The Clinton administration's proposal to increase the minimum wage has renewed interest in the question of how job creation would be affected. The president's proposal came at a time when academic interest in the effects of the minimum wage on employment was rekindled by David Card and Alan B. Krueger, whose book, Myth and Measurement: The New Economics of the Minimum Wage (Princeton University Press, 1995), challenged the consensus among labor market economists that increasing the minimum wage would reduce employment of teenagers and other low-wage workers.
Before publication of Card and Krueger's research, most economists regarded the issue of whether the minimum wage has adverse effects on employment as settled, although there were differences about the magnitude of the negative effects. Card and Krueger challenged this view, arguing that their research shows no significant adverse effects on employment. Other leading scholars, however, have defended the earlier consensus and developed new evidence that supports it.
The American Enterprise Institute held a seminar on June 1, 1995, to examine the evidence. Leading scholars presented their analysis, gave critiques of research approaches, and examined how disparate results should be interpreted. Marvin H. Kosters, resident scholar at AEI, organized the seminar and served as moderator. A summary of the presentations follows.
Alan B. Krueger, chief economist at the U.S. Department of Labor (on leave from Princeton University), indicated that his and Card's conclusion that the minimum wage has little or no effect on employment does not differ as much from earlier research as has sometimes been suggested. He argued that the negative effects on employment of increases in the minimum wage were fairly small in earlier studies and that their research suggests that those effects are even smaller.
Krueger's presentation concentrated primarily on the effects of an increase in the minimum wage on employment in a sample of fast-food restaurants. Their most widely publicized study compared employment changes in New Jersey, where the state legal minimum wage was increased in 1992, with Pennsylvania, where it was not. Analysis of the effects of this difference in policies was appealing as a case study because anticipatory employment adjustments might be expected to be small in view of uncertainty about whether the new minimum wage in New Jersey would actually go into effect as scheduled on April 1, because there was a high concentration of relatively low-wage teenage workers in fast-food restaurants, and because turnover was high so that employment levels could be adjusted relatively quickly.
Their results, presented in detail in their book, did not show a decline in fast food employment in New Jersey relative to Pennsylvania. The average fast-food restaurant in Pennsylvania had more employees than in New Jersey in early 1992, but later in the year--after the minimum wage had been increased in New Jersey--restaurants in both states employed about the same number of workers. Several ways of analyzing the data showed small positive employment differences in New Jersey, but since the estimates are not statistically significant, neither small negative employment effects nor small positive employment effects can be rejected.
Krueger responded to criticisms that have been raised about the validity of the data they collected by explaining the methods they used and discussing evidence on the reliability of their data in relation to data from other sources. He pointed out that reporting errors need not necessarily lead to bias in estimates of employment effects. Moreover, data from some other sources are consistent with the differences in employment trends they found between Pennsylvania and New Jersey.
David Card, professor of economics, Princeton University, acknowledged that studies of employment in a narrow sector of the economy, such as fast-food restaurants, have limitations. He described his and Krueger's analyses of differences across states in the proportions of workers who would be affected by an increase in the federal minimum wage. Wage levels and wage distributions for teenagers in different states are very different, and imposing a uniform standard when the federal minimum was increased in 1990 and 1991 would be expected to have quite different effects. Moreover, some states, such as California, had already raised their minimum wage to levels above the prevailing federal rate.
Their analysis was carried out by estimating the fraction of teenagers who were earning a wage between the old federal minimum and the new one. This fraction varies substan-tially among states, and their analysis shows that imposition of a national minimum pushes up wages more in states where more teenagers were initially below the standard. Although demand for teenage workers might be ex-pected to decrease in response to higher wages, employment rates of teenagers did not appear to be reduced more in states where their average wages were pushed up most. Efforts to take into account broader employment trends or regional differences did not change the picture much, nor did extending the time period for adjustment to minimum wage increases.
Card also discussed other analyses across states in which they examined total employment in the retail trade and restaurant industries, with results that were similar to those for teenagers. Estimates of the size and statistical significance of minimum wage effects on employment in these industry sectors were quite sensitive to other variables included in the analysis, but the estimated effects were small and positive. The imprecision of many of the estimates means that the possibility of small negative effects cannot be entirely ruled out, according to Card, but their studies lead them to conclude that the employment effects of the minimum wage were relatively small at the end of the 1980s.
Finis Welch, professor of economics, Texas A&M University, began by noting that economic theory has increasingly been brought to bear on analysis of labor markets and that one of the most robust predictions of economic theory is the law of demand. Applied to the minimum wage issue, the prediction is straightforward: higher minimum wages imply less employment. If this were a shortcut to improving incomes, why not just increase the minimum wage and save on investments in education, training, technology, and infrastructure?
The attention given to the Card and Krueger results and Welch's skepticism about their validity led him, with Donald Deere and Kevin Murphy, to examine the employment effects of the 1990 and 1991 federal minimum wage increases. The results of their analysis are broadly consistent with the earlier consensus of significant negative effects on employment. Welch then raises the question, If the law of demand is as robust as he and many other economists contend, why do some studies by Card, Krueger, and others get either anomalous or contradictory results?
One reason employment effects can be difficult to isolate is that the impact on labor costs of an increase in the minimum wage is often small in relation to other changes going on at the same time and in relation to sampling error. Welch illustrated this problem by calculating that the average cost to employ teenagers increased by about 1.5 percent when the minimum wage was increased in April 1990 from $3.35 to $3.80 (a 13.4 percent increase) and that the total increase from $3.35 to $4.25 (in two steps in 1990 and 1991) raised average teenage employment cost by only 3 to 4 percent. Since changes in teenage wages in a single month are typically on the order of 2 percent or more, however, the employment effects of a minimum wage change may not be easy to see in a simple comparison for a sample from one sector of the economy.
Kevin M. Murphy, George Pratt Shultz Professor of Business and Industrial Relations, University of Chicago, noted that minimum wage research has historically focused on teenagers because it would be extremely difficult to isolate the much smaller effect of a rise in the minimum wage on the price of labor for the economy as a whole. Even for teenagers, he argued, we need to look for the effects of relatively small changes in a world with a lot of noise, some of it systematic. Teenage employment rates, for example, were about the same in New Jersey and Pennsylvania in 1988, but the rate in New Jersey declined by more than 9 percentage points compared with Pennsylvania by 1991. This difference is so large compared with a difference of 1.6 percentage points (in the opposite direction) in 1992 that something else was obviously going on that was systematically different between these two states.
Comparisons across states of employment rates for twenty-five- to fifty-year-old men with more than twelve years of schooling--a group quite insulated from minimum wage effects--show that employment was growing much more rapidly in low-wage states, basically Sunbelt states in the South and Southwest, than in the rest of the country. Systematic differences among states need to be taken into account in order to trace differences in employment experience to causes such as an increase in the minimum wage. This problem is very similar to the need to take into account the influence of the business cycle and other factors in time-series studies, studies that examine the effects on employment of periodic increases in the dollar value of the minimum wage that interrupted gradual erosion of real minimum wage levels because of inflation.
Murphy also noted that raising the minimum wage does not make anybody better off without someone else paying for it. Many of those who need to pay the higher prices that result from a minimum wage increase have incomes very much like those who get a higher wage. Since many of the poor are not working, part of the benefit to low-wage workers of a higher minimum wage comes at the expense of nonworking poor consumers. Moreover, even if net disemployment effects are small, people with the hardest time getting a job will tend to lose out compared with those whose earning capabilities are a little higher.
Some of the favorable public discussion of Card and Krueger's research suggests that what they did represents a methodological breakthrough based on natural experiments. But the problems of bias and omitted variables that plague the time-series studies are also present in the new studies. Murphy concluded that the evidence from these new studies does not change his view about the negative employment effects of raising the minimum wage.
David Neumark, professor of economics, Michigan State University, with his coauthor William Wascher analyzed the effects of minimum wage changes across states. They found significant negative effects on employment. Their results for teenagers are sensitive to the way school enrollment is measured and to how school enrollment decisions are taken into account. This research led them to examine the effects of the minimum wage on both work and school choices.
Neumark explained that the small net employment effects of the minimum wage reflect partially offsetting shifts in school and employment status. Their results indicate that raising the minimum wage shifts employer demand from the least-skilled teenagers to those with slightly better skills. As a result, teenagers with somewhat better skills increase their working hours by going from being in school and employed to being out of school and employed. Some of the least-skilled teenagers, however, are priced out of the labor market, and their status changes from out of school and employed to neither in school nor employed.
By using methods that estimate the likelihood that a worker's wages are actually affected by an increase in the minimum wage, Neumark and Wascher's analysis helps to reconcile disparate results that have been reported. Their estimates show that in a high-wage state like California, where the probability of being affected by minimum wages is quite low, the estimated effects on employment should also be expected to be small.
Neumark and Wascher also analyzed data from payroll records for a sample of fast-food restaurants in New Jersey and Pennsylvania. Using these data, they estimated that a 10 percent increase in the minimum wage reduced employment by about 2 percent, a result consistent with their other studies. Neumark and Wascher conclude that the conventional economic model of low-wage labor markets and its prediction for the employment effects of minimum wages have not been contradicted by the new revisionist evidence.
Charles C. Brown, professor of economics, University of Michigan, began by reviewing the "old" economics of the minimum wage. He pointed out that time-series estimates were clustered in a fairly narrow range and that they were not very sensitive to whether hours of work or employment was examined or to different sets of control variables. He acknowledged that analyses that extend through the 1980s produced smaller estimated effects. Other puzzles left by that research include why the evidence did not consistently show black employment more strongly af-fected than white and why extension of coverage to industries that were not previously subject to the minimum wage did not show more noticeable effects. In most of the earlier research, the level and coverage of the minimum wage were combined in a fairly crude way, and estimated employment effects apparently reflected mainly the influence of changes in the minimum wage level.
Examination of a particular sector of the economy, or comparison between sectors affected differently by the minimum wage, is not a new analytic approach, according to Brown. In some of the studies examined by the staff of the Minimum Wage Study Commission, employment fell in sectors most affected by the minimum wage, but in others the results were mixed. Although these earlier studies were conceptually similar, those carried out by Card and Krueger used much more sophisticated analytical tools.
Estimates of minimum wage effects are apparently sensitive to the length of time allowed for adjustment to take place. Efforts to pick up more of the cumulative employment effects, however, are also likely to be more strongly affected by statistical noise and preexisting trends.
Estimates of the negative effects on employment of an increase in the minimum wage are fairly small. This should not be surprising because the effect on average wages of an increase in the minimum wage, even for groups most affected, such as teenagers, has also usually been quite small. In addition, the net change in employment reflects offsetting changes for gainers and losers. In his evaluation of the evidence, Brown concludes that the new minimum wage studies are subject to many of the same limitations and methodological problems as the older studies.
This summary was prepared by Marvin H. Kosters from a transcript of the proceedings.
