David R. Henderson  

Krugman's Priceless Economics

The last New Keynesian left al... Scoring Your Ukraine Predictio...

As regular readers of my posts on Econlog know, although I am often critical of Paul Krugman, I defend him when he's doing good economics (here, for example). His New York Times column yesterday, though, "Walmart's Visible Hand," essentially throws out basic price theory. Thus the title of this post.

The context is that Walmart has announced that it will raise wages for 500,000 employees. He uses this fact to reach this conclusion, which he states up front:

Second, and arguably far more important, is what Walmart's move tells us -- namely, that low wages are a political choice, and we can and should choose differently.

Already, that's an interesting statement for an economist to make: the fact that a private employer raises wages tells us the "low wages are a political choice." No, it doesn't tell us that. It tells us simply that a private employer raised wages. To establish that Walmart raised wages for political reasons, Krugman would have to make that case. Instead he moves on because he has bigger fish to fry.

Krugman challenges the idea that wages are set by supply and demand. After stating that conservatives believe this, he admits (he even uses the word "admit") that so do many economists.

But he doesn't think that's true. He writes:

Specifically, this view [that wages are set by supply and demand] implies that any attempt to push up wages will either fail or have bad consequences. Setting a minimum wage, it's claimed, will reduce employment and create a labor surplus, the same way attempts to put floors under the prices of agricultural commodities used to lead to butter mountains, wine lakes and so on. Pressuring employers to pay more, or encouraging workers to organize into unions, will have the same effect.

But labor economists have long questioned this view. Soylent Green -- I mean, the labor force -- is people. And because workers are people, wages are not, in fact, like the price of butter, and how much workers are paid depends as much on social forces and political power as it does on simple supply and demand.

Actually, this view not imply that "any attempt to push up wages will either fail or have bad consequences." We economists who believe that wages are set by supply and demand also believe that if the supply falls (shifts left on a graph of supply and demand) or if demand increases (shifts right on a graph of supply and demand), then an increase in wages will have good consequences. And that, after all, is the context for Krugman's article: not a government attempt to raise wages but an employer's decision to do so.

Also, wages are like the price of butter. This has nothing to do with the fact that the workers are people. Remember that the workers are selling their services--their hours--not themselves.

Then we get where he really wants to go: advocating higher minimum wages and more bargaining power (although he does not name the specific changes in law that he wants) for labor unions.

First, on minimum wages, Krugman, as is his wont, emphasizes only one part of the literature, the part that finds "little or no negative effect on employment" and treats that as an "overwhelming conclusion" of the so-called "natural experiments" with the minimum wage, rather than as the conclusion of one part of the natural experiment literature.

Second, on unions, Krugman argues that the "sharp increase in unionization" after World War II reduced income disparities. But here's where he is most "priceless." The bottom line of the literature on the effects of unions and wages, most of the literature coming from the 1950s and 1960s, and much of it summarized by the "dean of labor economics," H. Gregg Lewis of the University of Chicago, was that unionization in the one quarter of the labor force that was unionized raised wages for those workers by about 10 to 15%. But, of course, with higher wages, the amount of labor demanded fell. Where did these workers go? They went to the non-union sector. That shifted out the supply of labor there. So wages for the 75% of workers who were not unionized were 3 to 4% lower. The only way this would reduce income inequality is if the 25% of the workers who were in unions would otherwise have had wages below the wages of the 75% not in unions. Unfortunately, many of the non-union workers had lower wages to begin with. So the effect of unions on income inequality was, at best, not clear.

Krugman concludes:

What this means, in turn, is that engineering a significant pay raise for tens of millions of Americans would almost surely be much easier than conventional wisdom suggests. Raise minimum wages by a substantial amount; make it easier for workers to organize, increasing their bargaining power; direct monetary and fiscal policy toward full employment, as opposed to keeping the economy depressed out of fear that we'll suddenly turn into Weimar Germany. It's not a hard list to implement -- and if we did these things we could make major strides back toward the kind of society most of us want to live in.


Comments and Sharing

COMMENTS (36 to date)
Andrew_FL writes:

If wages are not set by supply and demand, what are we to make of the Keynesian theory of unemployment due to sticky wages? Surely the phrase "the wages get stuck and don't clear the market" is meaningless?

kingstu writes:

Mandating a minimum wage has real world consequences. While those still employed may be 10% better off, those who become unemployed are 100% worse off...


E. Harding writes:

Let's see what impact Krugmanism had during the Great Depression:
Oh, right! A two-year stagnation in industrial production. Krugmanism FTW!

fralupo writes:


And because workers are people, wages are not, in fact, like the price of butter, and how much workers are paid depends as much on social forces and political power as it does on simple supply and demand.


Also, wages are like the price of butter. This has nothing to do with the fact that the workers are people. Remember that the workers are selling their services--their hours--not themselves.

I'd go the the way the other direction, actually. All prices are socially determined because they are either set by people or by machines programmed by people. How else are you going to get curious results like sticky prices? That doesn't mean that deviating from a supply and demand ("simple" or complex) doesn't bring a cost, just that it happens.

ThomasH writes:

Not a particularly insightful column, I agree.

In terms of the argument on minimum wages (only one of the items on his list of recommendations) I take it he is saying that the actual proposal is so close to where the market is going, that the elasticity of employment wrt to the minimum wage increase is pretty small (and so will the effect on incomes of those who are employed). But he, unfortunately, dose not seek to teach his readers anything about microeconomics, here. If he did, he'd presumably argue that an EITC would be better.

vikingvista writes:

Yeah. It's time economics left the marginal theory of value on the waste heap of history along with the labor theory of value, to make way for Krugman's political theory of value. Central planning never seemed so sensible.

Tom Nagle writes:

It is interesting that no one has referred to what Walmart has said about its reason to raise wages: that with the economic recovery, its labor turnover has increased and that has hurt customer service. The president of Walmart demanded that his organization reverse the decline in service, which led to the decision to increase wages above the minimum to improve retention of people who have already been trained to do their jobs. This sounds like supply and demand to me.

JLV writes:

Dinardo, Fortin and Lemieux (1996) answers the "uncertainty" about union effects. Referencing literature from a time when regressions were computed on punch cards is cute, but unpersuasive (and arguably a non-sequitor).

And on minimum wages, I think the consensus of the natural experiments literature really is that we can rule out large disemployment effects.

I just don't see how Krugman is being unfair here. He references solid empirical evidence, and the response is essentially "but theory says....."

Don Boudreaux writes:

David is absolutely correct: that column by Krugman awful. It's anti-economics. It's written as though its author is unfamiliar with the introductory chapters of a principles-of-microeconomics textbook. And I'm not talking here about the conflict between the theoretical prediction that higher minimum wages cause job losses for some low-skilled workers and some empirical findings to the contrary. I'm talking instead of statements by Krugman such as this one: "and how much workers are paid depends as much on social forces and political power as it does on simple supply and demand."

What the heck does Krugman mean by "simple supply and demand"? No economist of any merit has ever taught that supply and demand are context-less, free-floating presences that are independent of the likes of "social forces and political power." For 33 years I've taught my principles students that, for example, occupational-licensing regulations (the product of political power) reduce the supply of workers in licensed occupations. Similarly, I explain that the strengthening of the social norm against smoking cigarettes (an instance of social forces) has reduced the demand for cigarettes.

But neither social forces nor political power - in the form of minimum-wage legislation - are likely to make N workers each more productive. And neither social forces nor political power can long ensure that employers who pay workers more than the value of their marginal products will remain in business.

Social forces and political power are real, and they affect supply and demand - but they do not affect supply and demand always, or even usually, in the ways that interventionists would like.

In short, to say that social forces and political power affect supply and demand and, hence, equilibrium wages is true. But it does not follow that any particular social force or any specific unsheathing of political power will bring about those happy equilibrium conditions that are intended by those people who fuel the social force or who wield the political power.

In the end, prices and wages are determined by supply and demand - a brilliant tool quite capable showing the influence on prices and wages of social forces and political power. Krugman, though, seems to suggest that social forces and political power play a role independently of supply and demand in setting prices and wages. That's anti-economics. That's the kind of talk one can excuse in a taxi driver or podiatrist who's never taken a course in economics or read a book in the subject. But it's a shame to hear it come from a professional economist.

Jon Murphy writes:

Dear God...There are no words.

If Krugman wrote that in my class, he would have failed.

Jeff writes:

It's even worse than that. Here are two consecutive sentences from Krugman's piece:

Paying workers better will lead to reduced turnover, better morale and higher productivity.

What this means, in turn, is that engineering a significant pay raise for tens of millions of Americans would almost surely be much easier than conventional wisdom suggests.

Granted, the first sentence ends a paragraph and the next starts one, but it sure looks to me like Krugman is trying to imply, without explicitly saying so, that raising everybody's wages will lead to reduced turnover and higher productivity.

If workers are paid their marginal product, then forcing wages higher will improve overall productivity because the least productive workers will be fired. Of course that's not what Krugman wants you to think. Rather, he's trying to get you to commit the fallacy of composition by thinking what works for Walmart will work for every business. But the reason it works for Walmart is that they are the only ones doing it, so they gain an advantage in the competition for good workers. That would not be true if everyone raised wages.

David R. Henderson writes:

Good point about the union studies and the limits of the time. I took a quick look at the article. It’s not one that I can judge with half an hour of looking. For those who want to look, it’s here.

Thad writes:

If wages aren't really set to supply and demand, then how can there be spillovers like he said at the beginning of his column?

Hazel Meade writes:

Krugman, as is his wont, emphasizes only one part of the literature, the part that finds "little or no negative effect on employment" and treats that as an "overwhelming conclusion" of the so-called "natural experiments" with the minimum wage, rather than as the conclusion of one part of the natural experiment literature.

There's a term for this sort of thinking.
I believe it is called "epistemic closure".

Ray Lopez writes:

Sigh, allow Ray to step in and clarify things. I am not an economist but I feel like one sometimes.

Krugman is right. Henderson is also right. It depends on how much "market power" (Google this) Walmart has. If WMT has lots of market power, like Krugman assumes, then Krugman is right. If not, then Henderson is right.

Ray Lopez writes:

Followup to my post: Ford Motor company in the early Model T days also had market power (due to a combination of better technology via River Rouge integration, lower prices, brand recognition, etc) Hence it could afford to pay workers more, which had several beneficial virtuous circle positive externality spillovers: it allowed lower turnover at Ford plants, it allowed workers the ability to purchase Ford autos, and it spilled over to the local economy, growing it, and drawing even more workers from the Deep South which made expansion easier for all Detroit auto makers and Ford too (driving out Ford's rivals). But it would not have been possible if Ford was a commodity auto producer, and indeed would have made no sense (Ford would be wasting their money on a form of charity, nothing more, since in this scenario Ford would be just a drop in the bucket).

JLV writes:

So here's a late addition, expanding on a previous comment. Many people are probably thinking of surveys like Neumark and Wascher (2007) when they think of literature that shows that minimum wages reduce employment. At that point, there were a fair number of studies, a la Card and Krueger that failed to find disemployment effects, and a lot of studies that showed modest and statistically significant disemployment effects (elasticities around -0.2).

I wouldn't say that's the current state of the literature. I'm thinking here of papers by Dube and co-authors (Allegretto, Dube, and Reich (2011) is the best example) that critique the econometric technique that produces disemployment effects. It boils down to this: allowing for within-state clustering in standard errors and controlling for state or region specific trends undoes the disemployment effect result (relative to the "canonical" two way fixed effects model).

This is beside the point. Its unclear if Krugman or Henderson has read the recent literature, but I'll give them the benefit of the doubt. Not mentioning the studies using the "canonical" model might be due to bias. But it is also consistent with recent developments in the literature. Why mention studies we now know are mis-specified?

Ray Lopez writes:

@JLV - "Its unclear if Krugman or Henderson has read the recent literature..." true, but it's unclear whether any of this literature is dispositive on the issue. "Controlling for state or region specific trends" will gut any trend that shows positive externalities from a market involving oligopolies (like Detroit and the Big 3), see my previous comment. It reminds me of an econ paper that made the claim mechanical refrigeration was no big improvement in quality of life over iceboxes. Probably true, as primitive man did fine without ice cream, and having a fridge probably contributes to obesity to a degree. Controlling for everything, 'primitive' man probably led a better life than sedentary civilized western man, assuming both survived into middle age, and I think skeletal studies show this is true in North America.

Jon Murphy writes:


Your explanation is wrong. Market power doesn't have anything to do with it. Rather, market forces does.

All businesses, even monopolies, must contend with market forces. Wal-Mart is raising wages because they have to, not because they want to. The labor market is tightening and now, in order to attract the talent they need, they need to pay more. Simple supply and demand analysis.

[comment edited with permission of commenter--Econlib Ed.]

Rick T. writes:

I know I am not up to date on the latest studies, or most of the early ones either. But before spending any time on them, I wonder how many years are they studying between an increase in minimum wage in one place, compared with some other place, and when they feel they have counted all the effects?

Economists, like most would-be scientists, like to end a study early so they can write their papers and move on. Unfortunately, the effects of most economic events that one might study play out over time, sometimes many years.

For example, initially an increase in minimum wages that results in a small retailer having to pay employees more than their economic value may not cause any layoffs. Why? Because the situation takes time to unfold. The store may be down to a skeleton staff and just can't lay anyone off and still stay in business. Why not go out of business? The store owner may be losing money and want to, but still has two years left on the lease which the owner has personally guaranteed. If the business closes, the owner still owes the remaining balance on the lease, so may consider lease expense a sunk cost. Staying in business at least brings in some revenue, and although losing money the owner is still better off staying open than giving up. When the lease ends, the business closes, and that is when all the employees lose their jobs.

My question is, do the studies go out more than two years and count those layoffs when the lease ends as something chalked up against the minimum wage increase? In theory it should, but then again, the longer the time period of the study, the harder it is to be sure that any differences in the employment picture of one place where the minimum wage went up and some other place where it didn't were caused by the minimum wage change, and not some of the thousands of other economic influences that cause one place to do better or worse than some other place.

This is why it is a waste of time to read most economic studies of this sort. They have no way of getting a correct handle on cause and effect because, done honestly and thoroughly, that is always too complex for their studies to consider without making the paper unpublishable.

At least opponents of minimum wages have good reasons why no business wants to pay an employee more than the employee adds in value. What reasons do proponents have why employers won't mind keeping employees that lose them money?

GaelanClark writes:

Okay, let me get this straight....Krugman insists that "raising the minimum wage"----INFLATION----and "setting monetary and fiscal policy to support full-time employment"----ABOLISH OBAMACARE----are our paths to success!

He has one right, even if he doesn't understand that that's what he actually means. (Hint....Obamacare)

Jon Murphy writes:

Even if we accept Krugman's explanation as correct, it still doesn't make sense given context:

The US economy is expanding.

US real wages are rising (nominal wages are outpacing inflation).

Job openings are at the highest level since the recession.

The Labor Supply/Demand Rate is at pre-recession lows (a good thing; that means there are fewer unemployed people per job opening).

I can go on, but all statistics indicate that the labor market is tightening, a period where "simple" economic analysis indicates wages would naturally rise.

If Wal-Mart's wage hike came in a period where rising wages were not to be expected (say, the depths of a recession), then ok, I could at least understand Krugman's argument. But this is a period where "simple" supply and demand analysis says wages should rise and, lo and behold, they are! It seems to me Krugman is merely trying to post hoc justify his biases rather than base it on any analysis.

David R. Henderson writes:

Why mention studies we now know are mis-specified?
Because we don’t know that they’re mis-specified. In a review of this literature, Robert Murphy writes:
What we should note is their [Neumark and Salas’s] general warning about Dube et al.'s desire to control for "spatial heterogeneity." Dube himself unwittingly demonstrates the huge potential pitfall when he argues, "[S]tates that have raised the minimum wage more over the past several decades are systematically different from other states." Then, as one example, he says that "they have experienced... sharper reduction in jobs involving routine tasks." Since this type of "state-specific trend" is quite plausibly the result of a state government raising its minimum wage relative to its peers, including these types of "controls" in the regression might mask the policy's true effect.

Cyril Morong writes:

Christina Romer wrote a critical article about minimum wage laws in the NY Times about 2 years ago, saying, among other things, that they can hurt the people they are supposed to help.


Here is something by Neumark from 2013 that refutes Dube


Denis Drew writes:

I think what Krugman means by "political" -- because it is what I would mean -- is a very general outlook that our labor market has accidentally and arbitrarily if you will slipped into an unnecessarily exploitative mode -- to wit:
[cut and paste]
If labor (and ownership) all over the economy were capable of extracting the max the consumer is willing to pay (as a practical matter labor can only do this through centralized bargaining I think), then, the spectrum of the physical output (including services) of the economy will be would be determined purely by consumer choice. (It's the market.)

Currently, in my outlook, America has a two tier market (as in de-unionized) where the context of physical output is distorted by labor not extracting that max: leaving many goods more price attractive than they would otherwise have been — leading to more demand for those cheaper products or more demand for other products with the left over money (how many Big Macs can we consume).

The end result seems not to be totaled up locally — at the original mass consumer level — rather the cash that is squeezed out of the bottom travels THROUGH (?) the middle and ends up on the top. If you squeeze a toothpaste tube on the bottom, it all comes out the top – where high end real estate prices escalate, high end services are in demand and sail boats are on sale.

PS. Up until recently too many economists in my observation would treat a rise in labor price as an equivalent rise in the whole product price — w/o considering that it was only a fraction. If Walmart's average wage rises to $100 an hour, prices only go up 50% :-) That happily seems a thing of the past as all get serious about the labor market.

Denis Drew writes:

I don't know what kind of specific labor law changes Krugman would like to see -- here's my latest favorite way to go:

[cut and paste]
The ultimate – federally prosecutable -- sweetheart labor contract may be no contract at all.

No one would doubt the criminality of a mob union boss and/or an employer threatening to fire workers for speaking out against a mobbed-up sweetheart contract – in order to obtain for themselves the pay and benefit moneys that might otherwise have gone to employees through fair bargaining practices. A for certain RICO or Hobbs Act target.

Why shouldn't the exact same extortionate activity be viewed in the exact same extortionate light when union busting “consultants” and ownership threaten to strip away workers' economic livelihoods should they dare to participate in a federally approved path to establish federally approved union bargaining rights?
US Attorneys -- Criminal Resource Manual 2403

B.B. writes:

I think Krugman (like Stiglitz and Sachs) is attached to the old "Cambridge" view of the labor market. Cambridge UK, that is.

In that view, the wage rate is the keystone of the system. It is the nominal wage, not money supply, that determines the price level. The setting of the wage rate is about class conflict and social norms and political power. Supply of labor is wage-inelastic, and the demand for labor is wage-inelastic. The production function is Ricardian fixed coefficients of production. There is little substitution for a given technology. Moreover, monopoly power and monopsony power are commonplace.

Cambridge accepted that if the nominal wage grew faster than productivity, the result was price inflation. The policy response is an "incomes policy," in which the government guided wage increases to keep inflation on targets. The policy kept real wages rising with productivity, which kept the labor share of income fairly steady. Monetary policy's job was to keep interest rates low and help finance the budget deficit. Fiscal policy's job was to manage aggregate demand to keep full employment.

These views were close to Keynes and his followers, Joan Robinson, Sraffa, Pasinetti, and in the US, with a lot of the Harvard crowd of the 1940s and 1950s, and it was the view of Galbraith in the 1960s. Samuelson flirted with these notions.

They are not my views at all. But there is a tradition behind Krugman's leftism. I wish Krugman would just rework his models into the old framework and give up on modern analysis. He wants to put old wine into new skins.

Ray Lopez writes:

[Comment removed for rudeness.--Econlib Ed.]

Roger writes:

As a non-economist, I wish you guys would establish the proper framework for the argument. If you frame it wrong, you can get oversimplified answers that lead policy makers astray.

The issue is not simply how much effect the minimum wage has on unemployment over the short term. It is also the effects:

1). Long term, after the effects of higher labor costs slowly work into long term technology decisions, job redesign, creation and demise of local businesses etc. This is something which would take decades to play out.

2). On hours worked per employee. Employers like having larger and more flexible work pools. This means less hours per worker. Higher minimum wages leads to the ability to sustain larger work pools, with fewer hours per employee. This is not less employment, but it does NOT help the worker as expected.

3). On skills and experience. Humans are not widgets. A higher wage can be offset easily by demanding a higher skilled worker and quickly redesigning the work requirements around the higher skills and experience. The net effect of this isn't necessarily more short term unemployment, but relative gains for the skilled and capable over the less fortunate. In other words, a class of consistently unemployed ( changing who is unemployed).

4). On benefits and work condition. Employers don't just pay a wage. We also pay benefits, and have terms and conditions of employment. An increase in wages can be greatly offset by reducing time off, employee meals, training, and so on, the employee does not gain to the extent this occurs.

5). On other business costs. When forced to raise one cost of production, the employer doesn't JUST reduce that input. It may be easier to reduce other expenses, or of course to increase prices. The business owner may keep hours worked the same, but instead cut back on lawn maintenance or advertising or whatever.

Economists should quit oversimplifying this about higher unemployment. The real debate is whether it makes labor markets less efficient, one possible outcome of many is lower short term employment. Most of the efficiency losses are unseen and unmeasured.

Mark Bahner writes:

It sure would be interesting to see Paul Krugman manage a Walmart store, giving him complete discretion in all areas, with the requirement that he meet some sort of average-or-better profitability for similar Walmart stores.

Bob Murphy writes:

Good post David. I make two additional points here, the latter of which might be astounding to some innocent readers.

Kevin Erdmann writes:

I just came here to post Bob Murphy's link, but Bob beat me to it. Hilarious.

TMC writes:

Bob Murphy, you should really have a KrugsVsKrugs tag for your post :)

Travis Allison writes:

In his article, PK cites a study from 2010. So he could claim that our understanding of the facts has changed:


efcdons writes:

I feel like you guys are talking past Krugman. You act as if the "market" is a naturally occurring phenomena so any deviations from the market are unnatural, bad, wrong, distorting, etc.

The point Krugman is making is the market is created through man made rules. These rules can favor capital or labor. These rules are "political" in that they are made (ostensibly) through a process of consideration and deliberation between various parties with different end goals.

What makes me think Krugman is correct is the share of the pie going to capital has increased at the expense of labor's share. This has happened for a number of reasons and all of those reasons came about through deliberate choices. Summers et. al. have begun to realize their story about globalization or skills deficits is wrong and the deliberate rule changes have been the main driver of the change in labor and capital's share of the national income.

David R. Henderson writes:

The point Krugman is making is the market is created through man made rules.
Krugman is one of the clearest writers in the economics profession. If you find yourself talking about the point he’s making and ignoring the specific claims he made that I’m criticizing, that’s a problem.

Comments for this entry have been closed
Return to top