4.5  Regression Analysis of Other Outcomes

We now turn to the impact of the youth minimum wage changes on the seven other outcomes of interest: hours worked conditional on employment, studying, unemployment, inactivity, receipt of benefit income, the logarithm of actual weekly earnings for wage or salary workers, and the logarithm of actual weekly income. Although the results in column (7) of table 4 provide statistical support for the inclusion of age-specific time trends, we are not confident that this is the correct specification. Both because of this and the sensitivity of the results for employment to this issue, in the analysis of the impact on other outcomes we present two sets of results that, respectively, exclude and include such trends. First, table 5 presents the results for employment from column (6) of table 4 in column (1), and the results from analogous specifications that exclude age-specific time trends for the other outcomes in columns (2)-(8).[29]

In column (2), the results for hours worked show that 16-17 year-olds increased their hours worked by a large and statistically significant 2.2 hours per week (or 10-15 percent) in 2002 and 2003, but there was little impact of the minimum wage changes for 18-19 or 20-21 year-olds. The results in column (3) show statistically significant falls in the fraction of 16-17 year-olds studying of about 3-4 percent in each year after the minimum wage increases. We also estimate, generally smaller, drops in study rates for the 18-19 year-olds of 1-2 percent: these estimates are statistically significant in 2001 and 2002. The results in column (4) imply the unemployment rate for 16-17 year-olds increased a statistically significant 2 percent in 2001 and was marginally and insignificantly higher in the latter years, while there’s no systematic evidence of any effect on 18-19 year-olds’ unemployment. Given the fluidity between unemployed and not-in-the-labour-force states for many youth, inactivity (defined as neither working nor studying) may provide a more reliable outcome measure than unemployment for these age groups. In column (5), we estimate there was a significant increase in the fraction of 16-17 year-olds who are inactive after 2001, around 2 percent in each of the three years. Again, there is no systematic evidence of effects on the inactivity of the two older age groups. The results in column (6) for the receipt of benefits show increased rates for 16-17 year-olds, which are significant in the latter two years. For 16-17 year-olds, the results for these four outcomes and employment collectively suggest that, following the increases in the minimum wage, there was a drop in study enrolment which was accompanied by increasing inactivity and benefit receipt with little change in employment, although there was a large increase in hours worked by those employed.

In columns (7) and (8) of table 5 we present the regression results for changes in log weekly earnings and log weekly income. Given that the minimum wage increases increased the average wages of teenage workers, and without adverse employment effects, it is not surprising that we find an increase in earnings for these groups. The earnings results for 16-17 year-olds show strong increases in earnings after 2001 from about 15 percent in 2001, to 22 and 29 percent in 2002 and 2003, while for 18-19 year-olds the increases were more modest due to the more muted average wage increase, rising to about 14 percent by 2003. As earnings are only available for those who are employed, we also examine the impact on individuals’ total weekly income, which is available for the non-employed and arguably provides a better measure of the welfare effects of the minimum wage change. In comparison to the earnings results, the results in column (6) show smaller, although still significant for 16-17 year-olds, increases in teenagers’ income of 7-16 percent for 16-17 year-olds and 0-8 percent for 18-19 year-olds.

In table 6 we report the results from specifications for each of the eight outcomes that include age-specific linear trends in addition to the earlier controls. The F-statistic for the joint hypothesis that the time trend coefficients are zero is significant at the 5% level or better for all outcomes except hours worked, suggesting secular trends in these outcomes is potentially important. Including these trends also has a dramatic effect on some of the estimated impacts. In particular, in addition to the stronger positive impacts on 18-19 and 20-21 year-olds’ employment seen in table 4, the significant adverse impacts on 16-17 year-olds study and unemployment seen in table 5 no longer hold. Also, the estimated impacts on 16-17 year-olds’ benefit receipt, earnings and income are larger and more significant than previously, and the positive impacts on 18-19 year-olds income disappear and become significantly negative in 2003.