9 February 1998

Example 9:  Generates a trend component from seasonally adjusted input.
            Instead of performing a seasonal adjustment, the program 
            estimates a trend component for the series assuming there 
            is no seasonality in the series (based on a suggestion by 
            Robert Templeton of Statistics New Zealand).

            Use X-12-GRAPH to examine an overlay plot of the trend and 
            original data.  In this case, the original data is the 
            seasonally adjusted data.  (Use the -g execution flag to 
            make the correct files available for use with X-12-GRAPH.)



# Example 9: strtsnsd.spc

# Generates a trend component from seasonally adjusted input.

series {name='USSTRT'
        title='Trend of US Housing Starts From (011)(101) Forcast Extension'
        file='USSTRTS.dat'
        format="2r"
#This file contain the seasonally adjusted Total U.S. Housing Starts
        precision=1
        span=(1984.3, )	 
}
transform {function=log}
arima {model=(0 1 1)(1 0 1)}
regression {
        variables=(ls1987.dec tc1990.dec)
# These are the automatically indentified outliers.
}
forecast {}
check {}
x11{ type=trend
     sigmalim=(.7 1.0)
     trendma=13
# These sigma limits and Henderson trend length are recommended
# by Estela Dagum and Israel's Central Bureau of Statistics for trend
# estimation from a seasonally adjusted series extended by ARIMA forecasts.
     print=(b1 c17 d12)
     final= ao
}