My own memories of the Phelps/Lucas administration

My own memories are similar to Paul Krugman's, but with one important difference. This is the bit I remember differently:

"The Lucas view took the economics profession by storm – not because there was any solid evidence for it, but because it was so clever, because it led to nice math, because it let macroeconomists give in to their inner neoclassicists."

All of that may be true, but it is also trivial, compared to what is left out. I was bad at math, and had been trained as a Keynesian. What I liked most about the Lucas (or Phelps/Lucas) view was that it seemed to provide an explanation of something I could not otherwise explain. And theoretical views which explain something that cannot otherwise be explained are supposed to take us by storm!

Sure, before Phelps/Lucas we could explain business cycles by assuming sticky prices, but we couldn't explain why prices were sticky. And it's not just that we couldn't explain why prices were sticky; the assumption of sticky prices seemed inherently implausible on the face of it, because they seemed to mean that people would do something that wasn't in their own interests -- leave dollar bills on the table by failing to execute all mutually advantageous trades.

The Phelps/Lucas revolution seemed to fill that gaping hole in macroeconomic theory. Of course it occupied our attention! Damned right it should!

And at the time, it was also not clear whether the Phelps/Lucas view replaced the assumption of sticky prices or whether it provided a theoretical explanation of the apparent stickiness of prices.

Also at the time, I remember thinking that the informational assumption in the Phelps/Lucas island parable (people have full information about the price of their own output, but no information about current aggregate prices) was a bit stretched. But hey, this was an infant theory, and maybe with time we might re-work the parable and it would grow into something more satisfactory. Well, it never did grow tall and strong, but that was not obvious at the time. And it's not even 100% obvious now that some future economist may not figure out a way to re-work it better in the future.

I took the New Keynesian path myself. But despite the very worthy attempts of New Keynesians, we still haven't got a satisfactory explanation of sticky prices. And we need one.

It isn't all about math, aesthetics, and satisfying our inner neo-classicists. It's about trying to explain the world. That's what we are supposed to be doing. No apologies.

[Off topic: and John Updike's "Memories of the Ford Administration" is a book that sticks in my mind too. I strongly recommend anyone who works in a university to read it. How quickly culture changed to the political correctness of the 1990s. And then realise how quickly it can change again. It's all just bubbles and Pet Rocks.]

Comments

[curious]

Krugman:

"But here’s the thing: at this point, the freshwater school no longer remembers any of that – largely because they purged Keynesian and even monetarist thought from their classes."

Would you agree?

[Jon]

Nick, sounds about right.

[Nick Rowe]

curious: I don't know. Sounds plausible. But older (60+) monetarists or proto-monetarists would either laugh cynically or else get very angry to hear Keynesians of all people complain of "purges". But many purges aren't really that. Each of us wants to teach what we think is right, and not teach what we think is wrong. And classroom time and thinking time is scarce. It is always hard for us to recognise our own fallibility and ignorance. And how times change (just like in Updike's book). The Austrians are right.

Jon: thanks.

[Stephen Gordon]

Since I did my MA where you got your PhD (UWO, for those who didn't know), it's not surprising that I too agree with this take. Of course, I went to grad school in the latter part of the 1980's, so we knew already that the Lucas/Phelps approach had already had its failure, in the form of the 1982-83 recession.

What I got out of it was that we had two stories about how how recessions could occur, each of which was based on somewhat dodgy assumptions (*Why* are prices sticky? *Why* is information asymmetric?), and each of which generated the same qualitative conclusions (vertical LR Phillips curve, deviations possible in the SR). The only real difference was about the size of the effect of the shock and how long it would take to get back to the LR trend. I thought that was essentially an empirical question, and so I ended up spending more time on econometrics.

[pushmedia1]

"Well, it never did grow tall and strong"

What do you mean by this? That price stickiness or the information story in particular haven't been studied since the Lucas/Phelps stuff in the 70s?

[Kien]

I did some undergraduate subjects in economics when I was studying law/accounting, and went back to university later in the mid 9s to do a postgraduate diploma in economics (part time). Postgraduate micro was fun, and then I went on to postgraduate macro, thinking, oh good, I can refresh my memory of the IS/LM models and remind myself how the macroeconomy works. I heard nothing at all about IS/LM, and instead learnt about something called the "real business cycle". Wonderful, I thought. This model is going to tell me why we get recessions and what we can/should do about business cycles.

Now, I may have misunderstood (and still misunderstand) what this model is about. But I remember being surprised to learn that the reason why we have business cycles is because of technology shocks. First, I could not see how that accounts for recessions. Do we somehow get negative technology shocks? Also, changes in technology seem to me (at an intuitive level) to follow much longer-term cycles, and it seemed unlikely to explain the more regular shorter term business cycles that we seem to experience. I thought about this, and asked the lecturer (now an eminent professor) whether a more satisfactory explanation of business cycles might be "sentiment". Now looking back, and with the benefit of knowing about this apparent divide between "freshwater" and "saltwater" economists, and the view that academic economists have on "behavioural economics", my question must have seemed either very ignorant or somewhat innocent! I never got a reply from the lecturer to my email query.

I still have a hard-cover copy of Barro's book "Macroeconomics". I am afraid it is very much in a pristine state. I have had no need to open its pages to help me figure out what is going on in the economy, why central bank decides to raise/lower interest rates, why we have long-term unemployment, etc. If there is anyone who would be happy to buy the textbook from me for a nominal sum plus postage cost, please let me know!

[Bruce Wilder]

My background is in industrial organization, and I never had any difficulty understanding how a variety of factors combine to make prices "sticky", and why "sticky" is good in some important ways, but never had much success explaining any of them to macroeconomists. I regarded the 1970? Phelps book as very hopeful since it seemed to recognize many features of the economy I recognized; I didn't understand the argument and didn't really try to understand it, since I was busy, but he seemed to gesture in directions I understood; Lucas always seemed . . . a really bad turn. The "inner neoclassicists" always seemed to wish that prices could be made less "sticky", and Lucas seemed to feed that wish in some weird way.

In any case, when I think of the problems of the economy, 1974-1982, I always think oil had something to do with it.

[Lord]

I guess the counterpoint to why things are sticky is why the changes required are so large. Sticky wouldn't be a problem if the changes required were small, gradual, and continuous, but often the pressure builds up until a market rupture destabilizes everything.

[Barkley Rosser]

Nick,

This discussion misses the point. It is not sticky prices that cause business cycles, although they do allow for unemployment in short-term, static models, hence their appeal. In the GT, while Keynes briefly mentioned sticky prices, the main part of his argument did not assume them. When your problem arises from collapsing bubbles, the problem was the bubbles themselves in the first place, which is too much price flexibility, not too little. There is indeed a large and sophisticated, if widely ignored, literature arguing that sticky prices can actually be more stabilizing then flexible prices, and right now we are seeing evidence for how and why that may be true.

[Nick Rowe]

Stephen: I started my MA at UWO in 1977. I still have my old notes in my office, and plan to look them out and will post here to see what we covered in macro and money (Laidler, Parkin, Howitt, Patinkin, and Fried).

pushmedia1: what I mean by "Well, it never did grow tall and strong", is about the same as Paul Krugman: that we could never get the numbers to come out even roughly right, empirically. Sort of an impossible trinity, trying to get: expectational errors that weren't implausibly big; fluctuations in real output that weren't implausibly small; and persistence of output fluctuations that wasn't implausibly short. Even back of the envelope calculations showed the numbers would always be out by an order of magnitude in one of the 3 desiderata. (Actually, it was tough trying to get even 2 of those desiderata to work.) Despite trying to re-jig the basic idea to get more action. So people just gave up trying.

Kein: I had exactly the same sort of response when I first saw Kydland and Prescott's (1980?) first RBC paper. Stunned incredulity. To their credit though, RBC theorists have been working on those problems. (They don't now need absolute declines in technology (engineers forgetting how to make stuff) to get recessions, for example). And they have been stricter than others about making sure the numbers they apply have some sort of external justification (rather than just choosing parameter values out of thin air to make the theory fit the data). There's probably some grains of truth in what they are doing, but I still don't buy it.

Bruce: Yes, IO-based stories of sticky prices seem worth pursuing. I did one once in the 1980's. Took Lucas' 1972 model, made just 1 firm per island (so monopolistic competition replaced perfect), added customer search costs (to get a kinked demand curve), and it morphed into an extremely Keynesian model. Fun to write, especially as satire, but the results were extremely fragile (not at all robust wrt very minor changes in the assumptions). I didn't see how it was going to grow up tall and strong, so abandoned the infant. Micro stories don't always work out when you aggregate over the whole economy.

Lord: what you are describing is essentially the New-Keynesian approach to monetary policy: since some prices don't want to change quickly, try to control AD so those prices don't need to change quickly. That way we stay in (or near) equilibrium.

Barkley: there is a debate within Keynesianism (as in Keynes' GT Ch 19(?)) over whether increased price flexibility would make things better or worse. But most New Keynesians can't get too excited over debating hypotheticals. "Since prices are in fact sticky, let's work with that, and try to figure out best policy given that fact, and not bother wasting time asking how the economy would behave if they weren't". I think that's the main reason that literature is (mostly) ignored.

[westslope]

Paul Krugman is an unapologetic blowhard.

"In any case, when I think of the problems of the economy, 1974-1982, I always think oil had something to do with it." -Bruce Wilder

Was oil more than just a catalyst? Maybe resoundingly losing the Vietnam War had something to do with it. I seem to recall Americans at each other's throats.

I'm constantly amazed at how Americans have managed to collectively revise the history of the Arab oil embargo. Presumably that helps fuel the popular urge to kill Muslim grandchildren and grandparents.

[Bruce Wilder]

"Micro stories don't always work out when you aggregate over the whole economy."

No, they don't. And, no single Industrial Organization story will ever be a general representation of price formation; an outstanding feature of the economy seen at the grunt's eye-level is its diversity.

But, as you say, "It isn't all about math, aesthetics, and satisfying our inner neo-classicists. It's about trying to explain the world." And, I will tell you that, in the world, many markets do not have a market-clearing equilibrium in price. They just don't. In the world, most observed prices are administered prices. From a great height, without squinting, those prices may look enough like market-equilibrium prices -- I'm not at a great height, so I cannot judge.

It is not just macroeconomists, of course. A lot of economists talk and teach as if they've never been in a supermarket with their eyes open.

For me, the problem with economics has always been its grave reluctance to face, absorb, or respect facts. About the rest, I am unqualified to judge.

[Nick Rowe]

westslope; the oil price rise no doubt had a macroeconomic impact. But that oil price rise might itself have been the (belated) consequence of monetary policy.

Bruce: Agreed. To my eyes, most firms face downward-sloping demand curves, and they pick a point on them by setting a price. But if you build a macro model with monopolistically competitive firms, you find that that, by itself, does not give you price stickiness. In many (but not all) respects, the model still behaves like a classical one. The LRAS curve is at lower output, but is still vertical.

[Alex Plante]

On a related note, you should read "Why Wages Don't Fall during a Recession" by Truman F. Bewley, Harvard University Press, 1999. He essentially went around asking business people why in a recession they prefered to cut jobs instead of wages, and the short answer is that cutting wages would hurt morale (and therefore productivity) much more than cutting jobs.

Sticky prices probably have something to do with sticky wages.

[Bruce Wilder]

"most firms face downward-sloping demand curves, and they pick a point on them by setting a price."

Typically, what they do is choose several prices, in order to price discriminate and capture consumer surplus.

Administered prices are prices held invariant pursuant to some scheme of management control of production and distribution. I would think administered prices are pretty much, by definition, somewhat sticky. Whether that form of "stickiness" is adequately modeled by, say, "menu costs" I couldn't say.

There are other considerations. Many products are dedicated capital equipment or consumer durables, and in typical market structures, their prices are not state-less, but are, instead, heavily path-dependent and/or affected by accumulated stocks.

Questions such as, "*Why* are prices sticky?" and
"*Why* is information asymmetric?" have lengthy descriptive answers, outside of macroeconomics. How factual answers to such questions become someone's "dodgy assumptions" remains a mystery to me.

[Nick Rowe]

Alex:
Suppose the "morale" theory is correct. Why should morale depend on nominal rather than real wages? Or on wages compared to past wages rather than wages compared to other firms' wages? Yes, I know there's a coordination problem here (who cuts wages first?). But once you start modelling that coordination problem explicitly, you generally find the results are very fragile.

Plus, if you believe the morale story, it needs to be built right into the model. Because if you have a model in which there's no morale in the production function, for example, and then bolt on a morale story of sticky wages, there's an internal inconsistency in the model.

Bruce: you can call it administered prices, or you can say they choose prices. But why don't they updated those choices, say, daily, or even hourly?

The dodgy assumptions question really refers back to the internal consistency of the model. If the rest of the model implicitly contradicts the reasoning behind the assumption of sticky prices, there's a problem. All the other predictions of the model may break down in unforeseen ways. Plus, inflation is one of the things we want to and need to explain. Prices do (eventually) adjust. What determines how quickly they adjust under various conditions.

[Patrick]

I just don't think most people think in terms of real wages. All they know is that yesterday they were paid X and today they are being paid less than X, and they're pissed. Unless I had a good pair of running shoes on and clear escape path, I wouldn't want to be the one telling the construction guys I see working across the street that their wage cut is meaningless in real terms!

[Matthew]

"I was bad at math..."

I don't think nay of us believe you were bad at math if you did your PhD in Eco at Western.

[Bruce Wilder]

NR: "you can call it administered prices, or you can say they choose prices. But why don't they updated those choices, say, daily, or even hourly?"

Because the price is just one variable in a system of control. The price is being held constant, pursuant to some scheme of management control, from which there are technical efficiency gains, and in which some other variables are adjusted in response to the vicissitudes of events.

McDonald's manages their restaurants, so that the hamburger costs the posted price, they carry inventories and schedule labor and otherwise operate in a way that meets demand at that price. They vary the price on schedule, for their own purposes. They might have a lower price on Tuesday, say, to price discriminate. But, it's all planned, and maybe advertised, as part of an elaborate scheme to manage demand, production and distribution. The whole operation is very tightly controlled, to minimize error and waste, achieving high levels of technical efficiency and productivity, and holding price fixed, or to scheduled changes, is part of the control scheme.

As a principle of control, it is not different in kind from a central bank holding interest rates fixed, while, or by, varying reserves or open market purchases.

Holding prices fixed can serve many purposes. Most people work for fixed salaries or wages, which, presumably, has to do with asymmetries of information and incentive between managers and their employees, and devising an incentive structure that enables a manager to tell the employee what to do, and have them do it, on time and in the manner prescribed (as opposed to buying a discrete service from a professional). It's hard to imagine a bureaucracy working well, if the wage changed every hour.

Movie theatre tickets are sold in a market, where no market-clearing equilibrium is possible. One can easily imagine varying the price of the movie seat with time of day, quality or popularity of the movie, etc. But, with the constant availability of empty seats at every showing, price competition could get out of hand. So, movie ticket prices are basically fixed (with a price discriminating schedule of available discounts), in agreements between theatre operators and distributors, which also specify how revenues and the cost of advertising and promotion are to be divided. Theatres are left to compete on popcorn and location and things they can manage, while constrained from what would be destructive price competition among themselves.

In the early days of the oil industry, prices varied wildly in the oil patch, until Standard Oil came up with a clever scheme for managing transportation, in way that dampened price. With the volatility of price, and the risks that entailed, out of the way, a giant bureaucracy was built that drove down the unit cost and price of kerosene as relentlessly as it expanded production and distribution.

The iPhone I carry around comes with fixed costs to purchase the phone and associated services, with Apple deriving revenue to recover the cost of the phone from both sources, in agreements with AT&T. I'm sure that the unit cost of manufacturing the phone declines fairly steadily, with the realized scale of production and experience, and the usual technical advances. And, the price of the phone does change, but on a planned schedule.

I don't pretend that any this has any obvious implications for macroeconomic modeling, per se, except, perhaps to emphasize the function of money and credit as "insurance" in the operations of risk-attenuating systems of control.

[Alex Plante]

Nick: When employers cut nominal wages it's perceived by employees as a betrayal. When employees are laid off, those who are laid off may also fell betrayed, but as they don't work for the company anymore, that doesn't matter as much to employee morale as a wage cut would.

Maybe morale should be included in the production function. In today's world most companies compete on the basis of quality of their services and products, as much as price, and in this kind of world any company with a demoralised workforce would soon find itself out of business.

[edeast]

To back up Patrick,
I'm a foreman in a treeplanting company, primary motivation is price, salary is piecework. This year the company downsized, and treeprices were lower. The lower prices did have an effect on morale. The consolation that at least I have a job doesn't help that much. What part is the real wage, tuition and living expenses are all up, so what goes go up if you make less? Is it vs peers?

[edeast]

For, Alex P.
I should add, that production wasn't really affected. Just morale, if that makes sense.

[www.facebook.com/profile.php?id=611985302]

From Nick

Bruce: you can call it administered prices, or you can say they choose prices. But why don't they updated those choices, say, daily, or even hourly?

Did you really write that with a straight face?

[Nick Rowe]

A firm's profit is a function of all the firm's control variables, which includes price. Draw profit as a function of price for each setting of all the other control variables (profit on the y axis, price on the x axis). Draw the envelope of all those curves. The top of that profit curve defines the profit-maximising price. If the firm operates under the simplified assumptions of perfect competition, that profit curve is a spike -- sort of inverted V shaped, only you fall off a cliff if you price above the market-clearing equilibrium. Under more general assumptions (like imperfect competition), the profit curve is smooth, like an inverted U.

The "PAYM insight" (Parkin [yes, Canadian Michael Parkin], Akerlof, Mankiw, Yellen) is that smooth objective functions are nearly flat near the top. (Sort of obvious after you've heard it.) This gives some traction for the small menu cost argument. A firm can be away from the summit of the profit function a first order of small amount, and yet the loss in profit will be only second order of small, and so second order of small menu costs could make it not worthwhile for the firm to change its price.

But when we aggregate over the whole economy, and build all these firms into a macro model, we hit a problem. The labour supply curve just doesn't appear to be very elastic. Put the economy into a recession, with increased unemployment, and it becomes a lot easier to hire labour at any given wage. The inverse-U profit curves tilt drunkenly, and even a small recession will put firms a long way below the new summit.

In other words, we hit exactly the same problem that the Lucas-Phelps view hit. You need large menu costs, or large expectational errors, to get get any sort of serious action.

Jim: yes, I did write that with a straight face. But it's not a rhetorical question. We need to know what determines the speed of price adjustment.

Matthew: when I arrived for my MA, I had never heard of matrix algebra. I just about managed to learn it, but have forgotten it since. UWO, like all economics departments, was less mathematical then. Last week a Carleton TA showed me the final exam of the math review/refresher class he had just taught the new MA students. If I had made no slips, I could have got about 50%.

[Patrick]

A slightly OT general question: In the few academic econ papers I've perused (it would be a stretch to claim I actually read them), it didn't seem that the authors actually 'solved' their models and compared it against reality. I know, experiments are hard to do in macro, but there is historical data to check against (i.e. see if the model can replay history).

Are there people running the models in computer simulation to see what they predicts and how it matches historical data?

[Stephen Gordon]

Of course.

[Nick Rowe]

Yep.

In some cases, economists estimate a model over one sample period, and then see if it still fits the data out of sample.

Another technique (known as calibration, and the RBC people have to be praised for having taken the lead here) is to take parameter estimates from somewhere else, say micro data, plug those numbers into the model, and see if the model fits the macro data without fiddling with the parameters to make it fit.

[Bruce Wilder]

NR: "A firm's profit is a function of all the firm's control variables, which includes price. Draw profit as a function of price for each setting of all the other control variables . . ."

I know that this might be shocking to you, but profit is unlikely to be a mathematical function of price (nor will output be a function of inputs), for any actual firm. The failure to think critically about such fundamentals troubles me.