Abstract:
This study examines the relationship between government expenditure on health and economic growth
in Botswana. It seeks to test the existence of cointegration and specification of the deterministic
components with special reference to the Pantula Principle. This helps to overcome the shortfall of the
method by Johansen, which may lead to spurious results by omitting the presence of deterministic
components in the analysis. The cointegration approach is used and tested using three methods by
Engle and Granger (1987) or EG, a procedure suggested by Johansen (1988) and error correction
model (ECM) approach proposed by Granger(1988) and short-run analysis is made using the pairwise
granger causality tests. Findings show that the correct model specification for testing long-run
relationships consists of one cointegrating vector with a constant which is the most restrictive
hypothesis according to the Pantula principle. Using the Johansen approach, total health expenditure
and recurring health expenditure have a cointegration relationship with growth while development
health expenditure and growth are not cointegrated. The ECM and the approach by EG confirm a weak
and/or no cointegration between the variables. Growth has no effect on government expenditure on
health in the short run, but a cointegration relationship suggests that it may marginally contribute to
an increase in health expenditure over the long term. The study clarifies the correct model to test for
cointegration and specification for the deterministic component. It confirms the existence of a
healthcare expenditure-led growth hypothesis. This requires Botswana to design a policy that targets
specific parts of recurrent and
