The minimum wage: theory before history

Greg Mankiw and Brad DeLong are having some back and forth over the minimum wage.  I’m willing to admit, unabashedly, that I form my judgments on this matter by theory more than "raw evidence."  When the evidence is unclear, or points in multiple directions, I favor the most plausible explanation.

Unlike like most market-oriented economists, however, I am not obsessed with the story of the downward-sloping demand curve for labor, to the exclusion of all other possible mechanisms.  I am more likely to see markets as extremely flexible and to look to the quality of job as a critical variable.  If minimum wages go up, I expect some mix of two scenarios:

1. The employer restores the previous net wage by worsening working conditions.

2. The employer upgrades the quality of job and thus marginal products, to meet the new level of minimum wage.

Now #1 is not much of an argument for boosting the minimum wage.  But is #2?

It sounds good but the employer had decided in the first place not to create those higher productivity jobs.  So those jobs must cost more and we should expect a negative effect on employment, albeit perhaps a slight one.

It is also the case that those jobs will go to the "most easily upgradable" workers among the low-wage working set.  I suspect those are the low-wage workers with relatively high human capital and high levels of adaptability.  Among the class of low-wage workers, the effects are probably anti-egalitarian.  That again does not make the minimum wage sound so great, even though the employment effects could be small or perhaps even zero.  I might add this also explains why the most articulate low-wage workers probably, for reasons of self-interest, favor increases in the minimum wage.

I don’t buy into the Card-Krueger monopsony scenario, at least not outside of rural Nebraska.  If you wish to defend it (does anybody? — even Krugman scorned it), comments are open.

I invite all participants to the debate to indicate the relative weights they place on "theory" vs. "history."  I’ll invent an imaginary, meaningless scale and opt in at "0.7" in favor of theory.  If the evidence were clearer, of course, my weights would change.


I'm nearly 99% theory because the empirical data is so noisy. The simplest theory is that when the price goes up, the demand drops. Proponents counter with "Ahhhh, but the price/demand curve is vertical, so demand doesn't drop". Perhaps, but if demand didn't drop, then you wouldn't see capital investment substituting for labor. And yet you can see that everywhere and always whenever the price of labor goes up.

Even if the price/demand curve was vertical, the higher price for labor would attract discouraged low-wage laborers back into the job market. So in order to not create unemployment from these newly-added workers, the price/demand curve actually needs to slope backwards, so that as employers are forced to pay more, they voluntarily hire more workers.

So when you go looking in the marketplace for unemployment, you have trouble finding it. That's because minimum wage increases affect so few workers (currently only 3% in the US) and are small in magnitude. Plus you have methodology problems. An employer listens to the news as much as anyone, and they weigh the probability of a future minimum wage increase against their future need for employees. It's unlikely that unemployment will jump on the actual date of the change in the minimum wage.

And, the theory predicts that a discriminatory employer can discriminate at no cost to himself, so that the effect of a minimum wage law is to increase minority unemployment faster than majority group unemployment.

Herr Cowen,

What good is theory without practical evidence?

Can you address the idea of the min. wage increase correcting for inflation or price increases in certain industries/products (i.e. gas, movie tickets, airfare)?

How much would a min. wage increase cost the private sector? Surely, there is a cushion at the very top of the pay scale that allows for discretionary adjustments to be made (i.e. smaller bonuses for the top 10%, a smaller corporate jet for the CFO or less alcohol at company events etc†¦). Can such cuts compensate for the increase in the min. wage? Would there still be a question regarding egalitarianism at the bottom quadrant and quality of jobs or products produced?

I'm on the evidence side, too. I read this a few years ago from Steven E. Landsburg:

"Here's how: Ordinarily, studies with large sample sizes should be more convincing than studies with small sample sizes. Following the fates of 10,000 workers should tell you more than following the fates of 1,000 workers. But with the minimum-wage studies, that wasn't happening. According to the standard tests of statistical significance, the results of the large-scale studies were, by and large, neither more nor less significant than the results of the small-scale studies. That's screwy. Screwy enough to suggest that the studies being published couldn't possibly be a representative sample of the studies being conducted.

Here's why that matters: Even if minimum wages don't affect employment at all, about five out of every 100 studies will, for unavoidable statistical reasons, appear to show a significant effect. If you could read all 100 studies, that wouldn't be a problem—95 conclude the minimum wage is pretty harmless as far as employment goes, five conclude it's a big job-killer, you realize the latter five are spurious, and you draw the appropriate conclusion. But if the 95 studies that found no effect were deemed uninteresting and never got published, then all you'd see were the spurious five. And then the next year, another five, and the next year another five. "

In other words, the evidence must be pretty darn strong that there is no effect, but those studies just aren't being published, probably because people want to believe in "teh Theery". Its like something that can't die because people refuse to believe its dead.

Now, I am relying on SL to tell the truth in this case, I don't know if he is. He seems respectable.

Ill ask the same question that I have been asking on all the GMU blogs recently. If there exists superior interventions that the government could employ to assist the working poor, then why advocate for a policy tool whose effects are highly uncertain?

You make an excellent point (which I don't recall seeing elsewhere) when you note that within the set of low-wage earners, minimum wage increases are effectively regressive. The most able might gain, but those who are at the low margin are the big losers. This inversion is one of the reasons I don't care for the minimum wage, and would prefer a more sensibly constructed intervention.
On the theory/evidence front: in principle, I'd prefer evidence, because I know that theory can and will fail to capture important details and effects. But right now the quality of available theory seems to exceed the quality of empirical data, so I'll say 5/10.


Milton Friedman's argument comes to mind in response to your opinion that the minimum wage looks "rather good."

Milton argued that if a $ 5.15 min. wg. looks "good", what about a $10 minimum, why not a $30 minimum?

Milton Friedman's argument comes to mind in response to your opinion that the minimum wage looks "rather good."

Milton argued that if a $ 5.15 min. wg. looks "good", what about a $10 minimum, why not a $30 minimum?

Would someone please put this argument out of its misery? Degree matters. If a 10mg dose of some medicine helps you does that mean taking 100mg must be better?

On the history/theory question surely we must begin with a very strong bias in favor of history, which is sometimes called "data." The burden must be on the theoretician to explain why the data conflicts with the theory. And the explanation better be good.

The paper is:

Card, David and Alan B. Krueger (1995). "Time-Series Minimum-Wage Studies: A Meta-Analysis

I haven't read the paper yet.


You wrote:

"Would someone please put this argument out of its misery? Degree matters. If a 10mg dose of some medicine helps you does that mean taking 100mg must be better?"

Of course degree matters. Does anyone seriously think that THE GOVERNMENT will produce better social results fine tuning the labor market than the market would produce via spontaneous order resulting from free contracting? The record of governments' fine tunings does not inspire confidence.

And if we think human freedom in contract, what a minimum wage does is to reduce this freedom. In this dimension, the greater the increase in minimum wage, the greater the reduction in this freedom.

Peter, your argument about tying minimum wage increases to immigration "reform" is both selfish and ill-considered. Why should we be concerned about the poor in America but throw others, just as human as those in America, to the dogs?

More importantly, though, you ignore that the idea of an "American Economy" is a fiction that exists only for political purposes. The econcomy is a worldwide thing, and the more productive that the world AS A WHOLE can be, then the higher a standard of living will be enjoyed across the globe. The bottom line is that by preventing immigrants to come mow my lawn, you're making me do it. And when I'm doing that rather than, say, building a new software system, the effects are rippling out to the entire community. Everyone *loses*.

And, to provide a fig leaf of on-topicness: in the absence of compelling evidence I have no choice but to be biased towards theory, say 0.8. When something is clearly shown empirically, then I may shift the other way -- but not until then.

...and Tyler nails the problem with Economists. They believe their threory over evidence. In most professions this would be considered a problem.

The economic conventional wisdom that took over in the late seventies fetishized incredibly simple models (people often talk of two freaking curves!), and it has been a horrible period for American workers, and at a time when there should have been an enormous peace dividend.

We should index Mankiw's wage from what he was making in 1975 to the real increase in the lower quintile since the same period.


Without getting into your broader arguments, let me just clarify.

What is under discussion is an increase in the minimum wage of a certain size. Your argument, which you attribute to Friedman, is that this can't possibly be a good idea because a very much larger increase is obviously a bad idea. This makes no more sense than claiming that two aspirin won't help a headache because taking 20 aspirin would be a bad idea.

You clearly oppose any minimum wage whatsoever. But not everyone agrees with you. Some people think a small increase would be a good thing. That an exorbitantly large increase, to $30/hr, would be a bad idea does nothing to refute their points. Nothing.

"I might add this also explains why the most articulate low-wage workers probably, for reasons of self-interest, favor increases in the minimum wage."

Call me too literal-minded, but I'm having a hard time picturing any low-wage worker speaking out against raising the minimum wage, articulate or not.

I'm willing to admit, unabashedly, that I form my judgments on this matter by theory more than "raw evidence."

What Ivory Tower?

This is exactly why all economists should be forced to go into business for themselves as part of a Ph.D program --- I'd love to see the consensus views on minimum wage, before and after, trying to run a small business.

Perhaps "recent history" is worth a division, but history is known for lending some considerable support for the theory, in this case. The Great Depression is a good place to look, both at Hoover's Wage Rates Conference and the NIRA. Also see apartheid-era South Africa.

I belong in the "federal minimum wage makes no sense" camp with commentor albatross. The interstate or even intercity cost of living differences are too large, and the ability of businesses to exploit this too damaging to the economy.

Eeek, I got to agree on the comments regarding theory vs. evidence. If the evidence doesn't back the theory, the theory has a problem.

Theory says a minimum-wage increase will create employment losses. History says maybe so, but they're small. Is that a conflict?
Well, if we parameterize the theory with plausible values, it predicts an effect about the size that history reports.

For example, minimum-wage workers in fast-food outlets are less than 20% of the total cost base (and an even smaller percentage in other retail businesses, like gas stations). So if there's a 40% minimum-wage increase, passed through entirely to consumers, it's less than an 8% price increase. The industry's price elasticity is somewhere around one (1), so that price increase would reduce retail volume by less than 8% ... presumably leading to job losses among the minimum-wage employees of less than 8%. (Considerably less, actually, since all the numbers above err on the high side.)

Theory and history seem to agree. What am I missing?



You seem to think that I am for open borders; not so. I have to agree with Milton Friedman that once you have a big nanny state like the US has, wide open borders do not make economic sense.

Peter Schaffer:

Does it matter to the people at the bottom whether they lose their jobs to immigrants, outsourcing, or automation? I mean, whether you can't get a job as a janitor because there's a Mexican doing the job at half your wage, or you can't get a job as a janitor because the descendants of the Roomba have taken over 90% of the work that needs to be done, you're still out of work.


Guess you are on the 'theory' side.

Couldn't find that paper on SSRN.

Of course we should require EXTREMELY GOOD evidence if we are even thinking about insinuating that the law of demand might have been overturned the way C&K did in their path-breaking piece based upon truly inferior data. Only extremely excellent evidence would convince this old boy of an exception to the law of demand.

Some thoughts:

1) There is no reason to care more about workers born inside an arbitrary line called the "US border" than workers born elsewhere. Indeed, I would argue that it is morally repugnant to do so.

2) The simple static model is stupid to apply here because the expectations of both employers and employeers matter a lot. Expectations about future minimum wages (and about future tightness in the low skill labor market) affect both capital investment decisions for employers (as with the semi-automated fryers at McDonald's mentioned above and workers' human capital investment decisions. There is evidence in the literature (see work by David Neumark now at UC Irvine) providing evidence that minimum wage increases are associated with increases in high school dropout rates, for example. The key, though, is that because capital is involved, the effects may take a long time to show up (which means looking for immediate changes as in the Card/Krueger NJ/Pennsylvania study and many of the other existing studies will miss them). Furthermore, because expectations are involved, the same observed change in the minimum wage may have different economic effects, if it affects expectations differently.

3) I was once widely ridiculed at a conference lunch for defending June O'Neill's statement "Theory is evidence too". At the margin, we substitute theory for data. That is, in some sense, what structural models are all about.

4) Does anyone care to defend Michigan's decision to raise its minimum wage in the face of massive auto layoffs and the second highest state unemployment rate?


Comments for this post are closed