EPA’s proposed Clean Power Plan uses a rarely used section of the Clean Air Act, Section 111(d) to regulate existing fossil-fired electric generating units (EGUs). This part of the Clean Air Act, like the more familiar provisions governing ambient air quality for more traditional pollutants, gives EPA the task of determining an acceptable target for emissions and leaves it to the states to figure out how to achieve that target. One aspect of EPA’s proposal that’s different from other comprehensive power sector regulations like the NOx Budget Trading Program or the Cross State Air Pollution Rule is that it’s not straightforward to figure out how the agency set targets for each state. For those rules, EPA, using the Integrated Planning Model (IPM), simulated the impact of a pollution tax that produced an acceptable level of abatement and then allocated responsibility for reductions in states based on their power plants’ responses to the modeled tax. In essence, EPA used a price to set the quantity. This approach was recently upheld by the Supreme Court.
For the Clean Power Plan, EPA describes four “Building Blocks” that it says, when combined, determine the target, called the Best System of Emission Reduction or BSER, that each state must meet. Nowhere in the 645 pages of the proposed rule does EPA spell out what marginal abatement cost it had in mind when it set the targets. So I set out to dig into the technical support documents for the rule to figure out how EPA set the goalpost, and if there was a straightforward way to measure the that goal against more familiar cap-and-trade or carbon taxation strategies. Here’s what I found.
Building Block 1: More Efficient Coal
EPA's offline estimate is that a 6% heat rate improvement will occur at a cost of less than $8 per ton of carbon dioxide (CO2).
Building Block 2: Redispatch From Coal to Gas
As many suspected, EPA imposes a carbon price in their IPM to get the redispatch they view as feasible (70% capacity factor at existing NGCC units). The tax is between $30 and $33 per ton of CO2. Note this is done after applying the demand reductions envisioned in Building Block 4 but not the renewable and nuclear generation envisioned in Building Block 3 (see below). There are 10% cost-savings when compliance is regional rather than state-by-state—smaller than I would have thought. EPA doesn't break out state-by-state impacts. One could easily imagine that there will be holdouts because a regional approach creates costs for them (see, for example, Bonneville Power). On the other hand, not having to deal with leakage issues may push states toward a regional approach.
Building Block 3: Build Renewables, Keep Nuclear
The IPM, when forced with the carbon tax produces only about 20 GW of additional renewables by 2020 and only 6 GW more by 2030 than the reference case. The effect of the carbon price is to accelerate deployment, not increase total capacity added by 2030. That means EPA must do something else to reach its targets.
For renewables, EPA calculates regional targets based on current renewable portfolio standards and assigns to each state an obligation to hit their share of the renewable energy target. The regional targets vary between 10 and 25%. Some states have already hit their targets and others are on track to do so early. EPA doesn’t supply a carbon price for these policies. Presumably it would be higher than $30 per ton or else the IPM would be building a lot more renewable generation in the policy scenario (It’s also possible – even very likely—that the cost assumptions for renewables are too high in IPM).
For nuclear, EPA estimates that keeping the approximately 6 GW of existing capacity that the Energy Information Agency (EIA) has characterized as "at risk" would cost $12–17 per ton. This is consistent with economic analyses of new units at the Vogtle plant in Georgia that assume it is in the money at carbon prices of $10–20 per ton. There's no discussion of how relicensing and expiration of second licenses for existing units plays into this calculation—I'm sure that there are at least some units whose second 20-year license will expire in the 2020s but I'm not sure how many. My guess is that 6 GW of at-risk nuclear is a serious underestimate, particularly when combined with the level of other renewables assumed in EPA’s targets.
Remember that neither the additional renewables or preserved nuclear is integrated into their dispatch model and so doesn't take account of electricity market equilibrium effects. It would be interesting to see how many of the nuclear units remain operational in IPM if the model is forced to implement the renewables target without some sort of explicit carbon price. Given that the rule allows but doesn’t require cap-and-trade for compliance, this is a real possibility.
Building Block 4: Energy Efficiency.
EPA takes the EIA electricity demand forecast (about 1% annual growth) and then modifies it on a state-by-state basis by adding in a reduction factor to account for demand-side efficiency programs. The assumed levelized cost of this program is 8–10₵ per kwh. Interestingly, this may be higher than for distributed generation on the timescale of the rule. Marginal abatement cost is estimated by using IPM output for reference and policy cases and comparing that to program costs and estimated energy savings. The bottom line number is $18–24 per ton of CO2. EPA uses the EIA's AEO as its reference demand forecast. This is likely to inflate the baseline for a variety of reasons. Also, EPA relies on questionable (albeit improving) EIA utility self-reported data to estimate the effectiveness of efficiency programs.
A lot of this information is in the GHG Abatement Measures TSD.
To me, this whole thing is reminiscent of California—lots of shadow carbon prices that are higher and lower than the visible power sector price that’s driving the cost-effective abatement strategy—redispatch. And it makes a fantastic argument for legislative action to implement a simpler and more cost-effective policy. Why can't we just have a carbon tax and cut my FICA withholding already?
