David R. Henderson  

Card and Krueger on Minimum Wage

Hummel on Fed Accounting... Scott Sumner vs. John Taylor: ...


How,' ask Card and Krueger early in the book, 'can the general public, most governments, and many other social scientists disagree with the negative view of the minimum wage that is so widely held by economists?' (p. 7). The reader might expect the authors to answer that question by looking at what the public, government officials, and other social scientists have said or written about the minimum wage. But that is not what they do. Instead, Card and Krueger analyze the thought processes of economists. '[E]conomists' views of the minimum wage,' they write, 'are based largely on abstract theoretical reasoning, rather than on systematic empirical study' (p. 7). True, but compare economists to the public, government officials, and other social scientists. Are the latter out there doing a 'systematic empirical study' of the minimum wage? Where are these studies published? Have we missed them? The simple fact is that virtually the only people systematically examining evidence on the minimum wage are economists. Moreover, the rest of us economists not doing empirical studies of the minimum wage are at least thinking about it in a systematic logical way--what the authors call 'abstract theoretical reasoning'. Are the authors saying that it's better not to reason, or are they saying that it's better not to reason abstractly? How do you reason non-abstractly?

The second reason economists think the way they do, write Card and Krueger, is that 'people have a tendency to see patterns that support simple theories and preconceived notions, even where they do not exist' (p. 8). Again, true. But are Card and Krueger saying that only economists are subject to this? Which is preconceived: the idea that the increased minimum wage will price people out of the market or the idea that when the minimum wage increases, everyone at the old minimum will get a raise and keep a job? If you had asked me, when I was 17 and hadn't read a smidgen of economics, what the effect of a minimum wage would be, I would have unhesitatingly answered that it would be to raise the wages of people making less than the minimum. That was my preconception. And I remember being stunned when, at age 18, I learned in a lecture by Ben Rogge, an economist from Wabash University, that the minimum wage hurts many of the unskilled black teenagers whom it is alleged to help. I don't think I'm unusual. Most people who haven't taken economics share what was my preconception.

This is from "Rush to Judgment," my 1996 review of David Card and Alan B. Krueger, Myth and Measurement: The New Economics of the Minimum Wage. It was published in Managerial and Decision Economics, Vol. 17, No. 3, 1996, pp. 339-344. Unfortunately, the gate opened for me for a while today and now it's gated.

Comments and Sharing

CATEGORIES: Labor Market

COMMENTS (19 to date)
Steve Roth writes:

Putting aside the quibbles and rhetorical tricks in this discussion, the distinction between the predictions of abstract theory and empirical results is important.

This especially when the theory has faulty or at least seriously questionable premises.


The best empirical studies we've got, in toto, say that at the relatively low minimum wages we've had over past decades, the earnings effect is significant, while the employment effect is so small as to be unmeasureable.

Here's the latest I know of, seems to be very well-executed:


The facts on the ground are not obviated by theory that resorts to the fallacy of the extremes: "if the minimum wage was $50/hour, there'd be less employment, right?" Yeah, right.

Jody writes:

Steve: Here's more recent data that suggests a significant negative impact on the minimum wage



David R. Henderson writes:

@Steve Roth,
I used no rhetorical tricks. Indeed, my purpose in highlighting this paragraph is to show that one can make rigorous arguments without math and by taking people's words seriously.
Also, if you scour everything I've ever written about the minimum wage, you will never find me saying anything like, ""if the minimum wage was $50/hour, there'd be less employment, right?" And my reason for not doing so is precisely the one you give.

Ricardo Cruz writes:
Steve Roth writes: The best empirical studies we've got, in toto, say that at the relatively low minimum wages we've had over past decades, the earnings effect is significant, while the employment effect is so small as to be unmeasureable.

Card and Krueger claim is that at the relatively low minimum wages we've had over past decades, higher minimum wages have raised wages and unemployment. Hence why they say their book flips conventional economics upside down.

Ricardo Cruz writes:

I meant to say: Card and Krueger claim is that [...] higher minimum wages have raised wages and employment.

Jacob Oost writes:
The best empirical studies we've got, in toto, say that at the relatively low minimum wages we've had over past decades, the earnings effect is significant, while the employment effect is so small as to be unmeasureable.

Assuming what you say is true (and it isn't), it's only because the studies merely look at who it considers participants in the labor force. When you overlook the homeless, the mentally retarded and developmentally disabled, ex-cons, hordes of male inner-city teens with no education, etc., people whose productivity or riskiness (I'm splitting hairs) makes them worth much less than the minimum wage, then you lessen the visible effects on employment.

In my last job working with the MRDD population, I saw people who were simply unhirable *at the existing minimum wage* but could have at least gotten started on the job ladder if they were legally allowed to work for two or three dollars an hour doing simple work.

Julien Couvreur writes:

Steve, the 50$/hr argument is not meant as an end-all argument, but to get the point across to people that minimum wage laws may have benefits for some, but also have negative effects (which are made obvious by the extreme).
People often fail to consider the negative effects, and thus the 50$/hr argument is useful.

Rob writes:


A very thought provoking review. It is a tenuous relationship between theorizing and testing theories in the data. Despite all the technical and computational advances of modern econometrics, the idea of rigorously testing theories derived on the black board with real world data remains elusive. As you aptly point out however, the typical partial equilibrium analysis of the consequences of a minimum wage increase requires no more mathematical sophistication than one would obtain by middle school. A vintage (i.e. pre New York Times) Paul Krugman would say, "The argument is simple but not simplistic." Clear economic analysis helps us organize our thoughts and impose necessary structure on the complexities of the real world.

On a deeper level my biggest criticism of the applied micro literature, which CK is indicative of, is there reluctance to adopt a theoretical model before doing the econometrics. This "letting the data speak for itself" approach has broadened the base of professions that economists take seriously. For instance, if you are a political scientist, sociologist, psychologist, etc. who knows the ins and outs of contemporary econometric tricks, applied micro people will generally take you seriously. The upside is that this gives us (by us I mean economists and economists in training) different perspectives on social phenomena. The down side is there is no null hypothesis so to speak. There is no reconciliation of theory with data. I think the down side dominates.

Cahal writes:


That research completely fails to take the recession into account.

I think the best way to model the min. wage is with an AD/LRAS curve. Min. wage increases AD (wages being an important component), but the impact it has on the LRAS can vary, because it increases productivity but also costs.

Thus, it depends on whether the productivity increases outweigh the cost increases. This is true at lower levels but there are diminishing returns as you go higher. It becomes (even more) obvious why the minimum wage shouldn't be $500/hour.

It also explains why studies show the minimal effect on employment but the expected increase in prices.

Maniel writes:

"There are three kinds of lies: lies, damned lies and statistics." Disraeli
The minimum wage disrupts the free market for labor; in particular, this law makes it a crime to hire help - or to work - at below that minimum, thereby destroying jobs (or simply preventing them from appearing in the first place). You may present all the statistics you want, but it does not pass the logic test.

Ben writes:

Maniel, you are assuming PERFECTLY COMPETITIVE labor markets. Once we have search frictions, we have a dynamic monopsony in the labor markets (where firms have wage making power, because labor faces search costs, and thus cannot easily find another job), once we have that, the minimum wage will increase employment. It's a model that assume zero search frictions that fails the logic test.

Chris Koresko writes:

I'm with Rob when he argues that empirical studies would be best presented in comparison with the predictions of models. Without a model, it's not possible to extend any conclusions beyond the data in the study itself.

Chris Koresko writes:

@Ben: Your assertion sounds interesting, but I don't understand it. Would you mind expanding on it? I don't see how search costs coupled with a minimum wage can increase employment. Does this account for the search costs of the employers as well? How does it change in the age of monster.com?

Thanks in advance.

Dale writes:


The argument goes pretty much like this.

If there are search costs for labor in excess of capital then employers can offer below the competitive market wage. They can do this because not accepting the offer has higher costs for the laborer than the employer.

Under this situation a minimum wage acts just like a maximimum price for a monopoly. Since the only way for the employer to reduce prices is to also reduce supply, an increase in the minimum wage keeps the marginal return on labor below the return to businesses yet increases the number of people who would be willing to work. Just as setting a maximum price for a monopolist will induce them to sell more product and lower prices (providing that said maximum price is binding)

Besides that there are a number of strange effects surrounding the minimum wage that seem to suggest, at the very least, that the simple model of job losses does not hold, even if we can generalize that it is the case for very high minimum wages. Of course we don't have those minimum wages and so such an argument (absent the current recession) is a bit specious.

Tom West writes:

I would also be hesitant to discount the psychological element of minimum wage. My experience with countries that have no minimum wage is that there seems to be an almost inevitable impulse to value people who work for next-to-nothing as being worth next-to-nothing. This seemed to apply both to employers and the population at large.

Maybe it's the endowment effect, but if you're paying someone $7.25 an hour, I think most employers will value them *more* highly than those doing the same work for $2.50. Just human psychology, I guess, but no less a real phenomenon for it.

Anyway, that's one reason that despite the harm that it brings by locking some members out of the market, I think minimum wage laws do a society more good than harm.

Mick Rol writes:

The studies of Card & Krueger focused on comparative studies of effects on employment in the fast food sector (a relatively homogeneous sector for labor) of different minimum wages across US states, focusing particularly in cases of raises in minimum wages in some states. The 'control groups' were states without raises in the minimum wage.

Their result was surprising, as it contradicted most of the weighty available empirical literature since a long time ago (Stigler and his famous study on the negative effects of minimum wages on minorities, Neumark, and many others). The Card & Krueger study has been challenged many times (for instance http://www.cato.org/pubs/journal/cj15n1-8.html ), but most of all it is surprising that telephone surveys of the fast food industry have given them such authority to erase decades of empirical investigation.

I would recommend to explore the case of Hong Kong without the minimum wage, and after its introduction. Or the case of France's very high minimum wage and its effect on employment for minorities in the banlieues.

To Ben: Classical reasoning does not necessarily assume a perfectly competitive model in all its assumptions. (It is a very unfair criticism of classical reasoning to imply that we believe that markets are perfect and adjust instantaneously.) In any case, the assumptions of the Joan Robinson monopsony model are much more restrictive and unrealistic even than the perfect competition model. At least the experience of Hong Kong or of informal sectors shows that monopsony and exploitation is a far-fetched assumption.

Chris Koresko writes:


Thanks for taking the time to explain.

Without yet having thought it through carefully, it seems to me that this reversal of the effect of the simple supply-and-demand-for-labor model is plausible, but only under a set of circumstances which are fairly narrowly defined.

If we accept the model and conclusion as you describe them, the it seems natural to suppose that if the search cost falls more heavily on the employer than the employee then the equilibrium wage will rise above the optimum level, and employment would be increased by forcing the wage rate down. Correct?

If that is so, it strikes me as a pretty weak argument in favor of a minimum wage law, unless one can demonstrate that the economy is on the right side of the search cost balance, and then argue that the benefits derived from the law outweigh the violation of the peoples' freedom of association.

Chris writes:


This is true. The easiest way to talk about (at least what I believe is) the most likely situation is that you have a number of markets some of which are fairly competitive and some of which are not. It is the mix of markets (or populations within markets) which gives us aggregate effects that are hard to find. I mean, if the entire situation was competitive it would seem like it would be very easy to see the predicted effects similarly as if the entire situation was monopsony.

That being said it is much more likely that employer power exists within the ranges that we are talking about. We can figure this just by thinking about what kind of marginal costs we could assume for each. And i have a hard time figuring that the relative marginal costs would fall harder on employers than employees. This would of course be for the normal situation. If a recession pushed the competitive equilibrium wage below the minimum wage no amount of monopsony power will matter.

That being said, i don't doubt that there are professions which would benefit from such a maximum wage. I just don't think they're really large enough to really worry about or homogeneous enough that we could have much of an effect.

Note also that I am not using this to argue for a minimum wage. Even if a minimum wage is not strictly negative in terms of aggregate employment it is almost certainly going to be distributionally inefficient.(by way of substitution effects*)

The best arguments for the minimum wage are that the first best solution to the problems (poverty etc) are simply politically untenable and that a minimum wage is all that we can get.

You could make an argument that employers don't account for the marginal effort of employees when paid more, but i don't think that one is quite as strong.

*substitution effects in this instance mean that employers are going to tend to substitute to more productive employees. This is one of the reasons that teenage unemployment studies are problematic. The industries involved tend to hire people who are no longer in high school as the minimum wage increases (I.E. the net employment loss for the market is lower than the net employment loss for the sub-population).

However, these substitutions imply that the resulting system, even if we are measuring the same macro output, is generating less welfare.

Though we should grant as well that we may simply not care about the welfare of specific damaged classes over specific advantaged classes.

bhagness writes:

has anyone thought about a situational minimum wage, one based on a persons life situations and expenses

Comments for this entry have been closed
Return to top