5 Concluding Discussion
In this paper, we examined the impact of large increases in minimum wages for teenage workers that occurred in New Zealand in the early 2000s. We believe there are four important conclusions that can be drawn from the results of the various analyses presented here. First, given the size of the minimum wage increases and the rightward shifts in the wage distributions (especially for 16-17 year-olds), we find no consistent and robust evidence of any adverse effects of the changes on teenage employment. In fact, in contrast to simple competitive model predictions of the effect of minimum wage increases, the only statistically significant point estimates imply possible positive employment responses to the changes. The estimated 2.2-2.3 percent increase in 16-17 year-olds 2001 employment, reported in columns (4) and (5) of table 4, imply an employment elasticity with respect to the minimum wage change of about 0.25; while the estimated 3.0 and 3.6 percent increases in 18-19 year-olds 2001 and 2002 employment, reported in column (7), imply an elasticity of about 0.08. Furthermore, we find compelling evidence that 16-17 year-olds increased their hours worked by 10-15 percent following the minimum wage changes in 2002 and 2003, although the strength of the increase depends on the presence or absence of age-specific time trends.
Second, although the individual-year estimates are reasonably imprecise, the results for employment in columns (5)-(7) of table 4 imply the impacts vary over time, and the overall impacts in columns (1)-(4) tend to disguise this variation. In particular, the individual-year estimates are consistent with there being a positive impact on teenage employment following the changes in 2001, no impact in 2002, and a negative impact in 2003.
Third, the results are sensitive to the decision to include age-specific time trends in the regression. On the one hand, including time trends in the employment regression strengthens the positive employment response following the 2001 reforms for 18-19 year-olds. However, although we find evidence of adverse effects on teenage study, unemployment and inactivity rates in the absence of trends, these results are not robust to the inclusion of time trends. That these outcomes have, to our eyes, the most apparent trends (see figure 5) casts doubt on the robustness of the strong results of adverse impacts presented in table 5.
Finally, there is consistent evidence of strong increases in the earnings and income for 16-17 year-olds, although the magnitudes of the estimated increases depend on the presence or absence of time trends: our sense is that the magnitudes in the absence of time trends are more believable. There is also some, less compelling, evidence of increases for 18-19 year-olds.
One important caveat to the analysis is that there has been an increase in either real or apparent non-compliance: although the density in the affected regions of the wage distributions fell following the minimum wage reforms, there has been a significant increase in the fraction of teenage workers reporting sub-minimum wages.
What sort of models can rationalise these findings? Card and Krueger (1995) discuss the limitations of, and several simple extensions to, the standard neoclassical economic model of minimum wage effects that allow for a richer set of impacts, including the possibility that employment may rise following a minimum wage increase. The extensions allow for various market imperfections excluded from the standard model, such as search costs, incomplete information, market power, and dynamic versus static responses. Dickens, Machin and Manning (1999) develop a general model in which labour market frictions provide employers with some monopsony power. Their model provides ambiguous predictions: depending on level of the minimum wage, it may have either conventional negative impacts on employment, or neutral or positive impacts. As the pre-reform youth minimum wages were comparatively low, we suspect that these types of extensions to the standard model provide the theoretical basis for understanding our findings. However, given the recent increases, whether such benign effects continue going forward remains a moot point.
