Forecasting by David Hendry

Author:David Hendry
Language: eng
Format: epub
ISBN: 9780300244663
Publisher: Yale University Press

Learning from past mistakes

We have considered the possibility that the equilibria towards which the forecasting model and economic system adjust may diverge over the forecast period due to shifts in the latter. However, this makes the somewhat heroic assumption that the two match during the sample. But, as we have seen, change is constantly occurring in the economy. Figure 7.8 shows UK annual wage growth and price inflation over the last 150 years with the solid line recording large shocks and mean wage inflation over 10 different epochs.6 There have been periods in which inflation is near zero or negative, and periods in which inflation has been well above 15% per annum, with peaks of 25% and troughs of −22%.

If we wished to forecast future inflation based on this historical data how would we go about it? There are a number of options. One is to throw up our hands in despair of ever being able to model the complexity of the historical record of inflation and simply to look at the recent history, in the hope that this will be a reasonable guide to future inflation movements. Or we might attempt to model the whole period, but allow the parameters on the explanatory variables to vary over time. There are various approaches for doing so, including allowing the parameters of a model to ‘drift’ in a manner that can be modeled; or supposing that there are a small number of distinct ‘regimes’ (e.g., high or low inflation), and that the data we observe at any point in time are generated from shifts between these regimes; or to try to capture the shifts in a general way. Before considering possible solutions, we need to explore the role of shifts in more detail.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.