Abstract
James A. Busuwski and Stuart Scott (1993)
"Some Issues in Seasonal Adjustment When Modeling
Interventions," Proceedings of the Section on Survey
Research Methods, American Statistical Association, forthcoming.
It has been demonstrated that applying intervention
analysis can aid in time series decomposition, in particular
seasonal adjustment (e.g., Buszuwski and Scott, 1988 and
Findley et al, 1988). However, there can be pitfalls in
modeling interventions, such as (1) unusual behavior
occurring near the end of the series, when the nature of the
intervention is unclear and (2) difficulties in selecting a
set of interventions for a series. for instance, illustrating
(1), if seasonality is changing, but one or more additive
outliers are modeled, intervention analysis will tend
incorrectly to reinforce the previous seasonal pattern.
Regarding (2), there may be conflicts between simplicity or
parsimony and diagnostic statistics form modeling and
seasonal adjustment. These issues are examined in the
framework of X-11-ARIMA seasonal adjustment with price index
series form the Bureau of Labor Statistics.
Last Modified Date: July 19, 2008
|