Abstract
Ph.D. (Economics)
This thesis comprises three empirical essays analysing the determinants and impact of remittances in South Africa. Except for the first and the last chapters, which set out the general introduction and conclusion, each of the three essays is self-contained and can be considered as a standalone piece of work. Two datasets are used to analyse the determinants and impact of remittances in South Africa. They are the National Income Dynamics Survey (NIDS), and unique matched data from a ‘customised household survey’ collected by the author in Hlokozi village in Southern KwaZulu-Natal and in more urban localities where connected migrants were working.
The first essay (Chapter 3) investigates the factors that can be shown to influence the likelihood and scale of internal remittances for migrant and non-migrant households in South Africa – it being understood that financial transfers may be received by households from individuals that households do not regard as members. It finds that these determinants consist of certain features of the receiving households, but also of some characteristics of the migrants who remit. More specifically, empirical investigation using approaches deriving from the double-hurdle regression model, the Heckman selection model and the control function model reveals that the probability and level of remittances are largely determined by common factors. Examples include certain characteristics of the household head (his or her age, race, education level and whether employed or not); what the income of the household is and in what type of area it is located; whether the migrant is male or female; how much the migrant earns; and how the money is sent back (self-delivery or other channel). However, certain determinants are not shared in the full sense: the probability of remitting appears to vary in the same direction as the size of the household and the household wealth, but the size of remittances seems to vary in the opposite direction to that in which these two determinants vary.
The second essay (Chapter 4) investigates whether remittances are causally linked to the reduction of poverty in South Africa. A fixed-effects vector decomposition estimator (FEVD), which allows for the estimation of the coefficient of the time-invariant variables and accounts for unobserved heterogeneity, is employed to estimate the poverty-reducing effects of remittances. Given the strong potential for endogeneity of the remittance variable, especially reverse causation (an increase in poverty might be associated with an increase in migration and...