4  Analysis of the Impact of Youth Minimum Wage Changes

Although attention generally focuses on the employment impact of minimum wage changes, several other outcomes are of interest, particularly in the case of youth minimum wages. A sizeable change in the minimum wage for young workers, by making work relatively more attractive, may affect the labour-supply decisions of non-participants. For example, for youth still in school, it may affect their decision of when to leave school and enter the labour market. Our analysis of the impact of the youth minimum wage reforms focuses not only on employment outcomes, but also on a variety of related outcomes.

The approach adopted to analyse the impact of changes in the minimum wage for youth workers compares the average outcomes of the age groups affected by the changes, namely 16-17 year-olds and 18-19 year-olds, to those of young adults (aged 20-25) before and after the policy reform in March 2001. Two crucial identifying assumptions underlie this approach. First, it assumes that young adults are not affected by changes to the youth minimum wage. This requires that there are no spillover effects of the changes in the youth minimum wage on the outcomes of those workers not directly affected by the change. This assumption would not be satisfied if, for example, employers bound by the higher minimum wage on youth workers respond by employing more experienced and productive older workers in preference to youth workers. Such a response would lead to the direct effect of the minimum wage increase on youth employment being overstated: the measured impact would consist of the direct effect on youth workers plus the indirect effect on young adult workers. Alternatively, in order to maintain a relative wage difference, a minimum wage increase for youth workers may cause wages for young adult workers to increase, and result in negative employment spillovers for these workers.

The second assumption is that any secular trends in the labour market outcomes for 16-25 year-olds are common across these ages. This assumption requires, for instance, that there are no differential age-specific trends. For example, our estimates could be confounded if demand for teenagers depends more on overall economic conditions than demand for young adults. Underlying trends in outcomes that differ by age group would also violate this assumption.

We begin our analysis with a discussion of the wage distributions for 16-17, 18-19 and 20-25 year-old workers, before and after the policy reforms. To the extent the reforms were binding on low wages, we would expect to see a fall in the incidence of low wages and an increase in the density at or above the new minimum wages. This analysis describes how the youth minimum wage changes affected the teenage wage distributions, together with a sense of the size of the group of teenage workers affected by the minimum wage changes. In addition, if the only changes we observe are in the sub-minimum ranges for teenage workers, this would provide some confidence in our identification strategy:

Following the discussion of changes in the wage distributions, we describe the trends in the outcomes of interest for the three age groups over the sample period. However, it is difficult to capture the complex relationship between age, year, and minimum wage effects on labour market outcomes in a graphical manner alone. We next formalise the analysis to quantify the impact of the youth minimum wage policy changes on these outcomes. We begin with simple difference-in-differences estimates of the effects of the changes in youth minimum wages on teenage employment and hours worked, that compares teenage outcomes with young adult outcomes, before and after the reforms. In order to control for possible confounding factors, we then extend this approach using regression analyses of the effects of the changes in youth minimum wages on these, as well as the other outcomes (study, unemployment, inactivity and benefit-receipt rates, and earnings and incomes).

4.1  The Wage Distributions

To the extent that the minimum wage is binding on youth workers, we would expect to see a reduction in density below and an increase in density at or above, the (new) minimum wage in the wage distributions for 16-17 and 18-19 year-olds following the 2001 youth minimum wage reforms. Figure 2 presents kernel density estimates of the distributions of log real actual hourly wages at each worker’s main job, separately for 16-17, 18-19, and 20-25 year-olds for the four years before (1997-2000), and the three years after (2001-2003), the minimum wage changes.[14] The solid and dashed lines in each figure describe the pre- and post-reform wage distributions respectively, while the two vertical lines indicate the 2000 and the 2002 minimum wage rates pertaining to each age group.

The top-left graph presents the kernel densities for the 16-17 year-old workers. Comparing the pre- and post-reform distributions, there is a significant reduction in the density at the lower end of the wage distribution following the minimum wage reform, and a large increase in the mass at wages just above the 2002 minimum wage is also apparent. Although not apparent in this figure, the initial (2001) increase resulted in a fall in wages both in the range between the 2000 and 2001, and between the 2001 and 2002, minimum wages.[15] The second (2002) increase resulted in a further reduction in the mass around the 2001 minimum wage level.

The top-right graph in figure 2 presents the kernel densities for 18-19 year-olds. In comparison to the densities for 16-17 year-olds described above, there are two points to note. First, the affected group of 18-19 year-old workers is relatively smaller than that of 16-17 year-olds. That is, the fraction of the 18-19 year-olds pre-reform distribution below the 2002 minimum wage is less than the corresponding fraction for 16-17 year-olds. Second, the reduction in density in the affected range for 18-19 year-olds is also less than for 16-17 year-olds. Nonetheless, there does appear to be a small reduction in mass below the 2002 minimum wage in the post-reform distribution of 18-19 year-olds.[16]

The bottom graph in figure 2 presents the kernel densities for 20-25 year-olds’ wages before and after the 2001 reforms. These graphs suggest that the wage distribution for 20-25 year-olds was quite stable over the period around the youth minimum wage reforms, which is reassuring for our analysis, although it seems there was a small drop in mass in the left tail of the distribution. This provides some confidence that the changes in the lower-end of the wage distribution for 16-17 and 18-19 year-olds were driven by the youth minimum wage reform, and that 20-25 year-olds’ wages were largely unaffected. This reform appears to have shifted the lower-end of the wage distribution to the right for both 16-17 and 18-19 year-olds relative to older adults. Also, the shift in the wage distribution is larger for 16-17 year-olds than for 18-19 year-olds,[17] partly because the minimum wage reform was less binding on 18-19 year-olds’ wages.

We next summarise relevant aspects of the year-specific wage distributions for the three age groups of workers over the sample period. Figure 3 describes the annual trends in the hours-weighted fractions of workers reporting wages below the current minimum wage, exactly equal to the current minimum, below the next year’s minimum, and the average wages. In each graph, we use dashed, solid and dotted lines to represent the 16-17, 18-19 and 20-25 year-olds’ respectively, and mark the 2001, 2002, and 2003 minimum wage change dates with vertical lines.

First, the top-left graph shows the fraction of wage or salary workers whose current main-job wage is below the current minimum wage. Assuming that no workers are excluded from the minimum wage coverage, and that there is accurate reporting of wages, this statistic measures the degree of non-compliance with the statutory minimum wage.[18] The fraction of workers in each age group reporting sub-minimum wages lay between 1 and 8 percent in the pre-reform years, was typically highest for 16-17 year-olds and lowest for 18-19 year-olds, and tended to decline over this period.[19] Following the minimum wage reform, the fractions of teenage workers paid below the minimum wage increased substantially, while the fraction of young adults was largely unaffected. In particular, the fraction of 18-19 year-olds affected increased from 1 percent in 2000 to 12 percent in 2001 and 2002, and 13 percent in 2003; while the fraction of 16-17 year-olds increased from 3 percent in 2000 to 6 percent in 2001, 9 percent in 2002, and 8 percent in 2003. Assuming that both the incidence of exemptions and the structure of measurement error in reported wages was reasonably stable over this period, these increases imply a significant increase in non-compliance with the statutory minimum wage.[20]

Second, the top-right graph in figure 3 shows the fraction of workers who are paid exactly the minimum wage in each year. This provides another measure of the extent to which the minimum wage binds. Prior to the youth minimum wage reform, a small (less than 1 percent) and declining fraction of workers in each age group reported earning exactly the minimum wage, further suggesting that these minimum wages were essentially non-binding. After the reform, in 2001, 1 percent of 16-17 year-old workers and 2 percent of 18-19 year-old workers reported exactly the minimum wage, and these fractions increased to 6 and 7 percent, and 2 and 4 percent, respectively in 2002 and 2003.[21]

Third, the bottom-left graph in figure 3 shows the fraction of workers whose current wage is below the next year’s (nominal) minimum wage. This provides a measure of the degree to which current workers may be affected by next year’s minimum wage.[22] Not surprisingly, given the constant nominal minimum wage and low inflation between 1997 and 1999, the fraction of workers paid below next year’s minimum wage is only marginally different from the fraction paid below the current minimum. However, there is a noticeable difference between the comparable fractions below the current and next-year minimum wages for 16-17 and 18-19 year-olds following the youth minimum wage changes. In 2000, 2001 and 2002 18, 16 and 25 percent of 18-19 year-old workers reported wages below the next year’s minimum, compared to the 12, 12 and 13 percent of workers reporting sub-minimum wages in 2001, 2002 and 2003. Ignoring measurement error issues, and in the absence of possible employment effects, this implies the compliance rate for the affected 18-19 year-olds of 25-50 percent. Similarly, that 12, 15 and 21 percent of 16-17 year-old workers report wages in 2000, 2001 and 2002 below the following year’s minimum, compared to the 6, 9 and 8 percent of workers who are report sub-minimum wages in 2001, 2002 and 2003, suggests a compliance rate for affected 16-17 year-olds of 40-60 percent. Although not part of our subsequent analysis, we have also presented the fractions of 2003 workers with wages below the 2004 minimum wage: these are 22, 23 and 7 percent of 16-17, 18-19, and 20-25 year-olds respectively.

Fourth, the bottom-right graph in figure 3 shows the average real wage at each worker’s main job in each age-group and year. After falling between 1997 and 1999, the average real wages of 16-17 year-olds workers increased steadily from $7.60 in 2000 to $8.40 per hour in 2003. In contrast, and somewhat surprisingly given the shift in the wage distribution, the average real wages for 18-19 year-olds actually fell between 2000 and 2002 from $9.38 to $9.11 per hour,[23] before increasing dramatically to $10.24 per hour in 2003. As expected the average real wage of 20-25 year-old workers was reasonably flat over the 1997-2003 period at around $12.50 per hour, and showed no increase between 2000 and 2003.