XIV. Modeling of Ongoing DWR "Indifference" Costs

DWR/Navigant computed the costs assignable to DA for the uneconomic portion of ongoing net purchase costs for the DWR portfolio of contracts (consisting of both contract and spot purchases) for the time period October 2001 through 2010.⁶² DWR describes these costs as: "(1) the net change in operating costs of the DWR contracts, i.e., costs of power purchased minus the resale value of any excess power; and (2) the portfolio effect of averaging fixed cost power from contracts with spot market purchases." (Direct Access Exit Fee Scenario Analysis in Support of Rulemaking 02-01-022, May 17, 2002, p. 2.)

The average cost of net short power to bundled customers is calculated separately for the July 1 and the September 21 DA cut-off cases. In each case, production costs for bundled load were determined using ProSym to dispatch utility-retained generation (URG) and DWR contracts to meet hourly loads. When the bundled customer loads exceed the URG and contracts, the model assumes power is purchased at spot market prices. When must-run URG and contracts exceed bundled customer loads, the excess power is sold in the market. Forecasted administration and general fixed costs are added to net power purchase costs to get total costs.

The increase in average cost of net short power to bundled customers (comparing the July 1 and September 21 cut-offs) is the amount of revenue required from DA customers if the net short power costs to bundled customers are not to increase.

There are two major groups of differences between the scenarios. The original analysis was based upon the DWR's revenue requirements underlying D.02-02-052. One set of scenarios illustrates the impact of updating assumptions and data to reflect changes since the DWR filed its revenue requirements underlying D.02-02-052. Scenario 1 reflects changes in generation, load forecasts, DA percentages, transmission and distribution losses, gas prices, and updates through late April 2002. Scenario 8 includes the effect of the renegotiated contracts.

The second set of scenarios reflects the sensitivity of the DA surcharge to various factors and assumptions as specified by parties at the workshop. These simulations were intended to provide parties with a quantitative data set as a basis to perform their own analysis and present testimony regarding the appropriate basis for computing DA CRS.

The data underlying the base case drew upon the DWR 2001/2002 revenue requirement implemented by D.02-02-052⁶³ for the period January 17, 2001 through December 31, 2002, allocated among customers in the service territories of the three utilities.

Of the various modeling scenarios performed by Navigant, parties basing their analysis on Navigant's modeling generally support Navigant's Scenario 8 as providing the most accurate basis for determining the applicable portion of DWR costs applicable to a CRS.

The longest DWR contract ends in 2013, although the vast majority of the contracted energy expires by the end of 2011. Most of the DWR/Navigant results are based on a 20-year recovery period, because this is the expected term of the bonds. The length of the period has a significant impact upon the level of the DA CRS. For example, for Scenario 8 the surcharge for 10-, 15- and 20-year recovery periods as calculated by DWR/Navigant is as follows:

Years DA CRS ($/MWh)

20 $25.79

15 $30.45

10 $40.09

A. Areas of Dispute Relating to Forecasts

Henwood disagrees with Navigant in two forecasting major areas. The first key difference is in the assumption about new generation construction in WSCC and California in the next several years.

1. Assumptions Regarding New Generation Additions

The level of uneconomic costs is sensitive to assumptions concerning the nature and extent of new base load plant coming online because more available generation translates to lower power prices, which in turn yields higher DWR shortfalls, leading to a higher DA CRS. (See Multiple Parties⁶⁴/Lauckhart, Exh. 31, pp. JRL-4 - JRL-5; CIU/Chalfant, Exh. 32, pp. 12-13.) The opposite is the case if less generation is available. DWR conceded the first day of hearings that it needed to remove 2,331 megawatts of planned capacity from its modeling. (DWR/Schiffman, Exh. 2, pp. 8-9) The impact of such a removal is higher generation prices and lower DA CRS. (DWR/Schiffman, 7/10, p. 64.) Further, it is not at all clear that these removals from DWR's assumed new generation cover the field. In Lauckhart's opinion, power plant construct will be delayed beyond the date in the Navigant model. (Multiple Parties/Lauckhart, 7/16, pp. 660-661.)

Navigant, in its modeling, assumed that significantly more new base load generation will be built than did Henwood. The Navigant assumption regarding new base load generation results in lower market clearing prices than does the Henwood approach. Market clearing prices thus drop to the point that new highly efficient power plants are not able to earn enough revenue to cover operating costs plus fixed O&M, or to provide any contribution to debt service and other fixed costs of the new power plant. Henwood claims that this assumption by Navigant is simply not credible, and that power plants will not be financed and built under these assumptions.

2. Assumptions Regarding Market Price for Surplus Power Sales

The second area of disagreement between Navigant and Henwood relates to the market-clearing price for sales of surplus power. Navigant assumes that the price that DWR will get for this sale is 50% of prevailing spot market prices for the hour of sale. Navigant provides actual DWR historical buy and sell data that shows sales prices are 50% of purchase prices. Opposing parties argue that the assumed spot price sale should be set at 100% of prevailing spot prices. DWR indicates that the impact of this price differential is to reduce the DA CRS by 0.286 cents/kWh.⁶⁵

Henwood challenges this assumption, however, arguing that the DWR purchases are primarily in heavy load/high priced hours while DWR sales are primarily in light load/low priced hours. This fact would indicate why spot sale prices by DWR will be lower than spot purchase prices. Since Henwood is forecasting spot prices hourly, it does not believe it is reasonable to take a low spot price in light load hours and then assume that DWR could sell any surplus at only one half of that low price. Furthermore, if Navigant assumes that DWR buys and sells power in the same hour and that sales prices are 50% of purchase prices in that same hour, it may well be that DWR would be making its sales at the hourly spot price, but that the purchases are being made at twice the spot price. While Navigant assumes that purchases are made at spot prices and sales made at 50% of spot prices, their analysis could also lead to the conclusion that DWR sales are made at spot prices while DWR purchases are made at two times spot prices.

Navigant modelers reduced the spot price projection by comparing not just spot prices to spot prices, however, but also spot prices to balance of month, weekly, quarterly, and long-term sales. (DWR, McDonald, 7/11, pp. 176, 191-195.) DWR admits that comparing such different products limits the usefulness of comparisons because products are being mixed. (DWR/McDonald, 7/11, pp. 191-92.) Another DWR witness agreed that to include such a mixture of products in reporting power sales, as is done in Table 1 of Exhibit 1 in the testimony of Christopher Smith, would be improper. (DWR/Smith, 7/10, p. 24.) Had DWR used a proper apples to apples comparison it would have found that the appropriate relationship was 100%.

Thus, parties argue that the Commission should therefore reject the 50% discount off the PROSYM forecasted price proposed by DWR. CIU urges the Commission to assume that power will be sold off system at 100% of the PROSYM forecasted spot price. (See CIU/Chalfant, Ex. 33, pp. 7-8.)

Discussion

The disputes over the validity of the Henwood versus Navigant modeling forecasts must be addressed in the context of how modeling data is to be used in this proceeding. Navigant only presented its data as illustrative. The modeling conventions presented in this proceeding involves highly complex and sophisticated simulation techniques. As noted by TURN witness Marcus, today's computer models require the estimation of all the parameters on the cost of the existing system and forecasts of fuel prices the forecast of new generation to be added in the Western U.S. Because generation and price are interlinked, it had become difficult to forecast, and relatively small changes in generation can result in relatively large changes in price.

Another, even more controversial parameter involves the simulation of bidding behavior, including ways in which bidders will not follow economic theory and will bid above marginal variable costs. A modeler must choose an expected capacity withholding and bidding strategy and will obtain a different market price depending on which strategy is chosen.⁶⁶

Within the caveats of the complexities of the assumptions underlying models such as Navigant's and Henwood's, we must determine to what extent we must rely on such models. We conclude that Henwood's assumptions regarding new generation additions appear more convincing than the assumptions made by Navigant. DWR/Navigant offered no substantive arguments to refute the alternative new generation additions assumed by Henwood, but only notes that any difference in new generation assumptions is not the sole or even primary cause of cost differences. Nonetheless, we recognize the forecasts are only as good as the underlying assumptions made. If those assumptions prove wrong in the future, the underlying forecasts will be wrong.

We have similar concerns as to the reliability of assumptions as to prices for surplus power as off-system sales. Henwood did not present convincing affirmative evidence that surplus power will necessarily be able to be consistently sold at full price.

Likewise, we believe that DWR/Navigant's assumption that such surplus sales can only yield a price discounted by 50% of the market price is unduly pessimistic. While Navigant based its 50% assumption on recorded transactions, recorded experience is not necessarily indicative of future results. Particularly once the utilities take over administration of the DWR contracts, there is reason to believe that a higher price can be realized on surplus power sales than has been DWR's experience up until now. DWR agrees that it is reasonable to presume that when the utilities take over the function of selling power and more players thus become involved, the market will tend toward more efficient operation. (DWR/Schiffman, 7/10, 75.) DWR admitted that it was reducing the spot price projection by comparing not just spot prices to spot prices but also spot prices to balance of month, weekly, quarterly, and long-term sales.⁶⁷ DWR admits that comparing such different products limits the usefulness of comparisons because products are being mixed.⁶⁸ Another DWR witness agreed that to include such a mixture of products in reporting power sales, as is done in Table 1 of Exhibit 1 in the testimony of this witness, would be improper.⁶⁹ We believe the most reasonable estimate, given the uncertainties involved, favors an off-system sales price closer to 100% than to 50%. For purposes of this order, however, it is not necessary to adopt a precise off-system DA CRS market price, since we are not relying on long-term forecasts to set CRS.

Under our approach to be consistent with the modeling underlying the implementation of the DWR revenue requirement in A.00-11-038 et al., we shall direct that the assumptions underlying off-system sales and new generation additions underlying the adoption of the overall revenue requirement be applied on the same basis in modeling the DA in/out scenarios. Since under the Water Code,⁷⁰ the overall level of the revenue requirement is determined by DWR, and not the Commission, we do not have the authority to adopt an overall DWR revenue requirement that is inconsistent with that determined to be "just and reasonable" by DWR. While we exercise authority to determine the manner in which those charges will be allocated between bundled and DA customers, there must be consistency in the treatment of DWR resource assumptions in determining the cost responsibility of DA customers.

If we were to apply resource assumptions solely for modeling DA cost responsibility that were inconsistent with those applied by DWR for charges applicable to bundled customers, we would produce a mismatch of charges applicable to bundled versus DA customers that would not yield a 100% collection of overall DWR revenue requirements. Instead, the use of inconsistent revenue requirement resource assumptions applied to DA versus bundled customer charges will lead to either gaps or double-counting in the collection of DWR power charges among bundled and DA customers. Accordingly, while we encourage DWR to take into account the findings of this decision with respect to the merits of Henwood's modeling assumptions, we ultimately recognize that DWR has ultimate responsibility for determining its overall revenue requirements and associated modeling assumptions. In any event, any forecast error regarding resource assumptions will only have temporary effects given the process that has been established for true ups of forecast data to reflect actual experience.

B. Categories of Costs to be Excluded to Measure "Bundled Customer Indifference"

1. Exclusion of 130% of Baseline Quantities

Parties' Positions

SCE and other parties representing DA interests disagree with the DWR/Navigant indifference calculation which excludes exempted load (i.e., usage below 130% of baseline by residential customers) in computing the applicable DA CRS unit cost assigned to DA customers relating to the DWR historic undercollection. The exempted load that DWR excludes was exempted by the Commission, pursuant to AB 1X, from the allocation and rate design for the 3¢/kWh surcharge adopted in D.01-05-064. AB X1 required that residential customers' usage below 130% of baseline not be made subject to any increases in electricity charges. The revenue shortfall was assigned to remaining bundled customers. If additional revenue shortfalls result from Commission adopted policies, these parties argue that the Commission will decide how to allocate and collect them. Therefore, SCE and these other parties argue that no adjustments for this exempted load should be made for purposes of DA CRS.

ORA and TURN support the DWR approach, however, inasmuch as the loads over which the bundled rate surcharge was calculated were those loads over 130% of baseline. A significant portion of the costs in excess of 130% of baseline were allocated to the commercial and industrial classes. They argue that under CIU Witness Chalfant's proposal, DA customers would escape those costs, even though they would be paid by bundled service customers in those same classes.

PG&E argues that residential usage below 130% of baseline should not be excluded from the indifference charges. PG&E argues that since overall electricity charges do not change, references to Water Code Section 80110 are not relevant.⁷¹ Moreover, if and when the Commission moves to "bottoms-up" charges for these customers, and if the result would otherwise be an increase in electricity charges for residential usage below 130% of baseline, then the Commission must address whether residential electricity charges must be modified because of Water Code Section 80110. Absent these conditions, however, PG&E claims there is no basis for excluding residential usage below 130% of baseline from the non-bypassable charges being considered in this proceeding.

Discussion

In its modeling scenarios, DWR's calculates the total shortfall during January 2001 through September 2001, then allocates that amount between DA and bundled customers, based on the percentage of load each placed on the system during that time period. In allocating the shortfall between DA and bundled customers, DWR excludes residential load below 130% of baseline. This allocates more of the shortfall to DA. The result is an allocation of just over $1 billion of the January 2001 - September 2001 shortfall to DA customers.

Because the treatment of that revenue shortfall and effects on the residential load are being addressed through issuance of bonds, this is a matter that has been addressed in the Bond Charge Proceeding (A.00-11-038 et al.).⁷² Thus, we consider the issue of whether the 130% of baseline quantities is to be included or excluded from the calculation of DA cost responsibility is moot for purposes of this proceeding.

2. Long-Term Contract Only Versus Incremental Short-Term Costs

Parties' Positions

SDG&E identifies DWR "stranded" costs only as including long-term contract costs (net of revenues from surplus sales) and associated financing costs incurred on behalf of DA load that left bundled service after July 1, 2001. SDG&E argues that the inclusion of spot market purchases increases the DA CRS because spot prices are substantially lower than long-term contract costs. With less bundled load as a result of the DA migration after July 1, 2001, the share of low-cost spot purchases in the DWR portfolio drops and the high-cost long-term contracts weigh more heavily in the smaller overall portfolio. DA customers should be responsible for what was incurred on their behalf and not for costs incurred (or not incurred) after they commenced taking DA service from ESPs. SDG&E argues that costs for spot market purchases made after these customers have departed, by definition, could not have been made on their behalf. Costs that were not or will not be incurred cannot be stranded. SDG&E argues that to include any costs other than long-term contract costs results in a cross-subsidy of bundled customers by DA customers.

Other parties (e.g., PG&E, SCE, DWR, ORA, and CLECA) propose to include not only long-term contract costs, but also spot market and fixed costs, that is, all DWR costs incurred. They argue that the calculation must include not only DWR's long-term contracts, but also the assumed purchases to meet the remainder of bundled customers' loads in order to achieve indifference.⁷³ Under current circumstances, these remaining purchases are likely to be the least expensive, on average. The increase in DA displaces this lower cost power out of the bundled portfolio.

Discussion

We conclude that it is appropriate to include short-term contracts in the indifference calculation to capture the "squeeze-out" effects identified by PG&E and others. If the effects of this "squeeze out" of lower cost power are not included in the calculation, bundled customers are not indifferent to the increase of direct access after July 1, 2001. They lose the benefit they would have received from having this lower cost power make up a substantial amount of power used to serve them.

Since this power is the marginal source that is squeezed out by the increase in DA above the July 1, 2001, level, the parties argue that it must be included in the calculation. This is so regardless of whether DWR purchases the power, as is the case currently, or the utilities' purchase the power, as may be the case after January 1, 2003.

In its initial testimony in this proceeding, SDG&E argued that only DWR long-term power should be taken into account in calculating the DWR-related DA CRS. SDG&E's calculation thus did not take into account the "squeeze out" effect just described. However, after cross-examination by SCE and TURN isolating and illustrating this squeeze out effect,⁷⁴ SDG&E conceded that its initial approach had to be "clarified," and that its DWR calculation had to be modified. We conclude that not only DWR's long-term contracts, but also the marginal price of short-term purchases to meet the remainder of bundled customers' loads, must be taken into account in order to accurately calculate bundled customer indifference.⁷⁵

3. Administrative & General Costs

Parties' Positions

Certain parties argue that DWR's A&G cost should not be included in the calculation of Indifference Costs. CIU argues that because the costs are fixed, the incremental direct access load is not responsible for any of these costs. SDG&E similarly argues that because fixed costs that do not change whether or not customers switch to DA or remain bundled, such costs should not be allocated to DA customers. Other parties disagree arguing that the increase in DA customers between July 1 and September 20, 2001 would result in fixed A&G costs being allocated to the fewer remaining bundled service customers. Without allocating a portion of the A&G costs to the incremental DA customers, the remaining bundled service customers would be forced to pay an increased billing amounts to cover these costs.

Discussion

We find that fixed A&G costs should be included in the calculation to produce bundled customer indifference. As noted above, by excluding fixed A&G costs from the calculation of DA CRS, these costs are entirely absorbed by the remaining bundled customers. Because fixed costs are being spread over a reduced base of bundled customers, the result is an increased per-kWh cost to bundled customers. Consequently, because their per kWh cost would increase bundled customers are not indifferent to the exclusion of fixed A&G costs. DA customers' costs responsibility is being determined on an "indifference cost" principle rather than an avoided cost methodology.

C. Utility-Specific versus Statewide Surcharges

1. Parties' Positions

SDG&E, together with certain DA parties, propose that the Commission adopt a uniform statewide-levelized charge for the DWR component of the CRS, based on the Commission's adopted revenue allocation for long-term DWR contract costs. SDG&E believes that maintaining DA CRS on a statewide basis offers greater certainty and stability to DA and bundled customers throughout the state. SDG&E's proposal also moderates the impact of the DA CRS, thus keeping DA an economically viable alternative, consistent with the Commission's stated goal.⁷⁶

SDG&E proposes an initial DA CRS of 1.22 cents/kWh, based on a 15-year levelized annual cost, utilizing DWR/Navigant's Scenario 8, averaged across the three utilities. SDG&E computes an equivalent utility-specific DA CRS of 2.76 cents/kWh, which would be more than twice as much as PG&E's 1.1 cents/kWh. Under a nonlevelized approach, SDG&E's 2004 DA CRS would be 5.5 cents/kWh compared with only 2.2 cents for PG&E.

SCE, CLECA, and PG&E, among others, advocate utility-specific DA CRS for DWR costs. These proposals include allocating spot market purchases zonally⁷⁷ and separate capped DA CRS for each utility.⁷⁸ SCE's proposal to allocate spot market purchases zonally results in higher DWR charges for SDG&E than for SCE and PG&E. This occurs because SDG&E has a relatively higher proportion of net short compared to DA load.

Under these proposals DA CRS will differ by utility either by level or duration. Where charges are either higher in one service territory versus another or applied for a longer period of time, SDG&E argues that there is an inherent inequity in the DA market where the tradeoff of DA for bundled customers in one service territory is likely a more viable alternative than for bundled customers in another.

SDG&E argues that customer-specific DA CRS, even though more cost based, only increase the level of instability and uncertainty for DA and bundled customers. This is a result of the significant variability in DA CRS that can occur across customer classes, particularly with CEC's proposal that offers customer-specific rates and different amortization periods. Regulatory objectives often must balance efficiency with simplicity, fairness and other considerations. In addition, SDG&E argues that customer-specific DA CRS are costly to implement and unworkable.

In its modeling runs, DWR/Navigant only developed utility specific DA CRS in Scenario 5 which allocates DWR power to the utilities in accordance with the methodology utilized in the development of DWR's 2001/2002 revenue requirement. That is, long-term contracts were allocated in proportion to each utility's net-short position, and additional spot sales or purchases were made zonally.

CIU also opposes utility-specific charges, arguing that the Navigant modeling is not precise enough to capture all of the relevant utility-specific variables. CIU argues that the imprecision is compounded by the great differences in charges that would result among the three utilities, with SDG&E charges more than 65% higher than those for PG&E.

2. Discussion

We find the arguments of SDG&E and others unconvincing as a basis to adopt a single uniform statewide rate for DWR power charges. The adoption of utility-specific rates is consistent with the manner in which bundled customer electricity charges are set, including charges for large industrial customers that take bundled service. We have already discussed above our reasons for declining to base DA CRS on levelization of long-term forecasts.

We conclude that the most material rationale underlying the proposal for levelized statewide charges is mitigate the effects of an excessively large DA CRS in the SDG&E service territory that would be significantly larger than for PG&E or SCE and that would create a greater risk of making DA uneconomic in the SDG&E service territory. We recognize this concern, but conclude that a more appropriate way of dealing with it is to set utility-specific charges, but mitigate their effects by imposing caps, as we adopt below. Utility-specific charges are more consistent with established principles of cost causation and will be less likely to mask the true cost of service associated with providing service.

⁶² Because the present value of the 2011 cost differential for the July and September cases is minimal and the differential was negative in an earlier version of the model, DWR elected to use 2010 as the end year in its calculation.