8  Summary

The Christchurch Health and Development Study (CHDS) is a panel data set of over 1,200 children born in 1977 in Canterbury area hospitals. These longitudinal data are used in this report to isolate any ethnic differences in the early labour market experiences of youth by the time of their 21^st birthdays in 1998. We test the specific hypotheses that Maori accumulate less work experience and face lower wages than non-Maori, once other relevant factors are held constant. The following results are obtained:

• Maori youth acquire fewer school and post-school qualifications, and have more years of potential work experience than non-Maori in the CHDS. Despite this fact, Maori accumulate more than one-quarter of a year less in actual work experience than non-Maori by age 21. Most of this difference in accumulated work experience occurs between mixed Maori and non-Maori. Yet, the ratios of potential to actual work experience are similar between sole and mixed Maori, and significantly lower than that of non-Maori.
• Substantial ethnic differences in personal and family background characteristics are found. Maori parents are relatively less likely to have post-school qualifications. Maori youth are relatively more likely to be raised in single-parented households and in families receiving social welfare benefits. Average real family incomes are substantially lower for Maori compared to non-Maori. Before the age of 16, Maori are more likely to have behavioural problems and inconsistent attachments to school. After the age of 18, Maori are more likely to be convicted of a criminal offence, diagnosed with an alcohol or drug problem or held responsible for the birth to a child. These mean differences in personal and family background experiences are generally larger between sole Maori and non-Maori than between mixed Maori and non-Maori.
• The distributions of the total time spent in education, training and work between the ages of 16 and 21 indicate that there is more heterogeneity among Maori than non-Maori. A substantially larger proportion of Maori youth accumulate very little time in these productive activities compared to non-Maori. These ethnic differences narrow considerably at the upper ends of these distributions. These data suggest that average differences in labour market outcomes between ethnic groups may be concentrated among a small, particularly disadvantaged group of both sole and mixed Maori. Ethnicity may play a role in the outcomes for the most disadvantaged in society. For example, among the youth who discontinue their education and training early in life, Maori acquire about one-half of the work experience of non-Maori by age 21.
• Regression analysis indicates that only about one-fifth of the observed gap in work experience between Maori and non-Maori can be explained by differences in gender, formal qualifications and number of children (ie, variables typically available from cross-sectional data sources). Once personal and family background characteristics are included in these regressions (ie, variables almost unique to the CHDS), nearly one-half of the observed gap in work experience between Maori and non-Maori can be explained. Even though measured backgrounds are important factors behind the relatively lower accumulation of work experience by Maori, they do not explain the entire ethnic gap in this labour market outcome.
• Adding ethnicity to the other explanatory variables does very little to enhance the predictive performance of our regression models on work experience. The estimated coefficients on the indicator variable for Maori in these regressions are negative, but insignificant at conventional test levels when family background factors are included.
• One of the more surprising results in this study is that ethnic differences in early work histories seem to be largely relegated to the relationship between mixed Maori and non-Maori. Although sole Maori have less work experience than non-Maori by age 21, these differences are not statistically significant either outside or inside our regression analysis. Mixed Maori, on the other hand, have significantly less work experience than non-Maori by age 21, and these differences remain significant even after we have controlled for other factors in our regression models. This general conclusion is robust to changes in the definition of mixed Maori. The previous report in this project used a more liberal definition of Maori that was based entirely on the self-defined ethnicities at age 21. All of the ethnic differences in the accumulation of work experience in this earlier analysis were relegated to the relationship between this larger group of mixed Maori and non-Maori.
• Ethnic differences in work experience are even larger among youth in the CHDS without formal school or post-school qualifications. For example, Maori have slightly more than one-quarter of a year less in work experience by age 21 compared to non-Maori. The equivalent gap between unqualified Maori and non-Maori is about two-thirds of a year. Regression analysis can account for only a fraction of the differences in work experience between the ethnic groups.
• Quantile regressions are used to assess whether ethnic differences in the accumulation of work experience can be attributed to specific sections of the experience distribution. We find that the observed gaps between Maori and non-Maori in the lower tails of these distributions remain when we control for only the gender, qualifications and children born to these youth. However, when we control for observed family backgrounds, these ethnic differences in the lower tails largely disappear. Our family backgrounds measures may be particularly important for explaining the relatively poor performance of Maori among the disadvantaged in society.
• At age 21, Maori workers receive higher hourly earnings than non-Maori workers. This is despite the fact that Maori in this sample have fewer qualifications and less work experience than non-Maori. Although this ethnic difference in wage rates is not significant overall, it is generally significant once we control for these other factors in our regression analysis. Once observable factors have been held constant, we find that Maori workers receive hourly earnings that are around 9% higher than observationally equivalent non-Maori.
• The possibility exists that Maori workers receive relatively higher hourly earnings because of sample selection effects (ie, only the most productive Maori are able to find employment). To test this proposition, we use our regression results to predict the hourly earnings that face all youth in our sample regardless of their current work status. Our results suggest that Maori, on average, face wages that are around 4.5% higher than those faced by non-Maori. It should be noted, however, that this analysis does not consider the possibility that this sample selection process on work status is related to unobservable factors that systematically reduce the wages faced by non-employed Maori relative to non-Maori.
• Attempts are also made to ’endogenise’ both educational attainment in the work experience regressions, and educational attainment and work experience in the hourly earnings regressions. The number, birth order and sex composition of siblings are used as instrumental variables. These results are disappointing. Firstly, there is loss in explanatory power in our regressions when educational attainment is measured as actual years in education and training. Secondly, the instrumental variables capture only a small fraction of the variation in educational attainment. Thirdly, the net result is an increase in the standard errors on these endogenous explanatory variables, and no appreciable change in the results on the ethnic differences in the accumulation of work experience and hourly earnings.