The Response of Hours of Work to Increases in the Minimum Wage.

Kenneth A. Couch [*]

David C. Wittenburg [+]

This paper examines the effect of minimum wage increases on the hours of work of teenagers (ages 16 to 19) using monthly data from the Current Population Survey. Our findings are consistent with the prediction from neoclassical theory that minimum wage increases have a negative effect on labor demand. However, the estimates we provide here for the elasticity of hours of teen labor demanded with respect to the minimum wage suggest that alternative estimates based on aggregate employment consistently understate the total impact of minimum wage increases on teenage labor utilization.

1. Introduction

While raising the minimum wage has long been thought to be associated with a reduction in the utilization of labor by profit-maximizing firms, measures of the responsiveness of labor demand other than employment to changes in the minimum wage are seldom examined. The most recent literature on the impact of minimum wage increases focuses almost exclusively on its impact on overall employment (Neumark and Wascher 1992, 1994; Card and Krueger 1995; Deere, Murphy, and Welch 1995; Burkhauser, Couch, and Wittenburg 2000a, b). Well-known surveys point out the potential importance of alternative measures of labor utilization (Hamermesh 1993, p. 62). In this paper, we examine the impact of minimum wage increases on the number of hours worked by teenagers using an estimation strategy from our previous research (Burkhauser, Couch, and Wittenburg 2000a, b).

The primary reason for examining hours is that changes in aggregate employment might obscure an increase or decrease in labor demand as measured by hours of work. Consequently, the elasticities of labor demand estimated from employment data might provide a biased depiction of the overall responsiveness of labor utilization to changes in the minimum wage. Whether such a bias exists, along with its direction and magnitude, is the topic of this paper.

In the remaining sections, we first highlight results from the recent literature on the relation between the minimum wage and employment. Then we discuss the data and estimation strategy used to calculate the impact of minimum wage increases on hours. Next, we present our empirical results and compare them to estimates of the effect of minimum wage increases on employment from our previous research. We show that hours of work are reduced proportionately more than employment when the minimum wage is increased, and accordingly, we conclude that elasticity measures based on employment data understate the total responsiveness of labor demand as measured by hours of work to changes in the minimum wage.

2. Literature Review

We confine our literature review to studies that examine the impact of minimum wage increases on state teenage employment levels using data from the Current Population Survey (CPS) (Neumark and Wascher 1992, 1994; Card, Katz, and Krueger 1994; Card and Krueger 1995; Deere, Murphy, and Welch 1995; Burkhauser, Couch, and Wittenburg 2000a, b). [1] A standard equation from this literature is:

[E.sub.it] = [[alpha].sub.0] + [MW.sub.it]B + [X.sub.it][[chi].sub.n] + [T.sub.t] [t.sub.t] + [S.sub.i][s.sub.i] + [[epsilon].sub.it] (1)

where [E.sub.it] is the employment to population ratio of the group of interest in state i in year t; [MW.sub.it] is a variable representing the minimum wage in state i in year t; [X.sub.it] is a set of explanatory variables; [T.sub.t] is a set of dichotomous time variables; and [S.sub.i], is a set of dichotomous locational (state) variables. The key coefficient of interest is B, which represents the effect of minimum wage increases on employment.

Similar to other papers in the minimum wage literature, the central papers using this estimation strategy draw conflicting conclusions. [2] Neumark and Wascher (1992) were the first to use the model described above and found that the impact of minimum wage increases on the employment of teenagers was negative and significant. Card, Katz, and Krueger (1994) and Card and Krueger (1995) reviewed Neumark and Wascher's analysis and found that when a more direct measure of the minimum wage was used, the impact of minimum wage increases on employment was insignificant, and, in some cases, positive. Finally, Deere, Murphy, and Welch (1995) found a negative and significant effect of minimum wage increases on employment using a nonlinear measure of the minimum wage, though, unlike the studies above, they excluded year effects from their specification.

Recently, we reviewed the above studies and showed that the estimates are very sensitive to the specification of year effects (Burkhauser, Couch, and Wittenburg 2000a, b). We demonstrated empirically that when year effects are included, all of the variation that originates from federal increases is eliminated from the variable that represents levels of the minimum wage across locations. [3] Hence, specifications that include year effects only capture the small amount of variation where the state minimum wage is higher than the federal minimum wage. We concluded that although the experience of states that have a higher minimum than the federal rate could be used as a proxy, it is questionable to measure the effects in this fashion because there is much more variation in the federal minimum wage over time. Across a wide variety of specifications, whenever year effects are omitted from the estimations, one consistently finds that raising minimum wages decreases teen employment.

Although the above studies provide important information on the impact of minimum wage increases on aggregate employment, they do not provide estimates for the effect of these increases on total labor utilization. A general methodological consideration is whether the measure of labor should be hours worked rather than overall employment. As discussed in Hamermesh (1993, P. 62), when labor is homogeneous and all workers are engaged for an identical number of hours, the choice between employment and hours is inconsequential. However, when hours are heterogeneous, the choice of employment over hours of work as a measure of labor demand may lead to biased estimates.

There are two potential biases in using employment as the dependent variable. Changes in employment will overstate the effect on hours if those who keep their jobs increase their hours after a minimum wage increase. This situation might occur because employers seek to increase the hours of a smaller staff to avoid fixed employment costs. Changes in employment will understate the effect on hours if employers decrease the hours of work of their remaining employees rather than laying off more workers. [4]

To investigate whether the impacts of federal minimum wage policies are similar in the dimensions of hours and employment, we estimate models similar to those found in our prior work on employment (Burkhauser, Couch, and Wittenburg 2000a, b) but instead examine hours of work. In fact, the only difference between the estimates found below and those based on Equation 1 from our prior work is that we measure the effect of minimum wage increases on the average number of hours worked by teenagers in each state, rather than on aggregate state teenage employment. To be consistent with our previous work, we also test the sensitivity of our results across alternative parameterizations of the minimum wage. It is important to note that we only report estimates for the effect of all minimum wage increases (state and federal) on hours worked. [5]

3. Data

We use monthly data from the outgoing rotation groups of the CPS. We create state-month observations on the basis of individual level information provided in each of the monthly CPS data sets. In each month, approximately 22,500 observations of working-age individuals are available for use in construction of the state-month observations. Sample weights are used to make the sample nationally representative.

In our original analysis, we constructed data extracts for the periods from January 1979 to December 1992 and from January 1979 through December 1997. The analysis from the first period (1979 through 1992) approximately replicated the period covered in the previous literature (e.g., Card and Krueger 1995). This sample contains 8568 state-month observations. We added the analysis from the second period to provide additional evidence for the impact of recent minimum wage increases in 1996 and 1997. [6]

We only report comparable results for hours of work in this paper using data from the 1979 through 1992 period because we are concerned that changes in the CPS questionnaire in 1994 might influence estimates from 1979 through 1997. [7] Although these changes do not affect how employment is recorded, they have significant effects on the number of hours reported. Thus, we omit the hours estimates in the later period from this paper.

Table 1 provides definitions of all the variables used in our analysis, as well as their weighted means and standard deviations. These variables include the log of the greater of the federal or state minimum wage in each state, the log of the average adult wage in the state, the proportion of teenagers in each state's population, the prime-age male unemployment rate in each state, state dummies, seasonal dummies, and a recession dummy.

The primary dependent variable for our analysis is the average number of usual hours worked by all teenagers on their primary job, rather than the employment-to-population ratio used in the previous literature. To construct this variable, we aggregate the weighted number of hours and divide by the relevant state population. We set the dependent variable to zero for those who did not work in the previous week.

4. Results

Table 2 contains the parameter estimates from regressions of average usual hours of work for Equation 1. [8] In the first column, we exclude state and seasonal controls and find that the minimum wage has a negative and significant effect on average usual hours worked. The associated hours elasticity is -0.481. This estimate indicates that a 10% increase in the minimum wage is associated with a 4.8% decline in the average number of hours worked by teenagers. In columns 2 through 4, we incrementally add state, seasonal, and recession control variables to the specification in column 1. Across all specifications, we find a negative and significant effect of the minimum wage on average teenage hours worked.

As the primary focus of the paper is comparing elasticity estimates based on hours with those based on employment, the last row of Table 2 contains comparable employment-based elasticity estimates from our prior work (Burkhauser, Couch, and Wittenburg 2000a, Table 3). The elasticities in the last row were calculated using the employment-to-population ratio as the dependent variable in an otherwise identical estimation procedure to that described in the paper for hours.

As can be seen by reference to the table, the elasticities based on hours consistently indicate that larger reductions in labor usage by firms occurs than is suggested by estimates based on employment figures. The additional response of the demand for labor based on the hours elasticities ranges from 13.3 to 28.4%.

In estimates not shown in the paper, we recalculate the elasticities shown in Table 2 using the alternative CPS measure of hours, the average number of total hours worked last week by teenagers at all jobs. We construct the variable for use in the estimation procedure in the same fashion as the usual hours variable. Our range of elasticity estimates (-0.509 to -0.594) is slightly higher than for usual hours and indicates that minimum wage increases have about the same effect on either "usual" or "total" hours from the previous week.

In our prior work, we also tested the robustness of the finding that the minimum wage reduces employment to alternative methods of representing the minimum wage in Equation 1. In Table 3, we provide the same sensitivity tests for our result that raising the minimum wage leads to reductions in hours of work. Here, we present results for the model that contains all of the control variables except for the recessionary controls as in column 3 of Table 2. [9] This specification typically results in an intermediate outcome in terms of the size of the estimated labor market impact and was the model used in our work on cross-group differences in minimum wage impacts (Burkhauser, Couch, and Wittenburg 2000b).

For comparative purposes, the first row of Table 3 contains the elasticity estimates from column 3 of Table 2. The second row of Table 3 contains an elasticity calculated where the minimum wage is allowed to have both a contemporaneous as well as a lagged effect that occurs a year later. [10] The elasticity based on those combined effects for hours is -0.765 versus -0.583 for employment. Finally, we consider a model where rather than using a single variable containing the value of the minimum wage for each state--month observation, we construct dummy variables representing each individual level of the minimum wage. For this nonlinear specification, the last two rows of Table 3 report the calculated elasticities for the discrete increases in the minimum wage to $3.80 (-0.443) and $4.25 (-0.491). The comparable elasticities from our prior work based on employment are -0.396 for the $3.80 increase and -0.406 for the $4.25 increase.

Across the methods considered for representing the minimum wage, we consistently find that raising the statutory minimum wage reduces demand for hours of work. The additional reduction in the demand for labor implied by the elasticities based on hours of work relative to those based on employment contained in Table 3 range from 11.8 to 30.9%.

5. Conclusion

This study examines whether the adjustment of hours of work by employers after minimum wage increases in the period from 1979 to 1992 was larger or smaller than their adjustment in the dimension of employment. The data and equations estimated in this paper are identical to those found in Burkhauser, Couch, and Wittenburg (2000a) with the exception of the choice of the dependent variable. This allows us to make direct comparisons between the estimates we make here for the elasticity of hours of work with respect to the minimum wage and our prior estimates of the elasticity of employment.

In all of our estimates, we find that raising the minimum wage reduces hours worked by teens. Our results also indicate that estimates of the elasticity of teen labor demand with respect to the minimum wage based on employment data consistently understate the effect of minimum wage increases on labor utilization by 10 to 30% relative to those based on hours of work. The understatement of the impact of minimum wages on labor demand that occurs when aggregate employment rather than hours is examined is the result of employers choosing to decrease hours of teen workers who retain their jobs.

(*.) Department of Economics, University of Connecticut, 341 Mansfield Road, Storrs, CT 06269, USA; E-mail [e-mail].

(+.) Research Associate, The Urban Institute, 2100 M Street NW, Washington, DC 20037, USA; E-mail [e-mail]; corresponding author.

Thanks to David Autor and Daniel Fecnberg for valuable comments that improved the paper. The views expressed are those of the authors and should not be attributed to the Urban Institute or its sponsors.

(1.) Extensive summaries of the recent literature are available (Burkhauser, Couch, and Glenn 1996).

(2.) For a more detailed review of these papers, see Burkhauser, Couch, and Wittenburg (2000a, b).

(3.) This occurs because federal minimum wage rates are raised in discrete steps and typically do not change over a calendar year.

(4.) Employers would also be expected to substitute adult for teen labor when the price of teen labor increases. Unfortunately, we cannot estimate the impact of minimum wage increases on prime-age adult employment using Equation 1 because one of our key control variables, the state unemployment rate, is systematically (inversely) related to prime-age adult employment.

(5.) We also estimated the impact of state minimum wage increases by including year dummies. Our findings are consistent with those of Burkhauser, Couch, and Wittenburg (2000a) and Card and Krueger (1995). We find an insignificant impact of state minimum wage increases on hours worked. This result is not surprising, given that over 90% of the variation in the minimum wage is lost when year fixed effects are included in the specification. Those estimates are available upon request.

(6.) Because of the improving economic climate from 1993 to 1997, these estimates also provided a sensitivity test for our results to a period of more robust economic growth.

(7.) Specifically, the CPS changed its instrument to more closely probe hours worked in the second job and hours worked by those with limited labor force participation. Consequently, post-1994 estimates of total hours worked last week will be much larger than pre-1994 estimates. The usual hours worked last week question was also changed to capture individuals who had variation in usual hours worked. We find that the changes in total hours of work questions had a larger effect on survey responses than the change in the usual hours questions. For example, between 1993 and 1994, the number of total hours increased by 23%, whereas the usual hours estimate decreased by less than 3%. Although empirically, both series appeared to have a seam problem due to the change in questions, the difference in the usual hours series at the point of the break appeared relatively small. We estimated our models using the usual hours variable for 1979 through 1997 and found that the associated elasticities were approximately equiva lent to those for the 1979 through 1992 period. Given the more robust economic growth in the years from 1993 through 1997, we would have expected the estimated impacts to have declined. We felt these anomalous results were potentially due to the altered CPS questions and chose not to present them here. Those estimates are available upon request.

(8.) The corresponding estimates for employment are in Table 3 of Burkhauser, Couch, and Wittenburg (2000a).

(9.) We made comparable estimates based on the alternative specifications contained in Table 3 for each set of control variables alternatively considered in the columns of Table 2. All of those estimates yield the same qualitative conclusion drawn in the text. We also tested the sensitivity of our results to corrections for state-specific autocorrelation, state-specific heteroscedasiticity, and across-state correlations. These alternative specifications do not change any of our substantive findings.

(10.) For a more complete discussion see Burkhauser, Couch, and Wittenburg (2000a, p. 668).

References

Burkhauser, Richard V., Kenneth A. Couch, and Andrew J. Glenn. 1996. Public policies for the working poor: The earned income tax credit versus minimum wage legislation. Research in Labor Economics 15:65-109.

Burkhauser, Richard V., Kenneth A. Couch, and David Wittenburg. 2000a. A reassessment of the new economics of the minimum wage literature using monthly data from the current population survey. Journal of Labor Economics 18:653-80.

Burkhauser, Richard V., Kenneth A. Couch, and David Wittenburg. 2000b. Who minimum wage increases bite: Results from the current population survey and survey of income and program participation. Southern Economic Journal 67:16-40.

Card, David, Lawrence F. Katz, and Alan B. Krueger. 1994. Comment on David Neumark and William Wascher, Employment effects of minimum and subminimum wages: Panel data on state minimum wage laws. Industrial and Labor Relations Review 48:487-96.

Card, David, and Alan B. Krueger. 1995. Myth and measurement: The new economics of the minimum wage. Princeton, NJ: Princeton University Press.

Deere, Donald, Kevin M. Murphy, and Finis Welch. 1995. Reexamining methods of estimating minimum wage effects: Employment and the 1990-1991 minimum wage hike. American Economic Association Papers and Proceedings 85:232-7.

Hamermesh, Daniel S. 1993. Labor demand. Princeton, NJ: Princeton University Press.

Neumark, David, and William Wascher. 1992. Employment effects of minimum wages and subminimum wages: Panel data on state minimum wage laws. Industrial and Labor Relations Review 46:55-81.

Neumark, David, and William Wascher. 1994. Employment effects of minimum and subminimum wages: Reply to Card, Katz, and Krueger. Industrial and Labor Relations Review 47:497-512.

Table 1. Definitions, Means, and Standard Deviations of Variables Used
in the Analysis
Variable            Description
Usual hours         Average of usual hours worked in primary
                     job last week in each state
Log state minimum   Natural log of the greater of the
                     state or federal minimum wage a
Log adult wage      Natural log of the wage of
                     prime age adults (25-64)
Share of teenagers  Share of teenagers in the
                     overall state population
Unemployment rate   Prime-age (25-64) male unemployment
                     rate in the state
Seasonal dummies    Dummy variables with a value of one
                     for each month of the year
State dummies       Dummy variables with a value of
                     one for each state
Recession dummy     Dummy variables with the value of one
                     when the economy was in recession
Number of states    (includes District of Columbia)
N
                        CPS
Variable             1979-1992
Usual hours            11.1
                      (21.9)
Log state minimum       1.23
                       (0.10)
Log adult wage          2.21
                       (0.21)
Share of teenagers      0.11
                       (0.02)
Unemployment rate       0.05
                       (0.03)
Seasonal dummies
State dummies
Recession dummy
Number of states       51
N                    8568
The means are based on weighted state averages from the relevant
surveys. Standard deviations are in parentheses.
(a) We calculate the wage rate using the CPS as the hourly wage rate
for those who report a wage rate, or, for salaried employees, as
earnings divided by hours.
Table 2. Effects of Minimum Wage on the Ratio of Usual Teenage Hours
(ages 16 through 19) to Teenage Population (January 1979 to December
1992) a
Explantory
Variables                   (1)         (2)         (3)
Log state minimum          -5.34 **    -5.57 **    -6.07 **
                           (0.57)      (0.60)      (0.46)
Log adult wage             -0.29       -0.05       -0.08
                           (0.96)      (0.35)      (0.27)
Share of teenagers         -0.68        4.26        3.42 **
                           (2.49)      (2.43)      (1.84)
Unemployment rate         -37.l **    -34.l **    -27.5 **
                           (1.41)      (1.44)      (1.10)
State effects                No          Yes         Yes
Seasonal adjustment          No          No          Yes
Recessionary dummy           No          No          No
[R.sup.2]                   0.10        0.23        0.56
N                        8568        8568        8568
Hours elasticity           -0.481      -0.501      -0.547
Employment elasticity b    -0.450      -0.413      -0.440
Explantory
Variables                   (4)
Log state minimum          -6.38 **
                           (0.48)
Log adult wage              0.09
                           (0.29)
Share of teenagers          3.09 **
                           (1.84)
Unemployment rate         -27.3 **
                           (1.11)
State effects                Yes
Seasonal adjustment          Yes
Recessionary dummy           Yes
[R.sup.2]                   0.56
N                        8568
Hours elasticity           -0.574
Employment elasticity b    -0.447
(**) Statistically significant at the 5% level.
(a) Standard errors are in parentheses.
(b) Employment estimates from Burkhauser, Couch, and Wittenburg (2000a)
 Table 3.
Table 3. Comparison of Estimated Elasticities using Alternative Measures
of Measures of the Minimum Wage for the Preferred Hours and Employment
Specifications
                            Usual Hours     Aggregate Employment
Elasticity Estimates      Worked Estimates      Estimates a
Log state minimum              -0.547              -0.440
Log state minimum-lagged       -0.765              -0.584
Nonlinear specification
 * Minimum wage of $3.80       -0.443              -0.396
 * Minimum wage of $4.25       -0.491              -0.406
(a) The aggregate employment estimates are from Burkhauser, Couch, and
Wittenburg (2000a); Table 3, colum 3 (log state minimum); Table 5,
column 3 (log-state minimum-lagged); and Table 6, column 3 (nonlinear
specification).