Determining the relative labour force status of Māori and non-Māori using a multinomial logit model

Using a stratified sample of 150,000-unit record data on the Māori and non-Māori working-age population from the New Zealand censuses of 1981, 1986 and 1991, this study estimates multinomial logit models in order to determine the effects of age, education, marital and parental status, local unemployment rates and type of location on the probability of being in full-time employment, in part-time employment, unemployed or
out of the labour force. This study found that differences in individual characteristics can explain a substantial proportion of the Māori–non-Māori differences in labour market outcomes (and, in 1981 and 1986, all of it for women). The beneficial effect of education is substantially larger for Māori than for non-Māori, putting an exceptionally high penalty on the majority of Māori who are without formal qualifications.