Bryan Caplan  

Minimally Convincing

Murphy on Private Production o... Confused by a price change...
Two high-quality studies of the disemployment effects of the minimum wage are getting a lot of attention.  The first looks at Seattle.  Punchline:
This paper evaluates the wage, employment, and hours effects of the first and second phase-in of the Seattle Minimum Wage Ordinance, which raised the minimum wage from $9.47 to $11 per hour in 2015 and to $13 per hour in 2016. Using a variety of methods to analyze employment in all sectors paying below a specified real hourly rate, we conclude that the second wage increase to $13 reduced hours worked in low-wage jobs by around 9 percent, while hourly wages in such jobs increased by around 3 percent. Consequently, total payroll fell for such jobs, implying that the minimum wage ordinance lowered low-wage employees' earnings by an average of $125 per month in 2016.
The second looks at Denmark.  Punchline:
On average, the hourly wage rate jumps up by 40 percent when individuals turn eighteen years old. Employment (extensive margin) falls by 33 percent and total labor input (extensive and intensive margin) decreases by around 45 percent, leaving the aggregate wage payment nearly unchanged. Data on flows into and out of employment show that the drop in employment is driven almost entirely by job loss when individuals turn 18 years old.
These look like careful studies to me.  But if I were a pro-minimum wage activist, neither would deter me.  As I confessed before the empirics were in:
If I were sympathetic to the minimum wage, I'd insist, "The worst the experiments will show is that high minimum wages hurt employment in individual cities.  That wouldn't be too surprising, because it's easy for firms and workers to move in and out of cities.  The experiments will shed little light on state-level minimum wages, and essentially no light on federal minimum wages."
In other words, I'd channel Ayn Rand villain Ivy Starnes:
She made a short, nasty, snippy little speech in which she said that the plan had failed because the rest of the country had not accepted it, that a single community could not succeed in the midst of a selfish, greedy world - and that the plan was a noble ideal, but human nature was not good enough for it.
The same goes, of course, for a group-specific minimum wage.  If I favored the minimum wage, I'd look at the Danish results and say:
"Fine, high minimum wages hurt employment for workers in the critical age category.  So what?  It only shows that when some workers are poorly protected, firms prefer to hire them.  The experiments shed little light on comprehensive minimum wages that protect everyone.  Plus, Denmark is a small country of 6M people.  If Denmark's minimum wage hurts low-skill Danes, we need a pan-European minimum wage, not deregulation."
What if the studies had come out the other way?  Opponents also have a plausible objection: The studies measure short-run disemployment effects, which we should expect to be mild.  In the long-run, however, employers have far more flexibility.  They can replace labor with capital.  They can replace low-skilled workers with higher-skilled workers.  And they can shed workers the guilt-free way - by attrition.

I'm not an Austrian economist; our profession's shift from pure theory to empirics has been a huge step forward.  But we also need common sense and a broad view of what empirical evidence counts as "relevant."  As I've said before:
In the absence of any specific empirical evidence, I am 99%+ sure that a randomly selected demand curve will have a negative slope.  I hew to this prior even in cases - like demand for illegal drugs or illegal immigration - where a downward-sloping demand curve is ideologically inconvenient for me.  What makes me so sure?  Every purchase I've ever made or considered - and every conversation I've had with other people about every purchase they've ever made or considered.

Research doesn't have to officially be about the minimum wage to be highly relevant to the debate.  All of the following empirical literatures support the orthodox view that the minimum wage has pronounced disemployment effects:

1. The literature on the effect of low-skilled immigration on native wages.  A strong consensus finds that large increases in low-skilled immigration have little effect on low-skilled native wages.  David Card himself is a major contributor here, most famously for his study of the Mariel boatlift.  These results imply a highly elastic demand curve for low-skilled labor, which in turn implies a large disemployment effect of the minimum wage.

This consensus among immigration researchers is so strong that George Borjas titled his dissenting paper "The Labor Demand Curve Is Downward Sloping."  If this were a paper on the minimum wage, readers would assume Borjas was arguing that the labor demand curve is downward-sloping rather than vertical.  Since he's writing about immigration, however, he's actually claiming the labor demand curve is downward-sloping rather than horizontal!

2. The literature on the effect of European labor market regulation. Most economists who study European labor markets admit that strict labor market regulations are an important cause of high long-term unemployment.  When I ask random European economists, they tell me, "The economics is clear; the problem is politics," meaning that European governments are afraid to embrace the deregulation they know they need to restore full employment.  To be fair, high minimum wages are only one facet of European labor market regulation.  But if you find that one kind of regulation that raises labor costs reduces employment, the reasonable inference to draw is that any regulation that raises labor costs has similar effects - including, of course, the minimum wage.

3. The literature on the effects of price controls in general.  There are vast empirical literatures studying the effects of price controls of housing (rent control), agriculture (price supports), energy (oil and gas price controls), banking (Regulation Q) etc.  Each of these literatures bolsters the textbook story about the effect of price controls - and therefore ipso facto bolsters the textbook story about the effect of price controls in the labor market. 

If you object, "Evidence on rent control is only relevant for housing markets, not labor markets," I'll retort, "In that case, evidence on the minimum wage in New Jersey and Pennsylvania in the 1990s is only relevant for those two states during that decade."  My point: If you can't generalize empirical results from one market to another, you can't generalize empirical results from one state to another, or one era to another.  And if that's what you think, empirical work is a waste of time.

4. The literature on Keynesian macroeconomics.  If you're even mildly Keynesian, you know that downward nominal wage rigidity occasionally leads to lots of involuntary unemployment.  If, like most Keynesians, you think that your view is backed by overwhelming empirical evidence, I have a challenge for you: Explain why market-driven downward nominal wage rigidity leads to unemployment without implying that a government-imposed minimum wage leads to unemployment.  The challenge is tough because the whole point of the minimum wage is to intensify what Keynesians correctly see as the fundamental cause of unemployment: The failure of nominal wages to fall until the market clears.

And don't forget the vast literature on labor demand elasticity...

Comments and Sharing

COMMENTS (15 to date)
Andrew_FL writes:
I'm not an Austrian economist; our profession's shift from pure theory to empirics has been a huge step forward.

This is a strange statement to make in a post which largely uses historical analysis to illustrate economic theory, not try to falsify it.

Philip George writes:

"And don't forget the vast literature on labor demand elasticity... "

On why elasticity is a bad measure for minimum wage studies:

"The elasticity of the rectangular hyperbola is equal to 1 everywhere. This suggests that at every point on the rectangular hyperbola, employment is equally responsive to a change in the wage rate. But we know that the rectangular hyperbola at points close to the origin along the x-axis is nearly vertical and that at points far from the origin along the x-axis it is nearly horizontal. In the nearly vertical portion a large change in the wage produces only a small change in employment. In the nearly horizontal portion of the rectangular hyperbola a small change in the wage produces a very large change in employment. So clearly elasticity is the wrong metric to measure the responsiveness of employment to a change in the wage rate. The correct measure to use is the reciprocal of the slope."

From Section E of The derivation of involuntary unemployment from Keynesian microfoundations in Real-World Economics Review.

Thomas Sewell writes:

Please don't confuse them with the facts, they're trying to do politics here.

Thaomas writes:

Sowell is partly right
Proponents of a higher minimum wage are trying for a political effect, raising the incomes of low wage workers when better methods like an EITC are not politically feasible. This still requires facts. Is the elasticity of demand in the specific market less than one?

Don Boudreaux writes:

Thaomas writes that assessing the effects of minimum-wage policies on poverty-reduction "requires facts." He's correct. But he is incorrect to assume - as he does - that the relevant fact is whether or not "the elasticity of demand in the specific market less than one." Other facts are relevant, such as 'Is the relevant test of a minimum-wage policy whether or not it increases the total income of a specified group of workers even if that policy reduces, perhaps to zero, the incomes of several of the members of that group?'

In one sense this latter question is normative, which is not to say irrelevant. In matters of government policy, facts are useful only insofar as mastery of them better enables us to achieve the goals that we wish to achieve. And the selection and ranking of goals is ultimately and unavoidably a value judgment.

Yet in another sense this latter question is factual: Do we, or do we not, agree that an acceptable outcome is achieved if a minimum wage raises the total income of low-skilled workers even if it reduces, perhaps to zero, the incomes of many individual low-skilled workers?

Drilling down just a bit, Thaomas overlooks the fact (!) that there is no objective - no factual - definition or description of the relevant group of workers over which the total-income effects of minimum wages is to be assessed. Does this group, in the U.S., consist only of workers currently earning $7.25 per hour or less? Or does this group consist of workers making, say, $7.98 per hour or less? Or, as a third alternative, does this group consist only of workers making, say, $8.00 per hour or less and who are heads of households?

Obviously, questions about what is the relevant group over which the elasticity of demand for its labor services is to be assessed factually must be asked and answered satisfactorily before any empirical investigation into the income effects on that group of a minimum wage can commence. The number of different ways to plausibly define this relevant group are many, and choosing which of these many definitions is the appropriate one cannot possibly be an exercise purely in determining 'the facts'; it must involve value judgments.

It's comforting to imagine that the formation of government policy can be reduced to an exercise in discovering, processing, and following only "the facts." But such an imagined method of forming policy is just that: imaginary. It can never be otherwise. That's a fact.

Daniel Kuehn writes:

re: "And don't forget the vast literature on labor demand elasticity..."

Labor demand elasticity is not what's contested. the relevant elasticity to talk about is the labor supply elasticity faced by the firm.

Julien Couvreur writes:

I think you're correct. The argument "X must be applied at maximum scale to work" has been applied in a number of contexts before (gun control, collectivism and various economic policies).

MikeP writes:

The argument "X must be applied at maximum scale to work" has been applied in a number of contexts before...

And health insurance.

The calls are reflexively for "single payer", not "n+1 payer" or "payer of last resort".

It's the statist mentality: that which is not mandated must be prohibited.

KevinH writes:

I'm not saying you are wrong, but the challenge you propose in point #4 at the end is pretty easy to counter. You'd only be right that there was a paradox if minimum wage approached the median wage. Wage stickiness affects the whole distribution, while minimum wage only affects the tail. It is pretty easy to argue that the effects at the tails on unemployment can be overwhelmed by other factors.

I think a fair reading of both studies is that the obsession in economics with linear models is endangering the field finding relevance going forward. Simply worrying about if a derivative is 0 or negative isn't going to solve these type of problems.

Eg, the Danish study shows that 18 yr/olds as a group, fare better under the increased wages (40% wage increase for 32% drop in employment). To make any sense of the resulting benefits or detriments to society, you can't treat money like a 1d scalar. However, if we remove that assumption and the ease with which money is added and subtracted as in a ledger, 99% of the economics work in the last 5 decades needs to be redone and reexamined.

KevinH writes:

I can't reply, but I thought of a followup to my last comment.

If I was an enterprising young economist I'd study graph theory. It should be pretty simple to find a situation where an economic law from the past few decades holds for a fairly realistic and intuitive network. Then analyze the networks in which the law fails to hold.

MikeP writes:

Eg, the Danish study shows that 18 yr/olds as a group, fare better under the increased wages (40% wage increase for 32% drop in employment).


1.4 * 0.68 < 1

Lewis writes:

More evidence for downward-sloping demand curves:
Americans Lured Off Job Sidelines May Also Limit Pay Gains

Eric Rasmusen writes:

Very nice post. Using regressions has huge limitations, and it's good to point out that people pick and choose, saying elasticities are zero in one context (minimum wage) and infinite in another (immigration).

Zenon Zygmont writes:

As Don Boudreaux and Mark Perry and others point out - there are other ways a firm can adjust to a higher MW than cutting workers. When I toss this idea out to my Principles students they immediately come up a dozen or more adjustments firms can make that do not lead to lay-offs but reduce all workers' overall utility. But how do you you proxy this in empirical work?

Pete M writes:

Mr. Caplan, you say regarding Seattle: "That wouldn't be too surprising, because it's easy for firms and workers to move in and out of cities." But what if the primary impact was on local restaurants and mom-and-pops? Such firms most likely don't have the ability to move away.


Return to top