Breaking it up into trend and seasonal and error components

 GDP=scan("/Users/stevensabol/Desktop/R/gdp.csv")
Read 258 items
> GDP=ts(GDP,start=1,frequency=4)
> dlGDP=diff(log(GDP))
> plot(stl(log(GDP),"per"))


This allows us to do a structural decomposition
log(GDP) = trend + season + error 

Here is the business.