The monetary transmission mechanism in South Africa: A VECM augmented with foreign variables

We investigate the monetary transmission mechanism in South Africa, a small open country, in terms of both the “timing and the effect” of monetary policy. We use a modelling approach that has not been utilised yet to do this. The trade shares of South Africa’s trading partners changed considerably between 1980 and 2009. The country’s trade with China increased from none before 1993 to a three-year moving average of 14% in 2009. China thereby became the largest trading partner of South Africa in 2009, while trade with Germany, Japan, the UK and the USA decreased in the last few decades. We thus use time-varying trade weights to create the foreign variables. This is novel, since previous models either used data from the United States as a proxy for the rest of the world or created the foreign variables by weighing the data with the latest fixed trade shares of the main trading partners. The next step is to incorporate this model in a global vector autoregressive (GVAR) model for South Africa, which includes similar models for each of the country’s main trading partners. We could then determine the impact of shocks in the rest of the world on the South African economy. We develop a structural cointegrated vector autoregressive (VAR) model with weakly exogenous foreign variables suitable for a small open economy like South Africa. This type of model is known as an augmented vector error correction model (VECM), referred to by VECX*. We compile the foreign variables with trade-weighted three-year moving average data for 32 trading partners. This accounts for the significant change in trade shares over time. We use the model to investigate the functioning of the monetary transmission mechanism in South Africa. We find three statistically significant long-run economic relations: the purchasing power parity (PPP), the uncovered interest parity (UIP) and the Fisher parity (LIR). These long-run restrictions identify the VECX* model. The model performs well according to the reduced-form error correction equations and the diagnostics statistics. The persistence profiles of the three cointegrating vectors converge to zero, suggesting that the model will return to its long-run equilibrium after system-wide shocks. The generalised impulse response functions (GIRFs) are in line with expectations. The effect of a shock to the monetary policy interest rate on the inflation rate suggests a monetary policy lag of around 24 months. A “prize puzzle” is only observed in the first quarter following the monetary policy shock. Thereafter, the inflation rate declines as expected in response to the interest rate increase. The VECX* model therefore shows the effective transmission of monetary policy in South Africa between 1979 and 2009. (Please note that the full paper will be loaded within the next two weeks. My UK supervisor, Prof Hashem Pesaran, is busy reviewing the model, where after the paper will be finalised.) Current work: Incorporate this model in a Global VAR (GVAR) model for South Africa, where all the foreign variables are endogenous. The aim is to determine the impact of financial shocks in the rest of the world on the South African economy.