Comparison study of methodologies for estimating the long-run exchange rate pass-through to import prices for South Africa

Abstract

Abstract The resilience of trade balances of the major industrialised economies such as the US and Japan to changes in their exchange rates following the switch from fixed to floating exchange rate regimes, triggered interest in the exchange rate pass-through relationship. Because of the importance of the pass-through issue particularly in economic policy formulation, a sizeable literature has developed over recent years. Comprehensive surveys of this literature include Menon (1995), Goldberg and Knetter (1997) and McCarthy (2002). However, not much attention has been paid to the comparison of the methodologies for estimating exchange rate pass-through. This research report aims to address this imbalance by comparing some of the exchange rate pass-through estimation methodologies via a Monte Carlo simulation study, based on the South African data set. The econometric results reported in this research report suggest that the Johansen type VECMs are superior to polynomial distributed lag models, exchange rate pass-through to South Africa’s import prices is incomplete (around 78%) and that the speed of adjustment to long-run equilibrium is low, about 7 per cent of disequilibrium in the previous month is corrected in the current month. We conclude that if we are not sure about the unit root properties of the data (as is normally the case), then the ARDL precedure is the appropriate model for empirical work.