New methods for forecasting inflation and its sub-components: Applications to the UK, USA and South Africa

The aim is to forecast the chief components of inflation (such as changes in fuel prices, food prices and prices of durable goods) for the USA, UK and South Africa, and to test whether the weighted sum of the component forecasts gives a more accurate overall forecast for inflation, than simply forecasting overall inflation itself. In the long run, the ratios of these prices to the overall consumer price index have altered because of technological changes and globalization, among other factors. For example, the prices of internationally traded consumer goods have fallen relative to prices of services. By building separate models for the components, the long-run information in the data and specific economic features likely to drive each component can be exploited. These models will test for asymmetries, such as the tendency of petrol prices to respond faster to rises than to falls in oil prices. The models should help better understand the causes of overall inflation through understanding the inflation trends of the underlying sectors. Modelling the components separately should also highlight where interest rate policy could be effective, and where other policies such as competition policy or price regulation might have complementary benefits.