---------------------- Forwarded by Zimin Lu/HOU/ECT on 03/31/2000 01:43 PM 
---------------------------


Anjam Ahmad
03/21/2000 04:10 PM
To: Martina Angelova/LON/ECT@ECT, Trena McFarland/LON/ECT@ECT
cc: Dale Surbey/LON/ECT@ECT, Stinson Gibner/HOU/ECT@ECT, Zimin 
Lu/HOU/ECT@ECT, Vince J Kaminski/HOU/ECT@ECT 
Subject: PPI Index Short-Term Models

Hi Martina & Trena,

I made a few minor changes to the models...  I reckon these models can be 
checked in Houston now - the raw data is included (see attached spreadsheets) 
and all the regression parameters and main statistics are displayed.  I 
became interested in how far DZCV and PLLU should be below RPI and whether 
this model is reflecting any reversion to the RPI level; I believe that the 
forecasts are accurately reflecting this.  Please see graphs below:



Both models really need our RPI curve to be linked (at the moment I have just 
copied the 2.3% number forward).  Because the auto-regressive error term is 
not very important, we can run the models forward with reasonable 
confidence.  As I mentioned, I don't think we can really run this model more 
than 12 months, in fact, I think we should run for 9-12 months and blend the 
next 3-4 months out with the long-term model.

Hope I can fix the long-term ones now with some new insight!

Regards,

Anjam
x35383

PLLU:    DZCV: