3 The specification
The following specification was estimated:
(1) ln Yik= ak + Σbjk X ijk + Σcjk Z ijk + d1k Nik + d2 k Nik2 + vik
where the dependent variable is the natural logarithm of annual income in current dollars, k stands for each of the four ethnic groups. X ijk represent the six binary educational qualifications variables for individual i, where the excluded educational qualification level is ‘no school qualifications’. The explanatory variables, Z ijk, control for occupation[5] and industry[6] (18 one digit variables), weekly hours of work, marital status (‘married’ or ‘de facto’ versus other categories), and locality of residence (‘major urban’ and ‘rural’ versus ‘semi-urban’). Variable N measures the potential years of work experience by educational qualification level, in the usual quadratic form.[7]
Equation (1) is estimated for both 1986 and 1996, and all models estimated utilise the White adjustment to correct for heteroscedasticity and for consistent estimates of coefficient variances. The model incorporates before-tax income levels.[8] The model is unrestricted by ethnicity, gender, and for each year.
While the expected sign for ‘hours of work’ is positive, the effects of ‘locality of residence’ and ‘marital status’ are a-priori not entirely clear, and examining their effect is of interest, in particular across ethnic groups. In addition, ‘urban’ living is expected to have a positive coefficient if ‘urban’ job and employment opportunities are greater than when living in ‘rural’ or ‘semi-urban’ areas, but the extent of it may very well vary for the Maori population. Finally, there is wide empirical evidence of a positive relationship for males between being married and income, which is likely to represent mainly supply side effects.
Notes
- [5]Nine one digit ‘occupation’ and nine ‘industry’ categories were controlled for. The 1996 and 1986 census occupation classifications were somewhat different and therefore different specifications for the two years are chosen. The occupation categories in 1986 were: (1) Managerial/administrative, (2) Professional, (3) Clerical, (4) Service, (5) Agricultural, (6) Production/transport workers, and (7) Sales. In 1996 the categories were: (1) Managerial/administrative, (2) Professional, (3) Clerical, (4) Service, (5) Agricultural, (6) Trade oriented, (7) Plant and machine operator, (8) Elementary/low-skilled, and (9) Technical. Importantly, the base occupation category, which is ‘Clerical’ is used across the two sample years in 1986 and 1996.
- [6]The industry categories in both years were: (1) Agriculture, hunting, forestry, fishing, (2) Mining and quarrying, (3) Manufacturing, (4) Electricity, gas, and water, (5) Construction, (6) Wholesale and retail trade, restaurants and hotels, (7) Transport, communication, (8) Business and financial services, (9) Community, social and personal services.
- [7]Years of experience is specified as ‘age - years of schooling - 5’ since school starts at age 5 in New Zealand.
- [8]The income information in the New Zealand census (and similar to Australia) is reported in 13 categories, based on an annual gross income. The mid-point of these categories has been used as a measure of income throughout the study. The lowest income category in the census is nil income or loss for which income of zero is designated. The rest of the annual categories were $2,500 or less, $2,501-$5,000, up to $100,000 or more, for which, based on a Statistics New Zealand estimate, a mid-point of $130,960 was assumed.
