Scott Sumner  

Wage stickiness as a macro problem

PRINT
Open Borders Meets the Writers... Highlights from the Asness Int...

A commenter recently made these observations:

Working in an office, I still don't see the libertarian position on laying people off versus sticky wages. What I have seen is firms tend to lay people off versus larger wage cuts because:

1) If a firm cuts wages for all workers, then the top workers will simply find another position with company that pays higher. (It might take 18 months.)
2) Firms not only have low marginal productive workers but low marginal productive departments. (Often the department were created to expand the firms core business and are not long term successful.) So it is easier for firms to cut a marginal department altogether versus cutting into productive core business.
3) Firms can measure the marginal productivity of a worker so it is easier to identify ones to lay off.
4) And finally, firms like the subtle message of creating fear with their workers that if you don't produce then you can be laid off.


I see this type of thing all the time. While these comments are interesting, they have very little bearing on wage stickiness as a macro phenomenon. Rather they explain why real wages might be sticky. But the macro problem is nominal wage stickiness, which is an essentially unrelated phenomenon. (Actually, both types of stickiness exist.)

When General Motors lays off lots of workers during a recession, it is not because the wages of GM workers are sticky, it's because the wages of nurses and schoolteachers and accountants and fast food cooks are sticky. Because most non-GM wages are sticky, a 10% fall total labor compensation leads to roughly 10% fewer hours worked in aggregate. That's a recession. During recessions, people tend to buy far fewer cars but roughly the same amount of food and haircuts and ER visits. Thus the sticky wages of nurses and grocery workers and barbers don't cause many of them to lose their jobs, whereas GM workers would lose jobs even if they reduced their nominal hourly wage in proportion to the fall in NGDP.

Thus you can't understand why wage stickiness leads to unemployment by thinking about wage stickiness at specific firms, it is aggregate wage stickiness that matters. (Microfoundations don't help here.) Similarly, it makes almost no difference if the wage demands of the newly unemployed are flexible, the problem they face is that the wages of the nurses and schoolteachers and accountants and fast food workers who still have jobs are sticky, and hence there's no money to hire the unemployed, even if they took a pay cut. If wages of existing workers are sticky, then only higher NGDP will put people back to work.

It's a giant coordination problem. Don't believe me? In 1993 the Mexican price level plunged by 99.9%, almost overnight. And yet there was no immediate upsurge in unemployment. Why not? Because all workers agreed to an immediate 99.9% reduction in nominal hourly wages. But that sort of thing is the exception, not the rule, and in the Mexican case required legislative action. (It was called a "currency reform.")


Comments and Sharing


CATEGORIES: Labor Market




COMMENTS (21 to date)
John Hall writes:

I feel like it took a long time in my economics education before I really grasped this point.

E. Harding writes:

But reducing wage stickiness at GM would also help a lot for keeping employment there high.

"and hence there's no money to hire the unemployed, even if they took a pay cut"

-??????? What??? I don't understand this at all.

Scott Sumner writes:

E. Harding, You said:

"But reducing wage stickiness at GM would also help a lot for keeping employment there high."

No it would not. The basic problem they face is that demand for cars falls sharply in recessions, flexible wages at GM can help, but only a very small amount.

It's wage stickiness at other companies that hurts GM workers.

Michael writes:

"It's wage stickiness at other companies that hurts GM workers."

So, then, to "wage-adjust" out of a recession would require wage cuts for people who are still employed, doing the same work they were doing before, and who aren't really at any significant risk of being laid off?

And they must agree to do this because GM cannot sell enough cars to keep its workers employed (even at reduced wages) if they don't?

Or there is, you know, NGDPLT.

Farid Elwailly writes:

It's interesting that some think demand problems are solved if only wages were not sticky. It's not just that there isn't enough money for wages, there isn't enough money for everything!

The nominal price of everything has to fall for there to be enough money to keep activity on an even keel. Don't think just wages...

NGDPLT solves this problem nicely by reducing the nominal price of everything if that's what's needed due to some shock.

Market Fiscalist writes:

Doesn't it depend upon the elasticity of demand for cars ?

People stop buying so many cars and instead build up their cash balances and that causes (say) NGDP to fall by 10% and demand for cars by 50%.

Car workers take a 10% cut. Demand for cars is very elastic so when everyone else sees that cars are now 10% cheaper in real terms , they buy twice as many - and demand for cars goes back to where it was before.

E. Harding writes:

"The basic problem they face is that demand for cars falls sharply in recessions,"

-Nominal or real?

"flexible wages at GM can help, but only a very small amount"

-Why? Wouldn't flexible labor costs at GM lower the price of cars, thus, increasing real consumption of them? See Market Fiscalist's comment.

ADD writes:

[Comment removed for supplying false email address. Email the webmaster@econlib.org to request restoring this comment and your comment privileges. A valid email address is required to post comments on EconLog and EconTalk. Throwaway or temporary email addresses are not accepted.--Econlib Ed.]

Michael Byrnes writes:

Market Fiscalist wrote:

Doesn't it depend upon the elasticity of demand for cars ?

People stop buying so many cars and instead build up their cash balances and that causes (say) NGDP to fall by 10% and demand for cars by 50%.

Car workers take a 10% cut. Demand for cars is very elastic so when everyone else sees that cars are now 10% cheaper in real terms , they buy twice as many - and demand for cars goes back to where it was before.

I assume car manufacturers cannot retroactively slash the prices of their other inputs (energy, intermediate goods, etc) along with wages. So either workers or manufacturers must take an extra hit.

And it would have to be a cut of greater than 10% to restore sales - if NGDP falls by 10%, and, say, 50% of workers in the economy see no change in their wages or demand for their services (teachers, nurses, doctors, liquor producers/sellers, etc), then to restore sales of final output everyone else would need to take a hit of 20%.

Plus, this assumes perfect information.

Scott, this seems like a really important and fundamental point, central to your macro view, that you haven't really explicitly explained until now. (Probably you have and I just missed it). I hope you post on this again!

Scott Sumner writes:

Michael and Farid, Yes, NGDPLT is the solution.

Market, Yes it does, but only a little bit. If GM wages fall by 10%, the price of GM cars will only fall by 2% or 3%, I'd guess, and that's not enough when demand for car sales is plunging in a deep recession.

E. Harding, Both nominal and real demand falls.

Michael Byrnes, Thanks, I've actually made these points before in several posts, but perhaps not for a while.

Market Fiscalist writes:

@Michael Byrnes

RE: "And it would have to be a cut of greater than 10% to restore sales - if NGDP falls by 10%, and, say, 50% of workers in the economy see no change in their wages or demand for their services (teachers, nurses, doctors, liquor producers/sellers, etc), then to restore sales of final output everyone else would need to take a hit of 20%."

My point is that cannot assume a 20% cut is needed. In your example 50% of industries see a 20% drop in demand at their previous prices.

Lets simplify things by assuming that the car industry is vertically integrated and the only input is labor (they mine their own ore, build their own machines etc) , and that cars rust quickly and the whole output has to be sold in the same period it is produced.

They therefore have to drop prices (and lower wages) until the market clears. As prices drop the real income of half the population rises and the other half falls. But without further information we simply cannot tell how much prices must fall to reach the new equilibrium. Common sense (which may be a bad guide here) would indicate it can't be much since otherwise there would a conflict between Scott's theory and the neutrality of money that indicates that when demand for money increases all other prices should fall equally if we assume price flexibility.

Of course when you add frictions like the fixed costs of already-acquired capital and the fact that car firms will increase inventory and reduce employment rather than cut prices/wages then things change in favor of Scott's theory, but on the other hand the whole 50% of industry takes the whole brunt of a fall in NGDP assumption is itself a huge simplifying assumption since even food sales and healthcare are affected by recessions - and this weakens his case a bit.


Jose Romeu Robazzi writes:

Shouldn't we explore aggregate supply shocks and aggregate demand shocks differently ?

Under aggregate demand shocks, probably money is tight and monetary easing probably can do the job, no relative prices will change meaningfully.

But under supply shocks, things are different. Prof. Sumner has defended the idea that modern economies rarely will show an adverse aggregate supply shock. I tend to agree. But there is the possibility that (important) industry sectors may suffer a negative supply shock. My view is that whenever that happens, we should start with the notion that the entire society got poorer. Relative prices will have to change. I reject the idea of "involuntary unemployment" not because I don't understand the macro wage rigidity, so brilliantly described in this post and comments, but because the word "involuntary" and the proposed solution (boost AD) in effect seem to disguise that there was a negative supply shock, and may fool agents on the idea that things could "be back to normal", i.e. "be back to how they were". Actually, please correct me if I am wrong, but under a sectoral negative supply shock, if the monetary authority succeeds in stabilizing aggregate nominal spending, real spending (and income) will be lower. But don't get me wrong, I do think that constant nominal spending is still a sensible policy proposition to acomodate negative sectoral shocks.

Michael Byrnes writes:

Jose Romeo Robazzi wrote:

"But there is the possibility that (important) industry sectors may suffer a negative supply shock. My view is that whenever that happens, we should start with the notion that the entire society got poorer. Relative prices will have to change."

I think Scott would agree with this (and usually describes an adverse supply shock as a decline in real wealth).

I think the point he is making here is that demand shocks will also drive relative price/output changes - everyone and every industry will not experience a major demand shock, such as 2008/09, in the same way.

If NGDP is on target, then relative price changes can be presumed to be caused by supply effects. If NGDP is way off target, then one can be sure that demand is driving relative price changes. But the intervention in that case is not to worry abut relative prices - it is to bring NGDP back to target and let relative prices do what they will.

James writes:

"hence there's no money to hire the unemployed"

That's only a problem if new hires insist on being paid before they start work, or if their work doesn't increase income to their employers. The former is unlikely. There may be people who are incapable of work which adds to the income of a potential employer but if that's the case, it's their lack of ability that accounts for their unemployment. The wage rigidity story is an epicycle and explains nothing that can't be explained by budget constraints, productivity and reservation wages.

Michael Byrnes writes:

James wrote:

The wage rigidity story is an epicycle and explains nothing that can't be explained by budget constraints, productivity and reservation wages.

Only if you asume perfect knowledge throughout the economy.

Market Fiscalist writes:

In Scott's example there is both a change in the demand for money and a change in the relative demand between 2 classes of goods.

Demand for goods of the first type remains constant while goods of the second type varies with the business cycle and is correlated to demand for money.

If there is some price flexibility then when demand for goods of the second type falls , their price will fall relative to goods of the first type and producers of the first good will become richer since they now get more of these type of goods in exchange for their own.

If there is a CB that adjusts the supply of money to match the supply then relative prices will never fluctuate. Producers of goods of the second type benefit from this policy, while Producers of goods of the first type lose since they have full employment in all situations.

So: In a world with some price flexibility where some goods are more highly correlated to the demand for money than others - is monetary policy in part a welfare scheme for producers of those goods ?

Michael Byrnes writes:

MF wrote:

If there is a CB that adjusts the supply of money to match the supply then relative prices will never fluctuate. Producers of goods of the second type benefit from this policy, while Producers of goods of the first type lose since they have full employment in all situations.

This just isn't true. Relative prices can shift for supply- or demand-side reasons. If the supply of money is adjusted to match the demand to hold money, relative prices will still fluctuate for supply-side reasons.

I don't think we should ant to encourage profit making opportunities based on flucutations in aggregate demand.

Market Fiscalist writes:

'This just isn't true. Relative prices can shift for supply- or demand-side reasons. '

I just meant in my simple model, where I assume that the only thing that changes relative demand is the change in the demand for money. Obviously in the real world loads of things are going to cause relative prices to change.

Michael Byrnes writes:

MF wrote:

I just meant in my simple model, where I assume that the only thing that changes relative demand is the change in the demand for money. Obviously in the real world loads of things are going to cause relative prices to change.

Why on earth would you want to see such price changes happen?

OK, people want new cars, so X number of new cars are made and sold in the US each year. A monetary shock occurs and number of cars sold falls precipitously (or, alternatively, the prices at which cars are sold fall precipitously). Either way, the industry and its employees are in chaos.

I see nothing whatsoever that is beneficial about allowing those price shifts to occur in response to a monetary shock.

It would be different if we were talking about supply effects. Tesla somehow rolls out an electric car that is superior to, and cheaper than, anything Detroit can produce? By all means, let those shock waves roll through the industry. The price of oil rises to $250 per barrel? Or falls to $10? Either of those would send shockwaves through the auto industry, and should.

The Fed persues an overly loose or overly tight monetary policy? That's a different story - I can't see any benefit to allowing either of those to percolate through the economy.

Market Fiscalist writes:

Well, my point is that in the situation described by Scott where some industry are far more affected than others by changes in NGDP, some industries benefit far more than others from monetary policy that stabilizes NGDP. In fact in some situations (and with the right assumptions) participants in industries not affected by changes in NGDP might actually be better off with no monetary policy since they will benefit from falling prices in industries that are affected.

James writes:

Michael Byrnes,

It seems that while you quoted a part of my comment, you didn't bother to read it. If you had, you would have noticed a list of sufficient conditions to explain cyclical unemployment without nominal rigidities.

Comments for this entry have been closed
Return to top