no context, have all public information stored, and instantly accessible, in their heads, have all the necessary expertise and learning to perfectly optimize all decisions that they face, and can do this analysis instantly and costlessly, costlessly in time, money, and effort.

If you give an explanation that breaks these assumptions, no matter how mountainous the evidence from reality for it is, they say it's not an explanation, even though the assumptions that they make have laughably less evidence for them.

Ok, so what explanations would I give for why there is the price and wage sticky preferences. They basically involve human psychology and culture, and these things come from evolution and history, which clearly are exogenous (unless we really can perfect the flux capacitor). I don't have the time, but I have no doubt I could write a paper with a mountain of evidence for this from psychology and sociology.

But there's also a lifetime of experience. There's no t-statistic attached. But I have to ask what are the odds I would have had the experiences I've had, read the things I've read, over a lifetime, if certain very basic things about human behavior and culture aren't true. The odds are microscopic with just the most basic assumptions like what I see is actually there, assumptions immensely more mild than the kind you typically see made in freshwater models, and even in standard empirical statistical analyses. Here's a quote I really like on this type of thing from the great growth economist Paul Romer of Stanford:

In evaluating different models of growth, I have found Lucas's (1988) observation, that people with human capital migrate from places where it is scarce to places where it is abundant, is as powerful a piece of evidence as all the cross-country growth regressions combined. But this kind of fact, like the fact about intra-industry trade or the fact that people make discoveries, does not come with an attached t-statistic. As a result, these kinds of facts tend to be neglected in discussions that focus too narrowly on testing and rejecting models.

Economists often complain that we do not have enough data to differentiate between the available theories, but what constitutes relevant data is itself endogenous. If we set our standards for what constitutes relevant evidence too high and pose our tests too narrowly, we will indeed end up with too little data...

My greatest regret is the shift I made while working on these external effects models...I suspect I made this shift toward capital and away from knowledge partly in an attempt to conform to the norms of what constituted convincing empirical work in macroeconomics. No international agency publishes data series on the local production of knowledge and inward flows of knowledge. If you want to run regressions, investment in physical capital is a variable that you can use, so use it I did. I wish I had stuck to my guns about the importance of evidence like that contained in facts 1 through 5.

– Journal of Economic Perspectives, Volume 8, Number 1, Winter 1994, Page 20.

[Declan]

I suspect that if you can figure out why people like round numbers, you'll be 90% of the way towards understanding why people like sticky prices. Or, in other words, what Richard said.

[K]

Ok. I think I understand. Thanks, Nick.  But take a case like Greece with P vastly (20%?) above equilibrium. It seems hard to believe that this is due to consumers not wanting a 20% price cut or retailers not wanting to relabel. It's not about comparison shopping! What seems more likely to me is that because wages are sticky
1) production costs are sticky too so producers can't cut prices
2) the employed can continue to purchase the same basket of goods as before at the same price

So aggregate supply has to adjust down by eliminating wages through wholesale layoffs. Wages are so sticky that it's much easier to fire a large chunk of the workforce than it is to reduce them. Prices can (to first order) remain constant with a reduced number people producing and consuming the goods.      

Do we not agree that wages are sticky? And if so, is the above mechanism not adequate for describing price stickiness for macroeconomic purposes?

[K]

Then of course we still have to explain wage stickiness. But as Richard Serlin said, employees get pretty disgruntled if you cut their wages. And you don't have to deal with ex-employees so it's easier to fire some of them than to reduce the wages of all of them. So you cut production instead of cutting price.

[Blikktheterrible]

What if we were to assume that adjusting a price is costly? That's very similar to the idea of menu costs, but I think it extends it a little. On top of menu costs, which often exist, there are implicit cost in deciding what a price should be, ie. the cost of the labor and information needed to adjust prices in a productive way. Because of that, there will be a threshold effect, with price setters only changing prices when the gains produced from the adjustment outweigh the costs of making the change. Those costs would presumably vary widely, depending on the cost of relevant labor and information. That should partially account for differences in "stickiness"of different prices.

Of course, there are also the effect of contracts and psychological factors, especially when it comes to wages.

[Adam P]

[email protected]:29pm: "Adam: I could see your explanation making sense if there were fixed costs of absorbing new information."

Yep, in the example I gave driving to two or more supermarkets to comapare prices would certainly have a large fixed cost component in the travel expense (time, effort, gas). Then there is the effort organize and absorb the information, remembering what the price/quality combos of the other place where when you're standing in the fifth store of the day.

So, the fixed costs are both in absorbing the new information and in acquiring it.

[anon]

Traders love volatility. They're also insane, generally speaking.

Most economic customers do not love volatility. It's unsettling.

Sellers know this. So where they have a reasonable choice between pure continuity and stickiness, they'll stick.

[G Voss]

While I haven't thought this through entirely, I thought I might throw in an alternative perspective. In many respects, it is an extension of the adjustment costs argument.

My own (limited) experience in setting prices leads me to believe that uncertainty faced by firms with some monopoly power is important. The point is easily seen in considering the information requirements needed to optimally adjust prices in a manner consistent with the theoretical flexible price ideal. A monopolistically competitive firm, the kind we see in NK models, needs to know the elasticity of demand for its product. (A competitive firm, of course, does not since the price is taken as given from the market place.) For many firms, even quite large ones, I suspect that information on demand elasticity - or to put it in less theoretical terms, how many consumers it will lose when it raises price - is hard to come by. (Think about the amount of effort that would be required by an academic to obtain and publish good quality demand elasticities for an industry; then imagine how difficult similar work would be for an individual firm.)

Throw in some risk aversion by managers (admittedly not common in macroeconomic models of the firm) and you may get a measured response to changes in demand conditions rather than the rapid adjustment to the new equilibrium price. Similar to the effects of uncertainty on monetary policy put forward by Brainard. Whether this effect is sufficiently important to generate "sticky prices" (and or persistence in inflation) or not, I have no idea. Nor have I thought through the general equilibrium effects and the implications for how monetary policy affects the real economy.

Empirically, any industry that has conditions that make it difficult to identify demand responses (whatever these conditions might be) are going to give rise to stickier prices, all else equal.

Finally, if such effects are important, it underscores the importance of anchoring inflation expectations, which current central bank orthodoxy considers to be so important. Since this allows such firms to easily incorporate such adjustments without concerns about demand effects.

[Bruce Wilder]

Professor Serlin, above, says, "They think it's not an explanation if it doesn't involve not breaking the assumptions that all people are completely rational, care only about maximizing a simple utility function that only considers current consumption with no context, have all public information stored, and instantly accessible, in their heads, have all the necessary expertise and learning to perfectly optimize all decisions that they face, and can do this analysis instantly and costlessly, costlessly in time, money, and effort."

Like Professor Serlin, I think the request for an explanation might come with implicit conditions that makes such an explanation impossible.

My usual answer starts with the observation that a great many prices are administered. That is, what we observe, is prices (and wages) set to some fixed schedule, and the bureaucracy that sets those prices both maintaining that price schedule by manipulating output supply and other variables, and acting, much of the time, "as if" the prices are fixed. (It's kind of paradoxical, since the firm acts "as if" something it itself sets, is as fixed as the stars in heaven. Take a zen moment on that one.)

This observation is not, in detail, consistent with the macro modeler's "menu costs" (which is not to say that the macro modeling choice is inadequate to modeling the macro effects). Prices may change frequently, and the cost of changing the price -- literally of changing the menu -- may be quite low, by design. But, prices are not responding independently to volatility in supply or demand. The movie theatre doesn't (usually) raise ticket prices for a popular movie -- at most, there might be slight tweaks to the price schedule, like disallowing "passes" for a time. The supermarket doesn't re-raise the price on a special that sells out; they might even give out "rain checks" enabling shoppers to get the price at a later time. Apple's price for the lastest iPhone is pretty much the same every year; they don't change the price when orders run ahead of supply for a time -- they just let the line get longer and call the Media to take pictures of it.

Prices are administered. That's just a fact. Is the wide-spread administration of prices and wages the same as the macro-economist's "sticky" prices?

I can explain administered prices, in micro-economic terms, if and only if I dispose of the assumption of profit-maximization by the firm. From our previous discussions, I don't think, as a macroeconomist, you are quite ready to let profit-maximization go. Ymmv.

When I say that the assumption of "profit-maximization" has to go, I mean that quite literally, in an abstract and theoretical sense. I'm NOT saying that we have to go all psychological and behavioral. What I am saying, is the administration of prices only makes sense if we relax some otherwise convenient assumptions about uncertainty, if we reason and model a world of incomplete information.

Firms -- real business firms -- operate as control mechanisms; a firm controls a production (and distribution) process. In the first lessons of textbook micro, we assume output-maximization to make the production function, a function. Output is a function of inputs, if and only if, output is maximal (or inputs minimal). Now, when (maximum) output is a function of inputs, that allows us to focus all of our attention on the problems of allocative efficiency. The technical problems of arriving at that maximum have been, literally, assumed away, and we are left with only the allocational efficiency problem.

Actual firms cannot assume away the technical, managerial and organizational problems. For an actual firm, output is NOT a function of inputs. That's such a basic and foundational aspect of the way economists think, that I think it can be hard to grasp fully, so I will repeat it. Output is not a function of inputs. Maximum output can be a function of inputs, but a real firm has to control the production process; it cannot "assume" maximization. It's not hard to see that output cannot be a mathematical function of inputs; just consult the definition of mathematical function.

And, when you say good-bye to output maximization, you say good-bye to profit-maximization. It can no longer be defined for an actual firm, for the same reasons.

Administered prices can be understood, readily enough, in the context of managerial control of a production and distribution process. I'm not going to do much to lay it out in a blog comment. I can sketch and point and wave my hands, a bit, if that helps. It's complicated and varies in many details across industries, and many of the factors cited by other commenters, like the role of culture and psychology as well as fixed and sunk costs, can play a part. But, if you are familiar with basic models of cybernetic control, you can work out the most basic elements. Prices are one variable in a system under control; only one variable can be controlled at a time, so price is held "constant", to facilitate control of other variables.

This administration of price -- holding price constant -- can be understood partly as an adaptation to a world of incomplete and emergent information, or asymmetric information, in which households and firms act strategically. Prices, for example, can be signals of quality. In a world of uncertainty and incomplete information, price is not enough. Every transaction is contingent. There are warranties and risks.

In the classic enumeration of the functions of money, money is:
1.) a medium of exchange
2.) a store of value (a means of insurance)
3.) a unit of account

If you think of the implications of "sticky" prices with your focus on money as a medium of exchange, your puzzlement takes a certain shape. But, administered prices only make sense in a world in which 2.) and 3.) are a big deal. That is, a world of uncertainty, incomplete information and incomplete markets, where almost every transaction carries contingencies and risks, and warranties and strategic committment matter.

It is also a world, lest we forget, where considerations of allocational efficiency can be completely dominated by considerations of technical efficiency. I don't want to lose sight of that. Economists have a patch over one eye -- they see only allocational efficiency and markets in an economy of markets AND hierarchies, in which technical efficiency is often much more important than allocational efficiency. Just the fact that $1.99 and $1.98 are more common observed prices than $1.62, is some pretty strong evidence, really, that pricing, and allocational efficiency, is subordinated to other considerations, considerations that I would hypothesize involve economies from technical efficiency gains.

Wages are often thought to be particularly, or peculiarly, "sticky", or sticky, nominally and more "sticky" somehow downward. Debt deflation is sometimes invoked, or supposedly irrational psychology, which somehow fails to grasp the movement of "real" wages, beneath the money illusion.

But, in a world of genuine uncertainty, in which workers are supplying behavior and not a uniform flow of generic labor services, and most workers are part of a hierarchical bureaucracy, which functions to control, in a technical sense, a production and distribution process, well, things are a bit different, aren't they? Wages, or wage-rates, are often fixed and warrantied as part of the implicit labor contract; the distribution of risk between employee and employer is critical to the incentives, per Knight.

Wages are fixed, there has to be a genuine loss to the employee of being fired and having to find another job, so that the employee is motivated to do as she is told (caricaturing life in a bureaucratic control hierarchy) and otherwise behaves responsibly. Again, I cannot lay all this out in a blog comment. Wave hands! Wave hands! I just trust that you are somewhat familiar with the arguments from either life or economics texts.

Coming back to the macro perspective, does the distribution of risk in structuring the incentives of a labor "contract" imply that the wage is "sticky" in a macro sense? I think it actually implies more than that, because it implies multiple relationships between money and labor transactions. Not just the medium of exchange / wage-rate relationship, but also the insurance / risk-distribution relationship and the unit-of-account / informational one.

since the risk of being fired plays a big part in many labor contracts, the macro rate of unemployment, to the extent that it changes the expected loss attendant on being fired, is going to matter, no? Similarly, the employer's credibility, during a general business downturn affecting the firm's output, or a financial de-leveraging, affecting the firm's claims to ownership, or equity-at-stake, could be important to behavior.

I try to imagine the business model surrounding Apple's iPhone model. It's highly visible in the business press as well as the technical gadget fan media, so many aspects are visible. There are big sunk cost investments, and a lot of calculation (unit-of-account). Big contingencies. And, everything depends on the incentives for behavior, which are directed, in large part by the implicit and explict distribution of risk to various actors, including consumers. Consumers are faced with a moderately complex, explicitly contingent contract, involving a "price" for the phone, and rates for service. And, they are asked to assume certain contingent risks -- a two-year committment to a service contract, cancellable with a schedule of penalty fees. The phone is warrantied for a year -- the warranty can be extended for another year; but AT&T won't sell insurance on the phone (though they do sell insurance on other phones). Etc.

(I'm not even thinking, yet, about the economics of sunk cost investment, increasing returns, experience, etc., in the actual production process, which implies a rapidly falling marginal unit cost of production.)

Do all those contingencies and all the attendant calculation add up to "sticky prices" in a macro sense? Can you accept such an explanation of sticky prices, if it entails giving up behavioral assumptions like profit-maximization? Or, the assumption that allocational efficiency considerations dominate economic behavior?

[Richard H. Serlin]

Mistake in my last comment: I referred to evolution and history as surely exogenous. They're instead givens. You could imagine a model where future evolution and future history (that affects relevant culture) are endogenous. But any effect on future evolution would be really slow and long term, so over periods of decades it looks like it would be pretty safe to consider it given unless you were discussing something pretty unusual. Culture also can be pretty resistant to things in a model, but there are important exceptions like this:

http://chartercities.org/blog/75/rules-and-culture-corruption-in-hong-kong

With regard to price stickiness/flexibility, it's different for different products, some are very flexible some aren't, it would take a while to describe in detail, and a while to explain the psychology and evolutionary and cultural basis. My complaint with many economists is they don't want to even make the effort. If the explanation has any non-super rational, super-calculator assumption they don't consider it an explanation, even if the evidence for it is immensely greater than the evidence that all, or the vast majority of, people are super calculators uncaring about anything else but current consumption without context considered.

For wages you seem to have a lot stronger and clearer inflexibilities. It's not just that people are highly resistant to a nominal pay cut from their current employers; they're highly resistant to a nominal paycut in their heads! to potential future employers. An unemployed manager who made $110,000/year at a highly specialized job, will be very reluctant to, after losing that job, take a new one paying $50,000 even if that's the best he can find. It's a matter of pride, a lower much lower prestige level than he's grown accustomed to over many years, and just accepting the idea that his earning power is now a lot lower. Eventually he would be forced to accept this, but he may resist and keep searching for like a year, maybe more, or he may retire early while still able to be very productive.

But again too many economists refuse to consider something an explanation if it involves pride, prestige, taking a while to come to grips with something, anything besides only caring about optimizing a super-simple utility function that only considers current consumption with no context (and optimizing it with perfect expertise and public knowledge, with no time or effort cost). As a great example see:

http://chartercities.org/blog/75/rules-and-culture-corruption-in-hong-kong

Another issue is that there is actually a long established and accepted explanation in academic finance for sticky wages: A firm can diversify idiosyncratic risk far better than an employee. And this is especially true if the firm is a corporation. Then, stockholders can really diversify away the idiosyncratic risks by holding a portfolio of the stocks of thousands of firms.

Therefore, a firm can take risk away from the employee in a win-win way by giving the employee a wage that's guaranteed not to change (at least not change downward) for a long contracted period, like a year or more. In other words, a firm can attract an employee with a fixed for a year wage of $70,000 just as much as it can with a highly variable wage, say, with an expected value of $90,000. It's cheaper and better for the firm to do the former, since the firm and its owners can diversify away the risk of variability a lot better than the individual employee can (This comes from the executive compensation literature in corporate finance, so perhaps many macroeconomists aren't familiar with it.).

Now, there are other factors, like motivation from linking the wage to performance within the contract period (as opposed to waiting until a new contract is drawn up after the old one expires for rewarding/punishing), but clearly this risk diversification factor is an incentive for stickier wages.

And, in addition, we get back to "humanity" explanations that too many economists don't consider explanations. They like to assume that people can plan instantly and with perfect expertise, with no time costs and unpleasantness. But in reality, having to constantly re-plan and change your lifestyle is something very unpleasant and time consuming for most employees, so it's another incentive for them to want stickier wages. The firm, on the other hand, can do a lot more pooling, diversifying, and statistical analysis with large numbers, to make this easier, than it is, in sum, for its employees.

In any case, getting back to my first point, mistakes, or not stating something well, happen, especially in casual conversation, which I consider commenting to be. But casual conversation can be a great, fast way, to learn, to work things out. If you're not willing to make mistakes it can be really hard to learn and achieve.

[Richard H. Serlin]

Sorry, the second link in my last comment should be:

http://www.aeaweb.org/articles.php?doi=10.1257/000282805774670392

[Adam P]

Nick, my two cents:

http://canucksanonymous.blogspot.com/2010/07/permanent-vs-temporary-changes-to.html

[Blikktheterrible]

In an earlier comment on this post, I shared my view that it's dangerous to overemphasize the question of why things are happening. I realize that this view seems absurd to many of those who (like me) are driven by a scientist's curiosity about the world. I would guess that a most economists studied economics out of curiosity about how economic systems function. I studied math because of an analogous curiosity. At the same time, I think the curiosity motive, like the profit motive, can lead us into suboptimal outcomes. More specifically, the value of an economic idea is not solely determined by the intellectual satisfaction that it gives. Why? Because economic assumptions drive real decisions in the real world.

I'll try to highlight this point with an absurd scenario. Hopefully, the principle of it carries over into the real world.

Imagine that gdp could be reliably predicted by the reading of entrails. Economists could dismiss entrail-based forecast as lacking micro-foundations or explanatory power, but would that matter? The theory that gdp follows entrail-based predictions would be a valuable one. The Chaiman of the Fed would be able to base monetary policy on the forecast of the entrails and the government could do the same with other economic policies.

[Adam P]

Blik, the problem with your example is that just predicting GDP is not really something econimist a interested in (except those employed by investors, who are interested). In fact econometric models are pretty decent at predicting GDP growth, if you read econbrowser you might have seen that last October Menzie Chin used an econmetric model to predict that the US would return to positive employment growth in April, it was a pretty good predicition.

However, what economists are interested in predicting is the effect on GDP of various policy interventions, some that may never been tried before. Just because reading entrails could let you predict GDP does not imply that, in the case the predicted GDP growth is low, reading entrails also tells you which of a number of policy interventions would be most effective in raising that growth and whether or not the policy will have any adverse effects. Plus you need some way of reasoning through what is the cost of those adverse effects. And if the policy has never been tried before how do you even establish that the entrails will give a reliable prediction, even if they always have in the past?

[Blikktheterrible]

My point is that what we're really looking for is a predictive theory (or mechanism). That's exactly what you just stated when you said, "what economists are interested in predicting is the effect on GDP of various policy interventions, some that may never been tried before." I'm just trying to be explicit in saying that predictive power is what we should be after. Perhaps a better model would be a magic 8 ball that answers our questions about economic policy. Perhaps the 8 ball wouldn't count as a theory, but my point is that we should be less concerned about whether something is a theory and more concerned about whether it can help us answer policy questions.

I think the place where you're disagreeing with me is that you think that we can only ask the "What would happen?" question if we first answer the "Why is this happening?" question. If so, you might be right about that. I don't know the answer to that and I'm not trying to address that question. I'm just trying to be explicit about what we should be looking for (things that help us answer policy questions).

I hope that makes sense.