********************************************************************* *** Introduction to ARIMA modeling in RATS. *** *** See Examples 16.1, 17.3, 18.1 in Pindyck and Rubinfeld (4th Edition, 1998) *** (P&R use data through March, 1996, ours ends in February, 1996) *** *** This version uses BJIDEN and BJFORE procedures. *** ********************************************************************* ********************************************************************* *** Set the CALENDAR and ALLOCATE range. CAL to start in January, 1960: *** and read in the Data: ********************************************************************** calendar 1960 1 12 allocate 1999:9 open data basics.wks data(format=wks,organization=observation) / rate ********************************************************************** *** Look at stationarity: ********************************************************************** * Graph the original series: graph(key=upleft) # rate source(noecho) bjident.src @bjident(diff=2,trans=none) rate * P&R settle on an ARIMA(8,1,4) with a constant: * Estimate and compute 24-period ex poste forecats: source(noecho) bjfore.src @bjfore(ars=8,mas=4,diffs=1,constant,rend=1996:2) rate \$ 1994:3 1996:2 rate_fore resids smpl 1994:3 1996:2 graph(key=below,header='Interest Rate Forecast') 2 # rate # rate_fore smpl * Eximine residual correlations: @bjident(diff=0) resids * 18-month ex ante forecast: @bjfore(ars=8,mas=4,diffs=1,constant,rend=1996:2) rate \$ 1995:10 1997:3 rate_fore18 resids smpl 1995:10 1997:3 graph(key=below,header='Interest Rate Forecast') 2 # rate # rate_fore18 smpl