                                                       9 February 1998

Example 10. This example was used in the article "Sliding Spans Diagnostics
            for Seasonal and Related Adjustments" (Journal of the American
            Statistical Association 85 (1990), 345-355) to show how stability
            (as measured by the sliding spans diagnostics) and smoothness (as
            measured by the two smoothness measures Statistics Canada
            introduced in X-11-ARIMA) can disagree on whether the direct or
            the indirect adjustment of an aggregate series is to be preferred.
            For this example, Total U.S. Single-Family Housing Starts, subject-
            matter considerations validate the choice of the more stable
            adjustment.

            Only the metafile and the spec file for the Total Housing Starts are
            given below for this example.  

            Some regions show signs of calendar month heteroskedasticity 
            (i.e., data from some months have more statistical variability 
            than data from other months).  For a more detailed example 
            showing X11 options to deal with heteroskedasticity, please 
            see example #2.

            For a discussion of modeling considerations for the Northeast and 
            Midwest regions, please see example #7.

            Suggested graphs:  overaly and history graphs comparing the 
            direct and indirect seasonal adjustment.  



Example 10: comp.mta  

Content of the metafile for comparing the direct and indirect adjustments of 
Total U.S. 1-Family Housing Starts via sliding spans and smoothness measures.

ne1
mw1    
so1    
we1    
tot1f    



# Example 10: tot1f.spc  

# The spec file for the indirect adjustment of Total US Single-Family
# Housing Starts, obtained by summing the adjusted starts for the 
# four different regions obtained from the four spec files (ne1.spc,
# mw1.spc, so1.spc, and we1.spc) listed above this one in the metafile
# comp.mta.

composite{
   title='TOTAL ONE FAMILY Housing Starts'
   name='US1FAM'
   print=(iap isp)
   save=(isf)
   savelog=(peaks indm7 indm10 indm11 indq2)
}
transform{function=log}
regression{
   variables=(
      ao1991.jan  
      TD
   )
}
arima{
   model=(0 1 1)(0 1 1) 
}
estimate{savelog=afc}
check{
   print=all 
   savelog=lbq
}
forecast{maxlead=24}
x11{
   seasonalma=(s3x9 )
   title=(
      'Composite seasonal adjustment of Single Family housing starts'
         )
   save=D10
   savelog=(m7 m10 m11 q2)
}
slidingspans{
   fixmdl=yes
   savelog=percent
}
