Economists Debate the Minimum Wage
By Robert P. Murphy
“That a branch of revisionist studies has arisen since the mid-1990s does not mean that the consensus has been reversed; it merely means that the prior consensus has been challenged.”
However, starting in the 1990s, this consensus began to unravel. A series of econometric studies, relying on new techniques for holding “other things equal,” challenged the existing orthodoxy.2 Once the researchers controlled for other trends, it appeared that in practice, modest increases in the minimum wage had a negligible impact on employment in the low-skilled and teen populations. Indeed, this revisionist literature has grown so influential that, recently, 75 economists—including seven Nobel laureates—publicly signed a letter to prominent federal politicians, urging them to raise the federal minimum wage to $10.10 by 2016.3
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.
Revisionist Studies Do Not Clinch the Case for Raising the Minimum Wage
Let us concede, for the sake of argument, that the new wave of research is correct, and that modest hikes in the minimum wage do not significantly impair teen (or other low-skilled worker) employment. Further, let us put aside any ethical or rights-based objections one might have to government interference in voluntary contracts between firms and workers. Even if we focus narrowly on low-skilled workers, it still does not follow that raising the minimum wage is necessarily a good idea.
In the first place, there is the matter of degree. Careful analysts will often summarize the new research in a nuanced way, saying “modest” increases in the minimum wage appear to have little impact on employment. But the proposed increase from $7.25 to $10.10 an hour is a 39-percent increase, which can be hardly be characterized as “modest.” Such an increase, therefore, could well destroy teenagers’ jobs, notwithstanding the revisionist studies.
There is a second, and independent, problem: Raising the minimum wage might represent a drastic harm to the most vulnerable and desperate workers if the specific employees who would be working for $10.10 an hour are different from those who would be working for $7.25 an hour. What could happen is that the higher wage would attract new workers into the labor pool, allowing firms to become pickier and, thus, to overlook the least-productive workers, who would remain unemployed or lose their jobs to more-highly-skilled workers.
I illustrate this point by using a supply-and-demand framework in which I have made the demand for low-skilled labor very inelastic.
In Figure 1 below, for diagrammatic simplicity, I assume that the initial market-clearing wage for low-skilled labor is $7.25 per hour. At this $7.25 wage rate, two million workers have jobs. Then, the federal government imposes a wage floor of $10.10 per hour. Because the demand for low-skilled labor is (by stipulation) very inelastic, the quantity of labor demanded falls by only 2,000 workers, a negligible drop of 0.1 percent. Economists running regressions on this episode would conclude—correctly—that raising wages by 39 percent had little impact on the absolute level of employment.
However, in our example, the supply curve (by construction) is a more typical shape, such that the large increase in the wage rate leads to a large increase in the number of workers seeking employment—500,000 in our scenario. There is now a significant amount of involuntary unemployment in the market for low-skilled labor; the unemployment rate would skyrocket.4
Even though (by construction) our hypothetical minimum wage has not significantly reduced total employment, it has, nonetheless, drastically impaired the functioning of the labor market. The “glut” of workers on the market means that non-price allocation mechanisms must come into play. Since there are now multiple applicants for a given job opening, employers can rely on other criteria, including racial and class background, to choose which worker gets the job. It is much more likely that an applicant will need to “know somebody” to get hired, and that teenagers from “respectable” backgrounds will be the ones to work at fast food restaurants, displacing teenagers who might be in more desperate circumstances.
These concerns are not merely hypothetical. Even many economists in favor of the wage hike agree that raising the minimum wage will affect the turnover of workers. For example, one of the leading revisionist authors, Arindrajit Dube, says that in one of his earlier co-authored studies “we… find that both hires and separations of low-wage workers (teens, restaurant workers) fall in response to [a] minimum wage increase, but employment levels do not change noticeably.”5
The Empirical Case on the Minimum Wage Is Far From Settled
Thus far, I have argued that even on its own terms, the revisionist empirical literature does not clinch the case for raising the minimum wage. Yet the situation is even worse for the proponents of hiking the minimum wage, for the empirical record is far from settled. There is currently a healthy ongoing debate in the empirical literature. The fact that a branch of revisionist studies has arisen since the mid-1990s does not mean that the consensus has been reversed; it merely means that the prior consensus has been challenged.
I will not attempt an exhaustive survey of the literature, but I do want to explain the main issues. First, a straightforward regression approach seems to validate the textbook treatment: hikes in the minimum wage are associated with a negative effect on the level of employment. This is intuitive, as it simply illustrates that demand curves in general slope downward. However, the revisionist studies argue that such a finding is spurious because of heterogeneity in the states that raise their minimum wage above the federal level. Once we refine the regressions to account for other factors that could affect employment, they argue, it is no longer obvious that hikes in the minimum wage retain any explanatory power.
To get a flavor of the disputes, we need to introduce some econometric terms. Introducing more independent variables into the regression for each of the items listed below isolates any remaining effect that could be explained by the minimum wage hikes.
Time trends: One can introduce a variable for the time period, so that national trends (such as the business cycle) will not distort the apparent effect of the minimum wage.
State-specific fixed effects: One can introduce a variable for each state, to account for permanent differences between them. For example, Florida has nicer weather than Minnesota, and so we would not want to try to explain faster employment growth in Florida solely by the minimum wage in both states.
State-specific time trends: One can introduce a variable for each state per time period, to prevent the minimum wage variable from apparently accounting for the changes in employment that are actually being driven by something else. For example, the loss of competitiveness by the Detroit automakers might cause a downward trend in employment in Michigan relative to other states over a certain time period. This is not a state-specific fixed effect because it is not a permanent feature of Michigan versus other states. Yet a national time trend would also miss it. Absent a state-specific time trend—where there are variables for Michigan-2008 first quarter, Michigan-2008 second quarter, etc., if we are using quarterly data—if Michigan just so happened to raise its minimum wage in the middle of the decline, then a regression might unfairly blame this policy move for too much of the (relative) decline in teen employment.
Construction of control groups: One of the most advanced techniques is to construct pairs of contiguous counties lying on opposite sides of a state border. This is the favored method of Dube et al. (2010) and subsequent articles; it is a generalization of the “case study” approach of Card and Krueger’s famous 1994 paper. Dube and his co-authors argue that regressions run on this smaller universe (rather than on the entire set of counties for which we have data) give a much crisper indication of the true effect of minimum wage hikes.
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.6
Neumark and Salas directly tackle the contiguous county control group method, but their arguments are difficult to summarize concisely. What we should note is their general warning about Dube et al.’s desire to control for “spatial heterogeneity.” Dube himself unwittingly demonstrates the huge potential pitfall when he argues, “[S]tates that have raised the minimum wage more over the past several decades are systematically different from other states.” Then, as one example, he says that “they have experienced… sharper reduction in jobs involving routine tasks.”7 Since this type of “state-specific trend” is quite plausibly the result of a state government raising its minimum wage relative to its peers, including these types of “controls” in the regression might mask the policy’s true effect.
In a very recent working paper, Meer and West (2013) make a different point that highlights the dangers in relying too heavily on a particular econometric specification. It is plausible that a hike in the minimum wage would show up primarily as a reduction in the growth rate of employment, rather than as a sudden drop in the absolute level. Why? For starters, it might hurt morale if a manager suddenly laid off ten percent of the staff following a hike in the minimum wage; instead he or she might decide to slow the rate of new hires and automate more, thereby hiring fewer workers than otherwise. Also, if the level of employment would have risen in the absence of the minimum wage hike, then a reduction in the level relative to the counterfactual trend might appear in the (coarse) data merely as a slowdown in growth, not as an absolute drop in the level.
Although it may seem counterintuitive, if the true effect of the minimum wage is to permanently lower the growth rate of employment, then a regression analysis looking for the impact on the level of employment may find virtually nothing. In their appendices, Meer and West give good illustrations of why this could happen.8 They revisit the data, including many of the control variables recommended by the revisionist authors—even including regional-specific time trends—and find that minimum wage increases do reduce the growth rate of jobs. However, if, instead, they look at the level of employment, they are able to reproduce the results in the revisionist literature—namely, that the minimum wage appears to have no significant effect once enough control variables are added to the regression.
Textbook theory and common sense suggest that hiking the minimum wage—particularly a substantial hike such as the proposal to raise it 39 percent over two years—will hurt job prospects for low-skilled workers, the very people allegedly helped by such a policy. However, since the mid-1990s, a branch of revisionist studies has argued that, empirically, this textbook argument is simply wrong.
Even if that is true, the results do not negate the arguments against such legislation because raising the minimum wage would still cause unemployment (which is not the same as a drop in total employment). The revisionist studies are consistent with the possibility that the most desperate of low-skilled workers—and the ones with the fewest connections and other advantages—will be displaced in favor of different workers, who are responding to the higher (legislated) wage. Furthermore, one should be very wary of accepting the results of the revisionist studies at face value. They depend strongly on the specification of the regression being run, and various modifications yield the original consensus finding.
In short, large increases in the minimum wage will likely harm many of the very people proponents wish to help. Certain empirical studies argue that the harm may be smaller than economists used to believe, but this remains very much an open question that is not at all settled—as some commentators would have us believe.
Allegretto, Sylvia A., Arindrajit Dube, and Michael Reich. (2011) “Do Minimum Wages Really Reduce Teen Employment? Accounting for Heterogeneity and Selectivity in State Panel Data,” Industrial Relations, Vol. 50, No. 2, April 2011, pp. 205-240.
Brown, Charles, Curtis Gilroy, and Andrew Kohen. (1982) “The Effect of the Minimum Wage on Employment and Unemployment,” Journal of Economic Literature, Vol. 20, No. 2, June 1982, pp. 487-528.
Card, David. (1992) “Using Regional Variation in Wages to Measure the Effects of the Federal Minimum Wage,” Industrial and Labor Relations Review, Vol. 46, No. 1, October 1992, pp. 22-37.
Card, David and Alan B. Krueger. (1994) “Minimum Wages and Employment: A Case Study of the Fast-Food Industry in New Jersey and Pennsylvania,” American Economic Review, Vol. 84, No. 4, September 1994, pp. 772-793.
Dube, Arindrajit. (2013) “Minimum Wages and Aggregate Job Growth: Causal Effect or Statistical Artifact?” Institute for the Study of Labor, IZA DP No. 7674, October 2013.
Dube, Arindrajit, T. William Lester, and Michael Reich. (2010) “Minimum Wage Effects Across State Borders: Estimates Using Contiguous Counties,” Review of Economics and Statistics, Vol. 92, No. 4, November 2010, pp. 945-964.
Katz, Lawrence F., and Alan B. Krueger. (1992) “The Effect of the Minimum Wage on the Fast-Food Industry,” Industrial and Labor Relations Review, Vol. 46, No. 1, October 1992, pp. 6-21.
Meer, Jonathan and Jeremy West. (2013) “Effects of the Minimum Wage on Employment Dynamics,” Texas A&M and NBER Working Paper, December 2013.
Neumark, David and J. M. Ian Salas. (2013) “Minimum Wages: Evaluating New Evidence on Employment Effects,” Employment Policies Institute, January 2013.
For example, Brown et al. (1982) surveyed a large number of time-series studies from the 1960s and 1970s and reported a consensus in the literature that a 10 percent increase in the minimum wage reduced teenage employment by 1 to 3 percent.
Some of the pioneering studies that challenged the orthodoxy were Card (1992) and Katz and Krueger (1992). The most famous study, involving a case study of fast food restaurant employment, was Card and Krueger (1994). Two recent studies, which epitomize the new approach and the econometric issues involved, are Dube et al. (2010) and Allegretto et al. (2011).
A press account of the letter is given in Jeanne Sahadi, “75 economists back minimum wage hike,” CNN Money, January 14, 2014. The full text of the letter itself is “Economist Statement on the Federal Minimum Wage,” Economic Policy Institute.
Of course, the reported unemployment rate would not necessarily increase because to be officially unemployed, someone must be not only out of work but also looking for work. Some low-skilled workers who lose their jobs will, after a while, give up looking. So the fact that the reported unemployment rate might not rise is small comfort to them.
This quote comes from Dube (2013), page 3.
The quotation comes from page 14 of Neumark and Salas (2013).
The quoted passages come from Dube (2013), page 4.
I walk through their demonstrations on this point of growth versus level in Robert P. Murphy, “Regression Pitfalls: Why Growth Rates versus Level Could Be Crucial,” Free Advice blog post, January 28, 2014.
*Robert P. Murphy is a Senior Fellow in Business and Economic Studies at Pacific Research Institute, and an economist with the Institute for Energy Research where he specializes in climate change economics. He is the author of The Politically Incorrect Guide to Capitalism (Regnery, 2007).
For more articles by Robert P. Murphy, see the Archive.