What are the relationships between aging, depression, non-communicable diseases and disabilities in South Africa?

This is the first study that offers a comprehensive analysis of depression among the old (60+ years) in South Africa, using the four waves of the South Africa National Income Dynamics Study (SA-NIDS) during 2008-2014. A state-of-art econometric methodology has been used to unravel the factors underlying depression among the old over the period 2008-2014. Our study methodologically builds upon the (sparse) extant literature on aging and depression in the following ways. (i) Available studies often use stepwise regression with frequent changes in the significance of the explanatory variables. In contrast, we rely on a comprehensive specification. (ii) Endogeneity of explanatory variables (eg non-communicable diseases (NCDs), limitations in carrying out activities of daily living (ADLs) or disabilities, and body mass index (BMI) categories) is often overlooked and the estimation bias is ignored in the interpretation of the results. We circumvent this problem by working with initial values of morbidity, disabilities, and BMI categories. (iii) As there are interrelationships between morbidity, obesity, and disabilities, we use three alternative specifications with initial value(s) of each in one specification. (iv) Depending on whether the dependent variable is binary (self-reported depression for =3 days in a week) or continuous (as in two indices of depression), we use random effects probit with Mundlak adjustment or simply random effects with Mundlak adjustment. Among the old, those more likely to be depressed are in their sixties, Black Africans, Coloureds or women. They are more likely to be suffering from multimorbidity, multiple limitations in ADLs, to be in lower asset quartiles, or to have recently suffered a family bereavement. Factors that attenuate depression include marriage, pension, affluence, and trust in a community and familiar neighbourhoods. An important feature of our study is the robustness of the key results.