Price setting conduct in South Africa 2002-2007 : implications of microdata for monetary policy

The objective of this research is to test the hypothesis that pricing conduct in SouthAfrica, revealed by studies of pricing microdata, can be shown to have an impact on themodeling and conduct of monetary policy. In order to discern stylised facts about pricingconduct in South Africa, use is made of two unique microdata sets, which are the unitlevel basis of South Africa’s measured CPI and PPI over the period from December 2001to December 2007. In particular, based on techniques which have been used incomparable studies in other countries, facts have been brought to light concerning interalia the frequency of price changes, the magnitude of price changes, the duration ofprices, heterogeneity in pricing, as well as evidence of seasonality, time-dependence andstate-dependence in pricing conduct. In order to understand the implications of suchpricing conduct, a basic closed economy theoretical model and thereafter an openeconomy New Keynesian DSGE model are used to compare the impact of various pricingassumptions. In general, but with some qualification regarding difficulties that arise incomparing pricing microdata with pricing conduct estimated in macro models, it is foundthat prices are more flexible than those estimated in the open economy DSGE model,implying sharper but less persistent interest rate responses to various shocks.Furthermore, the form of the New Keynesian Phillips curve used in the open economyDSGE model is found to be inconsistent with certain facts revealed through the pricemicrodata.