Bryan Caplan

Payroll Tax Hikes, Keynesianism, and the Recession: A Reply to Drum and Krugman

PRINT
Day Break as Social Exp... Econlog Makes WSJ's Top 25...
I recently pointed out that back in the good old days, Krugman would have graciously granted that a payroll tax hike is an especially bad idea when unemployment is high.  In response, Kevin Drum notes that the tax hike doesn't take effect until 2013.  That's news to me, and I thank him for pointing it out.  It's less bad to impose a payroll tax it in 2013 than it is to impose it today.  Still, it's facile for Kevin to remark, "if the recession isn't over by then we've got way bigger things to worry about than a minor increase in payroll tax receipts."  Recoveries take years, and some employers have been known to look ahead a year or two when they decide whether it's worth hiring someone today.

In contrast, Krugman reiterates Kevin's point about timing, then stops making sense:
Actually, it's even worse: Caplan frames the argument in terms of the nasty effects of raising labor costs. Um, we have a problem with demand, not supply; time to reread Keynes on wages.
Um, low demand does not cause businessmen to stop weighing whether another worker's marginal productivity exceeds his wage.  Yes, when demand is low, workers' marginal productivity is effectively lower.  A waiter in a half-empty restaurant brings in less revenue per hour for his employer.  But if the cost of keeping the waiter around in slack times goes up, employers are still going to want fewer waiters around.

Now if you check out Krugman's "Keynes on wages" link, he's making another argument that I already criticized: That wages cuts won't increase employment because they hurt (or at least don't help) aggregate demand.  As I explained before, this is logically possible, but extremely unlikely:

1. Cutting wages increases the quantity of labor demanded.  If labor demand is elastic, total labor income rises as a result of wage cuts. 

2. Even if labor demand is inelastic, moreover, wage cuts reduce labor income by raising employers' income.  So unless employers are unusually likely to put cash under their matresses, wage cuts still boost aggregate demand.
Not convinced by mere theory?  Scott Sumner shows that the experience of the Great Depression strongly contradicts Krugman.  Even when there were tons of idle resources, output sharply fell when labor costs spiked.

In any case, if Krugman were right in theory, than he shouldn't be crowing about the delayed arrival of the payroll tax.  On the contrary, he should want to impose the payroll tax immediately, and make it vastly higher.  I know this sounds like crazy implication to pin on a Keynesian, but it's true. 

How so?  Firms only have to pay the extra tax if they fail to give their employees health insurance.  If the penalty payroll tax were high enough, then, all employers would opt to buy health care for their workers.  And if, like Krugman, you believe that cutting labor costs reduces aggregate demand, you should also believe that increasing labor costs raises aggregate demand.  By this logic, now is a perfect time to make labor more expensive.  Is Krugman ready to bite that bullet?

P.S. In his link, Krugman seems to merely doubt that wage cuts increase aggregate demand.  In a related piece, however, he seems to believe that wage cuts actually decrease aggregate demand: "And soon we may be facing the paradox of wages: workers at any one company can help save their jobs by accepting lower wages, but when employers across the economy cut wages at the same time, the result is higher unemployment."


Comments and Sharing





COMMENTS (31 to date)
Mike writes:

Krugman is a partisan pundit, don't waste time pointing out he is wrong because it also wastes my time when I read it and it is boring.

david writes:
And if, like Krugman, you believe that cutting labor costs reduces aggregate demand, you should also believe that increasing labor costs raises aggregate demand. By this logic, now is a perfect time to make labor more expensive.

Surely Keynesians believe that cutting wages reduces aggregate demand by cutting short-term disposable income; how does making labor more expensive through tax increase short-term disposable income? The taxed money isn't going to wallets directly (and in a Keynesian world, consumers presumably do not immediately adjust spending upwards in anticipation of reduced future healthcare costs).

tom writes:

1. Odd Hominem: Is there a point where engaging Krugman is too ugly? He gets weirdly personally about you. First, he says Feldstein is a good enough economist to know better than to make his arguments, implying that you are not. I think you should take that insult and own it. Second, Krugman assumes that you would intentionally ignore the start date of the increase, when that would make no sense for you to do since it's so easy to refute.

2. Goldi-tax: These insurance taxes and mandates have surely been structured in amount and timing to be the perfect balance of employee/employer income distribution for purposes of stimulating our economy. There's a formula that Krugman would show you if you weren't so dumb and/or evil.

3. Krugman Summary: Caplan is an evil dummy who is not a good economist! Dummy!

Richard writes:

Krugman and other economists that believe in big government seem to be more interested in their own stake in the theoretical economic discussion than reality. Did they really believe TARP would lower unemployment or did they say that to take the intellectual hill which translates into more government grants?
Now that they have been proven wrong on their previous predictions they protect their position on the hill by continuing on this wrong headed tirade when there are simpler free market solutions that have worked before.
They say college molds the personality of a graduate engineer so that he will not spend much time proving he is right but will spend an inordinate amount of time proving he is not wrong. Could this be true for certain economists too?

spencer writes:

Scott is cherry picking data.

While he claims that higher wages drove down industrial production in the third quarter of 1933 the unemployment rate fell from 25% to 20%.

So if he is claiming that higher wages drove down industrial production why did it have just the opposite impact on employment?

It is obvious that Scott can not support his claims about the impact of higher wages on employment using the employment data so he finds some other data series that people who will believe anything blaming FDR on the depression will accept as evidence.

Scott does not prove anything.

Now, can you explain why employment rose sharply in the period when you claim higher wages drove employment down.

The employment data directly crontradicts your claims.

RD writes:

You'll do anything to block "progressive reform." You big meanie

spencer writes:

The economic data is overwhelming.

In the 1930s the correlation between wages and employment was very strongly positive.

When wages rose, employment rose.

When wages fell, employment fell.

I presume that is why Scott Sumner used industrial production rather than employment to make his point
because the employment data did not support his point.

I challenge you do show one single data series that demonstrates that the correlation between wages and employment was negative in the 1930s.

You have no evidence to support your theories.

David R. Henderson writes:

Dear Bryan,
I think this is one of your best posts ever.
Mike said, "Krugman is a partisan pundit, don't waste time pointing out he is wrong because it also wastes my time when I read it and it is boring."
I strongly disagree. Obviously Krugman is a partisan pundit, but he's not just any pundit. He's a Nobel Prize winner with a huge following and so, now that you've lured him into the ring, it's important to deliver the knockout blow, which you have. If Mike doesn't want to waste his time reading it, I think he knows that he has an option he doesn't mention.
Best,
David

AB writes:

"Cutting wages increases the quantity of labor demanded. If labor demand is elastic, total labor income rises as a result of wage cuts."

Chicago's mayor, Richard Daily, seems to currently realize this. In order to deal with Chicago's budget problems he's offering city workers lower pay in the form of unpaid furlough days, or he'll have to start laying people off. Of course it is possible that neither will happen if the union really gets their way.

bill woolsey writes:

Spencer:

Could you point me to the monthly employment data from Jan 1929 until December 1942.

Thank you.

323890234 writes:

@spencer

Isn't industrial production the more relevant metric? Who cares if we have millions of people employed as ditch diggers, to raise the standard of living we need material goods. You can argue that industrial production just benefits the greedy capitalist robber barons, but most factories make goods for the masses. If you own a shoe factory, in the end those shoes are going to be worn by the proletariat.

gnat writes:

"Cutting wages increases the quantity of labor demanded. If labor demand is elastic, total labor income rises as a result of wage cuts."

The statement is theoretically valid only in a partial equilibrium context--e.g., a movement along the labor demand "curve" as a result of a wage decrease--only if product price does not fall. If price and aggregate demand fall as a result of the wage decrease, the demand for labor may fall.

spencer writes:

the monthly data can be found at:
National industrial conference board,
published by G.H. Moore,
Business Cycle Indicators, Vol II

Here is the monthly unemployment rate when Scott Sumner was claiming that a "wage shock" was destroying employment.

Remember his first important time period of falling Industrial production was from
July 1933 to Oct. 1933.

unemployment rate
..........(%)

Jan-33...23.72
Feb-33...24.03
Mar-33...25.36
Apr-33...25.49
May-33...25.59
Jun-33...25.04
Jul-33...23.84
Aug-33...22.45
Sep-33...21.48
Oct-33...21.26
Nov-33...21.29
Dec-33...21.07
Jan-34...18.80
Feb-34...17.90
Mar-34...16.81
Apr-34...16.64

spencer writes:

P.S. This is the unemployment rate that does NOT, repeat NOT, include those employed by the government in "make work" jobs.

If you include those workers the unemployment rate in the mid-1930s the unemployment rate would have been 4 to 6 percentage points lower.

For example, 9.2% in 1937 as compared to the "official" rate of 14.3%.

Remember, wage were rising nicely from 1933 to 1937.

Radford Neal writes:

The economic data is overwhelming... In the 1930s the correlation between wages and employment was very strongly positive... When wages rose, employment rose.

I'm no expert on the economic history of the 1930s, but as a matter of logic, this isn't valid. When people talk about high wages having a bad effect on employment, they are really talking about the causal effect of some policy aimed at increasing wages above the level that would have been set in the market. A positive correlation between wages and employment says nothing about whether such a causal effect exists. Both wages and employment will tend to go up when the economy improves (for most reasons), and go down when the economy gets worse (again, for most reasons). This says nothing about whether employment would have been higher if policies that have the effect of propping up wages had not existed. To determine that, you need to look at what happens when policies change (and even then it's hard to determine, since usually several policies change at about the same time).

El Presidente writes:

Bryan,

Your theory and/or your application is incomplete.

1. Cutting wages increases the quantity of labor demanded. If labor demand is elastic, total labor income rises as a result of wage cuts.

. . . in the long run. How long is that? Production technology changes slowly in response to changes in input prices, especially in an economy like ours where there is a high ratio of capital to labor, and thus a large sunk cost of production. You wouldn't substitute labor for machines you've already purchased because wages declined, would you? Are laborers suddenly cheaper than electricity? That's the marginal decision you would be making in the short run. Why don't we all just mothball our computers and dust off our abacuses? The supposed substitution effect is lagged and represents a step backward as it requires the substitution of less efficient technology so long as demand does not increase. You wouldn't hire more workers to produce more output (with or without machines) unless there was excess demand, would you? And you would only keep additional less efficient workers (e.g. those without their ideal complement of capital) until they could be replaced/augmented with capital or until demand subsided. The supposed price effect is weak.

Krugman is right that you're working on the wrong side of the equation, but that's what neoclassicals are trained to do, so you're following the program. If your analysis was correct, when would we NOT want to cut wages, and what makes wage cuts especially stimulative NOW? You are basically arguing that the multiplier from domestic wages is lower and Krugman is arguing that it is higher. Since trade between the US and the rest of the world has fallen off sharply from its peak and our trade deficit has been decreasing of late, he's becoming more right, and you more wrong, every day.

2. Even if labor demand is inelastic, moreover, wage cuts reduce labor income by raising employers' income. So unless employers are unusually likely to put cash under their matresses, wage cuts still boost aggregate demand.

Who has a higher MPC: somebody making $25,000/yr or somebody making $250,000/yr in the same economy? Distribution matters. I know it doesn't show up in the major teaching texts and our discipline hasn't invested a great deal of effort in figuring out the intricacies, but this simple question is enough to demonstrate the flaw in your suggestion that stimulating demand is best accomplished by giving more money to the people who need it least since there are far fewer of them and they are far more likely to save a greater portion and spend a smaller portion of marginal income, and all the more so as incomes become more disparate. This would, at best, cause increased demand for luxury goods (i.e. the least productive with the smallest market) in the short run, but it would do so at the cost of a sustained decrease in demand for normal goods and a simultaneous increase in demand for inferior goods (negative income effect --> substitution). Again, terrible idea for stimulus. There may be a snowball's chance in hell of making the numbers pencil, but only if increased output is achieved by simultaneously reducing the median standard of living. What sort of pyrrhic victory are you advocating? Why would we want to push toward an even more elongated/polarized income and wealth distribution? Who does that help? The only way your suggestion makes any sense is with regard to investment, which invokes long-run analysis, not short-run. If you want to advocate long-run policy for short-run problems, go right ahead. Just be sure to tell people that's what you're doing.

spencer writes:

OK Bryan.

You keep saying that lower wages lead to higher employment.

My challenge to you still stands, show me a single example of your theory working.

Milton Friedman said the most important test of a theory was did it generate good forecast.

So satisfy this requirement.

David J. Balan writes:

I still think the main point to be made here is that it is perfectly sensible for Krugman to be willing to accept a certain amount of anti-stimulus in exchange for major health reform, and so Bryan's original complaint against him is not valid.

As for the question of the relationship between employment costs and employment in a recession, I would imagine that there must be a huge literature on this. I am totally ignorant of that literature, but it seems like this discussion would be enriched by references to it.

bil. A. writes:

Spencer - you should post data for unemployment rates for the years following the other shocks that Sumner mentions.

You are also neglecting the massive dollar devaluation from around March/April '33 to Feb '34 - the long run effects of which helps explain the fall in unemployment (as the devaluation was greater than the wage increase - so real wages were falling).
-also part of the high wage policy was forcibly reducing the work week (reducing real wages), which likely reduced the # of people unemployed while it could easily have reduced the total hours of employment.

So the story is clear:
-dollar devaluation lowers real wages by ~50% - increasing employment % and increasing production.
-high wage policy increases hourly real wage 20%, decreases weekly hours worked 20% - reduces total hours worked (below what it would have been - the devaluation swamps this, so you may still see increased hours worked), but increases employment % number and DECREASES production.

The big picture is that if you are working more, but producing less, your real wage is lower - so I would expect to see lower unemployment when real wages are lower.

bil. A. writes:

El Presidente -

1)Just to be clear, Bryan is talking about reducing the payroll tax, and hence reducing the cost of labor (in the short run) or increasing after-tax wage rates (in the long run). Deadweight losses are real yes? Decreasing them is good yes?

2)While labor demand is definitely more elastic in the long run, that does not mean it might not be elastic in the short run in some sectors.

3)Labor and capital are complements as well as substitutes, more so in the short run. No, we aren't going to shift from using a guy and a bulldozer to using 100 guys with shovels in the short run. In the short run the marginal cost of using a bulldozer is the wage of the operator (and fuel). At a wage of $50 an hour, it might not be worth building another road, the bulldozer sits in the warehouse. If the cost of operating a bulldozer falls from $50 an hour to $40 an hour due to payroll tax cuts, then there will be more employment and production.
If we are in a sticky above-equilibrium wage situation, reducing the payroll tax is a way of moving down the demand curve and reducing excess supply of labor - of moving towards full employment.

4)Lower labor costs -> greater returns to capital -> higher values for capital assets (see below).

As to your second point:
-The choice isn't one between increasing incomes for the poor versus increasing incomes for the rich - it is between increasing incomes for the poor AND the rich by reducing deadweight loss (from taxation and stickiness) versus not doing so.
-Higher employer income means the marginal firm does not go bankrupt, does not lay off all of its employees - this effect being strongest for labor-intensive firms.
-More than 50% of households own capital stock - so gains to employers means gains to most Americans, not just "the rich". Increased dividend income or higher valued stock portfolios mean healthier household balance sheets - this is a good thing.

Will Wilkinson writes:

Spencer,

An example of lower wages leading to higher employment is the decrease in unemployment during the 1980s when the nominal minimum wage was allowed to stagnate.

Here's Heathcote, Perri, and Violante in a survey of empirical evidence on inequality trends:

For lower-skilled workers, unions and minimum wage laws deflect some of the impact of declining labor demand from prices (wages) to quantities (hours). In the 1970s, when these institutions were particularly strong, declining aggregate demand (the “TFP slowdown”) and declining relative demand for unskilled labor (skill-biased technical change) translated into lower employment for low-skilled men (Figure 7) in addition to lower relative wages (Figure 4). The combined effect was rapid growth in earnings inequality at the bottom. In the 1980s, institutional constraints lost some of their strength as unions weakened with the decline of the manufacturing sector, while the real value of the federal minimum wage was eroded by inflation. Thus in this period, the impact of labor demand shocks at the bottom of the distribution shifted from quantities to prices: wages fell sharply, but hours worked partially recovered. The combined effect was slower growth in earnings inequality. In the 1990s, the real minimum wage stabilized, while aggregate productivity growth recovered. The net effect was broad stability at the bottom of the wage and earnings distributions.

http://www.fperri.net/PAPERS/redusa_jan29.pdf

Does that satisfy you?

Bill Woolsey writes:

Spencer:

I was asking for empoyment numbers, not unemployment rates.

Anyway, the unemployment rate should have fallen to 3%.

Why didn't it? Why didn't the economy recover?
It should have returned to its growth path of the twenties.

Sumner's view is that by leaving the gold standard, the Roosevelt administration created a policy that cause the expected price level to rise. This resulted in a rapid increase in spending. This spending resulted in increases in production and prices (fulfilling the expectation.) The increase in production raised the demand for labor. This increase in the demand for labor expanded employment. The increase in the demand for labor raised wages.

However, the policies of cartelitzation and wage hikes reduced the increase in production and employment, and instead generated higher increases in prices and wages than would have otherwise occured.

The increase in nominal income due to the growing expenditure was more inflation and less output growth More wages, less employment, than otherwise would have occured.

It is possible, of course, that those particular sectors of the economy where the Roosevelt administration administered negative aggregate supply shocks-- cartelization-- employment fell, whereas in the rest of the economy, employment rose. On net, employment rose. But, it didn't rise enough.

Anyway, if a policy results in rising demand, this will result in higher output, higher prices, and higher (nominal) wages, and higher employment.

Given demand, a policy of organizing cartels to raise wages or prices or both, will result in lower quantities demaded for products and labor.

If we combine an increase in demand with a cartelization program, then demand rises and supply falls. The net effect on employment and output is ambiguous, but prices and wages rise.

During the period in question, the increase in demand was greater than the reduction in supply, so quanties rose (and prices and wages rose.)

However, production and employment did not rise enough as shown by the high unemployment rate at the end of the period.

If the cartelization program had not occured, then the increase in prices and wages would have been less, and the increase in output and employment would have been more.

Sumner has shown that industrial production (the sectors where the cartelization was concentrated) actually fell during months where wages were hiked in response to cartelization. I think it would be remarkable if employment in those sectors rose despite the drop in production.

However, that would be consistent with employment rising enough in the noncartelized sectors (which enjoyed growing demand as well) and that growth in employment outstripping the losses.

But the point is that industrial production should have risen in response to the growing demand.

Sumner is a serious economist. He is not a "pundit" making political points about either the Roosevelt administration or else health care reform.

He is very positive about the Roosevelt administration leaving the gold standard and sees that as the key element starting the recovery.

However, the cartelization caused the recovery in spending to be more inflation and less output and employment than it otherwise would have been.

What is so complicated about this?

David C writes:

The criticism towards spencer seems unjustified. Radford Neal, spencer is not asserting causation, Bryan Caplan and Scott Sumner are. Spencer is merely pointing out that the data reflects the exact opposite result of the causation they are asserting. However, your argument that the cost of labor is completely unrelated to the demand for labor seems a rather unusual one.

bil.a and Bill Woolsey, your argument is that other factors are more significant than the cost of labor during the periods in question. Bryan and Scott are arguing that the cost of labor was hurting businesses to a greater extent than other factors were helping them during the four months after the wage increases. Two separate arguments.

Here is the complete unemployment rate for the periods in question:
http://4.bp.blogspot.com/_pMscxxELHEg/SlVu0t0KqXI/AAAAAAAAFxo/rzSqiWrqGuo/s1600-h/DepressionUnemploymentRate.jpg

The first period Scott Sumner refers to is from July of 1933 to some time in 1935. During this period, the unemployment rate is erratic. However, neither increase in wages during this period corresponds to an immediate fall in employment as Scott Sumner asserts. During the second period, the unemployment rate fell overall. Maybe the number of hours worked decreased during both periods, but why would businesses hire more workers while decreasing the hours of the workers they already had on staff? At any rate, the evidence Bryan Caplan presents does not seem to support the argument he is making.a

Sharper writes:

2013? Notice how in the Dem's 10 year plans, the new taxes come right after the elections, while the pork comes right before the elections?

Lance Cahill writes:

Spencer,

The issue is not that unemployment was at a greater level after government policies attempted to enact an implicit price floor for wages, but that recovery was stalled by that policy. No one can deny that employment grew, but that's hardly indicative that current policy is not hampering recovery, or that wage growth is a casual factor of employment growth.

Productivity growth rates, additional liquidity, lower real interest rates, and a growing money supply should have led to a more robust (and quicker) recovery.

Harold Cole and Lee Ohanian even showed that total hours worked did not recover prior to 1940 (even including persons in "make-work" programs):

"Total hours worked per adult, including government employees, were 18% below their 1929 level between 1930-32, but were 23% lower on average during the New Deal (1933-39). Private hours worked were even lower after FDR took office, averaging 27% below their 1929 level, compared to 18% lower between in 1930-32.

Even comparing hours worked at the end of 1930s to those at the beginning of FDR's presidency doesn't paint a picture of recovery. Total hours worked per adult in 1939 remained about 21% below their 1929 level, compared to a decline of 27% in 1933. And it wasn't just work that remained scarce during the New Deal. Per capita consumption did not recover at all, remaining 25% below its trend level throughout the New Deal, and per-capita nonresidential investment averaged about 60% below trend."

http://online.wsj.com/article/SB123353276749137485.html

As other people have commented: the unemployment rate is a poor indicator of recovery, as it does not show how much work is restored, does not indicate how many hours are worked per worker(I don't believe the federal government had a way of estimating "marginally attached" workers at this time), and may include selection-bias regarding the longer-term unemployed.

There is not a want of research regarding the effect of above-trend wages during the 1930s and its effect on recovery. Lee Ohnanian and Harold Cole have done some excellent work to that regard.

There has to be some explanation for why recovery was relatively anemic after 1929. Was it a failure of monetary policy? Well, the base expanded by 100% in the time period of 1930-1939. Was it a matter of too little government spending? Most research has shown the fiscal programs had limited effect, and their limited retrenchment in 1937 could not have led to the large drop (unless you have absolutely insane fiscal multipliers) in industrial production the US suffered in the 1937-8 recession.

I know the Real Business Cycle has somewhat limited appeal outside of the Minneapolis Fed, but their narrative of the troubles the US encountered following the start of the Great Depression is very persuasive, and deserve to be engaged.

Nevertheless, I recall reading a paper included in Edward Prescott's and Timothy Kehoe's book on depressions in the 20th century(http://www.amazon.com/Great-Depressions-Twentieth-Century-Timothy/dp/0978936000) that talked about the decline in real wages (plus increasing productivity) and a recovery in the numbers of hours worked. My recollection may be incorrect, however. So, if you're looking for correlation, that might be a paper you should read.

Bill Woolsey writes:

Lance Cahill:

Caplan was making a ceteris paribus micro argument. The higher wages due to cartelization reduced employment compared to what it would have been.

I think it is true as far as it goes, but as Krugman always points out, micro processes are about relative prices and real wages. I think the explanation needs to be in terms of the real quantity of money and the demand to hold it.

(I think the most likely scenario is that increases in labor costs due to health care mandates will in fact reduce employment amount the unskilled too.)

Sumner makes a historical macro argument. Aggregate demand rose, but the higher wages due
to cartelization had negative impacts on _industrial production._

He does't say that it fell over the entire period.
Just that it fell during particular months due to the wage hikes before returning to increases caused by rising demand.

Sumner assumes that these adverse impacts on industrial production reduced employment in industrial firms. And, by the way, it was industrial firms where cartelization was concentrated.

That leaves plenty of other sectors of the economy where employment could rise during those
months where employment in industrial production would fall. Of course, it is also possible for unemployment to fall in the cartelized sectors despite less output and employment. Perhaps there were more discouraged workers. But, I don't think that is what happened.

Your assertion that Sumner is claiming that the cartelization program resulted in production and employment falling from where it was in 1932 is false. He doesn't say that. He says that leaving gold resulting in growing aggregate demand which led to the increase in output and employment. He claims that looking at monthy industrial production numbers shows a patten of the wage hikes depressing industrial production. He said nothing about total employment falling or unemployment rising anywhere.

My understanding of his argument is that industrial porduction rose with the rising demand, but it had reverses due to periodic jumps in costs. I think he beleives that industrial production was less than it would have been. That is, later increases in demand would have caused industrial production to have risen even more rather than reverse and then exceed the temporary reversals he observed.

I provided a simple aggregate supply and demand analysis. Aggregate demand increases. With no cartelization, output, employment, prices and wages all rise.

Now, combine that with a decrease in aggergate supply, the decrease in aggregate supply being less than the increase in aggregate demand.

Output and employment still rise, but less they they did in the previous scenario. And prices and wages still rise, but more than in the previous scenario.

We know that output and employment had not recovered to full employment levels in 1935 or
1936. And so, a policy that restricted the increase in output was undesirable in returning the economy to full employment.

It may be, of course, that the increase in aggregate demand was inadequate. And so, even without the cartelization policy, the increase in output and employment would have left the economy producing below capacity. Still, without the cartelization policy, it would have been closer than with it.

The debate between Krugman and Sumner is that Sumner claims that leaving gold raised aggregate demand. Krugman is grasping at straws--claiming that the cost push inflation made real interest rates negative and increased expenditure. Sumner argues that the historical record shows that that the gold policy already increased expected inflation and generated a strong recovery in spending. As the wage hikes were implemented, the inceases in industrial production were partly reversed. Sumner believes that without those reversals, industrial employment would have risen more.

Sumner's complaint with Krugman (over and over) is that Krugman refused to recognize monetary policies that can beat a liquidity trap. In particular, with a gold standard, devaluation will do the trick.

Bill Woolsey writes:

Bryan:

Here is how nominal wage cuts expand aggregate employment. (I assume you undestand this.)

Nominal wages fall. Nominal costs fall. Prices fall. The real quantity of money rises. If amount of real money balances people wanted to hold was in equilibrium with the previous real quantity of money, then excess money is spent, and real expenditures rise. The real volume of goods firms can sell increases. Firms expand production. The derived demand for labor rises, so employment expands.

Real wages and realtive prices aren't part of the process. Nothing need happen to the distribution of income between capital and labor.

Krugman's arguments are generally in bad faith because this is the market process by which the real supply of money adjusts to the real demand for money.

So, where do real wages fit in?

Suppose there is shortage of money. The demand to hold money is greater than the real quantity. This could be caused by an increase in demand or a reduction in the nominal quantity.

The shortage of money is matched by a surplus of goods and services. If it is assumed that goods prices are flexible but nomimal wages are sticky (at least downwardly,) then the surpluses of goods lower their prices, but wages don't fall in proportion. Real wages rise and real profits fall.

The price level does not fall enough for the real quantity of money to adjust to the demand to hold money. The sticky nominal wages and so sticky costs keep prices from falling enough. But the firms are in equilibrium. The higher real wages and lower real profits mean that firms are maximizing profit (or minimizing loss) at a lower level of output and lower prices. Until wages drop too, the market process that adjusts the real quantity of money to the demand to hold it is stymied.

The lower real output and income lowers the demand to hold money eough, so that when combined with the increase in the real quantity of money, due to the lower prices, results in equilibrium between the real quantity of money and demand.

Of course, there is a surplus of labor. If nominal wages fall, then prices fall too, but less than in proportion. So, nominal wages fall, real wages fall, prices fall, and real profits rise. The lower prices increases the real quantity of money creating an excess supply of money at the low real demand due to the low real income and output. Real expenditures rise. Firms sell more. They produce more. And they hire more labor. This is profitable because real wages fell and real profits rose. However, the increase in real money balances due to the lower prices is what generates the increase in real expenditure.

If you go though the exercise assuming that all prices are sticky in equal measure, and that it isn't just wages, then the shortage of money results in a surplus of goods. Firms cut production to clear markets at those prices. Producing less, they hire less. Employment falls. With all prices being equally sticky (wages and prices, that is,) then nothing happens to real wages. Wage income falls with employment and profits fall with production, of course. But nothing happens to real wages.

The firms are not profit maximizing. They should be cutting prices in the face of the surpluses rather than just cutting production. This may not be realistic, but it helps show that the problem isn't really about real wages, it is about an inadquate real quantity of money and that at least some prices (here including nominal wages) fail to adjust enough.

I really believe that Keynesians (from Keynes on, including now Krugman,) know all of this. And they do often make arguments about how this can go wrong. But an excessive "micro" explanation results in them making various debating points that are red herrings.

Do the thought experiment where it is product prices that are sticky, (say with price floors,) while wages are flexible. Will lower nominal and real wages help? Well, maybe there will be changes in production techniques so that more labor and less capital is utilized. But this would hardly be a desirable solution. Reducing the productive capacity of the economy until it is equal to the level of depressed output that keeps the demand for money equal to the real quantity determined by the price floors and the nominal quantity? Sure, evenutally the entire labor force could be utilized in primative production techniques that would keep them all busy producing the decreased level of output. Great.

Given monetary equilibrium and aggregate market clearing, the micro-analysis of real wages is exactly correct. If there is monetary disequilibrium, then that is the problem--the monetary disequilibrium.

Of course, your point that raising labor costs by mandating benefits will reduce employment is correct--ceteris paribus. Ignoring the red herrings, Krugman's argument is that raising wages and prices causes inflation. While the higher price and cost level depresses output, the expectations of rising prices raises demand now. It is simply the point about the equilibration process is about the level of prices with the movements in prices having the opposite effect.

High prices dampen real demand and low prices raise. Rising prices raise real demand, and falling prices dampen it.


Bob Roddis writes:

Moderation in the pursuit of Krugman is no virtue.

El Presidente writes:

bil. A

If we are in a sticky above-equilibrium wage situation . . .

I thought Bryan was agnostic about this (for good reason), unless he's changed his position or I misunderstood. If he isn't, I am. I understand the theory, but the thing underpinning the downward rigidity of nominal wages is demand. If demand falls off a cliff and wealth plummets, wages become more flexible as unemployment increases. I think we can safely say that the prospect of prolonged unemployment is sufficiently ominous to affect wages at this point.

The persistent deadweight loss argument is fraught with peril. One must assume that the 'loss', which is actually a substitution for government purchases and transfer payments, is less efficient than the specific foregone production. We are substituting government purchases and transfers for private purchases. We aren't burning the money we collect in taxes or destroying capital as a means of exacting revenge on rich people. There is no inherent loss of value from the tax revenue; just an alternative allocation of resources. Deadweight loss is real, but the negative impact comes either from less productive allocation of resources (lower utility from the government purchases and transfers) or the idling and depreciation of existing capital due to a shock. I don't think the first cause is a given, and the second cause is temporary. Once depreciation consumes the idled capital, the market reaches a new equilibrium and the deadweight loss is no longer the mathematical albatross we might initially consider it to be. If this deadweight loss is brought on not by an increase in taxes but by a decrease in demand, then we can hardly hang it around the neck of the government. Otherwise we'd be saying that ALL taxation must be eliminated to eliminate deadweight loss; that we actually prefer zero government spending. We might argue that we receive less utility, but that would require an aggregate utility function.

I understand full well that there is a strong and persistent dogmatic bias which says that, no matter what, a dollar spent by the government is less productive than a dollar spent by the private sector, but it simply isn't true. I sign off on government contracts every day. For the most part, we buy the same products from the same vendors for the same or lower price as the private sector does (larger purchasing power drives some costs down), and we do it with minimal overhead and zero profit margin for the agency acting as contract administrator. If there is a difference, it isn't so much on the cost side as on the benefit side. And I can't begin to tell you how sorely people miss things like trash collection, serviceable roads, and public safety services when they think they aren't getting them. I'd say the vast majority of 'government' produces great benefit, even when it isn't fully captured in GDP.

2)While labor demand is definitely more elastic in the long run, that does not mean it might not be elastic in the short run in some sectors.

Good point. I'm not sure we can be certain there is a net positive effect, but I'm open to the argument.

The choice isn't one between increasing incomes for the poor versus increasing incomes for the rich - it is between increasing incomes for the poor AND the rich by reducing deadweight loss (from taxation and stickiness) versus not doing so.

Again, the deadweight loss argument presumes that upward redistribution reduces reengages idled capital by making labor cheaper moreso than downward redistribution would through increasing consumer demand. Bryan was very clear that he advocated a reduction in the employers' contribution to payroll taxes as a means of increasing after-tax earnings while decreasing wages. That was his express intent. Arnold went further by riffing on this idea and suggesting that what we really needed was profit-support. That's right. He wanted a way for government to transfer money from low-income individuals to high-income individuals (shifting the relative shares of tax/debt burden) so that profit margins and after-tax earnings would be sustained and the stock market would rally. I'm not making this up. I wish I was.

Stock values rising is good, if it reflects underlying health in the broader economy. If it is merely a form of forced saving with no security, then I think your deadweight loss concerns should be applied rigorously here as well. If individual choice is the driving moral principle behind your arguments, then forcing people to trust the employers of our economy to get them more utility than they could get on their own ought to be at least as repugnant as government spending is made out to be by the regulars on this blog. What's more, it isn't likely to work.

As Krugman rightly pointed out, the problem is being driven by the demand side of the equation more than the supply side. We had a tremendous bubble followed by a pop that has resulted in a negative wealth effect. If capital is idle, adding additional capital doesn't help. Without additional demand, that's what would result from cutting the employers' contribution. The snowball's chance in hell for Bryan's argument would depend on dramatically increasing foreign demand for domestic exports. I don't think that is likely in the near future. Alternatively, he may be advocating an increase in labor-intensive production. But, that doesn't really sound like Bryan, to me.

Lance Cahill writes:

Bill:

I'm not sure if you are addressing me, but I don't believe I made any claims asserting that Sumner's assertions were false.

Commentator writes:

Um, basic supply & demand economics explains unemployment. When price is higher than the market clearing price, supply (labor) will increase, while demand (for labor) decreases. Because supply is exceeding demand, price has to be lowered to clear the market. In the case of the great depression, the labor market eventually cleared because the government - the central economic planner they are - eliminated supply (labor) via world war. And so you have it, unemployment was 'literally' eliminated.

Comments for this entry have been closed
Return to top