This paper analyses household‐level determinants of the probability and level of domestic remittances
in South Africa over the period 2008 to 2014‐2015. We exploit all four waves of the National Income
Dynamics Survey (NIDS) data to analyse the determinants of remittances in a panel setting using
random‐effects Tobit, Heckman selection, and two‐part model approaches. The panel nature of
this data allows us to incorporate individuals’ unobserved time‐constant characteristics (or
unobserved heterogeneity) in the models, a step that enriches the analysis and yields more accurate
resultsthan if we were to use only cross‐sectional analysis. It also allows usto incorporate information
about the dynamics of remittance behaviour for the same households. However, data availability
restricts the analysis to determinants associated with the recipient households. We find the
determinants of the probability of remitting to be non‐identical to the determinants of the level of
remittances. Determinants of both include the age, race, education level, and employment status of
the household head, and the income and the type of area of the household. The gender of the
household head and the size of the household are also important determinants, but appear to have a
positive effect on the probability of remitting, yet a negative effect on the amount remitted. These
results shed light on the factors that affect whether or not families receive remittances and, if they
do, how much.