Seminars & Events

APL (Ajiken Power Lunch)

Examining the impact of insurance on healthcare utilisation and provider choice using a natural experiment in South Africa

APL (Ajiken Power Lunch) is a lunchtime workshop open to public, including IDE staffs, visiting research fellows, IDEAS students, outside researchers and graduate students. This workshop provides a platform for presentation of any work in progress where we can discuss in either English or Japanese.

Those who would attend a seminar are asked to announce yourself to receptionists on your arrival at the IDE and to obtain APL Organizers' signature on your admission card after the seminar.


October 20, 2016. (Thursday) 12:30-14:00


Examining the impact of insurance on healthcare utilisation and provider choice using a natural experiment in South Africa


In 2006 the Government Employees Medical Scheme (GEMS) was introduced and gradually expanded to previously uncovered public sector employees. Public sector employees who indicated that their employer provides medical scheme cover increased from 59% to 73% between 2007 and 2008, stabilising around 77% in 2011. Under GEMS all governmentworkers became eligible for health insurance subsidies and low-earning employees received a full subsidy for the most basic benefit package, Sapphire. Under this scheme, no co-payments were required when using network providers. By the end of 2012, GEMS provided health cover to more than 200,000 government employees and 1.8 million individuals.

The introduction of GEMS created a natural experiment allowing for the estimation of the causal effect of coverage on health utilisation and provider choices for public sector employees. We exploit the gradual expansion of GEMS to implement an instrumental variable approach in two data sets: a combined version of Labour Force Surveys (LFS) and General Household Surveys (GHS) and the National Income Dynamics Study (NIDS) panel.

As expected in South Africa’s polarised health provider market, insurance causes a large response in both utilisation and provider choice. Our estimates from the two strategies show similar results and provide evidence of a very large impact of insurance cover (62 to 71%) on the utilisation of health services. Insurance increases the likelihood of using private providers and in particular private doctors (66%).

This analysis presents the most reliable indication yet of how an expansion of insurance may affect provider choice by lower income households in South Africa. It reveals a strong preference for doctors over public clinics (typically staffed by nurses) under insurance. While the inclusion of networks of doctors will enhance user satisfaction, the observed magnitude of the shifts under insurance cover suggests that rationing measures will be critical for the envisioned National Health Insurance (NHI).


Dr. Anja Smith (Faculty of Management and Economic Sciences, Stellenbosch University)



Institute of Developing Economies, APL Organizers
IMAI Kohei  E-mail:kohei_imaiE-mail
TSUBURA Machiko E-mail:Machiko_TsuburaE-mail
ASUYAMA Yoko E-mail:yoko_asuyamaE-mail