20160713AVU to Staff 37.doc

AVISTA CORPORATION
RESPONSE TO REQUEST FOR INFORMATION

JURISDICTION: IDAHO DATE PREPARED: 07/12/2016
CASE NO: AVU-E-16-03 WITNESS: Tara L. Knox
REQUESTER: IPUC RESPONDER: Tara Knox
TYPE: Production Request DEPARTMENT: State & Federal Regulation
REQUEST NO.: Staff - 037 TELEPHONE: [phone]

REQUEST:

Please provide the Company's rationale for including autoregressive terms in its weather normalization models?

RESPONSE:

The autoregressive terms correct for non-quantifiable variables that affect use per customer resulting in positive or negative autocorrelation of error terms. I have observed that the addition of the autoregressive terms has a minor impact on the weather sensitivity coefficients and a slightly more noticeable impact on the intercept and trend coefficients used to estimate baseload usage. As an example Staff_PR_037 Attachment A shows both the stepwise regression and the AR adjusted regression for Idaho Residential Schedule 11. The winter sensitivity changed from 0.242825 to 0.242901 (when rounded to 3 decimal places this results in no change to the sensitivity factor and would have no impact on the weather adjustment) whereas the intercept changed from 247.3265 to 245.4333. This difference would have no impact on the weather adjustment.

Page 3 of 1

Page 1 of 1