COMMUNICATIONS

439

nonwhite teens, because there are relatively more males than females in the teen-age labor force both for whites and nonwhites. Besides the statistical tests already referred to, these descriptive features add considerable credence to the estimates.5

To gain an appreciation for the extent of the unemployment effect of changes in the real federal minimum wage, consider columns 2 and 3 of table 1. Based on the unemployment elasticities in column 1, the unemployment effects of increases in the federal real minimum wage as of December 1965 for each teen-age labor force classification have been calculated for 1 and 10 percent changes in the minimum. For instance, consider the unemployment elasticity for all teens in column 1, which is 0.362. This means that for a 10 percent increase in the federal real minimum wage, the unemployment rate for all teens would increase by 3.62 percent of the prevailing rate. In December 1965, the unemployment rate for all teens was 12.9, so if the minimum were raised 10 percent in this month, the unemployment rate would increase initially by 0.47 due to this increase in the minimum wage. The new unemployment rate would then be 13.37, which is given in column 3. The unemployment rates in columns 2 and 3 were computed similarly. These calculations assume no change in the price level. Increases in the price level tend to erode the unemployment effect of changes in the minimum wage.

The Lag Structure of Unemployment Effects

In this section the unemployment effects are determined after 8, 16, and 24 months by lagging the minimum-wage variable in equation (5). All the elasticities are significant at the 5 percent level and are listed in columns 4, 5, and 6 of table 1.

The elasticities show the same direction of change as the lag is increased from 0 to 24 months. The elasticities for all teens, white teens, and male white teens increase steadily as the lag is increased from 0 to 24 months. The elasticity for female white teens increases as the lag is increased from 0 to 16 months, then decreases as the lag is increased to 24 months. The

5 Other specifications of the model were used with different independent variables. For instance, instead of deflating the nominal minimum wage by the wholesale price index, it was deflated by the consumer price index. Also, instead of using the total index of industrial production to remove business fluctuations, the indices pertaining to manufacturing and final products were used. By using these variables, four models were generated. An inspection of the Durbin-Watson statistics showed the error terms to be serially correlated. An adjustment with partial first differences raised the DurbinWatson statistics to acceptable levels. See Theil and Nagar (1961) for a description of the technique. Of the 28 unemployment elasticities for January 1954-December 1965, only five were not significant at the 5 percent level. Models which used the minimum wage deflated by the wholesale price index yielded slightly higher elasticities for white teens while models using the minimum wage deflated by the consumer price index yielded slightly higher elasticities for nonwhite teens. The complete results are available from the author on request.

440

JOURNAL OF POLITICAL ECONOMY elasticity for nonwhite teens decreases as the lag is increased from 0 to 16 months, then increases as the lag is increased to 24 months. The elasticity for male nonwhite teens increases, decreases, then increases again as the lag is increased from 0 to 8, 8 to 16, and 16 to 24 months, respectively. The elasticity for female nonwhite teens decreases as the lag is increased from 0 to 8 months, then increases as the lag is increased from 8 to 24 months.R

The remarkable feature of the lagged estimates of the unemployment coefficients is that even after 2 years the unemployment effects of a change in the federal minimum wage are quite substantial and in most cases greater than the immediate effect. The unemployment elasticity is greater than the initial elasticity after 24 months for all teens, white teens, male white teens, and female nonwhite teens. The elasticities for nonwhite teens and male nonwhite teens are slightly less than the 0-month lag estimates. The general tendency for the elasticities for all teens, white teens, and particularly male white teens to increase as the lag is increased indicates the delayed response pattern of employers in adjusting their factor proportions to a change in relative factor prices-particularly the quantity of male white teens hired. Although the O-month lag elasticities for white teens, and in particular male white teens, are low relative to other teenage classifications, they display the largest percentage increase as the length of the lag is increased. For nonwhite teens, with much higher initial lag unemployment elasticities, most of the adjustment takes place within 8 months.

Since price changes have been accounted for by using real minimum wages, the slight decrease in the elasticity for male nonwhite teens and female white teens from their peak values after 24 months reflects productivity increases relative to the productivity of other factors. However, the slow rate of decline in the elasticities indicates that even for these groups, productivity increases quite slowly and significant unemployment effects persist after 24 months. Since the 24-month lag elasticity for all other classifications of teen-age labor is the maximum calculated, the unemployment effects of a change in the minimum wage for all other classifications also persist for a considerable time after 24 months. Since these unemployment effects assume a zero rate of inflation, a high rate of inflation could substantially reduce the length of time of the unemployment effect.

The Unemployment Effects through Time

In this section the unemployment elasticities are investigated over three time periods: (1) January 1954-September 1961, (2) January 1954

6 For the specifications mentioned in n. 5, all 84 lag estimates of 8 months or more are significant at the 5 percent level.

COMMUNICATIONS

441

September 1963, and (3) January 1954--December 1955. January 1954 was chosen as the beginning date because it is the first month for which unemployment data are available for the classifications of teen-age labor used in this paper. The terminal dates of periods (1) and (2), September 1961 and September 1963, coincide with changes in the federal minimum wage to $1.15 and $1.25, respectively, while the last date. December 1965, represents the last month for which the data were available. The results are listed in columns (1), (7), and (8) of table 1 and indicate that the elasticities increase steadily from periods (1) to (3). Increased coverage, stricter enforcement of laws, and the availability of cheap substitutes in production may account for this trend. More work needs to be done before assessing these causes.

Conclusion

The results of this study indicate that (1) increases in the federal minimum wage cause unemployment among teen-agers; (2) the effects tend to persist for considerable periods of time; and (3) the effects seem to be increasing through time. This result is not surprising, since one of the fundamental difficulties with legislative attempts to alter the economic structure of the economy arises from the apparent inability of legislators to see the full social and economic consequences of their interference.

In conclusion, it appears that the minimum wage is responsible for a considerable amount of teen-age unemployment. To the extent that a minimum wage creates unemployment among minority-group teen-agers, it inevitably will exacerbate the social unrest in urban ghettos. With this information in mind, serious consideration should be given to reducing, or removing, the federal minimum wage for teen-agers.

References

Brozen, Y. "The Effect of Statutory Minimum Wage Increases on Teenage Employment." J. Law and Econ. 12 (April 1969): 109-22.

Kaun, D. "Minimum Wages, Factor Substitution and the Marginal Producer." Q.J.E. 79 (August 1965): 478-86.

Moore, G. M. "The Effect of Minimum Wages on Teenage Unemployment." J.P.E. 79 (July/August_1971): 897–902.

Peterson, J. "Employment Effects of Minimum Wages, 1939-1950." J.P.E. 65 (October 1957): 412-30.

Stewart, C., and Macesich, G. "Recent Department of Labor Studies of Minimum Wage Effects." Southern Econ. J. 27 (April 1960): 281-90. Theil, H., and Nagar, A. L. "Testing the Independence of Regression Disturbances." J. Amer. Statis. Assoc. 56 (December 1961): 793-806.

U.S. Department of Labor. Bureau of Statistics. Youth Unemployment and Minimum Wages. Bull. 1,657, 1970.

7 The same relationships exist between the elasticities estimated with the other specifications of the model mentioned in n. 5.

Senator TAFT. We will be at recess for the moment.

Mr. FEDER. Senator, a couple of questions.

Senator TAFT. Yes, go ahead.

Mr. FEDER. Your figures on teenage unemployment, is that composite of teenagers looking for part-time and full-time work, or fulltime work?

Dr. ADIE. Full-time work.

Mr. FEDER. And you are suggesting 1.67 million teenagers holding full-time jobs will lose those jobs as a result of minimum wage?

Dr. ADIE. By raising it to $2 an hour, yes, sir.

Mr. FEDER. Second question is economic theory. If we assume there is to be an economy on which there is no job expansion, and also no job contraction, stable employment opportunities in the economy, what will be the impact of a youth differential on adult employment?

if

Dr. MOORE. I am not sure I understand your question. But

Mr. FEDER. If there are 82 million jobs this year and we still have 82 million jobs in the economy next year, but we have a youth differential intervening which creates jobs for teenagers, where are those jobs going to come from?

Dr. MOORE. If we are raising the minimum wage between this year and next year and the economy was not growing at all, I would suggest that we will not have 82 million jobs next year, we would have less than that.

Mr. FEDER. Assuming we had 82 million, and it stayed stable.
Dr. MOORE. Well, you have assumed the answer.

Mr. FEDER. I don't know what answer I have assumed.

Dr. MOORE. You have assumed

Mr. FEDER. In that situation, will adults lose jobs to teenagers if you have a youth differential?

Dr. MOORE. You have assumed the answer. You have assumed that the total number of jobs is fixed. You have assumed that teenagers will get more jobs. Obviously they must come from somewhere else.

Mr. FEDER. In other words, unless the economy is expanding at a rapid enough rate to account for adult rate, youth differential will increase unemployment for adult workers, that is the thrust of the question.

Dr. MOORE. Not necessarily, I disagree with your basic assumption that the employment will remain constant.

Senator TAFT. The question seems to make some unwarranted assertions. Also there is a question of job preservation.

Dr. MOORE. There is no youth differential_now, but there are some exempt areas where youths can move into. Presumably that is what they would do if you had a youth differential, there would be more of them would be—would be in covered areas and less in the exempt

areas.

The CHAIRMAN. Thank you very much.
Senator TAFT. Thank you very much.

STATEMENT OF FREDERICK L. WILLIFORD, DIRECTOR OF GOVERNMENT AFFAIRS, NATIONAL FEDERATION OF INDEPENDENT BUSINESS, ACCOMPANIED BY NEEL EDWARDS, GOVERNMENT AFFAIRS REPRESENTATIVE, NATIONAL FEDERATION OF INDEPENDENT BUSINESS

The CHAIRMAN. Now Mr. Frederick Williford and Mr. Neel Edwards. You gentlemen are both from the National Federation of Independent Business, am I right? Director of government affairs and government affairs representative.

Mr. WILLIFORD. The federation certainly appreciates this opportunity to present a summary of our testimony and requests our full testimony be entered into the record.

Mr. CHAIRMAN. It will be.

Mr. WILLIFORD. Mr. Chairman, much of my comments today will be based upon a special survey made by the federation which, if you desire, I will also submit for the record.

The CHAIRMAN. We would appreciate it; thank you. We will insert it in an appropriate place in the record.

Mr. WILLIFORD. The National Federation of Independent Business is the largest single member business organization in the United States. Our membership, which numbers 344,000 small and independent businesses, covers the entire spectrum of the U.S. economy. The majority of our members are proprietorships and partnerships. Over 85 percent of our members employ fewer than 20 workers.

The federation is unique in another major respect. No position is taken by the federation on any legislation until our membership is polled to determine their attitude and opinion on that legislation. These surveys result in what are called mandate votes. Mr. Chairman, these mandate votes, however, are not to be confused with the survey I spoke of earlier.

The CHAIRMAN. Do you have any samples of any of your polling questions, questionnaire?

Mr. WILLIFORD. We have a sample of the questionnaires that resulted in this minimum wage study. However, I don't have samples of the mandate votes, the individual votes our members filed on given legislation each month.

The CHAIRMAN. No, no, just on this; you do have it?

Mr. WILLIFORD. Yes I have it and I am quite willing to submit it to you. Therefore, Mr. Chairman, the federation's testimony is not abstract theorizing. It is based on the documented opinions of tens of thousands small and independent businessmen.

Surveyed during the 92d Congress. 84 percent favored retention of the hourly minimum wage at the current $1.60 rate. Fifteen percent voted in favor of an increase, while 1 percent indicated no opinion.

Because the minimum wage question has so many ramifications, the federation determined that more detailed information is required for a thorough analysis of the impact of an increased minimum wage. The federation retained Faculty Associates, Inc., a research firm under the direction of Dr. Richard M. Bailey, University of California at Berkeley.