V. Results from Modeling and Forecasting of DA CRS Levels

To provide a framework for analysis of potential future DA CRS obligations and the resulting effects of alternative caps, Navigant produced a range of separate modeling scenarios, incorporating the "total portfolio indifference" approach.⁷

In response to requests from parties for a range of forecast sensitivities in key variables, DWR/Navigant modeled three separate cases. These cases identified key resource variables under a "base case" corresponding to the variables underlying DWR's revenue requirement analysis for 2003. DWR then separately modeled a "high and low case" incorporating variations in key variables. The key resource assumptions subject to sensitivity testing among the three ProSym cases are as follows:

Key Variables in the DWR/Navigant sensitivity analysis

Future DA Load: High Case Assumes 10% higher than Base Case

Low Case Assumes 30% lower than Base Case

Natural gas prices: High Case Assumes 25% lower than Base Case

Low Case Assumes 20% higher than Base Case

New Generation: High Case Assumes 20% higher than Base Case

Low Case Assumes 20% lower than Base Case

CTC Benchmark: High Case Assumes 15% lower than Base Case

Low Case Assumes 15% higher than Base Case

The assumed range of differences in the variables among the three ProSym cases was arrived at through consensus among the participants at the Navigant modeling workshop. For each of the ProSym cases (i.e., high, low, and base), Navigant applied additional varying assumptions concerning OSS prices, interest rate accruals, and cap levels. Alternative combinations of the variables were assembled into eight possible scenarios. The eight scenarios were run under each of the three ProSym cases to produce 24 separate scenarios (i.e., eight combinations times three ProSym cases) for each utility.

The scenarios illustrate the effects of changing the following key variables: offsystem sales (OSS) price (at either 50% or 100% of market-clearing price[MCP]), interest rate on cumulative uncollected CRS balance (bounded between 4% and 9%), and the DA CRS cap (either at 2.7 cents or 4 cents/kWh). Appendix A summarizes the key results for each of the 24 assumed modeling scenarios for each utility. To assist in evaluating the effects of each scenario under the alternative cap levels on bundled customers, Navigant computes the maximum undercollection reached and years that would be required to pay off the accumulated undercollection.

In its modeling runs that Navigant produced in the earlier phase of this proceeding leading to D.02-11-022, the forecast DA CRS for SDG&E was significantly higher than that for PG&E or SCE. The difference was due mainly to the fact that SDG&E held a higher percentage of the highest-cost DWR contract power in its portfolio mix. In the current modeling performed in this phase of the proceeding, however, SDG&E has the lowest DA CRS requirements compared to PG&E and SCE.

In the most recent modeling scenarios for this phase, however, the payback period and undercollections for SDG&E are forecast to be the shortest among the three utilities. The change in SDG&E's relative situation since issuance of D.02-11-022 is due to revised assumptions underlying the effective date for applying the DA CRS. In earlier model runs supporting the record underlying D.02-11-022, Navigant assumed an effective date of July 1, 2001 for applying the DA CRS. In D.02-11-022, however, the Commission adopted an effective date of February 1, 2001 for assessing DA CRS. Consequently, Navigant has reflected this modified assumption in its modeling runs performed for this phase of the proceeding. The aggregate level of DA cost responsibility, however, continues to be based on maintaining bundled customer indifference between the change in DA load between July 1 and September 20, 2001. The substitution of the earlier effective date thus results in an increased volume of DA load absorbing the same fixed DA CRS obligation.

Consequently, by spreading the cost obligation over a greater volume of DA load, the per-kWh DA CRS declines and the projected DA CRS payback period for bundled customers occurs sooner compared with the earlier model runs. The reduced unit cost and earlier payback is more pronounced for SDG&E compared to PG&E and SCE because SDG&E's volume of DA load between February 1 and July 1, 2001 changes the most.

DWR states that the version of the model submitted in this proceeding, incorporates actual values for most of the volumes and costs from the fourth quarter of 2001 (when the DA CRS obligation first began to accrue) through the end of December 2002. Independent System Operator (ISO) charges did not reflect recorded values since they are lagged by 90 days. ISO charges represent only a small percentage of total charges. DA load values used by DWR for 2001 and 2002 reflect estimates received from the utilities. DWR has requested actual DA load data from each of the utilities.

Because there was no DA CRS in effect prior to January 1, 2003, the accrued DA CRS obligation from September 20, 2001 forward represents an accumulating undercollection due from DA customers. In its modeling scenarios, however, DWR did not quantify the actual undercollection applicable to DA CRS for 2001-02, but merely presented a range of hypothetical undercollections through 2002 corresponding to the scenario assumptions used to calculate the year 2003 DA CRS. Navigant applied the percentage ratio of DA in/out differences for each of its 2003 forecast scenarios to the historic 2001-2002 costs. This approach assumes that the DA-in/out scenario forecasts for 2003 apply equally to the 2001-2002 historic period. We discuss the problems with this approach below.

⁷ The "total portfolio" indifference approach as adopted in D.02-11-022 incorporates the total costs of serving bundled customer load both from DWR and URG sources, and solves for the DA CRS required to keep bundled customers indifferent between a DA suspension date of July 1 versus September 20, 2001. The DA CRS is based on ProSym model runs on a DA-in versus DA-out comparative basis.