An econometric analysis of the socio-economic determinants of household earnings in the North West Province of South Africa

Abstract

The majority of households in the North West province of South Africa reside in rural areas and are living in absolute poverty. Knowledge of the determinants of household earnings in the province may info rm policies aimed at poverty alleviation (such as expenditure reprioritisation by government) and may assist households' decision making (such as that pertaining to decisions to improve standards and levels of education). Given the current lack of understanding into the determinants of household earnings and earnings inequality in South Africa in general and in North West province specifically, this study set out to identify the determinants of household earnings in the North West province of South Africa. In order to identify these determinants, a structured questionnaire was developed. A multi-stage stratified cluster sampling design was used to draw a representative sample of 593 households from North West province. These households were interviewed by a team of enumerators between June and October 1997. The data collected in this manner were supplemented by relevant data pertaining to North West province. These were extracted firstly from the compute r data files of the 1993 household survey conducted by the South African Labour and Development Research Unit (SALDR and secondly from the October 1995 Central Statistical Services (CSS) survey. Three data sets, for 1993, 1995 and 1997 were thus available for econometric analyses. Using these three data sets, a standard Mincerian earnings function was firstly estimated to identify the determinants of household earnings. Based on a survey of the theoretical literature, this function was estimated in extended form. Since censored data were used, Heck.man's (1976b) two-stage estimation

technique and a double - hurdle model (in which binomial probit models were fitted) were used to obtain unbiased and consistent regression estimates of the coefficients. Secondly, the wage gap decomposition method proposed by Oaxaca and Ransom (1994) was used to estimate the white/ African and male/ female labour market discrimination coefficients. The regression results of the extended Mincerian function established that human capital (as proxied by education level and work experience) is the most important determinant of household earnings in North West province. The other determinants that were found statistically significant in the regressions were the following : squared experience, the interaction of education and experience, the type of industry wherein employed, the nature of the occupation, work hours, the type of area of residence, region, place of work, dependency ratio, a wealth index, marital status, family type, gender, race, employer and former province. Union membership was found to contribute to the segmentation of the labour market m North West province. The regression results from the decomposition model established that labour market discrimination against Africans do not seem to be statistically significant in North West province. It was however established that there exists labour market discrimination against males. Thus, the current earnings inequalities that exist in the province were found to be basically due to (a) significant productivity differentials between whites and other population groups, and (b) labour market discrimination against males. Labour market discrimination against males may be a consequence of the affirmative action currently being practised in North West province.

It was also established that there exists a significant productivity differential in favour of males. The "productivity differential" established here can be regarded as pre-labour market discrimination (or human capital discrimination) against Africans and females respectively. It was concluded that efforts by the government of North West province to improve the quality of investment in human capital should be encouraged. Affirmative action without policies to increase the productivity of Africans (especially females) may have only a limited and short-term impact on earnings inequality and poverty alleviation in North West province. Improvements in the socio-economic in frastructure of rural areas, especially those which belonged to the former Bophuthatswana homeland can also be justified on the basis of the econometric results obtained.