3. Greenhouse Gas Modeling of California's Electricity Sector

In June 2007, our consultant E3 began development of a model of GHG reductions in the electricity sector. The work was funded by the Public Utilities Commission and ARB as a component of the State's analysis to inform policy decisions surrounding implementation of AB 32. E3's GHG Calculator calculates the emissions, cost, and rate impacts of different scenarios relative to a Reference Case. The results can also be compared to a Natural Gas Only Buildout scenario, as further described below.

The GHG Calculator is a cost-based, bottom-up, scenario analysis model⁸ of what it would cost seven groupings of California retail providers to achieve different levels of GHG emission reductions between 2008 and 2020, relying only on existing technologies.⁹

In the Stage 1 GHG modeling effort (July 2007 through November 2007), the E3 team modeled the electricity and natural gas sectors assuming a load-based electricity and natural gas sector cap on emissions. Users of the GHG Calculator were able to select among demand-side and renewable energy resources for development, in order to bring GHG emissions in the electricity and natural gas sectors down to a target level in 2020.¹⁰ The principal output of the Stage 1 model included the electricity and natural gas sector cost and rate impacts of reaching the GHG cap by developing the selected resource mix. The model also estimated the incremental cost of GHG emissions reductions resulting from the selected resource mix.

Key Stage 1 Questions:

· How much will various policy options reduce CO2 emissions?

· How will these policy options affect electricity rates?

· Underlying question: At what electricity sector target level do incremental improvements get expensive?

During the Stage 2 GHG modeling effort (February 2008 through May 2008), the E3 team refined model assumptions about retail provider-specific resources to reflect the Energy Commission and Public Utilities Commission recommendations to ARB on GHG regulatory strategies contained in D.08-03-018.¹¹ One of the major changes in the Stage 2 model enables users of the GHG Calculator to select the California-wide price of GHG emission allowances in terms of dollars per metric ton of CO2 equivalent (CO2e) emissions from 2012 - 2020. Users also have a number of other options in the GHG Calculator regarding potential GHG policy regulatory regimes. The GHG Calculator was designed to analyze different sets of rules for the auction or administrative allocation of emission allowances to the electricity sector, and for the use of GHG offsets.

Key Stage 2 Questions:

· What is the cost to the electricity sector of complying with AB 32 under different policy options for California (including different market-based program designs)?

· What is the cost to different retail providers and their customers of these options?

· Underlying question: What option has the best combination of cost and fairness?

3.1. Methodology and Approach: E3 GHG Calculator and PLEXOS

The GHG modeling analysis uses two tools in combination. The spreadsheet-based GHG Calculator was developed by E3 for use by staff and parties to evaluate alternative resource plans that can meet target GHG emissions levels. This simplified tool allows input values to be changed easily with updated results displayed in seconds. In addition, all of the calculations are available to all stakeholders because all of the formulas are provided in the spreadsheet.

The second tool used by E3 is the production simulation model PLEXOS.¹² This tool contains a detailed zonal model of the entire Western Electricity Coordinating Council (WECC) area, including individual generators, transmission lines, loads, and fuel prices. The PLEXOS model dispatches the system at least cost using an optimization algorithm, subject to constraints such as transmission limits, and reports GHG emissions and generation for each plant in 2008 and 2020. The PLEXOS dispatch is used to estimate the least-cost transmission-constrained WECC dispatch that provides cost-based electricity market prices and emissions levels of generators. The PLEXOS dispatch is also used to verify that the dispatch is feasible and that sufficient resources exist on the system for reliable operation.

PLEXOS is used to provide underlying data that is then fed into the GHG Calculator in Microsoft Excel. In order for the GHG Calculator to be able to evaluate the many target cases chosen by users, it is designed to extrapolate from the PLEXOS dispatch model results over a large range of input assumptions. To check the validity of this extrapolation, the E3 project team tested an extreme case in the GHG Calculator, and found that the resulting statewide estimate of costs and GHG emissions were within 2% of California's emissions levels derived from PLEXOS results using similar input assumptions.¹³ This "cross-check" of the GHG Calculator demonstrates that its results are in line with the results of a production simulation dispatch model.

3.1.1. Limitations of the Analysis and Scope of the Model

The purpose of the GHG Calculator is to estimate the key impacts of reducing GHG emissions in California's electricity sector on California electricity consumers. The GHG Calculator does not estimate the impacts of GHG policy choices on energy producers or entities other than the seven groupings of retail providers (and their customers) identified in the model.

The GHG Calculator is a high-level policy tool designed to test policy scenarios and not a resource planning tool with which to make specific resource planning or project choices. A number of trade-offs were made to accommodate the wide range of policy choices and carbon reduction approaches that the Energy Commission and Public Utilities Commission needed the GHG Calculator to model. A few of these limitations are highlighted here:

· The GHG Calculator does not dynamically solve or optimize resource selections based on policy criteria, least-cost criteria, the price of carbon allowances, offset prices, or any other criteria. The model simply provides the user the ability to select which resources to develop in creating a user-defined scenario.

· The GHG Calculator uses four time periods per year, which are fewer than would be used for a detailed planning study.

· The GHG Calculator uses summarized production simulation information for 2008 and 2020 and uses an interpolation approach in intervening years.

All of these choices make the GHG Calculator more flexible as a policy tool for evaluating GHG reduction strategies, but the results should not be used to make or advocate project-specific procurement decisions. In addition, the GHG Calculator does not directly inform questions relating to how the electricity sector might interact with other sectors of the California economy under a statewide GHG policy or market-mechanism regime. Similarly, the model does not evaluate macroeconomic impacts of emission reduction measures. These types of questions require a different set of tools to address.

There are many input assumptions in the model including numerous inputs that are specific to each retail provider. The E3 modeling team has sought to use as accurate information as possible in the GHG Calculator. The retail providers are expected to have better or more specific information on their individual resources and forecasts for their service territories contained within their individual utility resource plans. However, the GHG Calculator contains the best publicly available consolidated set of information for California's electricity sector.

The project team interacted both formally and informally with stakeholders while finalizing assumptions. Parties were given the opportunity to file two rounds of comments on E3's approach and methodology, and the assumptions have therefore been thoroughly reviewed and subject to comment. As a result of stakeholder input, many corrections and changes were made that have improved the analysis. Some stakeholders raised additional concerns about the input assumptions and methodology in the final round of comments, but these comments either were similar to comments submitted in the first round, or would not alter the final results significantly if implemented. As a result, the model was not modified following the second round of comments.

The strengths of the GHG Calculator are that it is non-proprietary and available to all interested parties, and includes only publicly-available information. It allows the user to choose a multitude of input variables. The intent was to create a transparent modeling process, allow interested parties to run their own cases, and avoid, to the extent possible, the perception that the results, and any resulting policy choices, are coming from a "black box." The model also benefits from the "bottom-up" detail of resource cost and potential contained within this portfolio approach to scenario analysis. In addition, the GHG Calculator is built on the foundation of production simulation dispatch modeling results for the entire Western grid. This level of detail helps validate and ensure that the simplified GHG Calculator produces a feasible and reasonable estimate of operations of the Western grid.

3.2. Key Driver Assumptions

Understandably, not all parties agree with all assumptions used by E3 because not everyone has the same view of the future in 2020. Fortunately, in this analysis, not every assumption is a "key driver" that has a significant impact on the modeling results, even among reasonable ranges of values. Thus, some assumptions matter more than others.

In any long-range forecast designed to guide policy choices, it is important to isolate the key drivers of results from the myriad issues that may be important in some contexts but can distract from the task at hand. Therefore, the analysis was focused on issues that are considered key drivers that are important to overall results.

The following table provides the key drivers that were identified and the default assumptions for each of these key drivers that are used in E3's analysis. The robustness of the results was verified for these key drivers through sensitivity analysis and alternative target cases.

Table 3-1

Key Drivers and Default Assumptions

3.3. Electricity Sector Resource Policy Scenarios

For analysis purposes, E3 developed three main resource policy scenarios that bracket the range of likely low-carbon resource portfolios in 2020 for the electricity sector, which are summarized below and described in more detail in Table 3-2:

· Natural Gas Only Case. This case assumes no new development of low-carbon resources beyond the 2008 level, and the addition of only new natural gas generation to meet load growth. There are no new energy efficiency, rooftop solar photovoltaics, or CHP programs in this scenario. The characteristics of this scenario are similar to those for the electricity sector in ARB's Business-as-Usual case,¹⁶ and this scenario represents what would be referred to traditionally as a business-as-usual case.

· Reference Case. This case assumes that existing State policies for the electricity sector (for example, the 20% RPS) are continued to 2020, and that the objectives of these policies are met for renewable generation, energy efficiency, demand response, rooftop photovoltaics, and CHP.

· Accelerated Policy Case. This case assumes substantially more aggressive targets and incentives than those included in the Reference Case, and a corresponding increase in low-carbon resource development. This is the case generally recommended in this decision, with some augmentation as detailed in subsequent sections.

All of these scenarios assume a mix of emission reduction measures for the electricity sector that result from regulatory requirements alone, separate from the introduction of any cap-and-trade system. Users of the GHG Calculator can also create their own scenarios by changing a variety of input assumptions, including resource portfolios, cost and performance assumptions, and emissions trading architecture.

Table 3-2

2020 Resource Portfolios for Three Key Resource Policy Scenarios

3.3.1. GHG Reductions in the Resource Policy Scenarios

E3's analysis reveals that different resource policy scenarios result in very different levels of GHG emissions in 2020. Compared to 2008 electricity sector emissions of 107 million metric tons (MMT) of CO2e, the Natural Gas Only Case results in a 2020 emissions estimate of 129 MMT,¹⁷ an increase of about 21 MMT relative to 2008 levels; the Reference Case results in a 2020 emissions estimate of 108 MMT, a nearly flat emissions profile; and the Accelerated Policy Case results in a 2020 emissions estimate of 79 MMT, a decrease of about 29 MMT relative to 2008 levels. These results are shown in Figure 3-1 and Table 3-3 below. These emissions estimates do not include the effects of a cap-and-trade system that includes the electricity sector.

Figure 3-1

2020 GHG Emissions in Three Key Scenarios

The contributions of different low-carbon resources to the aggregate emissions reduction in the Reference Case and the Accelerated Policy Case are shown as "wedges" in Figure 3-1, with more detail provided in Table 3-3.

Table 3-3

2020 GHG Reductions in Reference Case
and Accelerated Policy Case

(MMT)

3.3.2. Impacts of GHG Reduction Policies on Costs and Average Rates

The E3 GHG Calculator estimates the impacts of GHG reduction policies on total retail provider costs (total revenue requirements for provision of electricity service to customers) and average rates, as shown in Figure 3-2 below for the Natural Gas Only, Reference, and Accelerated Policy scenarios in 2020. These cost and rate estimates do not include effects of a cap-and-trade system; those potential effects are addressed in Section 3.4, with more detailed discussion in Section 5 below.

Figure 3-2

Utility Costs, Customer Costs, and Average Rates in Three Key Scenarios

The GHG Calculator also estimates private customer costs in 2020 for the Reference and Accelerated Policy cases, as indicated for 2020 in Figure 3-2. Private customer costs are those costs that are not paid through utility rates but rather invested directly by electricity customers, such as the customer costs associated with the purchase of a solar photovoltaic system after receiving a rebate or incentive. The utility or retail provider costs of that system would include the portion covered by the rebate offered by the utility for the system. An analysis of private consumer costs is relevant for all of the policies that induce investment at customer premises, including rooftop solar photovoltaics, energy efficiency, and CHP investments. No customer costs are included in the Natural Gas Only Case, because no energy efficiency, solar photovoltaics, or CHP programs are included in this scenario. Customer costs in 2008 were not estimated and so are not reflected in Figure 3-2. The E3 estimates of consumer costs presented in Figure 3-2 are not reduced by the electricity bill savings that consumers will enjoy as a result of their investments in energy efficiency and other demand-side resources; instead, the related cost savings are reflected in the total utility cost calculations.

Potential impacts on utility costs, customer costs, and average retail rates based on the E3 estimates are summarized below, and are illustrative of potential future cost and average rate changes, not definitive forecasts.

· The modeling suggests that total utility costs will increase in excess of inflation in all three resource scenarios due to load growth and due to increases in the capital costs of renewable and conventional generation and of transmission and distribution facilities.

· The modeling suggests that total utility costs would be the highest in the Natural Gas Only scenario, with utility costs about about 4% lower in the Reference Case. In the Accelerated Policy Case, utility costs are estimated to be 7% lower than in the Natural Gas Only scenario. However, inclusion of incremental private customer costs indicates that the Accelerated Policy Case would be the most expensive (6% higher than in the Natural Gas scenario), and the Reference Case the least expensive of the three scenarios (2% lower than in the Natural Gas scenario).

· Average retail electricity rates also will vary depending on the electricity resource policies pursued. For the three scenarios studied, average electricity rates are estimated to be lowest in the Natural Gas Only case, with average rates about 1% higher in the Reference Case and about 14% higher in the Accelerated Policy Case.

· Energy efficiency is extremely important for limiting the economic impacts of GHG reduction on consumers and the economy as a whole.

· The modeling suggests that average utility bills would decline along with policies that reduce GHG emissions, reflecting the lower total utility costs estimated for the Reference Case and the Accelerated Policy Case, even while average electricity rates may increase. With greater efficiency achievements, less energy is required to achieve the same level of energy services and economic productivity.

· Average customer bills are estimated to be the lowest in the Accelerated Policy Case because total utility costs would be reduced due to high levels of cost-effective energy efficiency and distributed resources, which offset the higher costs of renewable generation. Average retail per-kWh rates are estimated to increase under this scenario, however, because customers would purchase less electricity over which utilities could recover their fixed costs.¹⁸ Because of energy efficiency investments at costs lower than supply-side alternatives, costs and average bills are actually lower when the aggressive levels of energy efficiency are achieved.

It is important to consider these costs in the context of the costs of reducing GHG emissions from other sectors of the economy. This analysis is being developed in ARB's Scoping Plan process, and will allow ARB to make informed judgements about the amount of energy efficiency, renewable energy, and other emission reduction measures that should be pursued meet the AB 32 goals.

3.3.3. Sensitivity Analyses

The cost and rate impacts of different GHG reduction portfolios are sensitive to changes in some of the key assumptions underlying these results. For California's electricity sector, the most important drivers are:

· Load growth,

· Energy efficiency achievement and cost, and

· Natural gas price forecast.

In the E3 calculator, users can change the input assumptions for these values when developing their own scenarios. The results of an E3 sensitivity analysis for load growth are shown in Figure 3-3. Using Reference Case assumptions and varying only load growth, a 2% per year decrease from the Energy Commission's forecast that load will grow 1.2% per year results in an average decline in electricity demand of 0.8% per year, an emissions reduction of 28 MMT, and average rate increases of 10% after accounting for reduced capital investments. The reason rates increase at the same time that costs are reduced is that there are fewer sales over which to spread the utility revenue requirement. Increasing load by 2% per year above the Energy Commission's load forecast used in the Reference Case results in an average load growth rate of 3.2% per year, an emissions increase of 37 MMT, and a rate decrease of 8% after accounting for increased capital investments.

Figure 3-3

Sensitivity of 2020 Emissions, Utility Costs, and Average Rates

to Load Growth Assumptions

The results of an E3 sensitivity analysis for energy efficiency are shown in Figure 3-4. Using Reference Case assumptions and varying only the energy efficiency assumptions, emissions increase by 6 MMT in the case with no incremental efficiency, and decrease by 9 MMT in the high efficiency case. The "low goals," "mid goals," and "high goals" energy efficiency scenarios are based on the Itron Goals Update report for the three major investor-owned utilities in California. For the other entities in the state, energy efficiency achievements in these scenarios were extrapolated from AB 2021 filings to the Energy Commission.

E3 relied on the Itron scenarios in part because Itron was able to estimate the cost of achieving energy efficiency goals for those scenarios for the investor-owned utilities. Although the Commissions and the ARB are considering energy efficiency goals up to 100% of economic potential for energy efficiency, which is slightly higher than the Itron "high" scenario, currently no data or analysis exists to estimate the costs of achieving that level of energy efficiency.

Figure 3-4

Sensitivity of 2020 Emissions, Utility Costs, and Average Rates

to Energy Efficiency Savings Assumptions

For a natural gas price sensitivity analysis, E3 tested 2020 prices between $6 and $12 per million British thermal units (MMBTU) in 2008 dollars. The original gas price assumption ($7.85/MMBTU in 2008 dollars or $10.56 in 2020 dollars) is based on the NYMEX forward price for natural gas as of March 2008. The prevailing market price approach is the best approach to develop an unbiased estimate of future natural gas prices because it is the price that a commodity trader could actually buy or sell gas today for future delivery. This price reflects all available information in the market by those with the best access to the information and ability to interpret it.

As of July 28, 2008, average NYMEX gas futures for 2020 delivery were trading at approximately $9.86/MMBTU (2020 nominal) or approximately $0.30/MMBTU less than in March 2008 when E3 established its input values for 2020. This fluctuation is well within the sensitivity ranges evaluated. Gas prices up to $12/MMBTU in real 2008 dollars (or $16/MMBTU in 2020 dollars) were evaluated.

Figure 3-5 below illustrates the findings of the natural gas sensitivity analysis. For each gas price, the cost-effective options in the resource plan were re-evaluated. The results across this range of natural gas prices at the reference costs of resources do not significantly affect carbon reductions in the electricity sector. In fact, at current resource prices, no additional clean energy resources are cost-effective until a price of $12/MMBtu in 2008 dollars enables some biogas to be cost-effective.

Figure 3-5

Sensitivity of 2020 Emissions, Utility Costs, and Average Rates

to Natural Gas Price Assumptions

3.4. Modeling of Greenhouse Gas Cap-and-Trade Market

3.4.1. Modeling of Cap-and-Trade Design Choices

Within the broad cap-and-trade framework described in D.08-03-018, there are many potential design choices that would have an impact on California electricity consumers and the amount of carbon reduction achieved by the sector. The E3 GHG Calculator allows users to change some of these key cap-and-trade design assumptions and see the impact on key metrics, including utility costs and average rate impacts by retail provider; the impacts of a variety of GHG regulatory approaches on the electricity sector; and GHG emission levels both within California and in the entire WECC area.

Most of the cap-and-trade analysis was done assuming that the carbon market would initially be California-only, meaning that only in-state electricity generation and imports into California would face a carbon price, and not generation in the entire WECC area. This was the policy assumption in the GHG Calculator. Additional analysis was also done in PLEXOS with all generators in the WECC area facing a carbon price, simulating a regional or federal GHG policy. See Section 3.4.3 below for discussion of these results.

The GHG Calculator includes policy inputs that define the market price for carbon allowances and offsets, any limits on the amount of offsets allowed in the system, the method for distribution of allowances (auction, administrative allocation to deliverers, or some combination), and potential methods for distribution of auction revenue (or allowances - see Section 5.3 below) to retail providers.

If a user of the GHG Calculator chooses to model an auction for GHG allowances in a multi-sector cap-and-trade system, the user also chooses a market clearing price for GHG allowances. E3 did not endogenously model the market clearing price for GHG allowances in a multi-sector cap-and-trade program because the price would be the result of a number of policy and economic variables that fall outside the scope of this utility sector model, including the overall multi-sector cap on emissions, which sectors are included in the cap, the availability and price of qualifying offsets, the auction design, and other factors.¹⁹

Users of the GHG Calculator are also able to select whether, and how much, administrative allocation of emission allowances to deliverers would occur in the electricity sector. There are two steps to defining administrative allocation to deliverers: (1) the quantity to allocate administratively, and (2) the manner of the distribution of emission allowances to individual deliverers.

E3 modeled the distribution of allowances to deliverers using one or a combination of output-based and/or historical emissions-based allocation methods. In the case of output-based allocation, the output in the year allowances are granted is used as the basis of the allocation. In the case of historical emissions-based allocation, the emissions levels in 2008 are used as the basis of allocations. Both assumptions are simplifications for the purposes of modeling and do not constitute policy recommendations. In reality, the output-based allocations may be based on a prior year's output, and historical emissions may be determined by averaging over several years to reduce the volatility caused by hydro variations.

If a user chooses a combination of both output-based and historical emissions-based allocations to deliverers, the model computes the administrative allocations by separating the available allowances into two pools based on the user-defined percentages and then allocating the allowances within each pool in proportion to the deliverers' output or historical emissions, as appropriate.

In addition, users can decide to model auction revenue (or allowance - see Section 5.3 below) distribution to retail providers. There are three steps to defining this policy in the model: (1) determining the amount of revenue to be distributed to retail providers, (2) selecting the basis for the distribution (sales-based or historical emissions-based), and (3) defining whether the auction revenue to return is a fixed share of the overall carbon market or is linked to the actual spending of the electricity sector in the carbon market auction. The model only considers distribution of auction revenue to retail providers, although in reality other alternatives are possible.

Similar to the market for GHG emission allowances, offset prices are also specified by the user. However, the model allows an additional control, limiting the percent of a deliverer's GHG compliance obligation that may be met with different types of offsets. The maximum amount of offsets that can be purchased by a deliverer is specified as a percentage of its total requirement. The offset prices and quantity limits are set independently for each of three types of offsets depending on origin: (1) a non-capped sector in California, (2) the region or the United States, or (3) international.

3.4.2. Modeling Results for a California-only Cap-and-Trade System

The GHG Calculator was used to analyze some of the impacts of a California-only multi-sector emissions allowance trading system, i.e., not a regional or federal system, but including allowances for emissions associated with imported electricity. By design, a California-only multi-sector cap-and-trade program (including electricity imports) would achieve emissions reductions to meet a pre-determined GHG cap. The trading component of the cap-and-trade policy would enable those GHG reductions to come from sectors or sources with lower marginal abatement costs than other capped sectors or sources. Analyzing the multi-sector impacts and interactions of such a multi-sector program lies outside the scope of E3's modeling, which was focused on electricity, primarily, and also on natural gas. Multi-sector modeling is being conducted by ARB.

E3 found that a California-only cap-and-trade system, modeled in the electricity sector with an exogenous price for GHG emissions on all electricity (including imports), is likely to increase costs in the electricity sector without achieving meaningful additional GHG reductions within the sector beyond the level of mandatory program reductions, unless one of the following or a combination of the following to a lower degree, occurs:

· Carbon prices reach high levels ($100/ton CO2e or more);

· Natural gas prices increase significantly (100% or more);

· Technology innovation drives down the cost of low-carbon electricity resources relative to natural gas or improves the performance of low-carbon technologies significantly; or

· Lower-cost opportunities are available from other sectors under the cap-and-trade program (though in this case the GHG reductions would come from those other sectors and not the electricity sector).

This finding assumes that lower-cost opportunities to reduce GHG emissions are available from other sectors under the cap-and-trade program, and underscores the critical need for including multiple sectors within the program and linking, to the extent possible, to trading systems beyond California's borders. A number of well-publicized analyses of carbon costs across sectors indicate that lower-cost opportunities may exist in sectors other than electricity. A multi-sector approach will be able to capture lower-cost opportunities in other sectors, but such results were not modeled by E3. Instead, E3's analysis focuses on the availability and costs of GHG reductions within the electricity sector.

Table 3-4 below shows the key findings of E3's simulation of the impacts on the electricity sector of a multi-sector cap-and-trade system implemented in California only.

Table 3-4

Impacts of California-Only Multi-Sector Cap-and-Trade Program

on the Electricity Sector

3.4.3. Modeling Results for a Regional Cap-and-Trade System

In contrast to a California-only cap-and-trade system, linkage with trading systems on a regional basis, including all jurisdictions in the Western electricity grid, is more likely to result in a change in generator dispatch, with coal-fired generators operating less.

Under a cap-and-trade program, the prices of GHG allowances and offsets increase the variable cost of electricity generation. Currently, the lowest variable cost fossil-fuel units in the West are coal units, which also have the highest GHG emissions. If a carbon price were applied to all generators in the WECC area and if the carbon price became expensive enough, it would become more economic to dispatch existing natural gas units instead of existing coal-fired units. However, California's in-state generation mix contains very little coal-fired generation and includes mostly low-carbon, low-variable cost units (hydro, nuclear) and higher-carbon, higher-variable cost natural gas units. Therefore, including a carbon price would not change the dispatch order of generators in the State because the plants with the highest GHG emissions are already dispatched last.

While the dispatch order of generators in California is not expected to change much under a cap-and-trade program, California imports a significant amount of coal-fired electricity. Under a California-only cap-and-trade policy, out-of-state generators would not pay for carbon allowances unless they deliver their power to California. Thus, the dispatch order of out-of-state generation is not expected to change based on the cost of California-only carbon allowances if the coal generation is still economic to serve non-California load. In the GHG Calculator, the user may select whether specified out-of-state coal contracts should be dropped if the price of carbon makes these contracts uneconomic. Unspecified electricity imports to California are modeled consistently with D.07-09-017: the default assumption is that all unspecified imports are assigned a regional default emission factor of 1,100 pounds of CO2e/MWh produced.

To evaluate generation operational changes in a regional or federal GHG policy scenario, E3 ran several scenarios in PLEXOS in which the WECC-wide dispatch included a carbon price in the operating costs for all of the generators in the WECC area that emit GHG, with results shown in Figure 3-6 below. These PLEXOS scenarios included GHG allowance price assumptions from $0/ton to $100/ton of CO2e, in $10/ton increments, plus scenarios with prices of $120/ton and $150/ton. This analysis provides an estimate of the GHG reductions due to operational or dispatch changes of the 2020 WECC generator fleet due to a region-wide market for carbon allowances.

Figure 3-6

PLEXOS Results for WECC Dispatch with WECC-wide Carbon Price

This analysis found that, at the natural gas and coal prices assumed in the Reference Case, natural gas would begin to displace coal at a carbon price of about $50/ton CO2e, and that there would be a significant shift from coal to natural gas at a carbon price of around $60/ton. Higher coal prices relative to natural gas prices would be expected to reduce the required carbon price that would change operations. The answer to Question A in Table 3-4 above would change under a WECC-wide cap-and-trade program. This analysis was not built into the GHG Calculator; however, the results were presented at the workshop on April 21, 2008 and parties subsequently had an opportunity to file comments on the results.

In addition, a WECC-wide cap-and-trade program would significantly mitigate the "contract shuffling" concern raised in response to Question B in Table 3-4 above. A transparent, well-regulated regional system, with robust reporting and enforcement mechanisms, could eliminate incentives for contract shuffling and the resulting emissions reductions that are only on paper.

Finally, in a WECC-wide cap-and-trade program, new low-carbon generation may displace either coal- or natural gas-fired generation depending on time and location. Therefore, the relative price-point of carbon allowances needed to make new renewables cost-effective posed in Question C above depends on the relative variable costs and emissions rates of coal and natural gas. The responses to Questions D, E, and F would remain unchanged under a West-wide cap-and-trade program.

These findings only serve to underscore the critical importance of California's participation in a multi-sector and multi-state cap-and-trade system, to reduce costs and increase GHG reductions from the program.

3.4.4. Analysis of Effects of a Cap-and-Trade Program on Retail Provider Costs and Average Electricity Rates

A cap-and-trade program would add a GHG emissions cost to electricity generation, which could affect both wholesale and retail electricity prices. In a system with organized wholesale power markets such as California, all generators participating in the wholesale power market receive a single market clearing price for their electricity based on the bid of the last or "marginal" generator needed to meet electricity demand. The expectation is that, in most circumstances, the marginal generator would pass through its carbon cost in the market clearing price.²⁰ Retail providers would also be responsible for carbon costs associated with generation they own or have under long-term contract. These increased costs for both purchased and owned electricity would tend to increase retail rates, but could be offset to greater or lesser extents if allowances are distributed for free to deliverers and/or retail providers, as described briefly here and in more detail in Section 5 below. Cost savings arising due to the cap-and-trade program itself may also reduce bill impacts relative to other GHG mitigation approaches.

In this section, we provide a brief overview of E3's analysis of potential effects of a California-only cap-and-trade market on total utility costs and on average retail rates, depending on allowance allocation alternatives. We look at E3's estimates of the effects of a cap-and-trade program assuming that the resource policies included in Accelerated Policy Case are implemented, because we are committed to pursuit of the resource policies in this scenario. The E3 analysis of cap-and-trade market alternatives assumes a carbon price of $30 per ton CO2e and no offsets.

Because of its focus on only the electricity sector in California, the E3 model does not capture the important potential financial benefits of a multi-sector cap-and-trade program and, thus, it tends to over-estimate electricity sector costs that may occur in a multi-sector cap-and-trade program. A multi-sector cap-and-trade program would allow entities with compliance obligations to identify least-cost GHG reduction opportunities among all of the covered sectors, which in turn could allow California to meet its emissions goals at considerable cost savings, relative to a GHG reduction approach that relied only on increased mandatory programs. A cap-and-trade program with a larger geographic scope could yield significantly greater costs savings, which also are not estimated by the E3 analysis. Nor does the E3 model quantify the additional emissions reductions that can be expected due to the presence of a price on GHG emissions, which would encourage additional conservation and investments in efficiency and low-GHG generation. Because of these limitations, we find E3's analyses of cap-and-trade scenarios most useful as a means to compare relative costs of various cap-and-trade design options, and less helpful regarding identification of total electricity sector costs in a multi-sector and/or regional cap-and-trade program.

Figure 3-7 compares E3's estimates of utility costs for three cap-and-trade scenarios if the Accelerated Policy Scenario is implemented. The three cap-and-trade scenarios considered are (1) all allowances are auctioned and no allowances (or allowance value) are distributed to retail providers for the benefit of their customers; (2) all allowances are distributed at no cost to deliverers in proportion to their historical emissions; and (3) all allowances are auctioned, with either the allowances or allowance value distributed to retail providers for the benefit of their customers.

Figure 3-7

Estimates of Retail Provider Costs

With a California-only Multi-sector Cap-and-trade Program

(2008$ in Millions)

Figure 3-8 compares E3's estimates of statewide average retail electricity rates for the same three cap-and-trade scenarios.

Figure 3-8

Estimates of Average Retail Electricity Rates

With a California-only Multi-sector Cap-and-trade Program

($/kWh, 2008$)

Of the three cap-and-trade approaches considered, these figures indicate, as we would expect, that the most expensive approach from the retail provider and customer perspectives would be if all allowances are auctioned but no allowances or allowance value are distributed to the retail provider for the benefit of consumers. As indicated in Figure 3-7 and Figure 3-8, assuming $30 per ton allowance costs, such an auctioning approach could cost California retail providers approximately $2.4 billion more in 2020, with resulting increases in average retail electricity prices of about $0.009 per kWh, in 2008 dollars, compared to an approach in which all allowances are auctioned with retail providers receiving the auction revenues for the benefit of their customers. These results illustrate clearly why we believe it is crucial that all or almost all of the value of electricity sector allowances that are auctioned be distributed to retail providers, to fund emission reduction activities and mitigate these potential rate impacts.

The other cap-and-trade scenario presented in Figure 3-7 and Figure 3-8 would have all allowances distributed to deliverers at no cost in proportion to their historical emissions, which E3 calculated based on 2008 estimated emissions. As indicated in the figures, E3 estimates that this approach would cost retail providers approximately $1.5 billion more in 2020, with resulting increases in average retail electricity prices of about $0.005 per kWh in 2008 dollars, relative to auctioning with retail providers receiving the auction revenues for the benefit of their customers.

As illustrated above, auctioning with retail providers receiving auction revenues would largely mitigate the potential effect of carbon costs on total utility costs and retail rates while still providing powerful incentives to reduce emissions. As explained in more detail in Section 5, auctioning of allowances would create limited windfall profits in the form of "rents to clean generation," because the increase in the wholesale price of electricity paid to low-carbon resources that utilities purchase through the wholesale electricity market would exceed their compliance costs. The clean generation rents would constitute a wealth transfer from electricity customers to low-carbon electricity producers. Higher returns to clean generation would encourage further investment in low-carbon resources, principally renewable generation. Moreover, while the clean generation rents would tend to increase electricity rates somewhat, this potential increase might be outweighed by the cost savings benefits of a multi-sector cap-and-trade program, which are not captured by the E3 model.

As explained in Section 5 and illustrated above, distribution of allowances at no cost to deliverers would result in large windfall profits to independent generators and marketers, including allowance rents and clean generation rents. While clean generation rents have some offsetting benefits, as noted above, allowance rents are particularly worrisome. In Section 5, we recommend that historical emissions-based allocations to deliverers not be pursued, because of these unacceptably large wealth transfers and retail rate increases.

While not included in the above figures due to modeling limitations, output-based allocations to deliverers may reduce wholesale price increases and windfall profits, to the extent that output-based allocations would reduce the incentive for deliverers to pass through the carbon price in the wholesale energy market. (See Section 5.2.1.2.)

As explained in Section 5.4.2, we recommend that a fuel-differentiated output-based method be used to distribute a limited portion of allowances to deliverers in the early years of a cap-and-trade program, to be phased to 100% auctioning by 2016, with allowances distributed to retail providers and the auction revenues used to benefit customers.

3.5. Parties' Comments on Modeling Issues

Twenty-four parties filed comments that address modeling issues. The majority of modeling-related comments focus on input assumptions: integration costs,²¹ transmission costs, resource costs, energy efficiency achievements, CHP operating characteristics, and penetration rates in the Accelerated Policy Case. There was also some discussion of the results. For example, SDG&E/SoCalGas and PG&E argue that the estimated rate and cost impacts are too low, while some of the advocacy groups argue that the estimated rate and cost impacts are too high.

Other modeling-related questions and issues raised in the comments include the following:

· What is the best metric for evaluating allocation scenarios: should we consider retail provider "normalized" cost impacts (such as utility costs relative to utility benefits, or relative to utility size) or cumulative impacts from 2008 or 2012 - 2020, rather than just annual costs in 2020? (SCE, SMUD)

· Does the model show any value to a cap-and-trade approach? (LADWP)

· How reliable is the theorized electricity market clearing price effect²² of an output-based allocation, and what is the best estimate of the magnitude of this effect? (SMUD)

· How much uncertainty is there surrounding the key assumptions for the Reference and Accelerated Policy Cases?

The following sections discuss model and input issues. Other modeling-related comments are discussed in other relevant sections of this decision.

3.5.1. Model Structure and Operation

3.5.1.1. Documentation

Several parties, including SDG&E/SoCalGas and SCE, state that the model documentation is insufficient and that the model is overly complicated. They also express concern with labeling within the model that they claim is poor, inconsistent, or misleading.

E3 made substantial improvements in the model interface in the final version, including consolidation of controls on the Resources and CO2 Market tabs, color coding of inputs, adding an input/output printable table, and including a map to the different tabs. On May 6, 2008, Public Utilities Commission staff held a WEB-EX workshop to educate stakeholders' technical staff on the model's architecture and how to run scenarios. E3 also made itself available via phone, email, and in-person to meet with various stakeholders to answer questions and address concerns about how to use the model. Even with those efforts, there is a degree of irreducible complexity in the model that reflects the subject matter and the types of analyses requested, and only familiarity through use, rather than documentation per se, will help users fully understand its function and results.

3.5.1.2. Price Elasticity of Demand

Some parties comment that the model does not dynamically account for the price elasticity of demand. As designed, the GHG Calculator has no feedback loop by which demand for electricity or natural gas is reduced in response to increasing electricity, carbon, or gas prices (or increased in response to lower prices). These price-induced demand effects will change the estimated cost effectiveness of carbon reduction measures. However, it was too complex to build the effects of price elasticity into the model. Instead, E3 handled this issue in the following manner.

E3 tested the sensitivity of results to average price elasticity assumptions and found that the impacts on emissions, costs, and rates are very small even with a fairly aggressive assumption for price elasticity (-0.3). While the model does not dynamically iterate to adjust demand interactively with price until an equilibrium is reached, if a user wants to see the impact of price elasticity, there is a control that can be used to adjust demand based on user assumptions about the price response.

We note that the effects of price elasticity at higher prices are not clearly understood and the differential impacts on energy-intensive elements of the economy have not been addressed in this assessment. While demand response to average prices may be low, the more energy-intensive elements of the California economy pay electricity rates well above the average rate. Hence, they would be more likely to notice and to respond to price increases. Similarly, a fundamental purpose of adding the price of carbon into the price of electricity (which is what a cap-and-trade system does) is to induce technology innovation throughout the economy. Users would not have to rely on utility programs to invest in technologies that would lower their bills; instead they are rewarded for searching out incremental efficiency improvements. Price elasticity is an economy-wide issue which ARB is working on modeling, and there is need for more analysis. As has been recently demonstrated in the transportation sector, it may take very high prices to induce individuals to make big shifts in their use of energy but, once started, the changes may snowball. On the other hand, high electricity rates can discourage high consumption from the grid (e.g., prohibitively high prices in the upper tiers of residential rates may encourage solar photovoltaic installations). We do not know these "tipping points" for different types of electricity users.

3.5.2. Input Assumptions and Results

GPI comments that, "the input assumptions used by E3 in both the reference case and the other cases it has prepared appear to us to be valid. E3 has done a good job of estimating inputs based on the current market, and it has done some good work in estimating future markets. One thing that may not be possible to model is a large change in the market, such as a change in technology. While E3 may not be able to model such a market change, it is important to keep in mind that such a change is possible, even probable given the amount of effort going into improving technology and finding new energy sources." (GPI Comments, p. 34).²³

SMUD states that it "commends the Commissions and E3 for the Stage 2 modeling effort. Although the model has weaknesses at the specific [retail provider] level, the model nonetheless provides real information and allows participants to adjust parameters and view the impacts of those changes." (SMUD Comments, p. 12.)

PacifiCorp states that the E3 modeling results appear to support similar modeling performed by the Electric Power Research Institute that examined the effects of different CO2 prices on the WECC power market, including natural gas being dispatched ahead of coal once CO2 is priced closer to $60/ton (i.e., reducing coal electricity imports into California). (PacifiCorp Comments, p. 47.)

PG&E contends that "model results should always be represented in an uncertainty band." Regarding the Reference Case outcome of an emissions level of 108.2 MMT in 2020 for the electricity sector, PG&E comments that "slight changes in assumptions would change this figure. For example, if load growth continues at the 1990-2000 historic levels, 1.5%/year, then the 2020 electricity sector emissions projection becomes 114.5 MMT CO2. A few small, realistic changes in inputs change the emissions outcome substantially, and so the ARB's implementation of AB 32 must accommodate the uncertainty inherent in the sectors' 2020 emissions forecast." (PG&E Comments, p. 101.)

We agree that variations are likely in the key drivers over time, and it is important to recognize these as policy is developed. The GHG Calculator was developed to allow evaluation of the effects of changes in key drivers and exploration of policy decisions that would accommodate a range of actual conditions over time.

3.5.2.1. Electricity Prices and Natural Gas Heat Rates

Some parties (Solar Alliance and CalWEA/LSA) contend that the natural gas market heat rates and electricity market prices in the model are too low. Referring to the Accelerated Policy Case, they state that, "The electricity market prices used in the model average $54 per MWh. Assuming variable operations and maintenance of $2.50 per MWh in the market price and dividing the remainder by the gas price results in a market heat rate of approximately 6,600 Btu/kWh. This is 5% below the `clean & new' heat rate of a new [combined cycle gas turbine] CCGT, and is inconsistent with typical market heat rates of 8,000 Btu per kWh observed in the California wholesale market in recent years." (Solar Alliance Comments, p. 10, and CalWEA/LSA Comments, p. 10.)

In the Accelerated Policy Case, electricity loads are approximately 88% of the forecast load levels in 2020. At these load levels, the PLEXOS model indicates that natural gas plants are not always on the margin, which causes the relatively low market heat rate that concerns these parties.

The "market prices" referenced above are based on the PLEXOS model output and include only the energy component of the electricity wholesale costs. Therefore, the reported average market prices do not include the costs of capacity. The model includes the capacity value of displaced new generation in the calculation of resource value and adds it to the energy values cited. The total value of new resources once capacity value is added for the Accelerated Policy Case is about $74/MWh annual average value of energy and capacity, which we believe is reasonable.

3.5.2.2. Wind Integration Costs

CEERT contends that the wind integration costs used by E3 are too high and recommends that we rely on costs produced by the Intermittency Analysis Project (IAP) and adopted by the Energy Commission. According to CEERT, "IAP estimated integration costs [are] at $0.69/MWh for wind in a 33% renewables by 2020 scenario [whereas] E3 assumes a range of $4.09 - 6.36/MWh." (CEERT Comments, p. 16.)

The E3 team evaluated the IAP project and found the wind integration costs at the extreme low end of the range in the studies available and used to develop wind penetration cost estimates. The IAP appears to assume that the State's hydro system can be used to provide increased ramp and regulation needs at zero cost. Said another way, in the IAP analysis there is no opportunity cost for redispatching the hydro system. In addition, the IAP only evaluates a single resource scenario and provides no mechanism to estimate differing integration costs for different renewable resource mixes as is required in the GHG Calculator.

EPUC/CAC contend that the renewable integration costs used by E3 may be too low because "the model did not include improvements to the bulk transmission system or the costs of managing congestion on the bulk transmission system. As a result, the analysis does not ensure that renewable and other resource additions can be delivered to the load for the levels of costs assumed in the model." (EPUC/CAC Comments, p. 19.)

The GHG Calculator includes incremental transmission costs attributable to new renewables in order to evaluate the relative impact of new renewables for any case defined by the user. In addition, the GHG Calculator adds an integration cost for wind that includes costs of system balancing, ramp, and regulation.

EPUC/CAC also question the ability of the electricity system to integrate large amounts of renewable generation. EPUC/CAC contend that reliability impacts have not been fully assessed: "... the analysis does not ensure that renewable and other resource additions can be delivered to the load for the levels of costs assumed in the model [and ...] the California grid could see too much generation in generation pockets and too little supply in load pockets." (EPUC/CAC Comments, p. 19.)

We reiterate that the GHG Calculator is a policy-level tool and not a detailed resource planning or system operations model suitable for evaluating renewable integration. While PLEXOS has the capability of performing detailed operations simulation, it was not run in a manner that would provide detailed renewable integration costs for all possible cases of potential interest. Such analysis is not possible in a tool that allows for such diverse system configuration and range of plans necessary for policy-level decisions. To estimate integration costs, the GHG Calculator adds a renewable integration cost as a function of wind penetration. E3 developed the integration cost function based on numerous intermittent cost studies that analyzed the details of system cost.

We acknowledge that there is a great deal of uncertainty regarding the integration costs for renewable energy and more work is ongoing. Factors contributing to the uncertainty include (1) the proportion of intermittent to firmed or baseload renewables developed for the state's renewable energy goals and voluntary Renewable Energy Credit (REC)²⁴ market; (2) changes made to the fossil fuel generators' ramping capabilities over the next 12 years; and (3) changes made to the amount of regulation support, short-term and long-term "storage," and the integration of Smart Grid technologies, among many other factors.

3.5.2.3. Resource Costs for Conventional and Renewable Generation

TURN contends that capital construction costs in the model may be too low and do not take into account recent cost increases.

The cost of new clean energy technology is important, but also hard to predict. In the GHG Calculator, the Reference Case assumption is that current capital costs stay the same in real terms between 2008 and 2020. Increased demand for raw materials or competition with other regions for clean technology could drive up clean generation capital costs, in real terms, between now and 2020. However, capital costs for clean technology could also decrease in real terms if the technology improves and/or production methods and manufacturing become more efficient over time. If the price of inputs such as steel rises for all technologies, the relative change in prices among technologies may be less pronounced than if some technologies make major efficiency improvements while others do not. However, if solar thermal technology capital costs were to fall 25% in real terms between 2008 and 2020 while other technologies' costs did not change, for example, far more solar thermal installations could become viable in the near term, reducing the cost to the electricity sector of compliance with GHG reductions policies.

NRDC/UCS state that the assumed capital costs for combined cycle gas turbines (CCGT) are too low:

The E3 model documentation notes that the model escalated capital costs for all generating technologies "by 25% per year for two years to reflect recent rapid inflation in construction costs, with the exception of solar, thermal and wind." Because the model's CCGT capital cost assumptions are based on plants built in 2004 and 2005, they also appear to have been excepted from the 25% per year cost escalation applied to other resources. For consistency, and to ensure that CCGT capital cost assumptions reflect current market reality, the CCGT capital cost should be escalated by a similar rate to other resources, or by a widely used power industry price index such as the Handy-Whitman index. (NRDC/UCS Comments, p. 49.)

The CCGT capital costs were escalated to reflect recent capital cost increases using the same approach as adopted in Resolution E-4118 in the Market Price Referent proceeding, R.04-04-026. Furthermore, there is not an inconsistency introduced by using different escalation rates for the costs of CCGT and new clean resources because the data sources are different. The CCGT costs are based on actual plants built in California while the costs of clean energy technologies are based on planning level estimates used in the United States Department of Energy's 2007 Annual Energy Outlook. E3 found the 2007 Annual Energy Outlook costs to be lower than the range of costs reviewed and documented in the Stage 1 analysis and therefore applied higher inflation rates to provide an estimate of actual installed cost on the same basis as assumed in the Market Price Referent proceeding.

3.5.2.4. Natural Gas Price and Other Fuel Prices

A number of stakeholders claim that the natural gas prices used in the E3 scenarios are too low. According to CEERT, natural gas prices may be closer to $17/MMBTU by 2020, a price which it asserts would have implications for the cost-effectiveness of new renewable resources. Environmental Council and Solar Alliance prefer to assume $15/MMBTU in 2020 in 2008 dollars. In addition, they state that coal prices should be closer to $3.03/MMBTU in 2020, instead of $1.01/MMBTU.

Taking another view, TURN states that the assumed natural gas price is too low, but that "... it is not clear that a reasonable increase in gas prices will make renewable energy economic compared to natural gas anyway." (TURN Comments, p. 30.) However, CalWEA/LSA contend that an increased starting natural gas price would lead to a decrease in the cost of GHG reductions: "If the starting natural gas price is increased to $10 per MMBtu [from $7.85/MMBtu], the cost of GHG reductions from a 33% RPS decreases from $133 to $106 per tonne." (CalWEA/LSA Comments, p. 9.) NRDC/UCS also have concerns about the low prices used by E3 in its scenarios. However, they also believe that adding renewable energy might reduce demand for natural gas resulting in between 2% and 15% downward pressure on price levels in the future. (NRDC/UCS Comments, p. 46.)

According to CalWEA/LSA,

"in the long-run, fossil fuel prices can be expected to exhibit a positive real escalation rate, as they become increasingly difficult to find and produce. In addition, the structure of the E3 model does not recognize the potential for renewable resource costs to decline over time, as renewable technologies improve. These differential escalation rates become particularly significant over the multi-decade timeframe in which the GHG reduction program will operate. Indeed, one of the primary benefits of renewables is that they substitute capital costs for fuel costs, and are a long-term hedge against future fuel price escalation. The E3 model's use of constant, 2008 dollar costs in all years ignores these significant benefits of renewables. CalWEA and LSA have re-run the E3 calculator, assuming that a natural gas price of $10 per MMBtu in 2008 increases at the historical long-term real escalation rate of 3.5%; using this rate, the natural gas price would exceed $15 per MMBtu in 2020 in 2008 dollars. This change in the profile of natural gas prices used in the E3 calculator results in a GHG mitigation cost for a 33% RPS of $43 per ton." (CalWEA/LSA Comments, p. 10.)

SCPPA asserts that "if gas prices are assumed to be at or beyond today's prices of nearly $12/MMbtu, even higher allowance prices would be required to alter the dispatch of coal-fired generation." (SCPPA Comments, p. 10.)

As discussed in the section on sensitivity analysis above, natural gas prices in 2020 are a key driver of model results. The Reference Case natural gas price forecast for 2020 is $10.56/MMBTU in nominal dollars (or $7.85/MMBTU in real 2008 dollars). This is the price of natural gas for 2020 that could be secured in the NYMEX forward market at the time of the analysis in March 2008. Spot prices could increase or decrease from this forecast, and E3 and other parties performed sensitivity analyses on natural gas prices. However, the NYMEX market prices reflect the best publicly available unbiased forecast of future gas prices. If 2020 natural gas prices were to reach the range of $19 - $21/MMBTU in nominal dollars (or $14 - $17/MMBTU in real 2008 dollars), the average all-in cost of wind would be competitive with the cost of installed natural gas units. Likewise, if 2020 natural gas prices were to reach the range of $21 - $24/MMBTU in nominal dollars (or $15 - $18/MMBTU in real 2008 dollars), the average all-in cost of solar thermal would be competitive with the costs of natural gas generators.

We note that, while increases in assumed natural gas prices make the cost of renewable energy more attractive, higher gas prices also make out-of-state coal generation relatively more cost effective. Likewise, higher gas prices increase overall utility costs, given the high degree of reliance that California utilities have on natural gas generation.

3.5.2.5. Energy Efficiency

Some parties are concerned about the achievability of the energy efficiency levels in the E3 scenarios and about the likely costs:

[T]he EE values proposed for use in Phase 2 of the GHG modeling are more realistically achievable than the EE levels used in Phase 1. However, SCE has concerns about EE levels used in E3's Mid and High Cases because these cases assume utility incentive programs based on 100% of incremental cost[footnote omitted], an approach that has never been used on a comprehensive basis in the real world. Use of a scenario based on current incentive levels would be a more realistic assumption until the efficacy of the 100% can be demonstrated based on empirical results. (SCE Comments, p. 49.)

The aggressive case is unprecedented, and ARB should not assume that these levels of EE and [renewable electricity] will be achieved in the scoping plan. Small changes to the load growth assumption change emissions substantially. (PG&E Comments, p. 101.)

Regarding energy efficiency modeling, SDG&E/SoCalGas state that, "non-intuitive results such as the aggressive energy efficiency case showing that utility costs of these programs may exceed the `total resource cost' [footnote omitted] creates questions of modeling accuracy of these assumptions." (SDG&E/SoCalGas Comments, p. 41.) In fact, in the "mid" and "high" energy efficiency scenarios, utility costs are correctly higher than the total resource cost by a few tenths of a cent per kWh. This is because in a few cases the Itron analysis assumed that the current utility rebates exceed 100% of full incremental measure costs.

A number of current incentive programs administered by the investor-owned utilities have paid 100% of incremental cost for energy efficiency measures.²⁵ For example, several small business programs have paid incremental costs, and have paid more than incremental costs for certain qualifying customers. Furthermore, the low-income energy efficiency programs, although not incentive programs, may provide 100% or more of incremental costs, and generally are more comprehensive than investor-owned utility incentive programs, dealing with building envelope as well as lighting and heating, ventilation, and air conditioning systems. Additionally, retrofit programs, which provide incentives for the replacement of technologies before the end of their useful lives, often provide more than incremental cost; they may provide a high percentage or even 100% of total cost.

In general, assumptions about the penetration and costs of achieving energy efficiency in the model are among the largest uncertainties in the analysis, as discussed in the section above related to sensitivity analyses. Several parties also assert that there is insufficient documentation of the energy efficiency costs in the model. Cost assumptions are all "best estimates" based on analysis of investor-owned utility costs performed by Itron for the Public Utilities Commission's IOU Goals Update Study.

3.5.2.6. Interaction of Cap-and-Trade and Renewables Assumptions

Several parties express concern that a requirement to participate in a cap-and-trade system may not induce the development of new renewables, or may encourage renewables only at very high allowance prices exceeding $100/ton CO2e:

Given the E3 results showing the potential inefficacy of requiring the electric sector to participate in a multi-sector cap-and-trade program except at very high allowance prices and given the current absence of evidence about the cost of GHG reductions in other sectors, it would be premature to force the electric sector into a multi-sector cap-and-trade program. Thus, SCPPA recommends that the Commissions revisit their Interim Opinion and, upon reconsideration, defer recommending that the electric sector participate in a multi-sector cap-and- trade program. (SCPPA Comments, p. 3-4.)

A comprehensive approach to renewables is fundamentally important if they are to play a significant part in GHG reduction. Renewables are a capital-intensive industry with long-term planning needs, both for the facilities themselves and the transmission infrastructure necessary to support them. It is unrealistic to expect the substantial investment needed for renewables to exceed the current 20% target based on a brand new pricing signal from a yet-to-be established cap-and-trade system, which, based on the experience of other markets, is certain to be somewhat volatile in its fledgling years. (CalWEA/LSA Comments, p. 2.)

Despite the relatively high cost of renewables based on current prices found in the E3 analysis, increased renewables development will remain a significant component in decarbonizing the California electricity sector to meet the AB 32 targets and more critically California's 2050 goal of 80% reductions below 1990 levels. Mandates for renewable energy will ensure that renewables are developed even if carbon allowance prices are lower than the level necessary to induce new renewables or if fossil generation is cheaper than renewable generation for other reasons.

As described in D.08-03-018, we recommend that the electricity sector be included in the cap-and-trade program because it could encourage greater innovation and cost reductions, including in the development of renewable generation. Additional development of renewables could occur in the voluntary market for RECs, if utilities surpass renewables mandates, or if there is increased self-generation using renewables that is not accounted for outside of a cap-and-trade market. Some parties ask that some number of allowances be set aside for the voluntary market, as discussed in Section 5.4.3.2 below. Although E3 took a conservative approach and assumed no market transformation , a higher market price for electricity and a higher carbon price could drive new technology innovation, resulting in new sources of emission reductions in the sector at lower costs. The GHG Calculator allows parties to model alternative future scenarios by substituting their own values for selected variables; a number of these scenarios were submitted in comments. On this point, the modeling itself or its methodology is not the issue; rather it is the differing assumptions about the future that drive different results. Will carbon prices reach and maintain a level of $100/ton CO2 or more? Will natural gas prices increase significantly? Will technology innovation drive down the cost of low-carbon resources or improve the performance of low carbon technologies? We believe that, over the long term, the potential opportunities that can be created by increased market pressure are likely to outweigh the costs to ratepayers imposed by including electricity within an emissions cap-and-trade system.

3.6. Scenarios Submitted by the Parties

Several stakeholders used the GHG Calculator to model different outcomes to inform their own comments:

· PG&E used the model to show the carbon impacts of its proposed alternative scenarios.

· IEP used the model to show the impacts of alternative producer surplus scenarios.

· SCE used the model to generate alternative metrics for evaluating the "economic harm" of allocation scenarios.

· WPTF used the model to submit alternative allocation scenarios.

· SMUD used the model to evaluate different allocation scenarios and developed its own metric for evaluating them.

· Environmental Council created a preferred set of input assumptions for the Reference Case.

· NRDC/UCS submitted alternative scenarios to support their comments.

· NCPA used the model to develop and verify its own allocation model developed by R.W. Beck.

These submissions are discussed where relevant in this decision.

⁸ The GHG Calculator is a spreadsheet that simplifies the multiple possible outputs of the PLEXOS model into a few parameters; namely, the relationship between load and GHG emissions rates and the relationship between load and electricity prices.

⁹ The groupings of retail providers modeled are: (1) PG&E, (2) SCE, (3) SDG&E, (4) SMUD, (5) LADWP, (6) a grouping of all other municipal utilities, direct access electric service providers, and other retail providers in Northern California called "Northern California Other," and (7) a grouping of all other municipal utilities, electric services providers, and other retail providers in Southern California, called "Southern California Other." The model also separates out the load and emissions associated with the California water agencies, including the Department of Water Resources, the Central Valley Project, and the Metropolitan Water Project, in a separate category.

¹⁰ The Stage 1 modeling default assumption was that the target emissions level for the electricity and natural gas sectors was equal to the 1990 sectors' emissions as reported in the preliminary ARB GHG emissions inventory, dated August 22, 2007. ARB revised the GHG inventory on November 19, 2007, which resulted in an adjusted 1990 emissions level for the electricity and natural gas sectors. This change to the ARB GHG inventory occurred after the Stage 1 model was released and so was not reflected in that version of the model.

¹¹ Originally, E3 was required to provide estimates of GHG carbon dioxide equivalent (CO2e) emission reductions under various "load-based" cap options, in which retail providers rather than deliverers would have the GHG compliance obligations. However, as result of D.08-03-018, the recommended point of regulation for GHG emissions in the electricity sector is the deliverer of electricity to the California transmission grid rather than the retail provider. This change required a number of significant modeling changes to the GHG Calculator.

¹² www.plexossolutions.com.

¹³ For more detailed information on the cross-check, see the May 13, 2008 E3 presentation, Slide 39, Verification with PLEXOS.

¹⁴ The following definition of market transformation generally captures its use herein: "Market transformation refers to a system of intentional actions to shift markets in terms of product availability and customer choice. It implies a greater consumer or demand-side influence on the development and dissemination of technology. It encompasses actions aimed at equipment performance (both stand-alone and in systems), market dissemination of products and actors' orientation towards new products. In the energy efficiency context, market transformation aims to shift away from products with inferior energy use patterns by moving improved products to market faster and widening their share of the market (IEA, 1997)." Source: International Energy Agency (IEA), Energy Labels and Standards, OECD, Paris, 2000. http://www.iea.org/textbase/nppdf/free/2000/label2000.pdf.

¹⁵ Energy efficiency technologies included in the GHG Calculator consist primarily of technologies currently receiving incentives from investor-owned utility programs. Other off-the-shelf technologies are not included, and ARB's Draft Scoping Plan Appendices suggest a number of additional measures that are not included in Itron's set of measures. There are also many other delivery methods for energy efficiency that will require further analysis and evaluation. The Itron Goals Update report can be accessed at: http://www.cpuc.ca.gov/NR/rdonlyres/D72B6523-FC10-4964-AFE3-A4B83009E8AB/0/GoalsUpdateReport.pdf

¹⁶ There are three main differences between the Natural Gas Only Case and ARB's Business-as-Usual case: (1) ARB estimates a slightly higher rate of electricity load growth than that used by E3; (2) ARB assumes that no coal contracts expire between 2008 and 2020, whereas E3 assumes that California will not have responsibility for GHG emissions from coal contracts after their currently set expiration dates; and (3) ARB's Business-as-Usual case assumes a lower level of renewable energy in California than that included in the Natural Gas Only Case.

¹⁷ The business-as-usual case in ARB's Draft Scoping Plan projects electricity sector emissions of 139 MMT in 2020, which is 7% higher than the 129 MMT obtained from the GHG Calculator's Natural Gas Only Case.

¹⁸ Statewide retail electricity sales are estimated to total 277 terawatt-hours (TWh) in 2008, and to increase to 377 TWh by 2020 in the Natural Gas Only case. Statewide retail electricity sales in 2020 are estimated to be 321 TWh in the Reference Case and only 274 TWh in the Accelerated Policy Case (slightly less than the sales estimated for 2008).

¹⁹ ARB is modeling different scenarios of multi-sector GHG regulatory regimes and how these scenarios affect the State using the Energy 2020 model. In contrast, the E3 GHG Calculator focuses exclusively on the impacts of GHG policies on the electricity and natural gas sectors.

²⁰ A possible exception to this generality may occur in a GHG allowance cap-and-trade system with allowances allocated to electricity deliverers in proportion to some measure of output, which may not affect electricity prices, or not by as much as other approaches. However, the output-based allocation approach has never been implemented in practice, so the expected impacts of this approach have not been demonstrated empirically. For a more detailed discussion of the possible implications of output-based allocation approaches, see Section 5 of this decision, on allocation policy.

²¹ Integration costs include the cost of reliably incorporating intermittent resources such as wind and include the costs of increased ramp and regulation, and increased capital costs to increase the ability of the system to accommodate larger variations in generation output.

²² The "market clearing price effect" refers to the increase in wholesale electricity prices due to the introduction of a carbon allowance cost for electricity deliverers.

²³ Cites to parties' comments are to their opening comments due June 2, 2008, unless indicated otherwise.

²⁴ The Public Utilities Commission has defined and characterized the attributes of a REC for California RPS compliance in D.08-08-028 in R.06-02-012.

²⁵ "Incremental cost" is the difference in cost between a "normal" inefficient product and the substitute high energy-efficiency product.