≡ Menu

An Anonymous First-Rate Economist On the Economics of the Minimum Wage

First-year GMU economics masters student Ash Navabi sent to me notice of a commenter at on this post at Economics Job Market Rumors.  According to Ash, the commenters “Economist 1C88” and “Economist E446” are the same person.  Whoever this commenter is, I tip my hat to him or her.  He or she says in these seven short comments at least as much as, and perhaps more than, I’ve said about the minimum wage in dozens of longer blog posts – and he or she says it more eloquently, clearly, thoroughly, and cogently.  I paste these seven comments with my own numbering, but without further edit or comment by me, below the fold.  Read these comments and treat yourself to the reasoning of a first-rate economics mind.  Every word below the fold (save for the obvious quotations from other commenters) is that of Economist 1C88/E446.  (For the record, I have no idea who this economist is.)



Boeing airplanes are tradeable goods made by high-skilled employees. Low-skilled service sector workers do not produce tradeable goods; and presumably the elasticity of demand for haircuts and cheeseburgers is much lower than that for airplanes.

Trade is only one margin; there are lots of other margins. For instance:

(1) Low-skilled workers can be replaced by slightly more-skilled workers once the wage difference between them has dramatically compressed.

(2) New workers can enter (or old workers can increase their labor supply) in response to higher wages; if this happens, people who might need the job will be disemployed even if labor demand is constant.

(3) Labor can be replaced by automation.

(4) Customers can substitute between businesses that provide the same basic service but with dramatically different business models. (e.g. less shopping at Walmart, which is now more expensive, and more shopping at Costco; only problem is that Costco has half as many workers per dollar of sales, so labor demand decreases dramatically)

… and so on.

In general, the exasperating thing about left-wing economic commentary is that it always declares victory when it’s cast doubt on one margin of unintended consequences, even when 10 other still-viable margins remain.

Furthermore, it’s specious assert that the price of an input (labor) increasing by 107% will cause the final product to increase in price by the same magnitude.

No, obviously the baseline expectation would be that an increase in the price of one input will cause the output price to increase by (share of input in total input costs)*(increase in that input’s price). And that baseline expectation tends to be validated by most studies on this question, which tend to find roughly 100% pass-through of costs. See Aaronson (2001) or the recent working paper by Harasztosi and Lindner.

Indeed, even the original Card and Krueger study – if you read it carefully – found that prices increased by roughly the amount you’d expect.


and that there will be no multiplier effect to partially or wholly offset a (potential and undemonstrated) price increase

You’re revealing your ignorance when you describe the price increase as “potential and undemonstrated”; it has been demonstrated many times, and is arguably the best-documented empirical fact about the minimum wage.

And the postulated “multiplier effect” is just stupid. Multipliers only make sense in a world where the economy is ZLB-constrained, or in the short term if you’re just one state in a fiscal union, etc. They do not make sense in the unconstrained long run in any model (and I say this as a macroeconomist who is generally perfectly happy with talking about “multipliers”).

And this is on top of the fact that if you try to do some basic quantification, the “multiplier effect” from paying higher minimum wages would be tiny anyway. Think about it: what fraction of the marginal earnings of minimum-wage workers goes to buying from businesses that employ minimum-wage labor, and what fraction of those sales actually accrues to the minimum-wage laborers. It’s not a very big number and can’t conceivably be enough to offset any sizable disemployment response. (Moreover, if you’re tempted to counterfactually claim that it’s a big number, remember that this is also a downside to the minimum wage: to the extent that higher minimum wages feed disproportionately into the prices of the consumption bundle purchased by low wage workers, those workers are hurt.)



This is a very weird debate.

Minimum wage skeptics point out that, as a general rule, demand curves slope down and supply curves slope up. If you set a wage floor, there will be unemployment – because the quantity demanded falls and the quantity supplied rises. Simple, right?

Now, at this point there are a million different quibbles and complications – and for minimum wage supporters, it’s tempting just to wave aside the whole supply-and-demand analysis as a desperately naive exercise in Econ 101. Do this, and sprinkle on enough condescension, and you can revel in your sophistication and superiority as an economist:

lol at these people who think labour markets are an upward sloping supply curve intersecting a downward sloping demand curve… Really, labour markets are *not* a good example of Economics 101.

But I’m afraid that “lol at these people” doesn’t quite clinch this argument. It’s easy to say that the world is more complicated than Econ 101 – no dispute there! It’s much harder to argue that Econ 101 doesn’t apply at all: that it’s not even relevant as a benchmark, that supply and demand don’t slope in the right directions, or that some other force swoops in to lift employment and save the day.

After all, if higher minimum wages do prompt firms to hire less low-wage labor, and do encourage new workers to enter the low-wage labor pool, then it’s almost a syllogism that more people will end up unemployed. If you want to argue otherwise, you have to specify where, exactly, this analysis breaks down.

And that’s where things get so strange. Aggregate supply and demand responses represent the accumulation of many different margins. When the price of low-skilled labor rises, firms might demand less of it because they find a way to replace it with medium-skilled labor. Or maybe they’ll replace it with automation, or consumers will demand fewer products intensive in low-skilled labor, or they’ll acquire substitutes through trade… or any number of other possibilities. Maybe you have a clever argument for why one of these margins doesn’t matter, but you need to be far more ambitious (reckless, really) to deny all of them.



And suppose that you do deny all of them: suppose that for every single margin, you’ve found a way to argue that the supply and demand responses are inelastic. Now take a step back and think about the hypothetical world you’ve created. In this world, both the supply and demand curves for low-wage labor are near-perfectly vertical. This means, of course, that any shocks to supply or demand must result in massive equilibrating movements in the market-clearing wage. Does this actually happen? Not even close: 10th percentile wages are surprisingly stable relative to 50th percentile wages, and the big changes that do occur tend to come from the minimum wage itself.

So the project of denying every supply and demand response at the low end of labor markets is doomed from the start. But this doesn’t stop many of you from attempting it. I call this (forgive the alliteration) the Selective Skepticism of Substitution Syndrome.

Whenever someone mentions a margin along which firms or households could substitute in response to changing prices, thereby creating unintended consequences from the policy you support, you’ll dig deep and find a way to argue (earnestly as ever) that the substitution response is actually close to zero. Then they’ll mention another margin; again, you’ll find a way to deny it, and so on it goes…

The problem with all this Selective Skepticism of Substitution is that it makes you look silly. Surely there are, in reality, plenty of ways in which agents substitute in response to changing relative prices. That’s how the economy works! It would really be quite a coincidence if substitution happened to be shut down in only those cases that matter for the minimum wage.



So you end up with contorted denials whose logic, if applied elsewhere, is sweeping enough to deny the price mechanism pretty much everywhere. To wit:

3) High skilled workers usually take high skilled jobs regardless of the low skilled wages

Well, yeah, people at the very top of the skill distribution aren’t too directly influenced by the wages of people at the very bottom. But that’s not the question.

I want you to look at the table of occupations from the BLS’s Occupational Employment Statistics, and sort by median wage. Think about all those occupations with median wages below $15 – and then also think about all the occupations with median wages a bit above $15 that still have 20 or 30 or 40 percent of workers making below $15.

If we raise the minimum wage to $15 and entry-level fast food or cashier jobs are just as easily available as they are today, do you really think that none of the kinds of people who currently train for the other jobs with median wages below $15 will be tempted to just pick the lower-end jobs instead? “Pest Control Workers” have a median wage of $14.74 – are you really so confident that none of them will say “f*** pest control, I’m going to earn the same wage working the counter at McDonald’s”? The 25th percentile wage for “rock splitters, quarry” is only $12.81 – are you really so sure that none of them will get tired splitting all those rocks and take the Safeway cashier job closer to home instead, once it pays just as much?

The answer is that of course some of them will be tempted to switch jobs (or choose different jobs in the first place). It won’t necessarily happen right away – there are costs to moving between jobs, and even bigger costs to switching careers, so you’re not going to do it in response to a fleeting, idiosyncratic, almost below-the-radar change in the minimum wage (like, erm, virtually all of the cross-state minimum wage changes used for identification in existing “cleanly identified” studies). But it will happen in the long run, and even a very small response would be enough to swamp low-end labor markets – which are quite small relative to the vast middle of the distribution – and displace the existing workers.



A lot of minimum wage supporters, of course, realize that Selective Skepticism of Substitution is a dead end: it’s not credible to argue that all elasticities are zero, and it doesn’t give you any chance to flaunt your ever-so-much-more-advanced-than-Econ-101 pseudo-sophistication either.

So then they come up with alternative ideas. These ideas are almost universally terrible, of course, but that doesn’t prevent their zombie-like recurrence.

– They’ll say that higher wages lead to higher productivity that undoes much of the cost. (But with higher productivity, you need to hire fewer people to get the same effective amount of labor services, so this is a force for disemployment in its own right.)

– They’ll say that some kind of Keynesian channel, with minimum wage workers spending out of their now-higher wages, boosts demand and undoes the negative effect. (This is deliciously innumerate: only a tiny fraction of the marginal consumption basket of minimum wage workers flows through to other minimum wage workers. Perhaps 189d is trying to argue otherwise by complaining that “preferences are not homothetic” – with the presumption, I suppose, that at the margin minimum wage workers happen to spend 100% of their income on McDonald’s, even though the inframarginal share is more like 5%!)

– They’ll say that it’s really a story of monopsony – where firms face upward-sloping supply curves for labor, and use this market power to pay workers less than their marginal product. If you force them to pay a minimum wage, the story goes, then monopsony considerations will disappear and firms will actually demand more labor. (Of course, unless the monopsony wedge is huge, it’s not clear how this story can justify an aggressive minimum wage – and if the monopsony wedge was really that big, it would imply massive returns to employee recruitment that are hard to reconcile with the evidence. Nor is it clear why monopsony is such a big deal in the low-end labor market – of all places! – or why market power justifies breaking the price mechanism in this one instance but not any other. Nor can anyone explain why firms raise prices in response to a minimum wage hike even when this story implies that their marginal costs should be going down.)

Not very impressive…



To top it all off, none of these stories can explain the key feature of the minimum wage literature that supporters are always citing – which is that estimates for the disemployment effect are generally near zero. Not positive or negative depending on the exact balance of the monopsony and substitution effects (which might vary predictably based on the features of the industry or occupation), but zero.

To be precise, the estimates cluster around zero, with these researchers never able to reject the null hypothesis of zero at a rate higher than you’d expect from chance alone, and the most precise point estimates getting closer and closer to zero (as you may have seen in all those funnel graphs). Taking all the evidence at face value, in fact, you must believe that we have a rather precise quantification of the effect of the minimum wage, at almost exactly zero.

How remarkable that the monopsony or Keynesian or whatever channels happen to precisely cancel out the substitution channel in every environment that left-leaning labor economists study! Truly, this is an economic miracle that should command our deepest respect and attention…

… or else maybe there’s something else messed up with the studies, and the zeros are spurious. Come to think of it, maybe the Selective Skeptics of Substitution aren’t so bad after all. At least their hypothesis, however ridiculous, is consistent with the evidence they cite.