June 2019 issue contents
Estimating a Phillips Curve for South Africa: A Bounded Random-Walk Approach

Alain Kabundia, Eric Schalingb and Modeste Somec


In this paper we estimate a Phillips curve for South Africa using a bounded random-walk model. Central bank credibility, the slope of the Phillips curve, the natural rate of  unemployment, and the central bank's inflation target band are time varying. We find that the slope of the Phillips curve has flattened since the mid-2000s-particularly after the Great Recession-which is in line with the findings in most advanced countries. Our results do not lend support to the hypothesis that the ability of the South African Reserve Bank to hit its inflation target has decreased. With respect to the faith in the IT regime as measured by the degree to the extent of which inflation expectations are anchored to the target, our results indicate that inflation persistence has increased from 1994 to 2001, remained constant from 2001 to 2008, and eventually increased around 2008. This pattern is different from that of advanced countries where expectations have become better anchored relatively early in the IT regime. Moreover, we find that the increased stability of inflation expectations after 2008-which coincides with the Great Financial Crisis-is not only a result of good policy but also of "good luck."

JEL Code: C51, E52, E58.

Full article (PDF, 26 pages, 1,048 kb)

 World Bank
b  University of the Witwatersrand and VU University Amsterdam
c International Monetary Fund