Self-Selection in Migration Between Developing Countries: The Role of Unobservable Skills of Gauteng’s Migrant Population Nwabisa Nontenja* and Umakrishnan Kollamparambil* ABSTRACT The present study investigates the presence of self-selection among internal and international migrants in Gauteng, South Africa and further disentangles the effects of observed and unob- served characteristics in the self-selection of migrants by conducting Oaxaca-Blinder decompo- sition on overall employment and self-employment outcome variables. Our analysis, based on system-GMM analysis of pseudo-panel data, indicates that international migrants experienced markedly higher levels of employment than both locals and internal migrants driven by higher rates of informal and self-employment. International migrants were also found to outperform their South African-born counterparts on various variables indicative of well-being like total household income, per capita household income, higher life satisfaction, and lower debt levels. Even though international migrants did not have higher levels of education, the Oaxaca-Blin- der decomposition provided evidence of the positive selection of international migrants to Gauteng, based on unobservable characteristics, to achieve higher levels of employment and self-employment. INTRODUCTION Literature on migration has pointed to the existence of self-selection bias, as individuals who choose to migrate may be systematically different from those who choose not to migrate. (Borjas, 1987). Migration represents an adjustment in an individual’s portfolio of human capital invest- ments, along with activities such as education, training and healthcare, to maximise their overall lifetime returns (Gabriel and Schmitz, 1995). Furthermore, Gabriel and Schmitz (1995: 461) state that “life cycle factors such as age, marital status, presence of children, education and acquired labour market skills will affect how individuals evaluate differences between destination and the country of origin”. It then follows that for a population that is heterogeneous in factors such as ability, motivation, education, and life cycle characteristics, the returns associated with migrating across different locations are likely to be different (Dostie and L eger, 2009). Those who make the decision to migrate will therefore be a self-selected subset of the population. Literature has also sought to disentangle the effects of observable and unobservable characteristics in the decision to migrate. Even in the case of migrants and non-migrants who appear similar in observable character- istics such as age or education, there are potential differences in latent attributes that affect the migration decision (Nakosteen et al., 2008). * School of Economic and Business Sciences, University of Witwatersrand, Johannesburg doi: 10.1111/imig.12454 © 2018 The Authors International Migration © 2018 IOM International Migration ISSN 0020-7985 Published by John Wiley & Sons Ltd.