Printer Friendly

The effect of government-mandated benefits on youth employment.

Recent years have witnessed persistently high rates of youth unemployment. For example, in 1993 the unemployment rate for all youths aged 16-19 was 19%, and for black teenagers it was 39% (U.S. Bureau of the Census 1994). The 1993 figures are not atypical and are fairly close to the average over the past twenty years. The magnitude of these numbers suggests that youths face significant problems in the labor market. The general impression has been that the labor market is not providing adequate opportunities for youths. Furthermore, there is a belief that many social problems, particularly those that plague the inner city, are related to youth joblessness.

The federal government has periodically proposed programs to alleviate the "youth employment problem." A good example is the Job Training Partnership Act, which contains many initiatives specifically targeted toward youth. The government has also responded to the problem by lowering the employer's cost of compensation; two examples of this strategy are tax credits for training and employment and the creation of a youth sub-minimum wage. In addition to explicit wage adjustments, the federal government allowed the real value of the minimum wage to erode throughout the 1980s. The economic justification for these compensation-based policies is straightforward: a reduction in the employer's labor cost will lead to an increased demand for young workers.

The effect on youth employment of changes in the employer's cost of compensation, however, is currently a much-debated issue. The debate has been fueled by two recent developments. First, several recent papers reported no correlation, and in some cases a positive correlation, between the minimum wage and youth employment. These papers raise questions about the structure of the low-wage labor market and the soundness of using the simple competitive model to predict behavior in this market. Second, the current administration recently proposed a universal health insurance plan to be financed by a payroll tax collected from the employer. The likely effects of such a tax on youth and low-wage employment have become a focal point in the health care debate (O'Neill and O'Neill 1994).

The only current source of evidence on how government-imposed employer mandates affect youth employment is that found in the minimum wage studies, which have not produced a consensus. My purpose in this paper is to provide further empirical evidence on that issue. In particular, I use state and time variation in the unemployment insurance tax and costs of workers' compensation insurance to examine the impact of these programs on youth employment. I also revisit the minimum wage issue. The findings from this study will, I hope, help inform policy-makers about the potential impact on youth employment of future employer mandates.

Background

The mere existence of high rates of youth unemployment does not necessarily indicate that there is a labor market problem. Zagorsky (1993) argued that official unemployment statistics tend to exaggerate the extent of joblessness among youths. Furthermore, Clark and Summers (1982) showed that a portion of youth unemployment is due to the normal workings of a dynamic labor market in which young people hold many jobs for relatively short durations.

Nevertheless, substantial economic research suggests that high rates of youth unemployment result in less human capital development and reduced future economic welfare. D'Amico and Maxwell (1994), Lynch (1989), and Ellwood (1982) all provided evidence that unemployment during the teen and young adult years results in reductions in future wages and employment opportunities.(1)

There has been considerable debate in the literature over whether youth unemployment is caused by demand- or supply-side factors. Finnie and Cain (1990) reviewed the various explanations of youth unemployment and provided a detailed empirical analysis. In addition, Ihlanfeldt and Sjoquist (1990) and Lynch (1989) presented empirical analyses of the determinants of youth employment.(2)

With regard to the employment impact of the minimum wage, standard economic theory has long held that minimum wage increases reduce employment. Several recent studies, however, have challenged that assumption - among them, three articles in the October 1992 issue of this journal (Card 1992a, 1992b; Katz and Krueger 1992).(3) As noted above, the debate surrounding that issue is one reason more empirical evidence is needed on the link between unemployment and various forms of government intervention in the labor market.

Employment Effects of Government-Mandated Employer-Provided Benefits

Only a few studies have investigated the employment effects of payroll taxes and mandated benefits, and none have focused on youths. In addition, few of the studies that examine the issue actually look at employment. In most cases, conclusions regarding the effect of these policies on employment are inferred from evidence related to the effect of payroll taxes and mandated benefits on wages. A simple supply and demand analysis may be used to justify statements about employment from observed movements in wages. The imposition of a payroll tax decreases the demand for labor and will reduce both employment and wages. The magnitudes of the wage and employment changes are determined by the relative wage elasticities of the demand and supply of labor. For example, if supply is highly elastic, the imposition of a payroll tax will result primarily in a decrease in employment and little change in wages. On the other hand, if supply is highly inelastic, the imposition of a payroll tax will result in relatively little change in employment and a relatively large decrease in wages.(4) A similar analysis may be applied to mandated benefits, except that in this case, the supply of labor is increased if the worker values the benefit (Summers 1989).

Most germane to the current analysis are the several studies that have examined the effect of workers' compensation insurance on wages, since workers' compensation insurance is a government-mandated benefit.(5) In general, the evidence presented in these papers suggests that increases in insurance costs (benefits) lead to lower wages and small decreases in employment. Moore and Viscusi (1990) concluded from their findings that more than 100% of the employer's total cost of workers' compensation insurance was offset by lower wages. This result suggests that employment would actually increase in response to an increase in costs (benefits).(6) Gruber and Krueger (1991) estimated that 86% of the cost of workers' compensation insurance was offset through lower wages, a result which also suggests that there would be little employment effects. The same study also found that workers' compensation insurance had virtually no employment effects.

Those results are consistent with the predictions from a simple supply and demand model. In response to the imposition of the mandate, there is a decrease in demand for labor and an offsetting increase in labor supply, leaving total compensation and employment relatively unchanged.

Unemployment insurance is similar to workers' compensation insurance in that it is a government-mandated benefit, but in this case the benefit is financed through a payroll tax. Topel (1983, 1984a, 1984b) extensively studied the employment effects of unemployment insurance, and presented persuasive evidence that unemployment insurance reduces employment by increasing the amount of temporary and permanent layoffs. The primary reason for this effect is the incomplete experience rating of unemployment insurance, which provides a subsidy to unemployment.(7) Topel (1984b) also examined the effect of unemployment insurance on wages, and found that higher unemployment insurance benefits reduced wages.

Previous studies of the effects of payroll taxes are few in number. One notable study examining the impact of the Social Security tax is Hamermesh (1979). The results of that study suggest that for men aged 30-45, 15-35% of the tax was offset by lower wages. For all men over age 30, the estimates of the wage offset were between 0 and 36%. The relatively small wage offsets Hamermesh found led him to conclude that changes in payroll taxes have relatively large employment effects, as employers are unable to shift the burden of the tax onto their employees.

In summary, past research has indicated that mandated benefits such as workers' compensation insurance are substantially offset by wage cuts and have small to no employment effects. In the case of unemployment insurance, there may be somewhat larger employment effects because of the way unemployment insurance is financed and the subsidy it provides to unemployment. The evidence regarding the employment effects of payroll taxes is limited, but the one study noted above suggests large disemployment effects.

These studies, however, were conducted on samples of adults. For youths, the presence of a minimum wage may limit the size of the wage offset, and as a result, most of the effect of mandated benefits and payroll taxes may be in employment. Furthermore, since youths generally are less familiar than adults with the labor market, they may not know the risk of injury on the job, or the probability of unemployment, and therefore may not value workers' compensation or unemployment compensation benefits. Under these circumstances, a mandated benefit is identical to a payroll tax, and we would expect larger employment effects than otherwise would be the case. The Present study is the first to examine this question.

Empirical Model

The empirical model used in this paper is a reduced form model of employment. The basic model is specified as follows:

(1) [E.sub.it] = [[Alpha].sub.0] + [[Alpha].sub.1] [MW.sub.it] + [[Alpha].sub.2] [MW.sub.it-1] + [[Alpha].sub.3] [UC.sub.it] + [[Alpha].sub.4] [WC.sub.it] + [[Alpha].sub.5] [U.sub.it] + [[Alpha].sub.6] [POP.sub.it] + [[Theta].sub.i] + [[Delta].sub.t] + [[Epsilon].sub.it],

where E is the employment to population ratio of youth, MW is a measure of the minimum wage, UC is the employer's cost of unemployment insurance, WC is the employer's cost of workers' compensation insurance, U is the unemployment rate for all men, POP is the proportion of the population comprised of youths, and [Epsilon] is an error term. Parameters to be estimated include the [Alpha]'s, [[Theta].sub.i], and [[Delta].sub.t]; i is an index of states, and t is an index of time.

Equation (1) is similar to the standard model found in many past studies of the employment effects of the minimum wage, and represents a reduced form model of aggregate youth employment (Neumark and Wascher 1994). Thus, the model contains variables that affect both the supply (POP) and demand of labor. The primary difference between the current model and models used in earlier studies is the addition of the employer cost variables UC and WC.(8)

Data

The data used in this study are a time series of state aggregates covering the period 1982-89. All 50 states and the District of Columbia are represented. The dependent variable is the annual average employment to population ratio for youth in a state in a given year. Three age groups are examined: 16-19, 20-24, and 25-34. The employment to population ratio is derived from the Current Population Survey (CPS) and is calculated using all 12 monthly CPS surveys.(9)

The use of the annual average employment to population ratio is a significant improvement over past studies that use state aggregate data. Past studies typically use the employment to population ratio derived from one monthly CPS survey. The use of the annual numbers reduces the measurement error in this variable, particularly for small states (Card, Katz, and Krueger 1994). In addition, the employer contribution variables (UC and WC) are also yearly measures and are more consistent with an annual employment to population measure.

The independent variables in the model are a minimum wage measure, the employer's cost of unemployment insurance, the employer's cost of workers' compensation insurance, the unemployment rate for all male workers, the proportion of the population comprised of youth (16-19, 20-24, 25-34), and state and year dummy variables. The minimum wage variable used in this analysis is a modified version of the Kaitz index, and is the same as that used by Neumark and Wascher (1992, 1994).(10) This variable is the coverage-adjusted relative minimum wage. The variable is constructed as follows:

(2) [MW.sub.it] = ([B.sub.it] * [MIN.sub.it]) / [WAGE.sub.it],

where B is the proportion of wage and salary workers covered by the federal minimum wage law, MIN is the greater of the federal or state minimum wage in effect as of the month of May, and WAGE is the average hourly wage for all workers in the state. WAGE is calculated using data from the May Current Population Survey (CPS).

Arguments concerning the advantages and disadvantages of using a coverage-adjusted relative minimum wage are presented in Card, Katz, and Krueger (1994) and Neumark and Wascher (1994). In light of those arguments, I use alternative measures in the analysis. In particular, the minimum wage and average wage are entered separately, and the coverage variable is excluded.(11) In addition, since the minimum wage variable is calculated as of May in a particular year, while the employment figures are annual averages, I include a lagged value of the minimum wage variable in the model.

The variables that measure the employer's cost of unemployment insurance and workers' compensation insurance are the yearly average employer's contribution to these programs in each state, and are expressed as a percentage of payroll. The unemployment insurance data are taken from the Social Security Bulletin, Annual Statistical Supplement, and are derived from state agencies. The employer contribution is calculated using taxable payroll. The workers' compensation insurance data come from the National Foundation for Unemployment Compensation and Workers' Compensation (NFUCWC). The NFUCWC has estimated the employer contribution from a variety of sources, and has data available beginning in 1982.

These two variables are obviously crude proxy measures for the true cost of these programs for youth employees and thus subject to some amount of measurement error. The average combined contribution rate for the sample is approximately 4%, but the relatively small size of this figure conceals the fact that for low-wage workers, the expected contribution rate is much higher than the average.(12) The presence of statutory minimum benefits for both workers' compensation insurance and unemployment compensation insurance creates a non-linear relationship between the wage and the employer contribution.

An important point to note about the employer cost measures is that they represent employer expenditures and therefore depend on both the price (that is, the tax rate) of workers' compensation or unemployment insurance and the quantity, or frequency of occurrence, of injuries and unemployment. For example, the average employer contribution to workers' compensation insurance will be larger in states with a relatively large number of risky jobs (injuries), even if the tax rate is the same in all states. Equation (1) is a reduced form model of employment, and as such, the relevant variables are the components of the price of labor, including the payroll tax and costs of mandated benefits. The distinction between the price and quantity of the employer contribution is important, because the employment to population ratio may be related to both components. This point is obvious in the context of unemployment insurance. The purpose of this paper, however, is to estimate the effect of a payroll tax and mandated benefit (that is, price) on employment. Therefore, it is necessary to include additional variables in the model that are correlated with the frequency of injuries and the quantity of unemployment.

To capture differences across states and over time in the frequency of injuries, I include in the model a set of industry and occupation share variables and a variable measuring the proportion of the labor force that is female. The industry and occupation data come from the Statistical Abstract of the United States and the female proportion of the labor force is from the Geographic Profile of Employment and Unemployment. The industry share variables measure the proportion of non-agricultural employment in eight industries: mining, construction, manufacturing, transportation, wholesale and retail trade, finance-insurance - real estate, services, and government. The occupation share variables measure the proportion of employment in ten occupations: professional, managerial, sales, clerical, craft, operative, transportation, laborer, services, and agricultural. These variables will control for gross differences in the number and type of injuries. The aggregate nature of these controls may not, however, be a perfect solution to the problem. To control for differences in the quantity of unemployment, I include the male unemployment rate, the insured unemployment rate, and the one year lag of the insured unemployment rate. The insured unemployment rate measures the quantity of unemployment that is covered by the unemployment compensation program, whereas the male (gross) unemployment rate measures insured and uninsured unemployment. The existence of unemployment insurance reserve funds breaks the mechanical short-run relationship between unemployment insurance collections (taxes) and benefit payments. Therefore, the current insured unemployment rate and one year lag of the insured unemployment rate are included in the model to control for state differences in financing unemployment insurance.

The empirical model includes both state and year effects. Thus, in order for an effect of mandated benefits on employment to be identified in this model, there must be significant within-state time series variation in the employer costs variables. More important, there must be significant variation in the underlying tax rates.

Burton and Schmidle (1992) provided a detailed analysis of the variation in employers' costs of workers' compensation insurance. They reported that nominal workers' compensation costs, as measured by adjusted manual rates, decreased slightly in the early part of the 1980s and increased by 70% in the 1984-89 period. Furthermore, they showed that the variation across states was growing over time. The cost of unemployment insurance, they found, also exhibited a significant amount of time and state variation. The average weekly unemployment benefit increased at roughly 3.5% to 4% per year during the 1982-89 period, but the growth rate differed significantly across states. The coefficient of variation in the state growth rates of average weekly unemployment benefits was approximately one. Furthermore, Card and Levine (1994) reported that unemployment insurance tax rates increased significantly during the 1980s and that the increase depended on the taxable wage base and benefit level, which vary over time within states.

[TABULAR DATA FOR TABLE 1 OMITTED]

The remaining variables in the model are the male unemployment rate and the proportion of the population comprised of youths. The male unemployment rate is used to control for aggregate changes in the state economy that will affect employment.(13) The male unemployment rate is taken from the BLS publication Geographic Profile of Employment and Unemployment. The proportion of the population that is comprised of youths, a supply side measure, was derived from the 12 monthly CPS surveys conducted by the BLS. The means for all variables are shown in the appendix.

The Effect of Minimum Wages, Workers' Compensation Insurance, and Unemployment Insurance on Employment

Tables 1 and 2 show the change in employment and wages of youths and young adults between 1985 and 1988 for two sets of states: those in which the average employer contribution to workers' compensation and unemployment insurance underwent a relative increase, and those in which it underwent a relative decline. A relative increase in costs is defined as an increase of more than 30% in the average employer contribution to workers' compensation insurance (11 states), or a decrease of less than 13% in the average employer contribution to unemployment insurance (10 states). A relative decrease in costs is defined as an increase of less than 3% in the average employer contribution to workers' compensation insurance (8 states), or a decrease of more than 40% in the average employer contribution to unemployment insurance (11 states).(14)

The figures in Table 1 indicate that both employment, as measured by the employment to population ratio (Emp/Pop), and wages (Ln Wage) increased less in states that had an increase in the average employer contribution to workers' compensation and unemployment insurance than in states that had a decrease in those measures. The relative differences (differences-in-differences) between the two sets of states, however, are not statistically significant.

For young adults aged 25 to 34, there appear to be no differences in the change in employment between states that had an increase in the average employer contributions and those that had a decrease. In the case of wages, the figures in Table 2 indicate [TABULAR DATA FOR TABLE 2 OMITTED] that wages increased less in states that had an increase in the average employer contribution to workers' compensation insurance than in states that had a decrease. The difference, however, is not statistically significant.

The results in Table 1 and 2 are mixed with regard to expectations. While employment of youths aged 16 to 19 decreased more than that of young adults in response to increases in average employer costs - a pattern that may be expected if the minimum wage is binding for this group - relative wages also tended to fall more for this group. The greater relative fall in wages for youths would not be expected if the minimum wage was binding, or if employment losses were greater. None of the differences in Tables 1 and 2 were significant.

While the descriptive analyses of Tables 1 and 2 provide preliminary evidence of the employment effects of workers' compensation and unemployment insurance, a broader multivariate analysis is necessary to simultaneously control for changes in both policy variables and other factors. Equation (1) is the basic model used to obtain estimates of the effect of workers' compensation and unemployment insurance on youth employment. All estimates of equation (1) are obtained by a weighted least squares procedure that assumes that the residual variance is proportional to the state population. The use of weighted least squares is a common practice when aggregate data are used, since the variables in the model are obtained from state samples that vary in size depending on the population of the state. Thus, some of the variation in the model comes from sampling error, and weighted least squares will yield more efficient estimates than ordinary least squares (Card, Katz, and Krueger 1994).(15)

Employment of Teenagers Aged 16 to 19

Table 3 contains the estimates for the 16-19-year-old age group. The estimates associated with the employer contribution variables (column 1) are negative and, in the case of workers' compensation insurance, statistically significant (p = .052).(16) The coefficient on the unemployment insurance variable does not quite attain statistical significance (p = .14). The estimate associated with the workers' compensation variable suggests that a one percentage point increase in the employer contribution rate reduces the employment to population ratio of youth aged 16-19 by 1.5 percentage points. A similar calculation for unemployment insurance implies a 0.5 percentage point reduction in the teen employment to population ratio.

The magnitudes of these employment effects are reasonably close to previously published estimates. Hamermesh (1986) reviewed several studies that explicitly estimated demand elasticities for youths and young adults, and found that most of the estimates ranged between -0.31 and -1.80. Although the estimates in Table 3 are not pure demand elasticities, they are a measure of the responsiveness of employment to changes in the cost of employment. A one percentage point increase in the employer contribution variables is roughly equivalent to a 1% increase in compensation, since the employer contribution variables are measured as a percentage of payroll, and changes in the employment to population ratio are mainly the result of changes in employment. Therefore, the estimates of the employment effects listed in Table 3 are consistent with previous findings.

It is also important to note, however, that the employer contributions are measured as statewide averages that have a mean of approximately 1.5% for workers' compensation and 2.5% for unemployment insurance. Thus, a one percentage point increase in these rates implies large relative increases in the underlying benefits.

The estimates in Table 3 can be used to evaluate how enactment of the administration's recent health care proposal would affect employment. O'Neill and O'Neill (1994) estimated that the administration's health insurance plan would increase employer contributions as a percentage of payroll by approximately 8%. If we use the coefficient on the workers' compensation variable as an estimate of the impact of the health plan, the teen employment to population ratio will be reduced by 12%. Of course, health insurance differs from both workers' compensation insurance and unemployment insurance, and there is no guarantee that these calculations will be correct. For example, teenagers may value health insurance more than the other two programs currently being examined. Assuming that the estimates would apply, however, the health insurance plan has the potential to significantly decrease youth employment.

Table 3 also contains estimates of the effect of minimum wages on teen employment. The estimates of the minimum wage effects in column 1 are negative, but only the estimate associated with the lagged minimum wage variable is significant.(17) The magnitude of the minimum wage effects is somewhat larger than that found by Neumark and Wascher (1992, Table 5), who used comparable data and a similar specification. In addition, the estimates (not shown) of the minimum wage effects are unchanged in models that omit the employer contribution variables. Thus, the hypothesis that the past estimates of the minimum wage effect may have been biased by the omission of these variables is rejected. It does not appear that minimum wage levels and state-specific employer mandate costs are systematically related.

One noteworthy point about the minimum wage estimates is that they are obtained using a model and data that incorporate several refinements suggested by Card et al. (1994).(18) Surprisingly, correcting for flaws in previous studies appears to result in minimum wage effects that are larger and more significant than those found in earlier work such as Neumark and Wascher (1992).

Returning to the main point, the estimates in column 1 imply that higher workers' compensation costs (benefits) and unemployment insurance costs (benefits) reduce employment. This result is consistent with two different supply and demand models. One explanation is that employees between the ages of 16 and 19 do not value the two benefits, and thus there is no supply shift in response to higher benefits. This will lead to a reduction in employment [TABULAR DATA FOR TABLE 3 OMITTED] and wages. Another explanation is that the minimum wage is binding and thus supply is perfectly elastic. In this case, employment is demand-determined and is reduced as demand decreases in response to higher employer costs. The wage, however, remains unchanged under these circumstances.

One way to determine which of the two explanations is correct is to interact the minimum wage variables with the employer cost variables. If the latter explanation is correct, the disemployment effects should be greater the higher the minimum wage. In columns 2 and 3 of Table 3, the minimum wage variable is interacted with the workers' compensation and unemployment insurance variables. Due to collinearity problems, only one minimum wage variable is used. Including both the current and lagged minimum wage and complete interactions of these variables with the workers' compensation and unemployment insurance variables results in substantial multicollinearity. The estimates listed in columns 2 and 3 do not support the hypothesis that the disemployment effects associated with higher workers' compensation and unemployment insurance costs are due to a binding minimum wage. None of the estimates associated with the variables of interest are significant, and the pattern of the signs of the estimates is not consistent with predictions.

The model specification underlying the estimates in columns 2 and 3, however, does place heavy demands on the data. identification of interaction effects comes from differences in the changes (that is, differences in differences) in minimum wages and employer costs over time within a state. There may not be enough variation in the data to identify such an effect.

Table 3 also contains the estimates from an additional model specification. In column 4, the coverage-adjusted minimum wage has been replaced by the statutory minimum wage, and the average state wage for all workers enters the model separately. This specification was included for two reasons: to test whether the estimates of the employer contribution variables are sensitive to the choice of the minimum wage, and to test whether the negative minimum wage effects observed in columns 1 and 3 are spurious due to a positive correlation between employment and wages.(19) The results indicate that the estimates associated with the employer mandate cost variables are not sensitive to the choice of minimum wage measures. Furthermore, the estimates in column 4 are not consistent with the hypothesis that the negative minimum wage effects are spurious. The estimates of the effects of average state wages are small and insignificant.

Thus, it is difficult to argue that the negative effects of the minimum wage observed in columns 1 and 3 are solely due to the effect of the aggregate state economy on wages and employment. The lack of significance of the statutory minimum wage estimate is not surprising, since this variable does not adequately measure the relative effectiveness of the minimum wage constraint.

One other point to note about the estimates in Table 3 is that the employment to population ratio of youth aged 16 to 19 is very sensitive to changes in the state aggregate economy. Increases in the male unemployment rate lead to large decreases in the youth employment to population ratio. The estimate associated with the proportion of the population that is 16 to 19 is never significantly different from zero.

Employment of Young Adults Aged 20 to 24

Estimates of the effect of the minimum wage, workers' compensation insurance, and unemployment insurance on the employment of young adults aged 20 to 24 are listed in Table 4. The format of Table 4 is the same as that of Table 3.

There are a few differences between the estimates in Table 3 and Table 4. In Table 4, the estimates of the effect of the minimum wage are mixed. The estimate associated with the contemporaneous minimum wage is negative and marginally significant, but the effect of the lagged minimum wage is positive and never significant. In regard to the workers' compensation and unemployment insurance variables, only the coefficient on the workers' compensation variable is significantly different from zero. For this age group, a one percentage point increase in the employer's contribution rate to the workers' compensation program is expected to reduce the employment to population ratio by two percentage points.

As in the case for teenagers, the estimates associated with the interactions between the minimum wage and the employer cost variables do not provide evidence that the disemployment effects associated with the workers' compensation variable are due to a binding minimum wage. The interaction between the minimum wage and the workers' compensation variable is never significant. In addition, there does not appear to be consistent evidence that the employer costs variables have a smaller (less [TABULAR DATA FOR TABLE 4 OMITTED] negative) effect on young adults aged 20 to 24 than on teenagers. This result would be expected if the primary reason for the disemployment effects is the minimum wage constraint. Young adults are less constrained than teenagers by the minimum wage. The disemployment effects of higher workers' compensation insurance costs are greater for young adults than for teenagers, although the disemployment effects of unemployment insurance are greater for teenagers than for young adults.

One final point about Table 4 concerns the population variable. The estimate associated with this variable is positive and highly significant. The positive estimate associated with the population variable is unexpected and may reflect reverse causality: increases in the employment to population ratio for the group may lead to growth in the group's relative population, due to migration.

Employment of Young Adults Aged 25 to 34

The last estimates regarding employment are listed in Table 5. In Table 5 the effects of the minimum wage, workers' compensation insurance, and unemployment insurance on the employment of young adults aged 25 to 34 are listed. The minimum wage has no effect on employment, and of the two employer cost variables, only the workers' compensation variable is significant. These results provide evidence that the disemployment effects associated with the employer mandates are not being driven solely by a binding minimum wage. The minimum wage has no effect for this age group, while at the same time there appear to be some disemployment effects associated with workers' compensation insurance.

The Wage Effects of Minimum Wages, Workers' Compensation Insurance, and Unemployment Insurance

The analysis of the employment impact of minimum wages, workers' compensation insurance, and unemployment insurance showed the following: increases in the minimum wage tend to have a negative effect on employment for teenagers and to a lesser extent on young adults between the ages of 20 and 24; increases in workers' compensation insurance costs tend to decrease employment for both teenagers and young adults; and increases in unemployment insurance costs tend to reduce teenagers' employment. Furthermore, the analysis of employment provided some evidence that the disemployment effects of workers' compensation insurance and unemployment insurance are not solely due to a binding minimum wage. This conclusion is based on two pieces of evidence: (1) the lack of significance and contradictory signs of the estimates associated with interactions between the minimum wage and the employer cost measures, and (2) the existence of roughly similar disemployment effects across the three age groups.

Both of those pieces of evidence, however, are subject to some qualification. The significance of the first piece of evidence may be discounted due to the demands that the fixed effect methodology places on the data. The second piece of evidence is questionable because there may be other reasons for differences across age groups in disemployment effects. For example, the demand for teen labor of a quality that is paid at or near the minimum wage may be quite elastic. Thus, the disemployment effects of an employer mandate may be great for this group, resulting in a small to moderate effect for the entire group of teenagers. On the other hand, the elasticity of demand for young adult labor may be less elastic than it is for teenagers, and therefore the disemployment effect associated with an employer mandate would also be small for young adults. Other scenarios can be envisioned that emphasize relative supply shifts, but identifying exactly what is causing the relationship between employment and the employer mandates is a difficult problem.

Another way to test whether the disemployment effect of the employer costs variables is due to a binding minimum wage is to directly examine wages. If the disemployment effect is due to a binding minimum wage, then wages should be unaffected by the imposition of or increase in a mandated benefit or payroll tax. On the other hand, if the disemployment effect is due to relative shifts in supply and demand, wages will be affected. There should be a wage offset associated with the imposition of or increase in a mandated benefit or payroll tax. Thus, an examination of the effect of workers' compensation insurance and unemployment insurance on wages can provide additional evidence on the cause of the observed disemployment effects.

To obtain an estimate of the wage effect of the employer mandates, I estimate a simple wage regression. The specification of the model is similar to equation (1), except that wages replace the employment to population ratio as the dependent variable. The wage data are derived from the CPS Outgoing Rotation Groups. The Outgoing Rotation Group file is a monthly extract of the CPS that includes only individuals who are in their last four-month period of interviews. The age group and state specific average annual wage is calculated as the average group wage over the entire 12-month period. Individuals who were self-employed or who worked in agriculture were not included in the calculation of the average wage.

[TABULAR DATA FOR TABLE 5 OMITTED]

Wages of Teenagers Aged 16 to 19

Table 6 lists the results from the wage regression model for teenagers aged 16 to 19. The coefficient on the workers' compensation variable in column 1 is negative, but does not quite attain statistical significance (p = .16). The coefficient on the unemployment insurance variable is positive and insignificant. These results are consistent with the hypothesis that the observed disemployment effects of a mandated benefit are due to a binding minimum wage. As the demand for labor decreases in response to higher employer mandate costs, employment is decreased and wages remain unaffected due to the minimum wage constraint.

In columns 2 and 3, the minimum wage variable is interacted with the employer cost measures. The purpose of this specification is to test whether the estimates of the effect of the employer costs variables on wages are smaller (that is, less negative) at higher and presumably more binding levels of the minimum wage. The estimates from these models are imprecisely estimated, however, and difficult to interpret. The estimates of the main and interaction effects associated with the unemployment tax measure are significant, but the marginal effect, evaluated at the mean, remains [TABULAR DATA FOR TABLE 6 OMITTED] approximately the same between columns 1, 2, and 3. The estimates associated with the workers' compensation variable are always insignificant. The data do not seem to have enough variation to allow precise estimates of the interaction effects, as evidenced by the standard errors, which have increased by a factor of five. In light of this problem, most of the remaining discussion related to wages will focus on the simpler specifications.

As can be seen in Table 6, the coverage-adjusted minimum wage is negatively related to the teen wage. This is similar to the results that Card et al. (1994) found in their replication study of earlier work by Neumark and Wascher (1992). This result may seem surprising at first glance, but the negative correlation is expected if there is a positive correlation between adult and teen wages in a state. The usefulness of the coverage-adjusted minimum wage, however, is not undermined by such a correlation, and Neumark and Wascher (1994) provided a detailed analysis of the properties of a coverage-adjusted minimum wage. The statutory minimum wage is positively related to teen wages, as expected. Most important, however, the effect of the employer mandates on wages is affected little by which minimum wage measure is used in the analysis.

Wages of Young Adults Aged 20-24 and 25-34

The effects of the employer mandate costs on the wages of young adults are listed [TABULAR DATA FOR TABLE 7 OMITTED] in Tables 7 and 8. The results for these two groups are quite similar. The coefficient on the workers' compensation variable is negative and significant in both Table 7 and Table 8. The magnitudes of the effects are large. For the younger (20-24) age group, a one percentage point increase in the employer's contribution rate to workers' compensation reduces wages by approximately 6%. This is a large effect and implies an increase in employment, as changes in workers' compensation costs are more than offset by changes in the wage. This result does not appear to be consistent with the results reported above that indicate a decrease in employment in response to an increase in workers' compensation costs. Similarly, for the older age group, a one percentage point increase in the employer's workers' compensation costs reduces wages by 4%. The coefficient on the unemployment insurance variable is insignificant for the younger sample, but positive and significant for the older sample. The positive sign is unexpected.

Discussion

The purpose of this paper has been to examine the effect of government-mandated employer-provided benefits on youth employment. Toward this end, I performed two empirical analyses. The first analysis [TABULAR DATA FOR TABLE 8 OMITTED] examined the effect of such benefits on employment. The results from this analysis suggest that government-mandated employer-provided benefits reduced the employment of teenagers and young adults, although the effects I found are not uniform. The employment effects associated with a measure of the employer's cost of workers' compensation insurance were significant for all three age groups, and the magnitudes of the effects were roughly comparable across the age groups. A one percentage point increase in the employer's cost of workers' compensation insurance reduced employment by approximately 1.5 percentage points.

That finding contrasts with results reported by Gruber and Krueger (1991), who also examined employment. Using state-level data for 10 industries, Gruber and Krueger found small, negative, and insignificant effects of workers' compensation insurance on employment. The difference between their results and mine may be due to the different samples and model specifications. The focus of this paper has been on youth employment across all industries, whereas Gruber and Krueger examined the employment of individuals of all ages in only ten industries. Furthermore, Gruber and Krueger did not include any additional control variables in their analyses, and examined changes in employment as opposed to the employment to population ratio.

In addition, I have presented evidence suggesting that the disemployment effects of workers' compensation insurance were not due to the existence of a binding minimum wage. There was no significant interaction effect between the employer's cost of workers' compensation insurance and the minimum wage.

Most of my estimates of the employment effect associated with unemployment insurance are not significant, although for the teen sample the effect is marginally significant. This result may appear surprising given the work of Topel (1983, 1984a, 1984b), who provided evidence of a significant effect of unemployment insurance on employment. The difference between the estimates may reflect the fact that in Topel's work the driving force behind the results is the degree of experience rating. Topel also showed that the experience rating mechanism affects temporary layoffs more than other forms of unemployment. The current analysis examines the impact of the level of unemployment insurance costs on annual employment. The degree of experience rating is omitted in the current analysis, although the inclusion of state dummy variables should control for permanent differences across states in the level of experience rating. In addition, the young age groups examined here are less likely to experience temporary layoffs than those in Topel's work because they have less job-specific training.

To try to identify the underlying cause of the employment effects noted above, I performed an analysis of wages. The purpose of this analysis was to investigate whether there was a wage offset in response to higher employer mandate costs. The results suggest that there was no wage offset for teenagers, but among young adults, wages and the employer's cost of workers' compensation insurance were negatively related. A one percentage point increase in employer costs reduced wages by between four and six percent. These are large effects that imply increases in employment in response to an increase in workers' compensation costs, a result not found in the analysis of employment. The estimates of the effect of the cost of unemployment insurance on wages are significant only among the 25-to-34-year-old sample, and in this case the effect is unexpectedly positive.

The results of these analyses do not support a neat conclusion. There are a number of interesting findings, but it is hard to bring them together under one simple theoretical umbrella. For example, employer mandate costs were negatively related to teen employment, but unrelated to teen wages. These two results fit nicely into a simple supply and demand model with a binding minimum wage. The problem is that there is other evidence that undermines the credibility of such a construction. First, the estimates of the effect of minimum wages on employment are not very robust. Second, there are no significant interaction effects between the minimum wage and the employer cost measures. Finally, the negative employment effects of the employer cost variables are not always greater among teenagers, who are more affected by a minimum wage than are other age groups.

The results pertaining to the young adults also appear at first blush to fit a simple supply and demand framework. For this group, the results indicate that higher employer costs of workers' compensation insurance reduced both wages and employment. These results suggest that workers' compensation insurance is similar to a pure payroll tax, in which workers either do not value the benefit or value it less than it costs employers to provide. This explanation rings true for young workers, who may not know the true risks of injury on the job, or who may discount future lost earnings at a higher rate than older workers. Qualifying this conclusion, however, is the further finding that the wage offset for this group was large and implied an increase in employment, not a decrease. In addition, I did not find similar results for unemployment insurance. Theoretically, the effect of unemployment insurance should be similar to the effect of workers' compensation insurance.

In summary, the results of this study provide preliminary evidence on an important issue and challenge some previous findings in the literature. Taken at face value, the results suggest that there are significant employment effects associated with employer mandates. Policy-makers should take these effects into consideration when formulating and imposing new mandates on employers.

[TABULAR DATA FOR APPENDIX TABLE 1 OMITTED]

1 In the past twenty years, youths' wages have been falling in real terms and there has been a growing disparity between the wages of relatively inexperienced workers (that is, youths) and more experienced workers (Katz and Murphy 1992; Bound and Johnson 1992).

2 For a survey of earlier research, see Freeman and Wise (1982), Rees (1986), and Freeman and Holzer (1986).

3 For a review of the "new" minimum wage studies, see Card and Krueger (1995). For a review of the literature prior to these studies, see Brown (1988) and Brown, Gilroy, and Kohen (1982).

4 A variety of possible outcomes can be derived depending on the relative shapes of supply and demand.

5 Following is a partial list of studies that examine the wage-benefit tradeoff associated with workers' compensation benefits: Arnould and Nichols (1983), Dorsey and Walzer (1983), Butler (1983), Ruser (1985), Viscusi and Moore (1987), Moore and Viscusi (1989). Moore and Viscusi (1990), and Gruber and Krueger (1991).

6 Moore and Viscusi (1990) examined the total cost of the workers' compensation insurance program. The authors noted that the marginal wage offset of increases in workers' compensation insurance costs is smaller than the average wage offset.

7 The fact that unemployment insurance benefits for certain periods were not taxed also provided a subsidy to unemployment.

8 The addition of the employer cost measures may have a significant impact on the estimate of the employment effects of the minimum wage if these variables are correlated with the minimum wage.

9 The employment to population figures were made available by the Bureau of Labor Statistics (BLS). The BLS publishes identical figures for 16-19-year-old individuals in the publication Geographic Profile of Employment and Unemployment. The BLS does not publish similar figures for other age groups.

10 David Neumark and William Wascher were kind enough to make their data on the minimum wage available for the current research. For additional details related to variable construction, see Neumark and Wascher (1992).

11 Card, Katz, and Krueger (1994) presented evidence critical of the federal coverage variable, and in light of this evidence I omitted the coverage variable. The coverage variable was also excluded from equation (2) in preliminary work. The results from these regressions are similar to those reported later in the text for the coverage-adjusted minimum wage.

12 Note that the unemployment insurance contribution rate is calculated using taxable wages, while the workers' compensation insurance contribution rate is calculated using total wages.

13 Including a measure of total or male unemployment in the model has been a standard feature of virtually all previous research, even though there is a non-trivial likelihood that measures such as these are endogenous.

14 As noted above, the employer contributions are measured as expenditures and are dependent on the tax rate and frequency of occurrence. This fact explains why the average employer contribution to unemployment insurance decreased between 1985 and 1988, a time when unemployment was falling.

15 In order to check the sensitivity of the results to the weighting procedure, White's correction was also used to estimate the standard errors. The results from this alternative procedure were very similar to those reported in the text.

16 All significance levels reported in this paper are based on two-tailed tests.

17 A Lagrange multiplier test rejects the hypothesis that the sum of the coefficients on the minimum wage measures is equal to zero.

18 The current model does not include a school enrollment variable, or other supply-side variables besides population. I use annual employment to population figures and weighted regressions.

19 See Neumark and Wascher (1994) for a more complete discussion of this point.

REFERENCES

Abowd, John, and Orley Ashenfelter. 1981. "Anticipated Unemployment, Temporary Layoffs, and Compensating Wage Differentials." In Sherwin Rosen, ed., Studies in Labor Markets. Chicago: University of Chicago Press, pp. 141-70.

Arnould, Richard J., and Len M. Nichols. 1983. "Wage-Risk Premiums and Workers' Compensation: A Refinement of Estimates of Compensating Wage Differential." Journal of Political Economy, Vol. 91, No. 2, pp. 332-40.

Brown, Charles. 1988. "Minimum Wage Laws: Are They Overrated?" Journal of Economic Perspectives, Vol. 2, No. 3, pp. 133-46.

Brown, Charles, Curtis Gilroy, and Andrew Kohen. 1982. "The Effect of the Minimum Wage on Employment and Unemployment." Journal of Economic Literature, Vol. 20, No. 2, pp. 487-528.

Burton, John F., and Timothy P Schmidle. 1992. "Workers' Compensation Insurance Rates." John Burton's Workers' Compensation Monitor, Vol. 5, No. 1, pp. 1-14.

Butler, Richard J. 1983. "Wage and Injury Rate Response to Shifting Levels of Workers' Compensation." In John D. Worrall, ed., Safety and the Work Force: Incentives and Disincentives in Compensation. Ithaca, N.Y.: ILR Press, pp: 61-86:

Cain, Glen G., and Ross E. Finnie. 1990. "The Black-White Difference in Youth Employment: Evidence for Demand Side Factors." Journal of Labor Economics, Vol. 8, Part 2, pp. S364-S95.

Card, David. 1992a. "Using Regional Variation in Wages to Measure the Effects of the Federal Minimum Wage." Industrial and Labor Relations Review, Vol. 46, No. 1 (October), pp. 22-37.

-----. 1992b. "Do Minimum Wages Reduce Employment? A Case Study of California, 1987-1989." Industrial and Labor Relations Review, Vol. 46, No. 1 (October), pp. 38-54.

Card, David, Lawrence Katz, and Alan 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, Vol. 47, No. 3 (April), pp. 487-496.

Card, David, and Alan Krueger. 1993. "Minimum Wages and Employment: A Case Study of the Fast Food Industry in New Jersey." American Economic Review, Vol. 84, No. 4, pp. 772-93.

-----. 1995. Myth and Measurement. Princeton: Princeton University Press.

Card, David, and Phillip Levine. 1994. "Unemployment Insurance Taxes and the Cyclical and Seasonal Properties of Unemployment." Journal of Public Economics, Vol. 53, No. 1, pp. 1-29.

Clark, Kim, and Lawrence Summers. 1982. "The Dynamics of Youth Unemployment." In Freeman, Richard, and David Wise, eds., The Youth Labor Market Problem: Its Nature, Causes, and Consequences. Chicago: University of Chicago Press.

D'Amico, Ronald, and Nan Maxwell. 1994. "The Impact of Post-School Joblessness on Male Black-White Wage Differentials." Industrial Relations, Vol. 33, No. 2, pp. 184-205.

Dorsey, Stuart, and Norman Walzer. 1983. "Workers' Compensation, Job Hazards, and Wages." Industrial and Labor Relations Review, Vol. 36, No. 4 (July), pp. 642-54.

Ellwood, David. 1982. "Teenage Unemployment: Permanent Scars or Temporary Blemishes." In Freeman, Richard, and David Wise, eds., The Youth Labor Market Problem: Its Nature, Causes, and Consequences. Chicago: University of Chicago Press.

Freeman, Richard, and Harry Holzer, eds. 1986. The Black Youth Employment Crisis. Chicago: University of Chicago Press.

Freeman, Richard, and David Wise, eds. 1982. The Youth Labor Market Problem: Its Nature, Causes, and Consequences. Chicago: University of Chicago Press.

Gruber, Jonathan. 1994. "The Incidence of Mandated Maternity Benefits." American Economic Review, Vol. 84, No. 3, pp. 622-41.

Gruber, Jonathan, and Alan B. Krueger. 1991. "The Incidence of Mandated Employer-Provided Insurance: Lessons from Workers' Compensation Insurance," In David Bradford, ed., Tax Policy and the Economy. Cambridge: MIT Press, pp. 111-44.

Hamermesh, Daniel. 1979. "New Estimates of the Incidence of the Payroll Tax." Southern Economic Journal, Vol. 45, No. 4, pp. 1208-19.

-----. 1986. "The Demand for Labor in the Long Run." In O. Ashenfeller and R. Layard, eds., Handbook of Labor Economics. Amsterdam: Elsevier, pp. 429-71.

Ihlanfeldt, Keith R., and David L. Sjoquist. 1990. "Job Accessibility and Racial Differences in Youth Employment Rates." American Economic Review, Vol. 80, No. 1, pp. 267-76.

Katz, Lawrence, and Alan Krueger. 1992. "The Effect of the Minimum Wage on the Fast Food Industry." Industrial and Labor Relations Review, Vol. 46, No. 1 (October), pp. 6-21.

Lynch, Lisa M. 1989. "The Youth Labor Market in the Eighties: Determinants of Re-Employment Probabilities for Young Men and Women." Review of Economics and Statistics, Vol. 71, No. 1 (February), pp. 37-45.

Moore, Michael J., and W. Kip Viscusi. 1989. "Promoting Safety Through Workers' Compensation: The Efficacy and Net Wage Costs of Injury Insurance." Rand Journal of Economics, Vol. 20, No. 4, pp. 499-515.

-----. 1990. Compensation Mechanisms for Job Risks: Wages, Workers' Compensation, and Product Liability. Princeton: Princeton University Press.

Neumark, David, and William Wascher. 1992. "Employment Effects of Minimum and Subminimum Wages: Panel Data on State Minimum Wage Laws." Industrial and Labor Relations Review, Vol. 46, No. 1 (October), pp. 55-81.

-----. 1993. "Evidence on Minimum Wage Effects and Low-Wage Labor Markets from Disequilibrium Models." Unpublished paper, University of Pennsylvania.

-----. 1994. "Employment Effects of Minimum and Subminimum Wages: Reply to Card, Katz, and Krueger." Industrial and Labor Relations Review, Vol. 47, No. 3 (April), pp. 497-512.

O'Neill, June, and David O'Neill. 1994 The Employment and Distributional Effects of Mandated Benefits. Washington, D.C.: American Enterprise Institute for Public Policy Research.

Rees, Albert. 1986. "An Essay on Youth Joblessness." Journal of Economic Literature, Vol. 25, No. 2, pp. 613-28.

Ruser, John W. 1985. "Workers' Compensation Benefits and Compensating Wage Differentials." BLS Working Paper No. 153. Washington, D.C.: U.S. Department of Labor.

Summers, Lawrence H. 1989. "Some Simple Economics of Mandated Benefits." American Economic Review, Vol. 79, No. 2, pp. 177-83.

Topel, Robert. 1983. "On Layoffs and Unemployment Insurance." American Economic Review, Vol. 73, No. 4, pp. 541-59.

-----. 1984a. "Experience Rating of Unemployment Insurance and the Incidence of Unemployment." Journal of Law and Economics, Vol. 27, No. 1, pp. 61-90.

-----. 1984b. "Equilibrium Earnings, Turnover, and Unemployment." Journal of Labor Economics, Vol. 2, No. 4, pp. 500-522.

U.S. Bureau of the Census. 1994. Statistical Abstract of the United States: 1994. Washington, D.C.: GPO.

Viscusi, W. Kip, and Michael J. Moore. 1987. "Workers' Compensation: Wage Effects, Benefit Inadequacies, and the Value of Health Losses." Review of Economics and Statistics, Vol. 69, No. 2, pp. 249-61.

Zagorsky, Jay. 1993. "Is Youth Unemployment Really a Problem?" Challenge, Vol. 36, No. 5, pp. 61-62.

Robert Kaestner is Associate Professor in the School of Public Affairs of Baruch College and an NBER Research Associate. He thanks John DiNardo and Ted Joyce for their helpful comments on an earlier draft of this paper and Ewa Wojas for excellent research assistance.
COPYRIGHT 1996 Sage Publications, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1996 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Kaestner, Robert
Publication:ILR Review
Date:Oct 1, 1996
Words:9164
Previous Article:Using event history analysis to model delay in grievance arbitration.
Next Article:Matching and mobility in the market for Australian rules football coaches.
Topics:

Terms of use | Privacy policy | Copyright © 2022 Farlex, Inc. | Feedback | For webmasters |