By Kathy G.
I’ve been remiss in replying to this post by Megan McArdle, but today I’ve finally gotten around to it. This will be a really long post, so don’t say I didn’t warn ya.
McArdle basically argues two things: that 1) the minimum wage has a disemployment effect, and 2) that monopsony is not a persuasive model for the labor market (or at least for the low-wage retail sector). First I’ll deal with the evidence on the minimum wage. McArdle mentions the famous 1994 Alan Krueger and David Card study which looked at the impact of a 1992 increase in the minimum wage on employment in fast food establishments in New Jersey. Krueger and Card found that in that case, contrary to what standard theory predicts, the increase in the minimum wage did not decrease employment.
Very reasonable criticisms of that study have been made. McArdle summarizes:
The original study was a phone study; when another study asked for actual payroll records, they found the same result the standard model would predict: fast food employment dropped in New Jersey. Additionally, as Kevin Murphy has pointed out, the survey started long after employers knew that a minimum wage hike was coming—he compares it to assessing a midnight curfew by comparing the number of teenagers on the street at 11:59 to the number on the street at 12:30.
In response, Krueger and Card did another
study that looked at the impact of that same minimum wage increase on employment in fast food establishments in New Jersey. To counter the previous criticisms from economists like Kevin Murphy who said that their data was problematic and that they’d got the timing wrong, this time they used a more reliable data source (employer data from the Bureau of Labor Statistics) and looked at the data over a longer time period. And guess what? This new analysis confirmed their original findings: the increase in the minimum wage did not lead to a decrease in employment.
There are a number of other reliable scholarly studies on the minimum wage that report similar results—such as this one, this one, this one, this one, and this one, for example. There are also quite a few very good studies that show the opposite—that an increase in the minimum wage does indeed bring about a decrease in employment. A fair characterization of the literature is that the minimum wage’s impact on employment is ambiguous. But the fact that the findings are mixed is fairly compelling evidence that there must be something wrong with the standard perfect competition model of employment. And that’s because the textbook perfect competition model predicts that an increase in the minimum wage will always and everywhere lead to a decrease in employment, no ifs, ands, or buts about it.
Another thing that must be pointed out: given the anti-regulation ideological bias of the economics profession as a whole, it’s not hard to imagine that studies that do find that the minimum wage has a disemployment effect are considerably more likely to be published. I’m not accusing anyone of scholarly fraud here. But the fact is, there are lots of different datasets you can use, lots of models to go with, lots of variables to include or leave out, and lots of ways to slice and dice the data. It’s not unheard of for researchers to opportunistically try different models and methodologies until they hit upon one that gives them the results they want.
Here is what economist Edward Glaeser had to say in a recent paper about researcher incentives and empirical methods:
Economists are quick to assume opportunistic behavior in almost every walk of life other than our own. Our empirical methods are based on assumptions of human behavior that would not pass muster in any of our models. The solution to this problem is not to expect a mass renunciation of data mining, selective data cleaning or opportunistic methodology selection, but rather to follow Leamer’s lead in designing and using techniques that anticipate the behavior of optimizing researchers.
Indeed, Krueger and Card have written a
paper that provides strong evidence that “specification searching and publication bias” have led to an overrepresentation of studies that find that the minimum wage has a statistically significant disemployment effect. The ideological character of much of the economics profession in the United States suggests that there are rewards for producing scholarship that confirms the idea that the minimum wage causes unemployment, and punishment for scholarship that finds otherwise.
David Card, a highly regarded economist at Berkeley (among other honors, he won the John Bates Clark Prize, a prestigious award given every two years to the most outstanding economist under 40), has produced many of the best studies taking issue with the old conventional wisdom about the minimum wage. But he stopped studying this subject, to a large degree because the reception his research got was so hostile in some quarters of the economics profession. He said:
I’ve subsequently stayed away from the minimum wage literature for a number of reasons. First, it cost me a lot of friends. People that I had known for many years, for instance, some of the ones I met at my first job at the University of Chicago, became very angry or disappointed. They thought that in publishing our work we were being traitors to the cause of economics as a whole.
“Traitors to the cause of economics as a whole”! Those are strong words, especially coming from someone who seems, on the basis of interviews at least, to be a fairly mild-mannered, non-drama-queen kind of guy. And if someone who’s a tenured full professor and one of the leading lights in his field took so much heat that he abandoned this line of research, what do you think the chances are that aspiring Ph.D.s and ambitious young assistant professors are going to touch this issue with a ten-foot pole?
I mentioned before that I found some of the criticisms by Murphy et al. of the 1994 Krueger and Card study to be quite legitimate. But they made other criticisms that have not been so reasonable. Here is Murphy et al. on what economic theory has to say about the minimum wage:
The implications of the theory are also simple and direct. The prediction that an artificial increase in the price of something causes less of it to be purchased is the most fundamental prediction of economics; it is called the law of demand.
Well, actually, it’s not so clear that an “artificial” increase in price will necessarily cause less of the good to be purchased. For one thing, it depends on the elasticity of demand for the good. If demand is perfectly inelastic, an increase in price would
not lead to a decrease in demand.
More importantly, though, it’s a huge mistake to view the purchase of a unit of human labor as being exactly the same as the purchase of a widget. What economics has done is to take the models of the supply and demand of consumer goods and apply them to the supply and demand of labor. This, I believe, is fundamentally wrong-headed. Human labor and consumer goods are categorically different, and it’s a big mistake to treat them as if they were interchangeable. There are a slew of institutions, norms, and other features of labor markets that do not apply to product markets.
When I first read that paper by Murphy and company, I was struck by the passages in it about “the law of demand” and how you can’t “repeal” the law of demand. It was so literal! Now, I should mention that I’ve taken one of Kevin Murphy’s classes and I am familiar with his work. I have great respect for him. He is a first-rate economist, a brilliant econometrician, a gifted teacher, and, so far as my limited dealings with him go, a really nice guy to boot. But he is a University of Chicago economist in every sense of the word. I’ve heard him speak, and he is quite contemptuous of the idea that regulation can ever improve anything, or that the government can ever do a better job of anything than the free market.
I also believe, based on his writings, that Kevin Murphy, like all too many economists, takes the models literally. He is so enamored of them that he sees them, I think, not as tools for understanding, but as God’s revealed truth, handed down to Moses on stone tablets. He’s an economic fundamentalist, if you will.
Fortunately, though, the old-fashioned theories about labor markets that Murphy and others hold are gradually being displaced. A 2000 survey showed that less than half economists agreed that an increase in the minimum wage will always increase unemployment among young and unskilled workers; just ten years before, over 60% of economists believed that. And in 2006, over 650 economists, including Nobel laureates Joseph Stiglitz, Kenneth Arrow, Robert Solow, Lawrence Klein, and Clive Granger, signed a statement supporting an increase in the minimum wage. They wrote, “We believe that a modest increase in the minimum wage would improve the well-being of low-wage workers and would not have the adverse effects that critics have claimed.”
In the interview I cited earlier, David Card gives a good description of how the new theories about the labor market differ from the old models, and what his research on the minimum wage has to do with this:
What we were trying to do in our research was use the minimum wage as a lever to gain more understanding of how labor markets actually work and, in particular, to address a question that we thought was quite important: To what extent does the simplest model of supply and demand actually describe how employers operate in the labor market? That model says that if an employer wants to hire another worker, he or she can hire as many people as needed at the going wage. Also, workers move freely between firms and, as a result, individual employers have not discretion in the wages that they offer.
In contrast to that highly simplified theoretical model, there is a huge literature that has evolved in labor economics over the last 25 years, arguing that individuals have to spend time looking for job opportunities and employers have to spend time finding employees. In this alternative paradigm a range of wage offers co-exist in the market at any one time.That broader theory is, I think,pretty widely accepted in most branches of economics. . . . The theory explains a lot of things that don’t seem to make sense, at least to me, in a simple demand and supply model.
For example, what does it mean for a firm to have a vacancy? If a firm can readily go to the market and buy a worker, there’s no such thing as a vacancy, or at least not a persistent vacancy. During the early 1990s, when Alan and I were working on minimum wages, it was our perception that many low-wage employers had had vacancies for months on end. Actually many fast-food restaurants had policies that said, “Bring in a friend, get him to work for us for a week or two and we’ll pay you a $100 bonus.” These policies raised the question to us: Why not just increase the wage?
From the perspective of a search paradigm, these policies make sense, but they also mean that each employer has a tiny bit of monopoly power over his or her workforce. As a result, if you raise the minimum wage a little—not a huge amount, but a little—you won’t necessarily cause a big employment reduction. In some cases you could get an employment increase.
I believe that that model of the labor market is correct. There are frictions in the market and some imperfect information.
Now, getting back to monopsony: first of all, for readers who are unclear about what the concept means, I suggest that you read this
post on my blog, which explains the basics and some of the policy implications. But here, I will just say this: just as monopoly means “one seller,” monopsony means literally “one buyer.” In the context of labor markets, it suggests one buyer of labor, i.e., one employer.
Except that’s confusing, because when economists use the term today and apply it to labor markets, they generally don’t mean literally one employer. Rather, they mean that the supply of labor to an individual firm is not infinitely elastic—i.e., if an employer cuts wages by one cent, all the workers at that firm won’t immediately quit. The monopsony model holds that because employers set wages, and because of important frictions in labor markets (such workers’ heterogeneous preferences, incomplete information, firm-specific human capital, and mobility costs), employers have some degree of monopoly power over their employees. Which means that they can set wages below the levels that would occur in a labor market where there is perfect competition.
Theory predicts that if a minimum wage is set in a monopsony labor market, it can actually increase employment. It won’t necessarily do that, though—if you set it too high, employment will decrease. But that’s a huge contrast with the textbook perfect competition model, which predicts that a minimum wage increase will always, inevitably decrease employment.
Some economists believe that the monopsony model is, for the most part, a better fit for labor markets than the old perfect competition model is. After all, how realistic is it that, if your employer cuts wages by one cent, everyone at your workplace will immediately quit? Btw, I have asked variations of that question numerous times on previous blog posts, and Megan McArdle has not yet answered it. To be fair, all models are stylized and radically simplified representations of reality, and leave out many features that are very important in the real world. Still, some models are better than others at getting the basics right.
Getting back to the 1994 Krueger and Card paper— it’s not clear that a monopsonistic labor market explains their results. In fact, in their paper they consider, but then reject, such an explanation. There have also been other explanations for why increasing the minimum wage doesn’t always decrease employment, including matching models (which emphasize search costs), efficiency wage models (“where firms suffer from diseconomies of scale in monitoring workers and, therefore, must increase wages when expanding their workforce to maintain the required penalty for shirking”), and training enhancing models (“where a binding minimum wage induces workers to raise their productivity to the level of the minimum by acquiring education which otherwise would not have been taken”).
The other theories have their strengths, but I think a monopsony model is a better fit for most sectors of the labor market in the U.S. McArdle rejects this, at least in respect to the fast food and low-wage retail sectors. First of all, she points to “extremely low search costs on both sides” in the low-wage sector. True, the search costs tend to be low. But it’s also true that information is not complete. Employees don’t always know what other low-wage retailers are paying, and once they’ve found a job they generally don’t go out of their way to seek out info on the going rates at other workplaces.
Mobility costs are another factor. For a lot of people, particularly in less densely populated areas, it’s important that their workplace is not too far from home and doesn’t require a long commute.
I think heterogeneous preferences can be important here, but McArdle says:
This seems unlikely. It’s not like you’re taking a lower wage at Wendy’s because they have a great dental plan and they let you use the pool. The labor is unskilled, the wages are undifferentiated, and the benefits are nonexistent. Maybe there are some people out there who love Wendy’s food, or Gap clothes, so much that they never want to consume anything else, making the employee discount super valuable. But I cannot believe that this group is sufficiently large to be driving the market.
Heterogeneous preferences go far beyond preferences for certain kinds of benefits or types of work. It can also mean preferring to work for a particular boss, or with certain co-workers, or on a certain schedule. When I was in college, for several years I worked at a crappy data entry job. The pay was low, and I could have gotten paid much more doing something else, but my work schedule, 8 pm to midnight every weeknight, was ideal, because it was compatible with school.
Similarly, I was a research assistant for a project that looked at the low-wage retail sector, and one thing we discovered was that scheduling policies were very important to workers. It’s common in the low-wage retail sector for employees not to know from one week to the next what their schedule will be—they are expected to be “available” at all hours. Needless to say, this can be extremely problematic for single mothers with child care issues, or for students seeking to balance school and work. But although unpredictability in scheduling was the norm, their were some bosses and some workplaces that enabled workers to make their schedules in advance. So if a worker who desires predictability finds a job that provides it, she’s going to be reluctant to switch jobs, even if the pay is higher.
McArdle also mentions that collusion and cartels are unlikely. I don’t disagree with her there; in fact, I have never once mentioned collusion or cartels as a cause of monopsony. Economists who are proponents of the monopsony model don’t make the collusion argument, either. For example, Alan Manning, who has written the standard work on monopsony in labor markets, mentions employer collusion only once in his book, and that’s in passing, and he dismisses it as an explanation, anyway. Which is not to say that employer collusion never happens; it sometimes does. But so far as I know, it is rare, especially in the low-wage retail sector.
Finally, McArdle says she doesn’t think that the monopsony model is a good fit for the low-wage retail sector, since there is so much turnover there: “empirical evidence,” she says, shows that “most people do not stay in only one industry, much less only one firm.” But monopsony doesn’t require that turnover rates be low. All it requires is that there is some—maybe only just a little—friction in the labor market, such that worker turnover is not 100% sensitive to the wage.
If McArdle believes that raising the minimum wage will always create unemployment, which is what the perfect competition theory predicts, then I’d love for her to explain the many studies (which I cite above) that show that it doesn’t. Yes, over the years there have been even more studies that show there is a disemployment effect, but that’s not necessarily inconsistent with monopsony. The problem that the economic fundamentalists have to deal with is that the perfect competition model makes a number of very strong, and highly implausible, assumptions. It assumes that employers don’t set wages, for one thing, and although in some settings that’s obviously not true, in most cases it is the norm.
More important, the perfect competition model assumes a frictionless job market, with complete information, no search costs, no mobility costs, no heterogeneous preferences, no firm-specific human capital. Indeed, they assume a job market so frictionless that if an individual employer cuts wages by even one cent, every employee at that firm will immediately quit. Which is a more realistic assumption, do you think—that we live in a world where there are certain frictions in the labor market that to some extent bind employees to their employer? Or that we live in a world so frictionless that if an employer cuts wages by a penny, all employees at that firm will immediately quit?
I’m still waiting for an answer for that one.