WORKING PAPER NO: 93/2
XII TIME SERIES DECOMPOSITION AND SAMPLING ERRORS, DECEMBER 1993
INTRODUCTION
There is considerable interest in the effects of sampling error on the components of time series decomposition, such as,
the seasonally adjusted, trend and residual data. For example it is useful to have standard errors on the seasonally
adjusted data, trend, seasonal and residual which represent the effects of the sampling process. An important question
is what proportion of the observed variability in a time series is due to sampling error. Then there are design
considerations, for example given a certain sample design and characteristics what is an "optimal" time series
decomposition. One of the main deficiencies in past analysis has been providing a realistic model of an officially used
time series decomposition method. This paper demonstrates a realistic model for X11(Shiskin et al, 1967) and shows
how such a model can be used to analyse the effect of sampling design on time series decomposition.
|
| 
(152KB) |
|
/845519AA9FF5D4B2CA256EE600030D9C/$FILE/discover_banner_3.gif)
This page last updated 10 August 2004 | 
