A Sequential Panel Selection Approach to Cointegration Analysis: An Application to Wagner's Law for South African Provincial Data.

The main aim of this study is to extend the recently introduced sequential panel selection method (SPSM) to a cointegration framework which is particularly used to investigate Wagner's law for 9 South African provinces. We particularly apply the SPSM to the PMG and ARDL cointegration frameworks which we apply to annual data spanning from 2001 to 2016. The main findings show that when applying single country/region estimates we fail to find evidence of cointegration whereas within panel regressions, cointegration effects are present for the entire dataset. In further applying the SPSM we observed significant Wagner's effects for panels inclusive of Gauteng, Eastern Cape and Kwazulu-Natal provinces and when these provinces are excluded from the panels, cointegration effects are unobserved. [ABSTRACT FROM AUTHOR]