Reprinted from THE JOURNAL OF POLITICAL ECONOMY
Vol. 79, No. 4, July/August 1971

1971 by The University of Chicago. All rights reserved.
Printed in U.S.A.

The Effect of Minimum Wages on
Teenage Unemployment Rates

Thomas Gale Moore

Michigan State University

The model developed has the unemployment rates of four classifications of teenagers and the unemployment rate of males twenty to twenty-five as a function of overall labor market conditions, the minimum wage as a proportion of average hourly earnings for nonfarm workers, and the coverage of the minimum wage. The model hypothesizes that changes in the minimum wage relative to average hourly earning have an increasing effect as employers gain more time to adjust; consequently a distributed lag is used to measure the impact of the minimum wage. As might be expected, the minimum wage most adversely affected nonwhite teenagers; females sixteen to nineteen were more adversely affected than males.

An increase in the minimum wage in a competitive labor market will result in some lessening of employment. Even in a monopsonistic market, increases in minimum wages may lead to reductions in employment. Few economists believe that monopsony is widespread, but even were it widespread a uniform national minimum wage would likely lead in some markets to a reduction in employment.

Assuming the labor market is relatively competitive, a rise in the minimum wage will result in a fall in employment for those with the least skills or smallest productivity. It may also lead to the substituting of skilled labor for the unskilled. Economists are generally in agreement about these effects of the minimum wage. The disagreement revolves around the magnitude of the unemployment effect of minimum wages. If a rise in the minimum wage leads to negligible declines in employment, then the cost in terms of unemployment may be much less than the gains from the better-paid labor force. Alternatively, if the unemployment impact is substantial, or, to put it another way, if the demand for labor is elastic, rises in the minimum wage will lead to reduced total wages.

The few previous studies of the effect of the minimum wage have been rather inconclusive. The United States Department of Labor has sur

897

898

JOURNAL OF POLITICAL ECONOMY

veyed low-wage industries in an attempt to measure the impact of changes in the minimum wage (1968a, 1968b, 1968c). They concluded the impact was small or negligible. David Kaun found that "in industries. where the minimum wage had the greatest impact, the subsequent adjustment in terms of the factor proportion would also be great” (1965, p. 482). Yale Brozen showed that when minimum wages have been increased, the number of household workers (who are exempt from the minimum wage) has risen contrary to longrun trends (1962).

This study examines the impact of the minimum wage on disadvantaged classes of workers. Theory would suggest that when the minimum wage is raised, the impact would be felt most strongly by the unskilled, the inexperienced, and those facing discrimination. Thus teenagers, especially nonwhite teenagers, would be expected to suffer most from a rising minimum wage.

Clearly, not only the level of the minimum wage but its coverage would also affect employment opportunities of teenagers. In addition, overall labor market conditions would help determine unemployment rates of teenagers.

As is well known, unemployment rates of teenagers relative to those of adults have risen in the postwar period. It is sometimes argued that this has resulted from a rise in the relative supply of teenagers in the labor force. Therefore, the model developed tests this hypothesis.

Now a rise in the minimum wage should lead to a substitution of capital and skilled labor for the unskilled. However, over time as wage rates rise elsewhere in the economy, the barrier to employment at a given minimum wage will decline. Thus a $0.75 minimum in 1954 might bar some from securing work in covered industries but would bar relatively far fewer in 1969. Employers will compare the minimum wage for an unskilled worker with the wage he must pay for a more skilled and productive worker. The higher the minimum relative to the wages of more skilled workers, the less unskilled he will wish to employ.

A rise in the minimum wage relative to other wages should have an increasing impact over time. It takes time to substitute capital for labor. Employers prefer to reduce their work force by attrition rather than firing workers. The possibility of substituting more skilled workers has to be explored and the skilled workers found and hired. Therefore, a rise in the minimum wage could be expected to have an increasing impact for awhile, but eventually the general upward movement in other wages would begin to offset the change in the statutory rate.

The model developed, then, has the unemployment rates of four classifications of teenagers and the unemployment rate of males twenty to twenty-five as a function of overall labor market conditions, the minimum wages as a proportion of average hourly earnings for nonfarm workers, and the coverage of the minimum wage. The teenage groups tested

MINIMUM WAGE EFFECTS

899

were nonwhites, whites, males, and females, all sixteen to nineteen years old.

Unemployment rates for these subgroups were taken from various volumes of Employment and Earnings; put out by the Bureau of Labor Statistics. Monthly figures on average hourly earnings for production and nonsupervisory workers have been published only for 1964 to date in the BLS Handbook of Labor Statistics. Since monthly changes tend to be regular and smooth, for years prior to 1964 monthly estimates were made by interpolation. The coverage variable was computed using some nonpublished BLS bench mark data and then adjusting the total employment in those sectors primarily covered by these basic bench marks. Thus, for the years 1954-60 the monthly sum of total employment in mining, contract construction, manufacturing, and wholesale trade was adjusted by the proportion that total bore to the bench mark data for 1955 and 1956. For later periods similar adjustments were made as coverage was extended.' All data were seasonally adjusted. The actual equation fitted was:

U1 = a + b1U25 + b2[0.75 + Σ(1 − B1*)AM ] / AHE.

+b3(LF nwt/LF 1) + b1C + b¿T + u,

=

where U, is the unemployment rate of one of the subgroups of teenagers or the twenty- to twenty-four-year-old males in the tth month; U251 is the unemployment rate of males twenty-five and over; $0.75 is the minimum wage in January 1954, the first month; AM is the change in the minimum wage in the kth month (most months AM 0); AHE, is the average hourly earnings of production or nonsupervisory workers on private nonagricultural payrolls; LF nut is the nonwhite teenage labor force; LF, the total labor force; C is the percentage of the employed labor force covered by the minimum wage; T is time; and u is the error term. The parameter B was iterated from .10 to .99 to secure the best fit. The results given in tables 1 and 2 are for a B of .95 which maximized the R2. Using a B of .95 results in 43.1 percent of the effect of changes in the minimum wage being felt after twelve months, 69 percent at the end of two years, and 83 percent after three years. The Cochrane-Orcutt method of adjusting for serial correlation was used.

Experimentation with these variables and with a trend variable indicates that minimum wages as a proportion of hourly compensation is significant in explaining unemployment rates of nonwhite teenagers even when time is added. However, the coefficient is smaller and less significant due to the multicollinearity between time and minimum wages. The best regression results are those based on the unemployment rate of males 1 I will furnish, upon request, the coverage variable and the underlying data.

900

JOURNAL OF POLITICAL ECONOMY

TABLE 1

RELATIONSHIP OF UNEMPLOYMENT TO RELATIVE LEVEL OF
MINIMUM WAGES JANUARY 1954-JULY 1968

NOTE.-Cochrane-Orcutt method of adjusting for serial correlation was used. N = 175 monthly observations; t-values in parentheses.

* Significantly different from zero at the .1 level.

** Significantly different from zero at the .01 level.

twenty-five years and older, the percentage of workers covered by the minimum wage, and the minimum wage as a proportion of hourly compensation. Tables 1 and 2 present the regression results which are based on 175 monthly observations running from January 1954 through July 1968.

These data and regression results may understate the impact of the minimum wage. Higher unemployment may lead to workers or potential workers withdrawing from the labor force. There is some evidence that this has happened in the teenage market. Since 1961, when minimum wages were increased significantly, the participation rate for nonwhite teenagers has dropped markedly. While the rate for white teenagers declined slightly after 1961, it recovered by 1966 to the previous level.

The previous secretary of labor, Willard Wirtz, in January of 1969 recommended raising the minimum wage to $1.80 in January 1970 and $2.00 a year later. He also proposed making coverage universal by 1975. Table 3 indicates the effect of his proposal to raise the minimum (but not to extend the coverage) on nonwhite teenagers based on table 2, line 2 coefficients. If coverage were to be made universal, this would add 9.7 percentage points to nonwhite teenage unemployment rates.

As a test of the model, the average unemployment of nonwhite teenagers for the months of May, June, and July 1969 were estimated. (The data used to estimate the coefficients went through July 1968.) The average predicted by the model was 27.7 percent. The actual unemployment rates were 27.8, 28.6, and 22.3 percent for each of the months, or

(2.576)

2.353

(6.249)

(10.675)

(2.968)

(10.839)