                                                3 February 1998

Example 11. This example shows two things: 
            (1) Forecast extension can reduce seasonal adjustment revisions. 
            (2) Initial trend estimates can undergo revisions that 
                are reduced rather little by forecast extension, but 
                revisions to the trend are small by Lag 3. 

            Forecast extensions with leads 0, 12, and 42 are considered for 
            the series of South Total Housing Starts.  This example is discussed
            in Section 3.2 of the article "New Features and Capabilities of the
            X-12-ARIMA Seasonal Adjustment Program" that will appeared in 
            the Journal of Business and Economic Statistics (1998, 16:127-177).



Example 11: soths.mta  

Metafile for running the three spec files with differing amounts of forecast 
extenion to produce revison histories for South Total Housing Starts.

sothsn0
sothsn12
sothsn42



# Example 11: sothsn0.spc   

# For producing seasonal adjustment and trend revision histories for 
# South Total Housing Starts for adjustment without forecast extension.

series{
   title='SOUTH Total Housing Starts (maxlead=0)'
   name='SOTOT'
   period=12
   file='start64.dat'
   format='1R'
   span=(1984.1, )
}
transform{function=log}
regression{
  variables=(td ao1990.1)
}
arima{model=(0 1 1)(0 1 1)}
estimate{ }
forecast{maxlead=0}
check{print=all}
x11{ }
history{
  estimates=(sadj trend)
  trendlags=(1 3)
  print=all
  savelog=(asa atr)
  start=1994.1
}
