David R. Henderson  

IMF's Loungani: Demand for Labor is Downward-Sloping

PRINT
The great sin, the even greate... Compulsory Attendance IVs Reco...

I was on the road from Sunday a.m. to late last night and thus my sparser than usual blogging. I taught classes in Patuxent River, MD on Monday, Norfolk, VA on Tuesday, and Arlington, VA on Wednesday, with lots of driving in between and a flight home last night. I'm one tired puppy.

Now to an interesting study that caught my eye.

Our study is the first to use data on minimum wage changes for over 2400 counties in China. We combine the information on minimum wages changes with employment data from the Annual Survey of Industrial Firms, which covers over 70 percent of China's manufacturing employment. While China instituted a minimum wage system in 1994, enforcement of compliance with the law was significantly tightened only in 2004; the results described below are based on post-2004 data.

So what does the evidence show? On average across all firms, we find that an increase in the minimum wage leads to a small decline in employment: a 10% percent increase in the minimum wage lowers employment by a little over 1% percent.


This is from Prakash Loungani, "Does Raising the Minimum Wage Hurt Employment? Evidence from China," October 23, 2014.

Loungani concludes:

But if raising the minimum wage lowers employment, and ends up excluding low-wage workers from employment prospects, it may have adverse effects on both welfare and efficiency.

Why does he say "may?" If "raising the minimum wage lowers employment, and ends up excluding low-wage workers from employment prospects," it will "have adverse effects on both welfare and efficiency." For that matter, why does he say "If?" Does Loungani not believe his own results?

Notice, by the way, that the results he gets accord with the consensus view among U.S. economists for the United States circa 1981.

HT to Mark Thoma.


Comments and Sharing






COMMENTS (19 to date)
Daniel Kuehn writes:

I can't explain the "if" other than maybe he's allowing for the fact that the literature varies, but presumably "may" is a claim about total welfare.

On the title - of course people who think the minimum wage will not reduce employment tend to agree that demand curves slope down.

David R. Henderson writes:

@Daniel Kuehn,
I think you’re right that “may” is a claim about total welfare. And that’s why I asked the question. If the minimum wage reduces employment, overall welfare, adding up dollar gains and losses, falls. Remember that consumers and employers count too.

Daniel Kuehn writes:

I see, yep.

I think most people think of minimum wage with a view towards putting different weights on the welfare of low income families, with the concern being whether or not it really increases the welfare of low income families. That's where my thinking was, but you're right.

David R. Henderson writes:

@Daniel Kuehn,
Thanks. I’m glad we agree on the overall effects, whatever weights we put on them.
But now let’s look at low-income families. As you probably know, most of the benefits of the minimum wage don’t go to low-income families. They go to low-wage workers, most of whom are in families above the poverty line. Think about Peter Dorman, who has, as a commenter here, defended the minimum wage of the 1960s for how much it helped him. He never claimed, and I’m pretty sure he wouldn’t, that he had been in a low-income family. So it’s not well targeted on low-income families. Moreover, at least some of the people who don’t get work because of the minimum wage are in low-income families.
Also, let’s not ignore the consumer. Many consumers of goods produced by low-wage workers are in low-income families. Most? Probably not. But the point is that it is by no means clear that an increase in the minimum wage makes low-income families on net better off.

Daniel Kuehn writes:

I wouldn't necessarily equate low-income with poverty line, though. Here are my thoughts on your NCPA paper on that issue:

http://factsandotherstubbornthings.blogspot.com/2014/01/the-minimum-wage-is-not-well-targeted.html

Tom West writes:

But the point is that it is by no means clear that an increase in the minimum wage makes low-income families on net better off.

Indeed, but then it's by no means clear that an increase in the minimum wage makes low-income families on net *worse* off, so "may" seems to be exactly the word one should use in an academic paper.

David R. Henderson writes:

@Tom West,
Not true. See the first 3 comments above.

Pajser writes:

Assume that I don't like my job. I'd prefer to have 10% higher wage and to reduce my own total employment time for 10%. I'll have same income and more time for things I like. Some of that time, spending time with children or working at home may have other beneficial effects.

Loungani writes that generally, effect of 10% minimum wage increase is 1% total wage increase with 1% unemployment increase. And for firms in the "lowest decile," wage increase is 2.5%, while employment reduction is 1.8%.

As little as we know, it looks as welfare increase to me.

vikingvista writes:
I'd prefer to have 10% higher wage and to reduce my own total employment time for 10%

Typically when you work fewer hours at the same job, you don't produce more output per hour, so you don't produce a higher hourly wage. You produce less total output and correspondingly less total income. You become poorer.

Brian writes:

David,

This is an interesting study. One of the points they make is that evidence from China might be more applicable to emerging economies than U.S. data. However, the size of their observed effect is almost identical to what has been seen this past year in the U.S.

The left-wing Center for Economic and Policy Research made a big splash this summer when they pointed out that states with minimum wage increases had faster job growth this year.

Unfortunately for their argument, if one looks at job growth versus the SIZE of the minimum-wage increase, the data show a negative correlation. The 13 states that increased their minimum wage started from an average base of $7.93 per hour, so a 10% increase would be about $0.80. The linear regression implies that this should give a 1.36% reduction in the job growth rate. The results from the China study give a 1.33% reduction in employment for a 10% increase--basically identical.

Consequently, a 10% wage increase --> 1% job decrease appears to be a well-supported rule of thumb regardless of the type of economy.

Glenn writes:

"You produce less total output and correspondingly less total income. You become poorer."

Not if a minimum wage constraint binds, which is the whole point. I work 10 hours a day for $10/hour, for $100 gross. The government passes a law raising my hourly wage to $11. The authors predict my work day will fall, on average, by 1%. I now work 9.9 hours/day, for $108.90 gross.

"Not true. See the first 3 comments above."

The first 3 comments do not show that total welfare gains - measured in dollars - amongst the poor are negative due to minimum wage laws. Neither do they show that total welfare gains - measured in utes - are negative amongst the aggregate.

People are generally willing to forego some output in order to avoid having to see children working 14 hour days for a few dollars. Likewise, the poor are willing to forego some work hours for a higher wage during the balance.

Is it efficient, from the perspective of output-maximization? No. Is it efficient, from the perspective of welfare-maximization? It must be, or output-rich countries wouldn't universally adopt minimum wage laws. Of course, to set these constraints intelligently requires that we acknowledge both their consequences (in terms of lost output and reduced employment amongst the poor), and the relative costs and benefits, which is often missing from the public debate in the contemporary US.

The other interesting question is where the marginal tradeoffs between lost output and gained "social pride" wash. The development of China - and diffuse internal regional developments - provides an interesting case study.

vikingvista writes:
people who think the minimum wage will not reduce employment tend to agree that demand curves slope down

I don't know which claim is less credible: labor demand is not downsloping, or minimum wage employers tend to be monopsonistic.

The only reasonable argument for MW lifting someone's compensation, is that the type of employment changes to a type producing greater output. For those willing and able to engage in the new type of employment, their compensation goes up (whether or not they are better off). For those who don't or can't, their compensation goes to zero. In a large population, the latter will always exist.

vikingvista writes:
Not if a minimum wage constraint binds...I now work 9.9 hours/day, for $108.90 gross.

And if it binds, the min wage could be set to $55/h and you could work 4.5 h/d for $247.50 gross. That sounds much better, no?.

It makes you wonder if in the new economics, compensation any longer has any quantitative relationship to production.

The reality is that the $10 job is different from the $11 job is different from the $55 job, because the jobs must be more productive to cover the higher HOURLY rate.

Sure, there will be people who can and will do those more productive jobs. There's a wide spectrum of production for which you can find people willing and able to be employed. Except that spectrum is effectively truncated at the low end by the MW. It is hard to see how removing options for employment is helpful, particularly for those unable or unwilling to do what it takes to produce at the higher rate.

Brian writes:

"Is it efficient, from the perspective of welfare-maximization? It must be, or output-rich countries wouldn't universally adopt minimum wage laws."

Not so. There's no guarantee that laws passed by the government maximize welfare. The argument would have some validity if the minimum wage developed via the free market, but that's not the case here. It's possible, for example, that the minimum wage can be set low enough not to matter much, but makes people feel like they're doing something nice for the poor.

There are several lines of evidence that this is what's happening. First, the minimum wage is generally not indexed to the cost of living. If it were really about providing a living wage, COLAs would be common. Second, if the minimum wage were really about helping people maintain a living wage, it would not be operating set at either the federal or state level. After all, cost of living varies significantly from state to state and from area to area. A universal value for the minimum wage is not likely to be set above what the cheapest areas can bear. It follows that the minimum wage must be set well below the cost of living for everywhere else.

David Friedman writes:

"The authors predict my work day will fall, on average, by 1%. "

If I correctly understand the summary of the article, the 1% is a decline in the firm's total employment, not in its employment of minimum wage workers. In the U.S. case, assuming that all of the lost employment was of minimum wage workers, that would mean about a 20% decline in employment for them, since they are about 5% of the labor force. More plausibly, employment of non-minimum wage workers would increase, since they are a substitute for the now more expensive input, so a 1% decline in total employment would mean more than a 20% decline in minimum wage employment.

Of course, I don't know what the corresponding figures would be for China.

Pajser writes:

Friedman: "In the U.S. case, assuming that all of the lost employment was of minimum wage workers, that would mean about a 20% decline in employment for them, since they are about 5% of the labor force."

It is unlikely assumption. One does not fire 50% of workers and keep 100% of foremen.


David R. Henderson writes:

Wow, David Friedman, good catch. That’s huge! It does undercut the point I made about relating it to U.S. data. Thanks.

Glenn writes:

"There's no guarantee that laws passed by the government maximize welfare."

Agreed in principle, but most (all?) output-rich countries are democratic (relatively?). Over large enough electorates and enough elections, we can start to assume the representative voter exists.

And this is not a quorum. It is a universal truth that every state on the planet with per-capita GDP over x enjoys minimum-wage laws. This may be some cosmic coincidence, I grant you, but far more likely this is a legitimate expression of the representative value. It need not be Pareto optimal, but it is almost certainly welfare maximizing.

Minimum-wage laws are costly, therefore output-poor states do not have them. However, they are also valuable, from the perspective of social welfare. Therefore, output-rich countries DO have them. This is a much more elegant argument than, "some governments impose them because they are sneaky, controlling bastards, and they also all happen to be rich and democratic".

"If I correctly understand the summary of the article, the 1% is a decline in the firm's total employment, not in its employment of minimum wage workers."

Agreed. I was being deliberately sneaky to make my point (or at least selective), and should have disclaimed the fact. In actual fact, the expected loss conditional on being a minimum-wage worker will be higher than the unconditional average, but lower than in your case. The article confirms this by showing that losses are concentrated amongst firms with a higher share of minimum wage workers.

The point was simply to demonstrate that such regulations NEED NOT be welfare-losers (though, of course, they MAY be, particularly if floors are stupidly set - see Seattle).

steffijude writes:

[Comment removed pending confirmation of email address and for policy violation. Email the webmaster@econlib.org to request restoring this comment. A valid email address is required to post comments on EconLog and EconTalk.--Econlib Ed.]

Comments for this entry have been closed
Return to top