1  Introduction

This report presents an empirical analysis of the early labour market histories of youth in the Christchurch Health and Development Study (CHDS). We want to know the extent to which Maori might be disadvantaged relative non-Maori over their first few years in the labour market. Are ‘observationally equivalent’ individuals treated differently in the workforce because of their Maori ethnicity? This is an important issue for policy makers, because systematic disadvantages faced by Maori during this critical transition period into the labour market could persist (or be accentuated) over their remaining working lives.

In particular, we estimate the extent to which Maori accumulate less work experience than their non-Maori counterparts by age 21. What proportion of any gap in work experience between the ethnic groups is eliminated by controlling for differences in the total time spent outside of education and training, the attainment of formal qualifications and other measurable personal and family background characteristics? This should be an important contribution of this study. Other conventional data sets that can be used for analysing the importance of ethnicity for labour market outcomes in New Zealand (eg, the Population Census, the Household Labour Force Survey (HLFS) and the Household Economic Survey (HES)), contain no information on the accumulated work experience of individuals.

We also estimate the extent to which Maori face lower wage rates in the labour market than non-Maori at age 21. What proportion of the overall gap in hourly earnings between the ethnic groups can be eliminated by controlling for observed differences in work experience, formal qualifications and other relevant factors? Again, individual data on actual work experience is critical if we want to hold constant the relevant productivity characteristics of workers.[1]

This study uses unit record data from the CHDS on the family backgrounds, educational outcomes and labour market histories of young adults up to their 21^st birthdays. The CHDS is a panel data set, which follows the subsequent progress of over 1,200 children born in Canterbury area hospitals between April and August 1977. Parents or custodial adults in the households in which these children resided were interviewed at birth and every year until age 14 of these children. Both parents and the CHDS children were interviewed when these children had reached their 16^th birthdays. The most recent interviews focused entirely on the CHDS youth when they had reached the age of 18 in 1995 and 21 in 1998.

There are several unique advantages and disadvantages in using the CHDS data for examining ethnic differences in labour market outcomes. Two attributes of this panel, in particular, should add value to the existing literature in this area:

• These data concentrate on the early labour market experiences of a cohort of individuals who had reached their 21^st birthdays. Most empirical studies on Maori-non-Maori disparities have relied on individual data across the entire working-age population. There are good reasons to focus our attention specifically on earlier labour market experiences. Any ‘scarring effects’ associated with the disadvantageous treatment of Maori youth could persist over the four or more decades remaining in their working lives. This is also the age range when government interventions are more prevalent (eg, education, training and job search), and the time when these programmes might be most efficacious and cost effective.
• The CHDS provides a wide range of information on the work, education and training histories of youth, as well as their personal and family backgrounds. This information is critical for accurate assessments of the exact nature of these labour market outcomes (eg, more-or-less continuous information on work and education histories), and the particular opportunities and constraints facing individuals that influence these outcomes. One of the strengths of the CHDS is the ‘quality’ of the data available on both the dependent and independent variables that will be used in this analysis.

The primary weaknesses of the CHDS are both the ‘quantity’ and ‘geographic diversity’ of the data available for this empirical analysis. Unlike the Census and HLFS data used in earlier studies with many thousands of individual observations taken from many cohorts across the entire country, we are constrained by the initial selection of families to participate in this panel (ie, a single cohort of children and their families taken from a specific geographic location). Due to attrition in this panel, slightly less than 1,000 respondents were interviewed in 1998. Small sample size makes it difficult to generate statistically significant effects. In addition, there is a lack of diversity in terms of time and geographic location. Our data come from a single cohort born within a five-month period in 1977. Although some of these individuals will have migrated out of the Canterbury region, most will have remained in this particular geographic area. This is particularly problematic in terms of this present study, because of the much lower representation of Maori in the South Island.

It is difficult to know a priori, whether the benefits of data quality will exceed the costs associated with a small, nationally unrepresentative sample. In the end, the key is that the CHDS data are relatively unique, and should at least provide an alternative perspective on the more traditional data sources for this type of analysis.

The remainder of this report is organised in the following way. Section 2 summarises and critiques the existing empirical literature on ethnic differences in labour market outcomes in New Zealand. Section 3 develops an analytical framework for estimating the determinants of work experience and hourly earnings of youth in the CHDS. Section 4 presents some descriptive statistics from this panel data set and discusses the key issue of defining ‘who is Maori’ for this analysis. Sections 5 and 6 present and discuss the implications of the main regression results. Section 7 reconsiders this regression analysis in light of the endogeneity of both educational attainment and accumulated work experience. Finally, Section 8 summarises our findings, and draws some broad conclusions about this analysis.