PPI Index Short-Term Models

---------------------- Forwarded by Vince J Kaminski/HOU/ECT on 03/31/2000
04:03 PM ---------------------------

Zimin Lu
03/31/2000 01:44 PM
To: Vince J Kaminski/HOU/ECT@ECT, Maureen Raymond/HOU/ECT@ECT
cc:
Subject: PPI Index Short-Term Models

---------------------- 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: