|
Abstract:
|
Many time series are measured as levels, but the main interest lies with the rate of change; the obvious example is price indices. There are well-known ways of measuring the trend of the level, but the trend of the growth rate is not well-defined. It is shown that the one-period growth rate of the usual level trend is not a satisfactory measure. I will investigate three questions: How do we measure the trend growth rate historically? What is the best current approximation to the historical growth rate? How can we give an error margin for the current estimate? The general considerations are illustrated by application to the growth rate of the British Consumer Price Index, which is the main target variable for the Bank of England's Monetary Policy Committee. In this context, the trend growth rate appears to be a useful additional interpretative tool.
|
