David R. Henderson  

Robert Murphy on the Minimum Wage

PRINT
Meet James Schneider... The Sucker Tax...
In this article, I explain why, even if the revisionist empirical studies are accurate, it still does not follow that the proposed hike in the minimum wage will be a boon for low-skilled workers. I also argue that, because critics have raised many troubling concerns about these studies, we should not accept them at face value. I conclude that economists should maintain the standard view that employers have a downward-sloping demand for low-skilled labor and that raising the minimum wage will tend to destroy job opportunities for many of those whom advocates of the higher minimum wage wish to help.
This is the third paragraph from February's Econlib article, "Economists Debate the Minimum Wage," by Robert P. Murphy. If you have been following the debate, you know that there's some pretty heavy econometrics involved. Bob works his way through it and tells you what the controversies are. Here are two paragraphs on some of the key econometric issues:
Now that I've reviewed some of the terminology and methodological issues, I can summarize some of the key arguments. Dube et al. (2010) concede that if we rely merely on general time trends and regional fixed effects, we will, indeed, see the old consensus: the minimum wage destroys low-skilled jobs. Yet if we include regional-specific trends indexed by time period, the influence of the minimum wage begins to disappear and, in particular, using their preferred control group method (of contiguous county pairs) completely obliterates the textbook finding. The minimum wage may even have a positive impact on employment.

On the other hand, Neumark and Salas (2013) provide a summary and critique of the revisionist studies. For example, they show that the results of Allegretto et al. (2011) depend on a very particular choice of time period and on a particular "functional form" of the state-specific time trend. If Allegretto et al.'s same regression were run on a different portion of the time period they chose (which had recessions at the start and end), then the minimum wage would appear to hurt teen employment after all. Moreover, even using Allegretto et al.'s original time period, Neumark and Salas merely allowed the state trend variable to be a higher order (not just linear), and, once again, the result was that the minimum wage hurt employment. As Neumark and Salas put it, "Thus, [Allegretto et al.'s] claim that underlying trends that vary by state generate spurious evidence of negative minimum wage effects on teen employment is clearly not true. Rather, only with a very specific form of controlling for this spatial heterogeneity" do the revisionist results hold up.


There's a nice Reference section at the end that gives titles, authors, and coordinates of the various articles on the issue.


Comments and Sharing





COMMENTS (28 to date)
Roger McKinney writes:

The odd thing is that the theory says the min must be above the market rate to have an effect. Do the studies measure the difference between the min and the market wages?

Just eyeballing the data, I would say the min wage has usually been below the market rate and when it is above it is barely above.

Wages differ by region, but in Oklahoma where wages tend to be below the national average (because the cost of living is so low) the wage for entry level jobs is about $9/hr, considerably above the min wage. In high wage state like CA the market wage will be much higher.

So raising the min to $10 should have no effect. Libertarians should shame the left for being so cheap. We should encourage them to raise the min to at least $20. I doubt they will accept the challenge because the really don't believe their own propaganda.

Otto Maddox writes:

Dube, et al sounds like their torturing the data until it confesses.

Daniel Kuehn writes:

Quite the opposite, Otto. They're using very standard evaluation techniques in a literature that never really applied them before, probably because you don't get "treatment" from the minimum wage the same way you do with other policies that DO expect the use of these standard evaluation techniques.

Daniel Kuehn writes:

Branding one side of the literature as "revisionist" was quite clever of Bob. It gives a very strong impression of what you're supposed to think is going on.

Charley Hooper writes:

Revisionism: The theory or practice of revising one's attitude to a previously accepted situation or point of view.

What would be a better word, Daniel?

gofx writes:

Daniel Kuehn,

I don't think Bob was being clever, it merely reflects basic economic theory, logic, common sense, and business experience. There are limits to what empirical studies can accurately measure and control. The burden of proof should be on those studies. Theory, experience, common sense and previous studies would seem to give a fairly strong prior that raising the minimum wage negatively impacts employment compared to what it otherwise would have been, so it should take a lot of data to make the posterior probabilities move. I am sure there are also papers purporting to show that price controls don't lead to shortages.

Daniel Kuehn writes:

Charley Hooper -
It has other insinuations than that - usually implies a sort of heterodoxy.

There's a big difference between calling them "standard" and "revisionist" as Bob has done, and calling them "fixed effects studies" and "difference in differences studies", which might have been what I would have done. After all, Meer and West and Neumark and Salas REVISE Dube et al.! And all of them are REVISING Card and Krueger (who of course themselves were revisiting other work). So when you have a scientific dialogue like this with contending results from different methods, "revisionist" is a little funny.

Look, I said it was clever not that it was awful or devious of him. It was careful word choice on his part, as we've all come to expect from Bob.

Daniel Kuehn writes:

gofx -
Whereas I think Charley gets a little too worked up by my point, I have to say you're dead wrong if you are implying that Dube et al. are lacking in common sense, logic, grasp of basic economic theory, etc. in identifying their results as trustworthy.

Anyway, if you think that then why have the debate? Why engage with people you think are illogical and are missing basic economic points?

David R. Henderson writes:

@Roger McKinney,
The problem is with your concept of “the” market wage. There are many market wages: they depend on productivity. The current minimum wage is well above the market wage for some people. I can well imagine internships for very bright students, for example, paying $2 to $3 an hour if there were no minimum.
@Charley Hooper,
Well said.
@Daniel Kuehn,
I don’t know Bob Murphy’s motives in using the word “revisionist.” I expect it’s simply what Charley Hooper says above. But a little history here. In the 1970s, libertarians of my vintage started reading about “revisionist” history regarding World War I, World War II, and even the Cold War--work by Harry Elmer Barnes and William Appleman Williams, for example. Many of us used the word “revisionist” as a badge of honor.

David R. Henderson writes:

@Daniel Kuehn,
I just read your latest. We must have been typing at literally the same time. I don’t agree with your claim that supplying a definition that shows that the word used exactly fits the situation amounts to getting “a little too worked up."

Daniel Kuehn writes:

David -
"Unnecessarily concerned with my comment" perhaps? The implications are often more than just someone that revises, and I'm stressing that I'm not accusing Bob of trickery or anything.

Surely you would agree that "revisionist" often carries negative connotations even if it didn't in a particular case in the libertarian community?

RPLong writes:

+1 on Roger Kinney's comment and David Henderson's follow-up. During most of my time in the work force, minimum wage has been above the market wage for every region in which I worked. I haven't received minimum wage since I was 16 years old, which is more than half a lifetime ago. I also agree that there is no "the" market wage. To Murphy's point, I think this makes empirical results highly inconclusive - at least so far.

I think Daniel Kuehn should spend more time addressing the economics of Murphy's article and less time on the vernacular. Are we talking about economics or vocabulary? I don't think Kuehn is the final arbiter for how words are used. Different people use words different ways. Even in this comments thread we can see that Kuehn and Henderson use the word "revisionist" differently, so let's be charitable with Bob's analysis and assume the best.

Besides, regarding the economics of it, it simply doesn't matter why Bob used that word. An economic discussion can be had, whatever his motives.

Daniel Kuehn writes:

RPLong -
I spent very little time on the vernacular until I was presented with three critical comments.

Over the last two weeks I've written about 12,000 words on the economics of the minimum wage (mostly the empirical issues), I've exchanged lots of emails with Bob on the technicalities of the studies in the run up to this and some other work he's doing on it. So I'm talking about the economics plenty.

I was happy to just have a two sentence comment on the vernacular, but that was apparently what people wanted to talk about and not my prior comment on the identification strategy.

I can't control which way the comment thread goes, RPLong.

Daniel Kuehn writes:

Anyway, RPLong, if you want to talk the economics I'm curious what you think of my response to Otto. It seems to me in the world of labor market policy evaluation the minimum wage has been really weird in that it's one of the few big policies that for a while was not evaluated using standard quasi-experimental methods and identification techniques.

The reason seems fairly intuitive I guess. It's not like you assign treatment to one worker and not another worker with the minimum wage the way you do with a lot of other policies. So thinking of it the way we usually think of estimating treatment effects is a little difficult.

Regardless, you would think the burden of proof would be on studies that don't use these methods. Our inclination should be to trust things like the contiguous county studies (or things like Card and Krueger, although that's obviously a much, much weaker identification) and put the burden of proof on the fixed effects studies.

That may be my bias. In my econometric education, identification and good quasi-experimental methods have been heavily emphasized. But I never thought that was unique - isn't that what everyone gets drilled into them?

What are your thoughts?

Daniel, I noticed the word with which you started your second comment, "Branding". I believe that word carries negative normative connotation, along with its empirical content. Would I be mistaken if I interpreted your choice of that word as intending to convey disapproval?

In my quest, I am wondering if it is possible for us humans to wash all norms out of our selection of words. I suspect it may be impossible. Perhaps a silly goal.

But here's a flag I will wave. Let's all come out of our closets, libertarians and statists too. I am libertarian. I am suspicious of the state, suspicious of people who find ways to argue for state intervention. You may expect to see my values in my sentences. I posit that we must expect bias in all humans.

Daniel Kuehn writes:

Richard -
By branding I just mean that it was intentionally chosen to convey something about the studies. I did not imply anything negative by it. It was clever on Bob's part. Lot's of information conveyed in a single word choice. It's not the word I would have chosen because I would not want to convey that information. I would have used a different sort of branding.

This discussion of that two sentence comment is getting boring.

What do you think of my thoughts at 10:16 AM, Richard?

RPLong writes:

Daniel - As of my writing this comment there are 14 comments under this post, seven of which are yours. Four of your comments deal specifically with Bob's use of the word "revisionist." So you have 50% of the total comments here, and about 26% of them are about vernacular. That looks to me like you very definitely can control which way the comment thread goes.

But on to your questions. You say:

[Y]ou would think the burden of proof would be on studies that don't use these methods. Our inclination should be to trust things like the contiguous county studies (or things like Card and Krueger, although that's obviously a much, much weaker identification) and put the burden of proof on the fixed effects studies.
I don't think that we should prefer studies based on their analytical elegance. In my professional experience, I have seen crude linear OLS trend lines out-perform elegant multivariate time-series analyses many times. I believe we should always default to the simplest available model that offers the best predictive results.

Meaning, I don't hold a model design bias; I hold a predictive value bias. I don't want a model that has a better fit to historical data, I want a model that has a better fit to future outcomes.

Despite the simplicity of his models, I believe that Kevin Erdmann at Idiosyncratic Whisk has been making a strong empirical case for the fact that minimum wage causes, not only teenage unemployment, but also a decrease in workforce participation. I think the many criticisms he has received (including yours) have been good, but Erdmann continues to follow-up on each one.

In the end, it is difficult for me to accept an elegant model that contradicts both economic intuition and a substantial body of empiric research based on simpler models that have maintained strong predictive power for decades.

That's not to denegrate the "revisionist" research out there. They might still convince me yet - provided, of course, they can do so by proving themselves the better predictive models. Your opinion may differ, and probably does, and that is fair enough.

Just my thoughts.

Daniel Kuehn writes:

RPLong
Right I'd agree with a lot of that, but nobody is arguing that a particularly study should be preferred for its elegance. That doesn't seem to be the issue at all.

One thing you didn't mention is the identification problem pose by what you're calling the "simple" models. So are you saying that identification is not important? And if you think it is important, why would someone that cares about that problem prefer the fixed effects studies to the quasi-experimental studies?

I don't really understand your point about out of sample prediction here. We're not forecasting employment in the restaurant industry or anything like that. We're identifying a treatment effect. What does prediction out of sample even mean in that context? I may need you to explain how that's related.

RPLong writes:

Sorry, Daniel, I have a difficult time thinking about this outside of a specific context. Which specific identification problems are you referring to? Maybe if you fleshed out an example, I might be able to better convey my preferences for model selection and explain why.

Regarding predictive power, the question for most of us non-academics is entirely predictive. To wit, what outcome can we expect from an increase in the federal minimum wage? When most of us debate minimum wage increases, that is the angle from which we approach the issue. So, yes, in a sense we are talking about forecasting employment in the restaurant and other low-skilled labor markets.

Yancey Ward writes:

I now view the word "clever" with negative connotations.

Daniel,
About your comment at 10:16 AM, I take your subject to be the evaluation of studies aimed to credit, or to discredit, intervention by the state into labor prices. You are now probably one of the better-educated people in this subject. I lack knowledge to follow you into this subject.

I also lack ambition to follow. I am not biased, as I suspect you may be, to find ways to justify state intervention, to find a sort of "intelligent management" guided by expertise such as yours. I am biased rather to free myself, and any who would join me, of the state.

LD Bottorff writes:

Clever: Referring to the "new" minimum wage studies that challenge orthodoxy as "revisionist."

Contentious: Referring to orthodoxy as "myth."

I don't know if Krueger and Card chose the title for their book; it sounds like something a publisher would do to encourage sales. I like to think that most serious economists (and that includes Krueger and Card) are more clever than contentious.

gofx writes:

Daniel Kuehn,

I think you are getting "a little worked up"! My point was in response to YOUR accusation about Murphy's motives, not an assessment of Dube. My baysesian statement was an assessment about the general preponderance of evidence, and the a priori theory that support the negative impact on employment of increasing the minimum wage. Its like heavyweight boxing. You have to beat the champ, you can't just tie him or hit him a few times.

You said, "Anyway, if you think that then why have the debate? Why engage with people you think are illogical and are missing basic economic points?"

Does this mean I shouldn't debate you anymore?

Mark V Anderson writes:

I read the linked article by Murphy quickly, but I didn't see anything about short-term versus long-term results. It seems every study I've read looks at the short-term effect of raising minimum wages, and yet, it seems obvious to me that the long-term results will be much more significant. Most of the reactions of business to a wage increase, whether it be of lowering the workforce, automation, or even re-marketing the product to be sold at a higher price, will be much more effective over a few years than over a few months. So what's the point of parsing these short-term studies so precisely?

Daniel Kuehn writes:

gofx -
I'm not sure what you mean when you talk about my accusations of Bob's motives. I only have the highest opinions of Bob's motives. I think whatever accusation you are thinking of is mostly in your head.

re: "Does this mean I shouldn't debate you anymore?"

If you think I'm illogical or missing basic economic points I can't imagine why you'd bother. It seems like a waste of time. Granted I don't see why you'd have grounds to think that.

Then again, I don't see why you'd think I'm making some kind of problematic accusation about Bob's motives either.

Daniel Kuehn writes:

Mark Anderson -
I don't quite understand the source of the concern. Dube, Lester, and Reich (2010) - the principle study under contention here - look four years out. Do you think the effects will take longer than that?

gofx writes:

Ok Daniel, we can still debate. But I warn you, there are a few times when I agree with you!

Floccina writes:

From the Bureau of Labor statistics:

In 2012, 6 percent of women paid hourly rates had wages at or below the prevailing federal minimum, compared with about 3 percent of men. (See table 1.)

You can make all the laws but that does not mean that you can get people to follow them.

I for one think that the benefits of employment are big enough that I would not risk even very small losses of employment for higher wages for those currently earning below the proposed MW. that is even though it would probably be good for and me and those who get a raise and keep their jobs.

Comments for this entry have been closed
Return to top