The Impact of Resource Inflows on Child Health: Evidence from Kwazulu-Natal, South Africa, 1993–98

Abstract

This paper investigates changes to the health status of young children (aged 0–5 years) in the Kwazulu-Natal province of South Africa during 1993–98. In our estimation we explicitly take into account the potential endogeneity of household resources in affecting child health. In particular, we examine whether the effect of resources is differentiated by the source, the age and the sex of the recipient. Finally, we also take into account the panel structure of the data and conduct (household level) fixed effects estimation of the determinants of child health. The estimation results show that the state of child health has experienced marked improvement following the dismantling of apartheid. Our results point to the role of household resources and health infrastructure availability in improving the health status of children.

Notes

Pushkar Maitra, Department of Economics, Monash University, Clayton Campus, VIC 3800, Australia. E-mail: [email protected]. Ranjan Ray, School of Economics, University of Tasmania, PO Box 252-85, Hobart, TAS 7001, Australia. E-mail: [email protected]. The authors would like to thank two anonymous referees for helpful reports on previous versions. The usual disclaimer applies. Funding provided by the Australian Research Council Discovery Grant Scheme (to both authors), the Faculty Research Grant Scheme, Faculty of Business and Economics, Monash University (to Maitra) and the Australian Research Council Large Grant (to Ray) are also gratefully acknowledged.

During the apartheid era, the South African government delineated four racial groups: black (or African), Indian (or Asian), coloured (or mixed race) and white (or Caucasian). Our use does not signify acceptance of this terminology or the system of racial naming.

For example, according to the World Development Report [ 1993 ], Table A6, 39 per cent of children in the age group 24–59 months in sub-Saharan Africa suffer from stunting.

During the apartheid era the white government forced black South Africans into ‘homelands’ which were desolate regions incapable of sustaining a livelihood based on agriculture thereby creating a massive pool of unemployed black workers who were employed in the mines and in the white-owned agricultural farms. The migrant workers were forced to live away from their families, remitting home cash and goods to support their families. This was known as the ‘oscillatory migrant labour system’. These ‘homeland’ regions were semi-autonomous but were dependent on funds from the South African government for infrastructure development. Naturally, the apartheid era was characterised by severe disparities between the homeland and the non homeland regions of South Africa.

The KIDS data set is the outcome of a collaborative project between researchers at the University of Natal, the University of Wisconsin–Madison and the International Food Policy Research Institute. Details of the KIDS data set have been described by its principal authors in May, Carter, Haddad and Maluccio [ 2000 ].

The referee also suggested that we further restrict the sample further to children aged between six and 36 months on the ground that ‘children older than three years generally grow at the same rate as children in developed countries’, as noted by Alderman and Sahn [ 1997 ]. Unfortunately, this suggestion could not be implemented since it sharply reduced the sample size leading to a large loss in degrees of freedom. Instead, we introduced in the regressions a child age dummy that distinguished between children in the 0–5 years range that lie between 6–36 months and the others (that is, < 6 months or > 36 months) and also interacted this dummy with all the explanatory variables. We conducted a joint test for the statistical significance of the age dummy and interaction terms and found it to be statistically insignificant for both the 1993 and the 1998 samples. What this essentially implies is that children aged less than six months or more than 36 months are not different from the rest of the sample. We, therefore, decided to base our estimation on all children in the age group 0 – 5 years.

Maluccio, Thomas and Haddad [ 2001 ] also use the same data and exploit its panel nature to study changes to child health in Kwazulu-Natal during the first five years of post-apartheid South Africa. However, unlike here, they do not investigate the determinants of that change, which is the main motivation of the present paper.

B-P test statistic distributed as χ² with 36 degrees of freedom. The associated values were 31.930 for the 1993 sample and 70.666 for the 1998 sample.

AGECAT = 1 if Age of Child < 6 months or Age of Child > 36 months, 0 otherwise.

Note that migration in this case would have to be correlated with the unobserved factors that are correlated with health in a location, such as availability of clinics, over and above other measures included in wage differentials. See Thomas and Strauss [ 1998 ].

Additional information

Notes on contributors

Pushkar Maitra

Pushkar Maitra, Department of Economics, Monash University, Clayton Campus, VIC 3800, Australia. E-mail: [email protected]. Ranjan Ray, School of Economics, University of Tasmania, PO Box 252-85, Hobart, TAS 7001, Australia. E-mail: [email protected]. The authors would like to thank two anonymous referees for helpful reports on previous versions. The usual disclaimer applies. Funding provided by the Australian Research Council Discovery Grant Scheme (to both authors), the Faculty Research Grant Scheme, Faculty of Business and Economics, Monash University (to Maitra) and the Australian Research Council Large Grant (to Ray) are also gratefully acknowledged.