The Cato Review of Business & Government

---

Sense and Nonsense on the Minimum Wage

Donald Deere, Kevin M. Murphy, and Finis Welch

Donald Deere and Finis Welch are professors of economics at Texas A&M University. Kevin M. Murphy is a professor of economics at the University of Chicago.

In his recent State of the Union Address, President Clinton urged passage of another increase in the minimum wage. He stated that he had studied the issue and is convinced that modest increases in minimum wages do not decrease employment; in fact, Clinton argued, they may attract new workers into the workplace. The president’s proposal to increase the minimum wage to $5.15 an hour follows on the heels of a trial balloon announcement by the secretary of labor, Robert Reich. Both Clinton’s and Reich’s descriptions of minimum wage effects paraphrase the conclusions of a series of case studies that have culminated in a new book, Myth and Measurement: The New Economics of the Minimum Wage, by Princeton University economists David Card and Alan Krueger. Professor Krueger is currently on leave from Princeton and is Robert Reich’s chief economist at the Department of Labor.

Both economic common sense and past research contradict the Princeton studies, and an examination of the evidence surrounding the 1990-91 increase in the federal minimum wage shows that rumors of the death of the conventional economic wisdom are premature.

 Effects of the 1990-91 Increases in the Federal Minimum Wage

 On April 1, 1990 the federal minimum wage was increased from $3.35 per hour to $3.80; a year later it was increased to $4.25, for a combined increase in excess of 25 percent. The latter increase was the 17th since the Fair Labor Standards Act of 1938 established the first national minimum wage of twenty-five cents an hour. Though the original minimum applied to less than half of all workers, the federal minimum now applies to more than 90 percent of private employees (some seasonal, domestic, and a few other types of workers are still exempt). Most states have minimum wage provisions, but they are usually no higher than the federal minimum. The period immediately before the 1990-91 increase is an exception. During the nine years between January 1981 and April 1990, the federal minimum remained at $3.35, and by 1989 12 states and the District of Columbia had forged ahead with higher state minimums. The current federal minimum of $4.25 took effect in April 1991, and a few states such as New Jersey and Alaska have subsequently opted for a higher minimum. The descriptions of wages and employment that follow are based on the Current Population Surveys conducted by the Bureau of the Census and used by the Bureau of Labor Statistics to calculate monthly unemployment statistics.

During the 12 months before the 1990 increase in the federal minimum, 76.1 percent of hourly employees who were paid less than $5.00 an hour received a wage below $4.25, and 16.9 percent of workers making less than $5.00 an hour were paid exactly the prevailing minimum of $3.35, while 10.8 percent fell below the minimum. During the 12 months after the 1991 increase to $4.25 an hour, 36.3 percent of the workers paid $5.00 an hour or less received exactly $4.25, while 25.1 percent received less than the minimum. This shift in wages is much larger than the 9 percent average increase in all wages. The relative increase in wages of low-wage labor, together with the concentration at the applicable minimums, implies that there is substantial compliance with the wage laws and—what amounts to the same thing—a substantial increase in the cost of employing those who would otherwise earn less than the minimum. The increase in the cost of employing relatively unskilled workers leads us to expect that their employment will decline.

This examination of the 1990-91 changes begins with the hypothesis that minimum wage increases have the largest effects on employment where effects on wages are also largest. The analysis focuses on changes in employment rates for high- and low-wage groups of workers defined by age, education, ethnicity, marital status, gender, and state of residence. The study compares employment changes within each group to the fraction of workers who earn low wages (less than or equal to $4.25 in the year before the 1990 increase). For example, while only 8 percent of all men and 13 percent of all women received low wages, 44.5 percent of teenage men and 51.8 percent of teenage women were paid $4.25 or less in 1989. Also, 20.7 percent of men and 35.4 percent of women without high school degrees earned $4.25 or less.

Table 1 shows the proportion of low-wage workers in various age groups and the change in the percentages of all individuals in those groups employed just before and just after the minimum wage hike. Teenagers as a group have lower wages, and they experienced the greatest employment loss. The percentage of all teenage men with jobs fell from 43 percent before the minimum wage increase to 36.3 percent after—a 6.7 percentage point decline in employment of teenage men. Another way to express this is that compared to all teenage men with jobs before the hike, 15.6 percent fewer had jobs after the hike. Employment in the 20 to 24 age range also fell relative to employment of older workers, who typically earn higher wages.

As is true with age, when we divide the population on the basis of education, ethnic group, or marital status, we find that the group with the highest percentage of low-wage workers is also the group in which employment shows the greatest drop. Table 2 shows the percentage paid low wages and the associated employment decline for four groups with relatively large fractions of low-productivity workers. The numbers for all workers ages 25 to 64 are provided for comparison. Since the employment losses for all workers are smaller than the losses among the low-wage groups, it is clear that the mix of employment shifted in favor of those who would earn the most in any case. Notice, however, when we compare the columns in Tables 1 and 2 for men and women we find that, although a larger percentage of women earned lower wages, they generally experienced smaller employment losses. Similarly, when we distinguish between residents in states with high and low wages (not shown) we do not find larger employment losses among workers in states with disproportionately large numbers of low-wage workers. We do not view our gender and state findings as repealing the law of demand; instead we take them as a warning that minimum wages are not the only factors affecting employment, and that other factors must be considered before we can correctly assess the employment effects of minimum wages.

Our research shows that the gender differences are dominated by the broader trends of increasing labor market participation of women. Similarly, the residence comparisons show that relative employment expansion in the South and Southwest, where wages are also lower, during 1989-91 swamps the minimum wage response. It is true, of course, that when men and women are examined separately, the minimum wage increases coincide with shifts toward less employment for lower-wage workers, and it is also true that employment of low-wage workers in both high- and low-wage states fell relative to employment of workers who typically earn more.

The simple comparisons show relative employment reductions of low-wage workers in most cases, but not in all of them. In the exceptions we find clear explanations of results that might otherwise appear to be surprising. It follows that alternative explanations ought to be considered for the cases where employment responses are not surprising. The most likely explanation is the recession that began in 1990, because employment losses in recessions are typically greatest for low-wage workers.

To adjust for the recession, we predict low-wage employment based on the aggregate employment of men. From 1985 to 1990, when the minimum wage was increased, aggregate employment of men increased, while the increase in low-wage employment was proportionately greater. Following the 1991 rise in the minimum wage, aggregate employment was falling because of the recession. Based on the decline in aggregate employment, we calculate the more-than-proportionate employment loss for low-wage workers that would have been symmetric with the earlier expansion and compared this to actual employment losses for low-wage workers. The comparisons are conducted at the state level so that differences in state economic performance can be taken into account. Similarly, the comparisons for women account for the trend toward increased employment of women relative to men. The difference between the actual and implied loss in employment measures the effect of the 1990-91 increases on low-wage employment.

Table 3 gives estimates of the net (of recession) employment losses for six low-wage groups. Three of the groups are teenagers: men, women, and blacks (both men and women). The other three groups are adult (ages 20 to 54) high school dropouts: men, women, and blacks.

Based on the pattern of aggregate employment, we estimate that during the year of the $3.80 hourly minimum, 4.8 percent fewer teenage men were employed than would have been if the $3.35 minimum had been retained. The corresponding reductions for teenage women and teenage blacks are 6.6 percent and 7.5 percent, respectively; while for adults who did not finish high school, the reductions are 1.5 percent, 2.5 percent, and 4.4 percent, for men, women, and blacks, respectively. When the minimum wage was raised a second time, to $4.25 per hour, employment of low-wage workers fell a second time. The employment reductions over those implied by the $3.35 minimum and the decline in aggregate employment are: 7.3 percent, 11.4 percent, and 10.0 percent for teenagers; and for adult high school dropouts they are 3.1 percent, 5.2 percent, and 6.7 percent.

Our conclusion is simple and direct: to the extent that increased minimums raise the cost of hiring low-productivity workers, fewer of those workers will be employed. Note, however, that our finding of greater job losses for teenagers is due to the fact that a large fraction of teenagers have low wages, and it is not because the teenagers who earn low wages are necessarily more likely to lose their jobs than are the adults who also earn low wages. Teenagers constituted only 32 percent of those earning $4.25 or less in the year before the 1990-91 increases. Hispanic workers constituted 20 percent, blacks 16 percent, and adult women without a high school degree 12 percent of this low-wage population. We have not asked whether increased minimums reduce the employment of low-wage teenagers more or less than the employment of low-wage adults. The question of which workers earning low wages are most harmed by increased minimums is more subtle and demands further research.

The Conventional Wisdom on Minimum Wages

First among the reasons given for raising the minimum wage, indeed for having one in the first place, is to assist the working poor. Though it is quite appealing to suggest that everyone should earn a decent wage, it is difficult to imagine that it is this easy. If it is, then why do so many countries have workers earning so little? Have these countries just missed the boat on appropriate legislation? If it is possible to mandate high wages, then why not also have low prices for food, shelter, clothing, and everything else that is good?

Minimum wage laws focus on wages, not employment; if someone is employed, then he will receive at least the guaranteed wage. The law sets the terms of whatever employment happens to occur. The reduction in employment that results from increases in the minimum wage, which is concentrated among those workers with the fewest skills, is the cruel "dark side" of such legislation.

As a general rule, people do as well for themselves as they can; where fringe benefits, chances for advancement, and pay are concerned, more is preferred to less. But when wages are increased by fiat, there is no accompanying increase in productivity. It follows that it does no one a favor to pass a law that says he cannot work if he cannot earn more.

To understand the perversity of such a law, explore the alternatives available to employee and employer when the minimum wage is increased. The employee’s only choice is to find a job that pays more or become a nonemployee. The employer and the consumer, who is the employer’s employer, have a broader range of options. The employer can replace low-wage workers with more-productive workers; after all, the vast majority of workers have wages that exceed even the most aggressive proposal for wage floors. A second alternative is to outsource, to subcontract activities performed by low-wage employees by going abroad or to self-employed contractors, since the government has as yet been unable to devise a scheme for imposing minimum wages on the self-employed. A third alternative is to automate, to substitute machines that do not have legislated minimum prices. Finally, there is the alternative of just cutting back. If minimum wages accomplish anything, they increase the employer’s costs, causing the purchasers of his product or service to search for lower-priced alternatives. This may include choosing the same products from abroad or switching, in whole or in part, to different products.

If the world were simple, then the implications of increases in the minimum wage would also be simple. For example, if wages were the only form of compensation, if there were no fringe benefits or job amenities, and if all workers were of uniform quality, then everyone would get the same wage. A minimum that attempted to raise the wage would reduce employment.

But workers are of varying productivity, so that wages also vary, and a minimum that attempts to raise the wage of those with the lowest productivity will have effects similar to those described in the uniform-quality case. Some will get jobs at higher wages and others will lose jobs. Among those who would earn less than the minimum without the legislation, those with wages closest to the wage floor are the ones whose continued employment will cost employers the least, and they will be most likely to keep their jobs. Within the low-productivity group, the minimum will function like a tax, from poorer to poor—the winners will be those who would have fared best in any case.

As employers explore alternatives to the continued employment of their now more expensive low-skilled workers, they may turn to more-productive workers newly attracted to the labor market by the prospect of higher wages. This substitution of higher-quality for lower-quality workers has two important consequences. First, it illustrates again how the minimum wage has its harshest impact on those who are least productive. Second, replacement of lower-skilled workers—such as a 17-year-old replacing a 16-year-old or a young high school graduate taking over for an older high school dropout—is difficult to discern in the data. The findings above give the net employment declines for broad demographic groups resulting from the increased federal minimum. The certain substitution of more-productive for less-productive workers causes the net employment losses to understate, perhaps severely, what is happening to the least-skilled workers.

A popular idea among those who favor increasing minimum wages is that firms will respond to an increased minimum by "getting more out of" their employees. But how would this be accomplished? If greater productivity is achieved by substituting higher-quality workers, then low-skilled, and thus low-wage, workers take it on the chin. If, on the other hand, the same workers are now working harder, there is still a problem. Work has two dimensions, time on the job and effort at the job. At the previous wage the employer and his employees had reached an understanding about what constituted an hour’s worth of work for an hour’s pay. They could have reached an agreement on a greater amount of effort per hour and a greater hourly wage. But the fact that they did not suggests that the agreement they did reach was preferred to other potential agreements involving higher wages and greater effort. If low-skilled workers must put forth more effort just to keep their jobs and earn the higher wage, then they have actually taken a step backward. Their earnings have risen, but not by enough to make up for their increased effort. This must be true, otherwise the employer could have been "getting more out of" his workers all along simply by paying the higher wage.

Returning to the simple world of wage-only compensation and a uniform quality of worker, if the only change affecting the labor market is the increase in the minimum wage, then the predictions for employment are clear. On the other hand, even in this simplified world, if the labor market is continually buffeted by changes in firms’ hiring practices or variations in the prices of raw materials, then economists must be careful to hold "other things equal" in making predictions about how increased minimums will affect employment. As the results above show, employment is affected by factors other than the minimum wage, such as economic growth within the state, and these factors must be acknowledged in order to avoid obscuring, or missing entirely, the employment losses from increased minimums.

Only a small fraction of workers earn low wages: about 8 percent of men and 13 percent of women. Even for low-wage workers there are many other factors that affect employment, including the business cycle, education, gender, race, ethnicity, and region. In grappling with real data, it is difficult to sort all of this out, to "hold other things equal." Assessing the employment effects of minimum wages is not as simple as the theory implies. For example, women’s employment has been increasing relative to men’s for the past 45 years. A naive comparison of men’s and women’s employment after the 1990-91 minimum wage hikes that ignores this fact might conclude that minimum wages increase employment, since women’s wages are lower and more likely to be affected by the minimum. A simple comparison of employment changes in low-wage and high-wage states is also misleading. Just as with comparing men and women, there are larger forces at work that make it difficult to isolate the employment losses from increased minimums.

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.

 The New Economics of the Minimum Wage: Myth or Reality?

 A handful of recent case studies by David Card and Alan Krueger conclude that increases in minimum wages do not reduce employment; Card and Krueger suggest that employment may even rise as a result of legislated wage increases. The pursuit of this "free lunch" is the foundation upon which Clinton’s proposal for another increase in the federal minimum wage is built.

We examined four Card-Krueger studies that have appeared previously in the academic literature. The first study analyzes the impact of the 1990 minimum wage hike across the 50 states and the District of Columbia. The other studies concentrate on particular states; one examines California, while the other two look at employment changes in some fast-food restaurants in Texas and New Jersey. The case study of Texas fast-food restaurants was coauthored by Lawrence Katz, a Harvard economist and Robert Reich’s former chief economist at the Department of Labor. Each of the studies is aimed at measuring the effect of a different hike in minimum wages: the 1988 increase in California’s state minimum to $4.25, the 1990 and 1991 increases in the federal minimum, and the 1992 increase in New Jersey’s state minimum to $5.05.

The first study, and the only one to examine minimum wage effects across all of the states, uses data and methodology similar to ours. It compares changes in teenage employment across states to the incidence of low wages among teenagers in the state. The employment change is calculated around only the April 1990 increase in the federal minimum from $3.35 to $3.80. Based on this single grouping of teenage workers by state of residence, the study concludes that there is "no evidence that the rise in the minimum wage significantly lowered teenage employment rates." But this study did not adequately control for the fact that economic activity after 1989 was weaker in high-wage states than in states with lower wages. The employment losses for teenagers in low-wage states were not as great as in high-wage states, where during 1990 employment of men earning well in excess of the minimum was also falling.

The second study investigates California’s increase in its minimum wage to $4.25, which occurred in 1988, almost three years before the federal government raised the minimum wage to this level. Although this study could find no effect on employment, two scholars at Carnegie Mellon University, Taeil Kim and Lowell Taylor, have reexamined the California experience. The Princeton study contrasts changes in teenage employment in California with changes in a few other states that are assumed to be comparable. The Carnegie Mellon study compares changes in employment with changes in wages across different industries within California’s retail sector; it also compares changes in retail-sector employment with changes in retail-sector wages across each of the California counties. This study finds that the greater the increase in wages due to the increased minimum, the greater the loss of employment.

Our examination of California’s employment pattern suggests that, as a result of the July 1988 increase in California’s minimum wage, the employment of teenagers fell below what it would have been otherwise. It appears that California’s experiment with an early increase to $4.25 did for it what the federal minimum would later do for the country as whole. In contrast, Card and Krueger would have us believe that neither the increase in California nor the federal increase to $4.25 reduced employment.

The two remaining studies focus on the 1991 increase in the federal minimum, from $3.80 to $4.25, and New Jersey’s subsequent 1992 increase to $5.05. They rely on special telephone surveys of fast-food restaurants. The effects of the 1991 increase in the federal minimum to $4.25 are assessed by looking at what happened to employment at 100 fast-food restaurants in Texas. The effects of the increase in New Jersey’s minimum to $5.05 on April 1, 1992 are measured by looking at what happened to employment at about 320 fast-food restaurants in New Jersey and another 80 fast-food restaurants in Pennsylvania.

From discussions with some 500 fast-food outlet managers in three states, the authors are willing to conclude that minimum wages do not reduce employment. Though the technical criticisms of these studies yield a long list, we will restrict ourselves to the following three observations.

• The theory makes no necessary prediction about how employment at any one firm is affected by an increase in the minimum wage. For example, a restaurant with unionized employees all earning at least $6.00 an hour actually gains a competitive advantage from an increase in the minimum to $5.05. Why else would unions always be first in line to favor a high minimum, when all of their members earn more than the minimum? Likewise, franchised fast-food outlets do not necessarily suffer a competitive disadvantage when the minimum rises for all restaurants, including local pizza and sandwich shops.

• The reports of what happened to employment at a handful of fast-food restaurants notwithstanding, teenage employment rates in Texas and New Jersey fell between 1989 and 1992. Increases in the minimum wage cannot escape blame.

• The baseline used to infer that employment rose after the minimum went up is calculated just before the higher minimum takes effect, and long after employers knew of the legislated increase. To conclude that the change in employment over this time frame gives a complete view of the minimum wage effect is like comparing the number of teenagers on the street at 11:59 P.M. and 12:30 A.M. to measure the effect of a midnight curfew. Finding no difference does not mean that the curfew has no effect.

Finally, it is important to consider the broader trends of teenage employment in New Jersey and Pennsylvania. The secretary of labor has often stressed this comparison in support of the notion that increased minimums increase employment. Table 4 gives the teenage employment rates for each state from 1988 through 1992. Since the minimum wage hikes took effect on April 1, each year in the table begins with April and continues through March of the subsequent year. Comparing 1991 and 1992, the years on either side of the effective date of New Jersey’s hike to $5.05, shows that teenage employment in New Jersey fell by less than in Pennsylvania, where the minimum remained at $4.25. Recall that the increase in the federal minimum was debated and passed in 1989 and that the 1992 increase in New Jersey’s minimum to $5.05 was passed in early 1990. Now note that the teenage employment rates in New Jersey and Pennsylvania were virtually identical in 1988. By 1992 teenage employment in Pennsylvania had fallen 9 percent, while in New Jersey it had fallen 28 percent. Moreover, subsequent data show that teenage employment began to rebound in Pennsylvania during 1993 as it continued its downward spiral in New Jersey. Quite simply, higher minimum wages go hand-in-hand with substantial declines in the employment of low-productivity workers.

So what should be made of the new minimum wage research? Each of the four studies examines a different piece of the minimum wage/employment relationship. Three of them consider a single state, and two of them look at only a handful of firms in one industry. From these isolated findings Card and Krueger paint a big picture wherein increased minimum wages do not decrease, and may increase, employment. Our view is that there is something wrong with this picture. Artificial increases in the price of unskilled laborers inevitably lead to their reduced employment; the conventional wisdom remains intact.

The evidence on minimum wages and employment does not vindicate the president’s view that minimum wages increase employment. But the minimum wage debate has focused attention on the problems of older low-wage workers. Another "new" finding in this renewed debate over minimum wages is that teenagers are a minority of those earning low wages and thus of those most affected by minimum wage laws. While this information is hardly revelatory—teenagers constituted 32 percent of those earning $4.25 or less in the year before the 1990-91 increases and 38 percent of those earning at or below the federal minimum in May 1973—the attention now being paid to low-skilled workers other than the young is welcomed. Secretary Reich has stated that it is now more important to increase the minimum because of the many nonteenagers earning low wages. To the contrary, precisely because a majority of low-skilled, and thus low-wage, workers cannot easily outgrow this condition, it is important to prevent further declines in employment opportunities for these workers. It is not that there are too many low-wage jobs, but that there are not enough jobs for low-wage workers; and minimum wages make things worse.