VIII. Allocation of Aggregate DWR Revenue Requirement Among the Utility Service Areas

A. Procedural Background

As noted previously, DWR computed a separate allocation of its revenue requirements to each of the three respective utility service territories in its initial submittal to the Commission. In its earlier submissions, DWR's suggested approach was to allocate its revenue requirement among customers on a uniform pro rata basis in relation to the net short position of each utility (i.e., the "postage stamp" approach).

In its most recent November 5, 2001 update, however, DWR refrained from computing any particular allocation of revenues. DWR acknowledges that the determination as to how the revenue requirement is to be allocated among utility customers is the responsibility of the Commission.

The assigned ALJ's Draft Decision to implement an allocation of DWR Revenue Requirements was mailed for comments on September 6, 2001. The Draft Decision proposed an allocation on the basis of where energy supplies were delivered into the power grid. Energy sources procured north of Path 15 were allocated to PG&E customers, while sources procured south of Path 15 were allocated to customers of SCE and SDG&E. This approach caused a disproportionately higher cost per kWh to be allocation to PG&E customers in comparison to southern California customers.

Parties filed comments on the Draft Decision on September 12, 2001. Upon review of the Draft Decision, PG&E filed a motion to compel production of computer models used to compute revenue allocations in the Draft Decision. In its motion, PG&E also requested evidentiary hearings on computer modeling and revenue allocation issues, arguing that the Draft Decision utilized computer modeling information that had not been made available for parties.

On September 19, 2001, Commissioners Lynch and Brown issued a Joint ACR, noting that DWR had agreed to provide confidential access to its computer models subject to a nondisclosure agreement. Parties executed a nondisclosure agreement with DWR for confidential access to DWR's computer models. On September 26, 2001, an ACR granted PG&E's motion for evidentiary hearings on the issue of DWR revenue allocation. The previously issued Draft Decision was withdrawn in anticipation of the evidentiary hearings, and a schedule was set for evidentiary hearings.

On October 5, 2001, a computer modeling workshop was held, to address questions regarding modeling assumptions underlying the DWR revenue allocation. Parties also submitted data requests to DWR relating to revenue requirements modeling. DWR agreed to provide responses on October 19, 2001.

Evidentiary hearings on DWR revenue allocation issues were conducted over five days from November 13 through 19, 2001. The parties sponsoring testimony were PG&E, SCE, SDG&E, ORA, TURN, AGLET, and CLECA. Opening briefs were filed on November 29, and reply briefs were filed on December 5, 2001. In addition to the parties noted above, briefs were filed by the Utility Consumer Action Network (UCAN), the County of Los Angeles and the City and County of San Francisco. Oral arguments were presented before the Commission on December 11, 2001.

DWR is not a party to the proceeding, and did not actively participate in the revenue allocation hearings. Nonetheless, the DWR November 5, 2001 revenue requirement submittal formed the basis for computing the allocations sponsored by parties. The DWR revenue requirement submittal was marked for identification as a reference item, and thus made a part of the record in this proceeding.

B. Parties' Positions

1. Overview

DWR revenue allocation proposals were made by PG&E, SCE, SDG&E, TURN, and ORA. DWR offered no allocation proposal in its most recent revenue requirement submittal, and was not a party of record. For purposes of our deliberations on revenue requirement allocation, therefore, we will review the testimony presented by parties. We will make reference to statements made by DWR, as relevant, in the context of its use by parties in the evidentiary exhibits admitted into evidence.

Parties' proposals can be categorized into two generally opposing points of view. One group generally favors allocating statewide procurement supplies on a pro rata basis in proportion to the net short position of each utility, except for certain limited utility-specific adjustments. This approach has been generally referred to as a "postage stamp" allocation method. Another group favors allocating discrete costs to specific utility service territories by attributing specific supply sources to specific utilities. ORA offered a third alternative which would average the results obtained from the "postage stamp" and the location-specific cost approaches.

In general terms, the location-specific allocation proposals result in a greater proportion of costs being allocated to PG&E's customers relative to the southern California utilities when compared with the "postage stamp" approach. The effects of these parties allocation proposals in terms of cents per kWh charges are summarized as follows:

Summary of Parties' Allocation Proposals
Pro-Rata ("Postage Stamp") Approaches
Party								PG&E	SCE	SDG&E	Total
PG&E (Ex. 181)						Revenue Reqt ($000)		$4,802,609	$3,608,734	$1,592,118	$10,003,461
						Rate (cents/kWh)		10.126	10.126	10.126
TURN (Ex. 172)						Revenue Reqt ($000)		$4,765,407	$3,674,066	$1,563,989	$10,003,461
						Rate (cents/kWh)		10.047	10.309	9.947
Geographically Differentiated Approaches
Party								PG&E	SCE	SDG&E	Total
SCE (Ex. 151-A)					Revenue Reqt ($000)			$4,831,668	$3,825,626	$1,346,167	$10,003,461
					Charge (cents/kWh)			10.187	10.734	8.561
SDG&E (Ex. 168)					Revenue Reqt ($000)			$5,088,000	$3,593,000	$1,322,000	$10,003,000
					Charge (cents/kWh)			10.727	10.082	8.408
PG&E (Ex. 181)					Revenue Reqt ($000)			$4,287,022	$4,142,546	$1,573,894	$10,003,461
					Charge (cents/kWh)			9.039	11.623	10.010
Averaging of Pro-Rata and Geographically Differentiated Approaches
Party								PG&E	SCE	SDG&E	Total
ORA				Revenue Reqt ($000)				$4,973,279	$3,513,501	$1,516,682	$10,003,461
				Charge (cents/kWh)				10.486	9.858	9.646
DWR Sales (GWh, over 24-month record period)								47,430	35,639	15,724	98,793
Percent of Sales								48.0%	36.1%	15.9%	100%

Percent of DWR Revenues Allocated to Each Utility
Pro-Rata ("Postage Stamp") Approaches
			PG&E	SCE	SDG&E	Total
Proposal
PG&E (Ex. 181)			48.0%	36.1%	15.9%	100%
TURN (Ex. 172)			47.6%	36.7%	15.6%	100%
Geographically Differentiated Approaches
Proposal			PG&E	SCE	SDG&E	Total
SCE (Ex. 151-A)			48.3%	38.2%	13.5%	100%
SDG&E (Ex. 168)			50.9%	35.9%	13.2%	100%
PG&E (Ex. 181)			42.9%	41.4%	15.7%	100%
Averaging of Pro-Rata & Geographically Differentiated Approaches
Proposal			PG&E	SCE	SDG&E	Total
ORA			49.7%	35.1%	15.2%	100%

2. Parties Supporting the Pro Rata Allocation Approach

PG&E and TURN both present allocation proposals based upon variations of the "postage stamp." This general approach was also supported by Aglet, CLECA, and UCAN. Proponents of the "postage stamp" approach support allocation of revenue requirement on a uniform basis proportionate to net short position of each of the utilities. Under this approach, no cost differential is recognized in the allocation on the basis of where a particular source of power supply originates, or from which contract it was procured.

Proponents of the postage stamp approach argue that DWR's purchasing decisions were not driven by differentiating the individual power needs of each utility on a stand-alone basis. DWR did not procure three separate portfolios of supplies, but rather, pursued a statewide purchasing strategy to procure one overall portfolio as a result of a statewide energy crisis. PG&E's witness Kuga testified that because DWR's legislative mandate was to purchase power on a statewide basis, it is reasonable to allocate costs uniformly among the utilities on the basis of their retail net short position (i.e., DWR's retail metered sales). Witness Kuga notes statements made by DWR that any attempted cost allocation by service territory would be "an artifice which would result in an arbitrary allocation of costs that would not necessarily result in any more logical or accurate cost causation."²⁰

Although PG&E opposes a location-based allocation method in principle, PG&E argues that certain adjustments would be warranted in the event that the Commission adopted such an approach over PG&E's objections. The four adjustments PG&E makes to the DWR location-based allocation reflect various ways in which the DWR model may overestimate PG&E allocated costs. The four adjustments are: (1) decreasing the net short purchased to fulfill the Western Area Power Administration (WAPA) contract, (2) accounting for utility self-provision of ancillary services, (3) crediting PG&E with the benefits of operating the Helms pumped storage plant, and (4) eliminating the double counting of unaccounted for energy (Ex.163, pp. 2-8).

While TURN supports a postage stamp allocation approach similar to that supported by PG&E, TURN proposes modifications to recognize certain utility-specific impacts. TURN's proposed utility-specific adjustments are generally similar in principle to the alternative cost-based adjustments proposed by PG&E. Unlike PG&E, however, TURN has incorporated these utility-specific adjustments into its primary pro rata allocation recommendation. TURN's proposed allocation reflects: (1) a slightly changed definition of net short from that assumed by DWR's model; (2) exclusion of pumped storage generation and loads; and (3) a different allocation method for ancillary services to better credit utilities able to self-provide such services.

Aglet and CLECA also presented testimony in support of the postage stamp allocation methodology. Aglet's witness Weil testified that allocation based on equal dollars per MWh is fair and reasonable because: (1) actual flows of electricity over the state's transmission network, and consequently the assignment of costs to individual utility customers, are complex and difficult to predict or measure (Weil, 41 RT 618621-6187:24); (2) DWR contracts have stabilized the power market, to the benefit of all ratepayers; and (3) the State of California's credit quality is superior to that of PG&E and SCE, possibly leading to reduced risk premiums by generators and marketers, which benefit all customers.

The DWR costs also include ancillary services, demand-side management, administrative and general expenses and capital-related costs, financing costs, and potential accrual of cash reserve requirements. Most of the costs and cash flows are not specific to any utility. Witness Weil further testified that the postage stamp allocation will also facilitate similar treatment of off-system sales revenues. Generation need not be dedicated to specific off-system sales customers, and assignment of revenues based on geographic area would be arbitrary. More generally, it is difficult to identify specific DWR contracts and incremental dispatched resources that are meant to serve specific incremental customer loads, even within large areas separated by transmission line constraints. If we were to allow retrospective adjustments to inter-utility cost allocations, the computer modeling required would be difficult and contested. Power flows over constrained transmission systems are very much dependent on ephemeral conditions like weather, ambient temperatures, local loads, and other factors. (Weil, 41 RT 6187:6-24.)

3. Parties Supporting Geographically Differentiated Cost Allocations

SCE and SDG&E each propose variations of an allocation methodology on the basis of the location where the power was procured. SCE and SDG&E translate cost differences associated with power deliveries by location into differences in the cost of providing power to customers in each of the service territories of the three utilities. SCE and SDG&E reject the postage stamp allocation arguing that it is overly simplistic, ignores cost-causation principles, and results in an unfair and economically inefficient allocation of DWR costs among utility customers.

Although SCE and SDG&E differ in their proposals, they each propose to allocate some portion of DWR procurement costs by measuring regional cost differences. SCE and SDG&E each argue that its proposal is consistent with Commission policy favoring the allocation of revenue based on cost-causality principles. SCE notes that the Commission has established the cost of service principle for allocating revenue requirement in the context of rate design for customers of the utility, and argues that there is no reason to allocate costs between utility service territories differently from costs among customer groups within a single utility service territory. SCE and SDG&E argue that failure to allocate DWR revenues based on cost causation principles would be inequitable, discriminatory, and economically inefficient. These principles are set forth in Exhibit 153:

a. SCE's Proposal

SCE proposes that the DWR costs to be allocated be separated into two components for differing allocation treatment. SCE characterizes the procurement costs of DWR fixed long-term (90 days or longer) contracts as costs incurred to meet the joint net short position of all three utilities. Because these long-term contracts provided a benefit to the entire State of California by lowering electricity prices on the spot market, SCE proposes that such fixed contract costs be allocated pro rata based on each utility's net short position. For short-term purchases (less than 90 days), however, SCE proposes that supply costs be allocated between PG&E and southern California utility customers based on the separate zonal cost of supplies using Path 15 as a dividing point.

SCE is in partial agreement with the position of PG&E insofar as it proposes to apply average pro rata costs for long-term contracts without regard to the location of the energy supplies procured for customers of each utility. SCE is in partial agreement with the position of SDG&E insofar as it proposes to allocate short-term power costs separately for utility customers north or south of Path 15, as explained below.

SCE's proposal for the treatment of long-term contracts differs from PG&E's, however, in terms of the level of detail involved in measuring the allocation. To the extent that each utility's net short position varies on an hourly basis and DWR contract costs vary hourly, SCE proposes a proportional allocation of hourly contract costs based on the hourly net short positions of each utility. Revenues from DWR off-system sales would be allocated the same way. PG&E, by contrast, proposes use of only monthly average data for pro rating the allocation.

SCE argues that any allocation of actual energy costs on anything less precise than on an hourly basis would bear little resemblance to costs that are actually incurred for customers of each utility. Because the hourly cost data necessary to make the allocation calculations are not currently available in the record, SCE proposes that the Commission make an interim allocation of revenues on a monthly net short basis for now. SCE also proposes that the interim allocations based upon aggregate monthly sales data have a provision for an after-the-fact true-up using the hourly cost data once it becomes available from DWR.

The data that DWR would need to produce in order to provide for an hourly allocation under SCE's proposal is DWR's:²¹




· MWh purchases under long-term contracts;




· average hourly long-term contract costs;




· short-term MWh purchase prices in NP 15 and SP 15; and




· short-term hourly purchase prices in NP 15 and SP 15.

SCE also proposes that other miscellaneous DWR costs that do not vary with energy consumption or by time period, (e.g., A&G, DSM, and financing-related costs) are appropriately allocated on a monthly basis in proportion to each utility's net short position.

SCE draws a distinction between long-term contracts where the delivery point is irrelevant for allocation purposes and short-term power purchases to meet the residual net short position of each utility once long-term contract power purchases have been allocated. SCE views this second category of costs as being incurred in separate zones to meet specific utility needs in those zones. As such, SCE argues, the location of the related short-term power purchases becomes a relevant consideration for allocation purposes.

SCE's witness Stern acknowledges that in most instances, the costs associated with meeting these residual net short needs will not vary by location. But the price of power within northern California versus southern California will be different if a transmission constraint exists. When transmission constraints exist, SCE recommends that DWR costs of short-term and spot purchases be allocated separately for areas north and south of Path 15. Under this zonal allocation approach, contract costs for power delivered into the transmission grid in zones north of Path 15 are assumed to be 100% attributable to PG&E customer loads. Only the amount of power north of Path 15 in excess of PG&E loads is assigned residually to utility customers located south of Path 15.

SCE's distinction between unavoidable and shorter-term costs is based on the principle that "[t]he entity causing DWR to incur a cost should pay that cost." (Exhibit 150, SCE's witness Gary Stern, p. 1, line 7.) SCE assumes that: (1) DWR incurs long-term costs to meet the combined needs of all utilities, and (2) during times of transmission constraints, zonal power displaces each utility's URG. (Exhibit 150, Stern, p. 3, line 21.)

SCE also believes that each utility should be responsible for the share of DWR's ancillary service costs that it causes. Each utility's share of allocated costs would thereby be directly reduced by the amount of such costs that it provides for itself. To the extent there is congestion in real time causing a price difference between zones for ancillary services, SCE proposes that the allocation be done on a zonal basis, with separate allocation on an hourly cost basis.

b. SDG&E's Proposal

SDG&E agrees with SCE in its emphasis on cost causation as an important principle to apply in allocating DWR costs, but differs with SCE on how those principles should be applied here. SDG&E supports allocating all DWR procurement costs on a zonal basis, irrespective of whether they relate to fixed contracts or short-term purchases. In this respect, SDG&E differs with SCE that at least the fixed contract costs serve the joint needs of all three utilities, and thus should be allocated on a statewide pro rata basis without regard to supply zone. SDG&E states that it did not have the data or the time necessary to develop a comprehensive DWR revenue allocation proposal based on cost causation principles. However, SDG&E claims that the methodology developed by the Commission's Energy Division as reflected in the ALJ's Draft Decision moves the Commission as close as possible to a cost-based allocation under the circumstances.

As incorporated in the ALJ's Draft Decision, the Energy Division derived an allocation of the DWR's revenue requirement, differentiated between whether the energy is delivered over facilities in northern California or in southern California. As the geographical dividing point, the Energy Division used Transmission Path 15. Energy sources procured north of Path 15 were allocated to PG&E customers. Energy sources procured south of Path 15 were allocated to customers of SCE and SDG&E. SDG&E argues that the Energy Division method recognizes that differences exist in the cost to serve consumers in different parts of the State. SDG&E advocates the use of this method, arguing that the Commission cannot wait for a full-blown cost allocation study recognizing all relevant cost differences and constraints, for which there is neither the data nor the time to implement.

SDG&E proposes certain adjustments to the Energy Division allocation methodology, to provide what it considers to be a more accurate allocation result. First, SDG&E's proposed allocation reflects a price differential of electric energy supplies between the NP 15 and SP 15 regions. The adjustment, presented in Exhibits 155 and 157, would result in a shift in cost allocation from southern to northern California, the supporting calculations of which are set forth in (confidential) Exhibit 158. SDG&E's witness Nelson (Ex. 157) computed a revenue requirement reduction of approximately $186 million in SDG&E's share of the DWR allocation as a result of the price differential between NP 15 and SP 15 for the period January through July 2001. SDG&E assigns the higher NP 15 prices to PG&E customers on the premise the power associated with the higher NP 15 prices was consumed exclusively by PG&E customers. SDG&E's witness Nelson testified that one of the principal causes of the price differential was the congestion charge assessed by the ISO on the north-to-south power flow. For periods when no congestion existed on Path 15, prices were the same both for NP 15 and SP 15.

As a second adjustment, SDG&E recalculates the net short allocated to SDG&E customers for February and March 2001. The DWR PROSYM model allocates 14% of DWR's purchases to SDG&E for February and March 2001. SDG&E claims that DWR, however, separately presented data that indicated the correct percent of DWR's purchases during those months was 12%, not 14%, resulting in $51 million more being allocated than should be to SDG&E's customers. (Exh. 155, p. 4.) Although the difference in percentage is small, SDG&E argues that the amount at issue is significant because DWR's costs for power during this period averaged $269/mWh, higher than any other quarter. (Id.) SDG&E argues that its customers did not cause DWR to incur this cost and therefore should not be allocated that cost.

SDG&E makes a third adjustment for the lead/lag accrual to cash which accounts for the timing difference between the provision of services and the receipts of cash for them. In its August 7 update, DWR erroneously attributed to SDG&E customers certain purchases made in January 2001. When DWR corrected this in its November 5 filing, SDG&E claims the effect was to increase the revenue requirements to its customers. During January, DWR incurred substantial expenses but had not received any revenue in order to pay for them. This resulted in a large lag in accrual to cash in January, reducing revenue requirements in January that DWR allocated based on retail sales. Since DWR did not have any retail sales to SDG&E in January, SDG&E did not share in that reduction.

In later months, DWR's cash payments exceeded its accrual, resulting in a lead in accrual to cash. In those months, that added to DWR's revenue requirements and SDG&E's customers bore part of that addition. The net effect to SDG&E customers was an added cost due to their not sharing in the January lag (reduction to revenue requirement) but bearing part of the subsequent leads (increased revenue requirement) that were caused by the other utilities. SDG&E proposed to adjust for this net effect in order to avoid SDG&E's customers being allocated revenue requirements that they did not cause. SDG&E calculates a $65 million credit to the revenue requirement allocated to its customers due to this adjustment. (Exh. 155, pp. 4-5.)

SDG&E's proposal differs from SCE's in terms of the importance that SDG&E places on forecasted (i.e., ex ante) costs, as opposed to actual costs for allocation purposes. SDG&E argues that allocations of fixed long-term contract purchases should rely upon ex ante sales forecasts provided to DWR by the utilities, and that such allocations should not be trued-up for actual sales. SDG&E argues that because DWR made purchase commitments based on the sales forecasts provided by the utilities, each utility should bear responsibility for the allocation of costs that results from DWR's use of such forecast. SDG&E further proposes that variable price or spot (imbalance) purchases should be allocated based on actual consumption relative to what was purchased under the fixed contract, and that gains or losses from sale of surplus energy should be allocated based on the same differential between forecast and actual consumption.

SDG&E acknowledges that certain types of DWR costs (e.g., overhead and A&G) cannot be attributable to specific service territories. SDG&E does not oppose such costs being allocated to consumers on a uniform statewide pro rata basis.

4. ORA's Proposal

ORA recommends the Commission allocate DWR costs by using a simple average of what it characterizes as the most conservative versions of the postage stamp model and the Energy Division's cost-based model (Ex. 161, pp. 1-2). ORA argues that such an average will capture the relative strengths and weaknesses of both models. An average acknowledges different regional costs, but also biases cost responsibility toward the assumption that DWR's primary concern was to satisfy a statewide need. ORA believes that the evidence supports both assumptions. An average also reflects ORA's belief that a "cost-based" method is conceptually more accurate and appropriate than a postage stamp method. At the same time, an allocation average of the two methods reflects and acknowledges the lack of necessary information needed to develop an accurate cost-based allocation.

ORA's method utilizes utility specific input data from the DWR revenue requirements model in a manner similar to that described in the September 4 Draft Decision. ORA extracts data from the PROSYM input model runs made by DWR in an attempt to obtain DWR costs by utility in order to display different spot prices for PG&E and SCE that are attributed to transmission constraints. However these PROSYM input results do not match the actual DWR energy bill in 2001 or the expected energy bill in 2002. To obtain the energy bill which DWR estimates has actually occurred to date and forecasts for 2002, these model run results have to be increased. The average revenue increase required for 2001 is 26%; for 2002, the estimate is an 8% increase. ORA therefore simply increases the number the PROSYM input results for each utility by 26% in 2001 and 8% in 2002 to correspond to DWR's estimate of required energy revenues. In addition, ORA assumed a 7.43% percent increase to cover ancillary services applied uniformly to all three utilities following DWR's assumptions.

Another allocation option suggested by ORA is for the Commission to adopt a postage stamp allocation for 2001, and to adopt a zonally based allocation for 2002. ORA believes this averaging effect would also provide appropriate dispatch signals for the utilities' own retained generation decisions. Of course, this assumes real time communication between DWR and the utilities regarding the price of DWR's hourly purchasing opportunities.

C. Discussion

1. Statutory Basis for Allocation Methodology

We first address the contention of certain parties that AB1X, Water Code Section 80002.5 mandates a postage stamp allocation method by law. The pertinent language reads:




"It is the intent of the Legislature that power acquired by the department under this division shall be sold to all retail end use customers being served by electrical corporations...Power sold by the department to end use customers shall be allocated pro rata among all classes of customers to the extent practicable."




Cal. Water Code Section 80002.5 (emphasis added).

We do not interpret the statute as requiring any particular revenue allocation approach as a matter of law. As noted by SCE and SDG&E, the statute addresses how power sold by DWR is to be allocated, but does not prescribe how DWR's revenue requirement is to be allocated. Allocation of power on a pro rata basis relates to physical deliveries of MWhs, and simply means that DWR has to supply all customer classes with power on pro rata or proportionate basis (e.g., not giving residential customers priority and curtailing industrials, or vice versa). Also, this section addresses allocation among all classes of customers, not allocation among service areas. Furthermore, "to the extent practicable" recognizes that there may be practical reasons why DWR must allocate power (not costs) differently among classes of customers (not service areas). Accordingly, we shall determine the allocation of revenue requirements based upon the merits of the factual record, rather than relying merely on a legal interpretation of the statute.

2. Cost-Based Principles as a Basis for DWR Allocation Methodology

This Commission has traditionally recognized the principle that utility revenues should be allocated by assigning cost responsibility in relation to cost causation. Cost-based rates promote economic efficiency because customers pay for what they consume, and thus properly adjust their consumption to match what the product really costs (Ex. 153, p. 6). SDG&E witness Croyle notes that cost causation and cost allocation principles are "standard fare" for the utilities and not a new idea. (Ex. 153, pp. 11-12.) Cost-based allocation and rate design promote efficient utility planning.

SCE's witness Stern adds that not allocating spot energy purchases to utilities' service territories on a cost basis gives false signals to the utilities on how best to dispatch their own resources. For example, if the costs allocated to its service territory are higher than actual cost, the utility might erroneously dispatch one of its own resources that is less expensive than the allocated cost, but more expensive than the actual cost, which is not an economically efficient practice (Ex. 151, pp. 14-15). However, Dr. Stern agrees that allocation decisions made now for sunk costs, those already incurred by DWR during 2001, has nothing to do with economic efficiency (Stern/SCE, RT 5854). Croyle further argues that, had the DWR not been purchasing power on behalf of the utilities, the utilities would have had to face the market themselves and been exposed to all the factors that cause regional differences in pricing (Ex. 153, p. 7). SDG&E argues that the costs the DWR incurs should not be allocated on a different basis just because DWR is an interim provider.

We agree that the cost allocation principles adopted for DWR revenues should reflect the cost that the customer imposes on the system. The more difficult question is how to implement an allocation that best achieves that objective. In D.96-04-050, which decided revenue allocation and rate design issues in SCE's general rate case, the Commission stated:




"[W]e reiterate our primary goal of ratemaking, namely, to achieve rates which reflect the cost that the customer imposes on the system. This approach not only results in an equitable distribution of [SCE's] revenue requirement, but also provides the most accurate price signals to the customer regarding his energy consumption."

In D.96-04-050, cost causation principles were applied to compute marginal costs in the context of allocating revenues between different customer classes and designing rates within a single utility's service territory. In this proceeding, however, we face just the opposite situation. We are not allocating DWR revenues based on customer class distinctions nor designing retail rates. Rather, we are allocating revenues in the aggregate among three different utility service territories.

We agree that cost responsibility should be assigned in relation to those factors that cause the costs. We also agree that DWR revenues should be assigned on the basis of cost causation to the extent that clear drivers of cost can be identified and measured. Yet, in order for a cost-differentiated revenue allocation to be applied separately among the three utilities, there must be a discernable cause-and-effect relationship between the cost incurred and a cost driver.

SDG&E and SCE portray the choice of allocation methods before us as a dichotomy between either cost-based (i.e., the SCE/SDG&E approaches) or non-cost based (i.e., the PG&E/TURN approaches). We disagree with such a characterization of the alternative proposals. We view all of the allocation alternatives presented by parties as forms of cost-based allocation. The differences relate only to how accurately the proposed cost drivers under the alternative proposals reflect cost causation, not whether cost causation is an appropriate standard.

The proposals to allocate costs pro rata to each utility merely represent another form of cost-based allocation where the cost driver is the net short position of each utility. The share of costs assigned to each service territory differs in relation to the size of the net short. Thus, under all the proposals, including the pro rata allocation alternatives, the greatest portion of DWR costs is allocated to the PG&E customers, reflecting the fact that the largest portion of net short power procured by DWR is sold to customers in PG&E's service territory. None of the parties propose that the Commission apply an allocation approach that intentionally subsidizes a particular customer group, nor one that allocates a profit premium to certain customers beyond the straight cost incurred by DWR.

3. Allocation of Administrative and Financing Costs

Parties generally agree that DWR administrative and financing costs cannot reasonably be attributable to specific customer groups, and may be allocated on a statewide pro rata basis in relation to the net short position of each utility. These costs include administrative, DSM, and financing costs. Recognizing that there is essentially no dispute over the allocation of such miscellaneous costs, we shall allocate such costs on a statewide pro rata basis in relation to each utility's net short position.

4. Allocation of Power Procurement Commodity Costs

We next turn our attention to the dispute over the allocation of DWR's commodity costs associated with procurement of power. The dispute centers on whether costs of power supply sources in northern California (NP 15) are separately attributable to customers in PG&E's territory, or whether all utility customers statewide should be assigned a pro rata share of those costs. To determine whether the costs of specific sources of supplies incurred in northern California should be exclusively allocated to PG&E customers based on cost causation principles, we must determine whether a cause-and-effect relationship exists between these costs and the use of energy exclusively by those customers.

a. Statewide Versus Regional Procurement and Cost Measurement

Prior to DWR's entry into the role of purchasing the utilities' net short position, each utility procured its own separate portfolio of supplies to serve customers in its service territory. SDG&E witness Croyle argues that had DWR not been purchasing on the utilities' behalf, the utilities would have had to face the market themselves, thus exposing them to all of the factors that cause regional cost differences in prices.²² While Croyle's observation is correct, it does not lead to the conclusion that DWR procurement can be disassembled into the equivalent of three separate streams of costs incurred by independent entities. Where each utility makes its own separate purchasing decisions, discrete costs can readily be set by service territory based on each utility's own separate purchases. The cause-and-effect relationship between the costs incurred and the charges applied in a given service territory is self evident in such a case. Likewise, where each utility makes its own purchasing decisions, there is no need to "allocate" revenues among utilities, because the costs incurred by each utility are a discrete, objectively discernable quantity. Allocation becomes an issue where there is a common pool of costs that relate to multiple entities.

With the entry of DWR into the power procurement market for the three utilities, a single entity (i.e., DWR) made aggregate purchases of the net short position, rather than three discrete entities each procuring its own separate portfolio of supplies. Yet, because DWR did not procure a separate portfolio of supplies for each utility, DWR's purchasing practices are not directly analogous to the separate purchasing of discrete supply sources by each separate utility only for its own customers.

As noted by DWR itself: "DWR has had minimal flexibility in its choice of power providers. Therefore, it has not been possible for DWR to undertake separate solicitations for each of the IOU service areas."²³ SDG&E witness Croyle agreed in cross-examination that the DWR contracted for electricity on behalf of all three utilities and did not conduct separate solicitations for the three service areas (Croyle/SDG&E, RT 5997 and 5998). TURN witness Marcus concluded that, based on his reading of every long-term contract, DWR "was basically trying to do anything it could to alleviate problems for the summer of 2001 and, to a lesser extent, 2002, in the period from February through May...they were trying to get anything they could get."²⁴ DWR itself has stated that its service territory is "statewide. The energy associated with the net short contracts is not directly assigned to the IOUs or to a specific area."²⁵

DWR thus has not maintained separate portfolios to meet the net short positions of each utility. Any allocation of power purchased under the DWR contracts and spot market purchases for each respective service area by assuming an analogy to utilities' separate sources of supply by service territory is not consistent with the way DWR constructed its portfolio of supplies. The integrated procurement of energy supplies on a statewide basis, was not analogous to three distinct entities incurring separate and unique regional costs.

b. Transmission Congestion Criterion

In the absence of separate portfolios, we must consider whether any other factors resulted in different prices being incurred for energy delivered to customers in each of the three utilities' service territories. As noted by CLECA witness Barkovich, electricity that is bought on behalf of a group of customers that flows through the same grid should, under the laws of physics, be available to all of those customers that are served off that grid, unless transmission congestion prevents that occurrence.²⁶

SDG&E and SCE point to transmission congestion over Path 15 as a constraining factor causing DWR to procure supplies delivered north of Path 15 specifically for northern California customers (i.e., the PG&E service territory). SDG&E argues that because DWR assumed the transmission system would be congested, it therefore made purchases north and south consistent with that assumption. DWR's response to data request PG&E-8, cited in Exhibit 157 (p. 3) stated that there is "no factual basis" for assuming that transmission constraints do not exist.

There is general agreement that during the first few months of 2001, a price differential existed between power to be delivered north of Path 15, and power to be delivered south of Path 15. However, parties disagree as to whether DWR paid the differential exclusively due to servicing PG&E load demand.

SDG&E argues that north-to-south transmission constraints caused pricing differentials that represent a major cost driver relevant to the DWR allocation between northern and southern service territories which justify its proposed $186 million reduction in costs allocated to its customers. SDG&E acknowledges a difference in the size of the price spread between the public data cited in Nelson's testimony and the confidential data in his workpapers.²⁷ However, Nelson's workpapers (Ex. 158) reflect the costs that DWR actually incurred in buying power in the northern and southern zones. The public data only was provided only to illustrate the market environment in which DWR operated, since the confidential DWR data could not be made public.²⁸

Witness Croyle claims that the DWR could not possibly have ignored the cost differences imposed by the transmission constraints along Path 15, and had to ensure that each region would have enough electricity when the transmission constraint is operative. SDG&E claims that DWR acknowledges making purchasing decisions on the basis of this split between north and south. In response to Data Request SCE-01, DWR listed various procurement objectives under AB1X, and stated: "These objectives include, but are not limited to, the following...Match intrastate regional electric needs (north and south of Path 15 transmission constraints) to locations of supply." ²⁹ SDG&E thus argues that it is both factually correct and reasonable to recognize Transmission Path 15 as a "geographical dividing point" for allocating costs. SCE witness Stern likewise testified that in times of actual transmission constraints on Path 15, DWR was forced to purchase power in the zone north of Path 15.³⁰

We find no basis in the record to assume that actual transmission constraints were constant over time, or that the physical flow of power delivered into the grid from NP 15 sources was exclusively consumed by northern California customers of PG&E. Presumed NP 15/SP 15 transmission constraints at times were only anticipated to occur, but did not ultimately materialize. PG&E witness Kuga testified that at times, congestion was anticipated in pricing power in the day-ahead market, but in real time there was no congestion. Thus, actual power flows over Path 15 could and did physically flow north to south. DWR may have purchased power in one zone at a higher price than in the other zone even when there wasn't an actual transmission constraint, but where one was expected. That is, real price differentials could occur simply based on the expectation of congestion, even if that congestion fails to materialize. This has been referred to as "phantom congestion."

Therefore, the payment of a price differential for NP 15 power did not always equate to a physical constraint preventing NP 15 power flows to SP 15 destinations. Moreover, while one of DWR's objectives was to "match regional electric needs to locations of supply," there was no strict division segregating the source of deliveries to PG&E versus to southern California customers. DWR states that in fact, "energy associated with the net short contracts is not directly assigned to the IOUs or to a specific area."³¹ To the contrary, DWR stated that "power purchased under many contracts will in fact be used to meet the net short in more than one service area, directly or through swaps, exchanges or otherwise. The allocation of power will change continually over time."³²

Therefore, the existence of a price differential for congestion charges over Path 15 does not form the basis for any specific identification of supply sources with specific territories served. Similarly, even where transmission congestion constrained north-to-south deliveries, DWR might still be able to arrange a power exchange with other SP 15 supply sources to provide the benefits of NP 15 supplies to SP 15 customers.

Moreover, Path 15 does not, in fact, represent a boundary between PG&E and SCE, but rather falls within PG&E's service territory. Furthermore, sometimes the zones are separated by congestion on Path 26.³³ Thus, even if we agreed in principle that costs should be allocated based on a strict north-to-south transmission boundary, the measurements offered by SDG&E and SCE based on Path 15 would not be congruent with PG&E's service territory. The SCE/SDG&E approach would arbitrarily allocate the higher NP 15 costs even to those PG&E customers residing south of the Path 15 transmission constraint.

SDG&E argues that whether or not the system was constrained becomes moot in terms of cost causation since the cause of purchases north and south of Path 15 was an expectation of system constraint and inability to move purchases north and south. To the extent the congestion was phantom in nature, it indicates at least from a physical perspective, that powers sources procured in NP 15 locations could and did flow south for consumption by SCE and SDG&E customers. The only remaining question is whether the NP 15 price differential associated with the expectation of system constraint was attributable exclusively to PG&E customers, even where the actual flow of power was not constrained to the north.

Transmission constraints that limit service from specific generators to incremental customer loads, for example temporary Path 15 constraints, are unstable over time. Power flows over Path 15 when the line is constrained depend on weather conditions, and transmission constraints on Path 15 are not in effect all the time. Thus, attempting to model transmission constraints as a variable in DWR cost allocation would result in volatility, and unfairly magnify the effects of DWR's empirical modeling corrections on utility ratepayers.³⁴

Moreover, higher prices paid by DWR for power delivered into the transmission grid north of Path 15 that were paid during the early months of 2001 might have been caused in part by other factors besides just congestion, exclusively. For example, various contract prices DWR agreed to are a function of when DWR signed the contracts rather than where the power was ultimately consumed. Prices are also a function of the structure of the contracts, for example whether they include a separate capacity component, are indexed to natural gas prices, or call for delivery only during specified times of the day. Prices can be a function of the term of the contracts, as well. SDG&E and SCE did not adjust out such extraneous factors, but simply assumed the entire price differential was due to transmission congestion and thus assignable only to PG&E customers. No party presented evidence on the extent to which factors other than transmission congestion contributed toward the higher price of NP 15 power.

We conclude that the causes of the price differential cannot fairly be attributed exclusively to customers in the PG&E service territory. As noted by PG&E, we agree that congestion costs were a reflection of a statewide dysfunctional market during the early part of 2001, rather than a product of the physical configuration of the system. After FERC adopted measures to help minimize or eliminate the market flaws in California, the pricing across Path 15 changed substantially. Price differentials between north of Path 15 and south of Path 15 power have been diminishing, or have practically disappeared. Witness Nelson testified that New York Mercantile Exchange prices for 2002 deliveries suggest that NP 15 prices might be lower than SP 15 prices in the future.³⁵

To the extent that flawed market rules were due to statewide dysfunctionality of the market, the impacts of those rules cannot reasonably be isolated only to one geographical sector of California consumers. This finding is consistent with D.01-05-064 where we stated, "no customer is causing the exorbitant electricity prices faced by the utilities and CDWR. Thus, it would be unfair to attribute the current wholesale market prices as caused by any particular type of customer.... The price of wholesale energy bears no relationship to the cost of production, but is rather a function of what price can be extracted from the California market through manipulation."³⁶

In the same way that we cannot attribute dysfunctional price increases to particular customers, by virtue of their type, likewise, we cannot attribute such prices increases only to certain customers, simply by virtue of their location. Therefore, while PG&E customers certainly should absorb some share of the NP 15 congestion charge differential, they should not shoulder the entire burden. The statewide pro rata approach to allocation fairly assigns a share of these costs among all ratepayers.

Even if theoretically, the costs of supplies that were used to serve PG&E customers were systematically higher than for southern California customers, the underlying data to compute cost differentials is unreliable. Development of differential allocation methods has been impeded by the difficulties faced by parties in gaining access to modeling information, including the PROSYM input data set that underlies the DWR model.

SDG&E's witness Mr. Croyle admits that the quality of the data is less than optimal, but believes that his proposed allocation moves toward a cost basis that is more robust than alternative methods (Ex. 153, pp. 2-3). CLECA witness Barkovich testified that it is not possible for other parties to verify the results of DWR's modeling efforts, which are a function of unverifiable input assumptions and the algorithms contained in the model. The production of locational prices, the aggregation of these prices to ISO congestion zones, and the connection of these zones to the service areas of the three utilities are all open to question (Ex. 159, pp. 3-4).

The use of a uniform pro rata allocation approach on a statewide basis is also consistent with how DWR's production cost model works. DWR uses the PROSYM production cost model to simulate the operation of the western regional electric system, and to estimate DWR's total power purchase costs to serve a single statewide service territory. DWR has also developed a financial model which takes output from PROSYM and determines DWR's needs for utility customer revenues on a statewide basis, taking into account estimates of purchase volumes, ancillary services, and financing costs.³⁷

Thus, we are unpersuaded that the price differentials across Path 15 as computed by SDG&E can reasonably be attributed as higher costs to serve only PG&E customers to the exclusion of southern California utility customers.

Because any price differential between DWR's costs for power delivered north of Path 15 and for power delivered south of Path 15 was likely to have been the consequence of dysfunctional statewide market rules, there is insufficient basis to allocate a disproportionate share of NP 15 costs to PG&E customers based on the theory of cost causation.

c. Distinctions in the Allocation of Fixed Price Versus Short Term Purchases

We find no objective basis to apply different allocation methodologies based merely on whether a cost relates to a long-term fixed price or a short-term purchases, as proposed by SCE. SCE witness Stern testified that "[DWR] did not distinguish the delivery location in their process of procuring those long-term contracts." SCE distinguishes, however, between (a) long-term contracts used to serve the joint needs of all customers with no regional differences and (b) short-term power purchases presumed to meet the separate needs of each utility from distinctly different sources of regional supply.

From an operational perspective, however, we find no special significance in contract duration as a criterion for determining how much power DWR procured for each separate utility service territory. There is no evidence that DWR's intentions regarding service territory deliveries are different depending on whether the source is contract power of less than 90 days duration or long-term contracts. (Weil, 41 RT 6179:13-24.) There is no record information about the shorter-term contracts. (Barkovich, 41 RT 6141:28-6142:2.) SCE witness Stern acknowledged that DWR has not provided any information associated with the specific reasons for entering into individual short-term contracts, for example, whether DWR was motivated by transmission constraints or price factors, for example. (Stern, 39 RT 5864:25-5865:14.) Moreover, to the extent URG resources were fully committed, then DWR short-term power would not substitute for URG power. Instead, it would replace some other resource or simply increase reserve margins.

Observations as to the pattern of DWR's purchasing mix between short- and long-term purchases over time lend support to the conclusion that there was no distinction in the destination of power based on the contract term. DWR purchased only shorter-term electricity products during the first three months of 2001, then began incurring long-term contract costs in April 2001.³⁸ Therefore, under SCE's proposal, all DWR costs during transmission constrained periods in January, February, and March 2001 would be allocated zonally, and most would be allocated zonally until late in the year.

In the first few months of 2001, however, DWR was "scrambling" to obtain whatever resources were available in order to meet its procurement goals. (Barkovich, 41 RT 6164:11-22.) DWR's shorter-term costs began to decline after April 2001, and ancillary services costs declined significantly during the summer months. (Stern, 40 RT 5966:7-5967:11.) By autumn of 2001, long-term contract costs comprised a larger share of DWR's total purchases. The percentage of DWR's long-term contract costs overtook that of shorter-term purchases in September 2001. (Weil, 41 RT 6193:16-22; see also Exhibit 151-A, Stern.) While DWR's shorter-term purchases in the early months of 2001 had different terms than later long-term contracts, but they served a similar purpose in supplying the joint needs of the customers of all three utilities. (Weil, 41 RT 6190:26-6191:23.)

Accordingly, we find no basis to allocate the fixed and short-term purchases of DWR on different bases. We shall therefore apply a pro rata statewide average allocation basis to DWR's revenue requirement in relation to the net short position of each utility, with adjustment for the utility-specific impacts, as noted in Section f. below.

d. Allocation Based on Monthly Versus Hourly Cost Data

SCE has proposed that the DWR revenue requirement be pro rated based upon cost data disaggregated into hourly increments. Since hourly DWR cost data is not currently available to parties, SCE proposes an interim allocation based upon monthly net short data, with provision for a true-up using hourly data once DWR makes it available. SCE argues that anything less precise than hourly data will not provide for an accurate allocation of costs.

PG&E opposes the proposal for hourly allocation of data, arguing that it is too administratively complex and burdensome, and offers only a false sense of precision. The use of hourly cost data would entail maintaining 720 separate hourly cost reports per month. PG&E claims that if the hourly data is not well maintained, the cost allocation controversies over the hourly data will be endless.

SDG&E agrees in principle with the goal of precision that SCE seeks to achieve with hourly allocation. SDG&E questions, however, the practicality of implementing an hourly allocation given the complexities involved. SDG&E witness Croyle also observes that if all load in a block contract is priced at the same price in every hour, it is not necessary to allocate costs across the individual hours. The same result is obtained by allocating the cumulative energy among the utilities in aggregate.³⁹ PG&E likewise argues that hourly data would not provide a true reflection of cost causation for that hour because contracts typically use an average price for power provided across several hours of a day, perhaps for many days across months, seasons, and even years. In instances where a contract price averages the on peak and off peak prices, the average hourly price in the contract causes on-peak costs to be understated, and off-peak costs to be overstated. Thus even an hourly allocation approach would not capture the true avoided costs for each on-peak or off-peak hour, and the resulting hourly allocations would not give an accurate picture of actual hourly cost causation. Furthermore, PG&E argues that such an hourly allocation would not send price signals that could be relied upon to ensure efficient statewide dispatch of power resources.

In theory, we agree that the use of hourly cost allocations could provide more precise measures of cost causation as a basis for revenue allocation as contrasted with monthly cost data. Even if the hourly prices in DWR's contracts may be constant over several hours or reflect an average of on-peak and off-peak avoided costs, an hourly allocation would still more accurately correspond the net short position of each utility which varies on an hourly basis. An hourly weighting of the each utility's net short position would provide a more precise weighted average for cost allocation than would a monthly average. Although DWR has expressed a willingness to provide the requisite data needed to make the necessary hourly allocations, the data has not been provided for the record at this point. Accordingly, it remains uncertain as to how problematic it would be to obtain the necessary hourly data by each utility, and to agree upon its accuracy and reasonableness. We are not persuaded at this time that an adequate case has been made that the potential administrative complexities, litigiousness, and burden associated with an hourly cost allocation are offset by the potential for more precise measurement of cost causation in allocating DWR revenues.

Accordingly, we shall not make a final judgment regarding the use of hourly cost data for allocation purposes for prospective DWR allocations. We shall provide SCE or any other party the opportunity to make a further showing in the next DWR update proceeding. By that time, hopefully, DWR would have made available the requisite data, and parties will be able to provide a more empirical analysis about the practicalities of computing hourly-based allocations. For purposes of this order, we shall use monthly data for determining the allocations, but shall leave open the possibility of prospectively allocating DWR costs based on hourly data in the event we subsequently determine to use such data in a future proceeding.

e. ORA's Averaging Approach Criterion

We decline to adopt the averaging of two mutually contradictory approaches proposed by ORA for allocation purposes. Although ORA seeks to incorporate the purported advantages of two opposing allocation methods, ORA also imports the attendant disadvantages of each method. Moreover, ORA's method further complicates the issue by introducing a new allocation variable, namely, the percentage of weighting to assign to each of the two opposing methods that ORA uses. ORA provides only an anecdotal comparison of the relative merits of the two methods, but offers no quantitative rationale why a 50/50 weighting of the two alternatives is preferable to a 25/75 weighting, or some other weighting. Because ORA provides no basis to conclude that the comparative net advantages of each of the two methods are equivalent, its 50/50 weighting appears to be arbitrary. We conclude that whatever method is adopted, it should be based upon a consistent set of allocation principles and assumptions. ORA's method does not fit this criterion. We therefore decline to adopt it.

f. Adopted Allocation Methodology

In view of the various criteria considered above, we conclude on balance, that the pro rata statewide allocation approach offers the most objective, equitable and economically defensible methodology. Both PG&E and TURN have offered different allocation calculations based generally on the pro rata (postage stamp) approach to allocation. Of these two proposals, we conclude that TURN's is preferable in that it takes into account certain utility-specific adjustments that reflect more specifically the costs related to each utility. No party provided persuasive arguments as to why those adjustments should not be adopted. We find those adjustments promote a more cost-based allocation and reflect cost causation. Accordingly, we adopt those adjustments.

(1) Total Net Short Versus Retail Net Short

TURN proposes an adjustment to provide for a more consistent definition of net short between recorded versus forecasted costs. The DWR model calculates net short on a recorded costs basis reflecting the total net short, but omits certain components to derive something closer to a retail net short on a forecasted cost basis. In particular, DWR's definition of retail net short excludes two items included in the total net short: line losses and PG&E's purchases on behalf of WAPA.⁴⁰ TURN argues that for the sake of consistency, total net short should be used to allocate the DWR revenue requirement for the entire period, rather than using total net short for part of the period and retail net short for the remainder.⁴¹ DWR assumes that losses are different among the three utilities in its calculations of total utility load and total net short. By using retail net short, the DWR's postage stamp method prevents these differences from being considered in the cost allocation.⁴²

In particular, TURN is concerned that DWR has not properly treated the WAPA-PG&E contract. DWR's use of retail net short would allocate to other utilities the costs which PG&E incurs to serve WAPA.⁴³ This is inconsistent with the specific terms of the WAPA contract. The contract is treated as part of PG&E's retail customers' obligation as an ongoing purchased power contract. Prior to restructuring, it was included as a retail cost in PG&E's Energy Cost Adjustment Clause (ECAC) proceedings. During the transition period under AB 1890, the contract has been included as a purchased power cost in PG&E's Transition Cost Balancing Account (TCBA) accounts. In other words, PG&E's retail customers buy the power to serve WAPA and receive the revenues from WAPA (which do not cover the full cost of the power that is purchased). In short, TURN argues, WAPA is a PG&E retail contract, and therefore any calculation of retail net short for PG&E should not reflect any adjustment for WAPA.

We agree that TURN's adjustment leads to a more consistent definition and application of net short. We shall therefore adopt TURN's proposal to use total net short (subject to the further adjustments described below) to allocate DWR costs among utilities.

(2) Helms Plant Adjustment

TURN proposes an adjustment to recognize the role of PG&E's Helms Pumped Storage Plant in providing the proper economic incentives to maximize overall efficiency. A pumped storage generation resource, such as Helms, is a net consumer of energy due to the inefficiency of pumping. The underlying premise is that by consuming electricity during off-peak periods when it is relatively cheap, then producing electricity during peak periods when it is relatively expensive, there is a net benefit to the plant's operator and, by extension, to the ratepayers bearing the operating costs in regulated rates.

The underlying premise has been undermined, at least since January 17, 2001. PG&E continues to use off-peak energy to pump Helms in order to generate during on-peak periods. However, the utility is charged the full average net short rate as the cost of the off-peak energy, as well as the on-peak energy. The result is that the economic signals associated with operating the plant are skewed, and ratepayers are required to pay excessive amounts associated with the plant's operations. Under the current conditions, PG&E's ratepayers would be better off had the Helms plant not been used at all for the past 11 months, even though the state as a whole would have been worse off.⁴⁴

Accordingly, TURN proposes the following revenue allocation adjustments to avoid penalizing PG&E customers by eliminating that loss that would result from applying average costs to Helms-related consumption and production:




a) Reduce PG&E's net short loads in the January-June 2001 period based on recorded monthly operations of Helms. This requires subtracting both Helms-related generation output and pumping electricity consumption; in other words, treating Helms as if it did not exist. This will save PG&E ratepayers from bearing higher costs for Helms' operation during a period when the plant's operation was necessary to prevent blackouts.




b) A similar adjustment is proposed for the consumption and output of Helms from July 2001 and forward. However, because DWR's data are inadequately disaggregated to calculate the forecast operation of Helms, TURN proposes that Helms pumping be subtracted from PG&E loads and Helms generation be subtracted from PG&E generation when truing up any balancing account entries to actual net short kWh and allocated costs starting in July 2001.

One of these adjustments is a credit to PG&E to reflect the benefits of the Helms pumped storage facility. PG&E also computed a similar adjustment for Helms, though the calculation is different (compare Ex. 166 and 171). Both PG&E and TURN seek to make PG&E indifferent in regard to the operation of Helms. Exhibit 166, which illustrates PG&E's adjustment using non-confidential hypothetical numbers, describes a series of computations whereby a credit reflecting the pumping losses is made to PG&E, resulting in Helms revenues and costs being equal (See PG&E/Alvarez, RT 6326).

TURN's objective is to completely remove the effects of Helms from the PROSYM outputs, as shown in Exhibit 171. PROSYM is an hourly chronological production cost model, and fully capable of dispatching Helms in the most efficient way. Access to PROSYM would have allowed parties to determine precisely the net benefit of Helms and assure that it is allocated to PG&E rather than relying on methods which merely make PG&E indifferent.

We shall adopt the adjustments for Helms, proposed by TURN in order to promote more economically efficient price signals in the operation of the Helms facility. The TURN approach seems more straightforward to implement than that of PG&E, and allows for prospective true-ups.

In its testimony, TURN also states that a similar treatment should be adopted for the pumped storage resources of SCE at Balsam Meadows. In the interests of consistency, we agree. The same efficiency considerations that warrant the adjustment for the Helms Pumped Storage likewise apply to the Balsam Meadows pumped storage resources. We shall accordingly authorize a similar adjustment to reflect the efficiencies of the pumped storage resources of SCE at Balsam Meadows. Because no party provided sufficiently detailed information to the record to provide for a recalculation of the allocation ratios to reflect a Balsam Meadows pumped storage adjustment, we shall defer making any numerical adjustment in this order. Instead, we shall provide for parties to make a true up adjustment at the next update proceeding for DWR to reflect the effects of the pumped storage resources at Balsam Meadows on the same basis as we have adopted for Helms, effective from January 17, 2001, forward.

In its comments on the proposed decision, SCE argued that if an adjustment is to be made for Helms pumped storage, a similar adjustment should also be made for its exchange agreements, claiming they operate similarly to pumped storage operations insofar as on-peak energy is taken in exchange for a greater quantity of off-peak energy. We decline to include an adjustment for exchange agreements similar to that applied for pumped storage facilities as argued by SCE. SCE's exchange agreements provide fixed requirements, as noted by TURN, and do not have the economic dispatch issues connected with pumped storage. Accordingly, the rationale supporting the allocation adjustment for pumped storage facilities does not apply to SCE's exchange agreements.

We direct the utilities to work with DWR to maximize coordination and ensure that pumped storage facilities and other hydro resources are operated and dispatched efficiently.

(3) Adjustment to Reflect Differences in Self-Provided Ancillary Services

TURN proposes that the allocation be adjusted to recognize differences among the utilities in the amount of ancillary services that each utility provides for itself. There has been a major downward decline in the ancillary services market in recent months, such that the vast majority of ancillary service costs incurred by DWR occurred during the first two quarters of 2001. Self-provision of ancillary service costs was less of a factor during that period.

TURN provided a table showing its proposed adjustment factors relating to self-provision to be applied to gross load. TURN used these adjustment factors to allocate ancillary service costs in Ex. 170. TURN proposes the following adjustment percentages to apply to gross load, relying on data received from the utilities regarding the total system ancillary services self-provided by each utility:⁴⁵

January-June July-December

PG&E 51% 20%

SCE 75% 64%

SDG&E 100% 100%

As recommended, we shall apply TURN's recommended adjustments by total gross load for the first two quarters of 2001, and by adjusted gross load for the period from the third quarter of 2001 to the end of 2002 to reflect the greater self provision starting in July 2001.

(4) Effects of Lag in Cash Payments in
January 2001

As noted above, SDG&E has proposed a reduction in its allocated costs of $65 million to adjust for the effects of DWR lead/lag differences between expense accruals and cash payments. PG&E objects to SDG&E's $65 million adjustment, arguing that it improperly allocates a benefit to SDG&E for the month of January 2001, at the expense of PG&E and SCE, even though DWR did not purchase any power for SDG&E during the month of January 2001. PG&E argues that SDG&E is therefore not entitled to a $65 million credit since it had no transactions with DWR, either positively or negatively during January 2001.

We conclude that SDG&E's adjustment of $65 million to correct for DWR's lead/lag accrual to cash has merit. While it is true that DWR did not purchase any power on behalf of customers of SDG&E during January 2001, DWR did begin purchasing for SDG&E customers beginning in February 2001, and thereafter. Accordingly, while SDG&E did not participate in the savings associated with the lag in cash payments that reduced revenue requirements in January 2001, it did bear the burden of the additional cash payments that increased the revenue requirements in subsequent months when the effects of the January 2001 lag reversed. The acceleration (i.e., "lead") in cash payments in months subsequent to January 2001 thus offset the delay (i.e., "lag") in payments that had been experienced in January 2001. As testified by SDG&E witness Croyle, SDG&E customers were thereby being allocated costs to repay the benefit that SCE and PG&E realized in January 2001.⁴⁶ If SDG&E is to share in the burden of the increased cash requirements needed to pay for the lead in cash expenditures in subsequent months, it is equitable to allocate a share of the offsetting savings from the related lag that had occurred previously in January 2001.

The fact that DWR did not purchase power on behalf of SDG&E customers during January 2001 thus does not negate the fact that the January lag in cash payments were related to subsequent months' payments that were charged to SDG&E customers. Accordingly, we shall adopt SDG&E's proposed credit of $65 million for lead/lag adjustments as an offset to the payments to DWR after January 2001 to recognize the timing differences between the provision of services and DWR's receipts of cash for them. The adjustment results in a $65 million reduction in allocation to SDG&E and an increase in the allocation to PG&E and SCE of $36.397 million and $28.603 million, respectively.

(5) Effects of Differences in First Quarter 2001 Allocation for SDG&E

We decline to adopt SDG&E's proposed $51 million adjustment for the first quarter of 2001. SDG&E claims that DWR erroneously calculated the allocation for SDG&E customers during the first quarter of 2001 as 14% when it should have been 12%. PG&E disputes this adjustment, arguing that the $51 million reduction is biased and selectively changes parameters in certain months while ignoring offsetting changes in other months that would have tended to increase SDG&E's allocation.

We find that SDG&E's proposed $51 million adjustment fails to reflect the correct relationship of SDG&E sales to total DWR sales over time in a consistent manner. SDG&E does not dispute that, on an annualized basis, DWR correctly calculated SDG&E's percentage as 14%. SDG&E's dispute is essentially whether the percentage allocation for the first quarter of 2001 should be separately calculated for each month. SDG&E derives a 12% allocation for the first quarter of 2001 only by including the month of January 2001 in the denominator in calculating the allocation percentage. Yet, such an approach overstates the denominator by including sales from January 2001. Since DWR made no purchases for SDG&E in January 2001, there is no basis to include January sales in determining a quarterly allocation percentage for SDG&E for sales that were only made in February and March.

The inclusion of January 2001 sales in SDG&E's calculation of the first quarter allocation merely serves to inflate the denominator (and artificially understate the quarterly percentage) with an extra month of sales that had nothing to do with SDG&E. When SDG&E's sales are considered as a percentage of the total for each separate month of 2001, it can be seen that DWR's allocation correctly reflected the share of costs applicable to SDG&E customers. Thus, for January 2001, SDG&E is correctly allocated 0% of total sales. In February 2001, the monthly allocation is 14%. Likewise, subsequent months are correctly calculated. Thus, the $51 million adjustment proposed by SDG&E has no sound basis, and is accordingly rejected.

(6) Adopted Revenue Allocation Amounts

Consistent with thee adjustments adopted above, the resulting allocation of DWR revenue requirement and associated percentages that we adopt are as follows:

$000's

Utility Revenue Allocation % Allocation

PG&E	$ 4,327,511 47.8%
SCE	$ 3,373,764 37.3%
SDG&E	$ 1,344,187 14.9%
Total	$9,045,462 100%

²⁰ Kuga Direct Testimony (Ex 163) pp. 1-3, quoting DWR memorandum to CPUC dated August 9, 2001. ²¹ In a memorandum from Peter Garris, DWR's Acting Deputy Director, to Commissioner Geoffrey Brown, dated November 28, 2001, DWR committed to providing the hourly cost information to the Commission on a timely basis. ²² SDG&E/Croyle, Ex. 153, p. 7. ²³ See Ex. 163, pp. 1-3; quoting DWR memorandum to the CPUC, August 8, 2001; Response to Comments on DWR Revenue Requirements, p. 3. ²⁴ See 43 RT 6371-6372/Marcus. ²⁵ See Ex. 163, pp. 1-2; footnote 2. ²⁶ Ex. 159, Barkovich Testimony; p. 8. ²⁷ See, 41 RT 6126, lines 2-9 (and earlier comments on 6125).

²⁸ See, Exh. 157, 41 RT 6105, lines 5-15.

²⁹ See excerpt from DWR's response to SCE's Data Request SCE-01, as cited in Exhs. 155 and 157. ³⁰ SCE, Stern, 39 RT 5902. ³¹ See DWR's Response to PG&E Data Request 34, as referenced in Ex. 163, footnote 2. ³² See DWR's Response to Data Requests dated August 1, 2001, as cited on page 10 of CLECA Ex. 159. ³³ See Ex. 159 (Barkovich) pg. 4 footnote. ³⁴ See Ex. 160, Weil Testimony, page 4. ³⁵ Ex. 157, p. 4. ³⁶ D.01-05-064, mimeo, p. 18. ³⁷ See Ex 163, pp. 2-3. ³⁸ See Exhibit 151-A, Stern; the exhibit is confidential, but the cited fact is not. (See Stern, 40 RT 5966:7-10.) ³⁹ SDG&E, Croyle, Tr. Vol. 40, p. 6003. ⁴⁰ Marcus Direct Testimony (Ex. 169), pp. 6-7. ⁴¹ 43 RT 6369, witness Marcus. ⁴² In making this point, TURN is not stating that it specifically agrees with the loss factors used by DWR; rather, that if those different loss factors are used as part of the load forecast that underpins the DWR revenue requirement, then they should also be included in the load forecasts used to allocate costs among the utilities. Any differences between actual and forecast losses would be captured along with other differences in load, when truing up each utility's cost responsibility. ⁴³ Ex. 169, p. 6. When he appeared to testify in support of his prepared testimony, Mr. Marcus noted that DWR may have fixed this problem in its November 5, 2001 presentation of revenue requirement and inter-utility allocation. 43 RT 6374-75. ⁴⁴ Ex. 169, pp. 7-8. ⁴⁵ These figures are 100% minus the percentage of total ISO ancillary services (excluding self-provided marketers and municipals) self-provided by each utility, with PG&E set at 20% to reflect that it does not self-provide 100% of all services. (Ex. 170.) ⁴⁶ 40 RT 6070, SDG&E/Croyle