Like most other would-be thoughtful commentators, we’re plowing through the thousands of pages of documents known collectively as EPA’s final Clean Power Plan, and are by no means ready to make a final pronouncement on its good or bad parts. But there is one dimension of the plan that strikes us as remarkable, and since nobody else has mentioned it, we thought it deserved some comment.

It has been widely noted that EPA abandoned Building Block 4 of its proposed rule, which would have required states to achieve substantial energy efficiency improvements to meet the proposed targets. What no one seems to have noticed is that the final rule–which has more stringent emissions targets overall than the original–assumes that virtually all of those emission reductions will result from energy efficiency. Or, more precisely, from EPA’s assumptions about how energy efficiency improvements by states will translate into reduced electricity demand. Here are the relevant numbers from Table 3.11 of the CPP Regulatory Impact Analysis (RIA); all generation is in Terawatt-hours, with the much higher EIA 2015 Annual Energy Outlook (AEO) Reference case–to which EPA’s Base Case is apparently related–for comparison:

Screen Shot 2015-08-18 at 11.43.21 AM

* Total Generation does not equal the sum of coal + gas + renewables because it also includes nuclear, CHP and sources that are not impacted by the CPP.

** CPP Gas = Existing NGCC + new NGCC + Combustion Turbine + Oil/Gas Steam

*** CPP Renewables = Hydro + non-Hydro + “Others” (MSW and fuel cells)

Thus EPA expects that the CPP will result in relatively small changes in both renewable and gas generation beyond what it thinks would happen anyway; about 5% more renewables and 5% less gas. Coal, on the other hand, takes a 22% hit, purely as a result of the overall presumed decrease in demand. Assuming that those retired coal plants emit at an average rate of 2,000 lb CO2/MWh, reducing coal generation by 322 TWhr reduces emissions by 290 million metric tons (“MMT”). Add in the reductions from lower gas (65 MMT) and we’re at 355 MMT, which is close to the total 374 MMT in reductions EPA expects from the rule.

While those hypothesized demand reductions might conceivably follow from major increases in electricity costs as a consequence of the CPP, that’s not what animates EPA’s demand reductions. The agency believes (although we don’t) that the CPP will have no effect on retail electricity prices at all [RIA, pp. 3-39].

Accordingly, there is absolutely no reason to believe that reducing demand by the amount postulated in the CPP is even possible. Here’s the relevant trend in electricity generation (with a comparison to EIA’s AEO, which provides a significantly different perspective). All figures are in Terawatt hours:

Screen Shot 2015-08-18 at 11.46.19 AM

In other words, EPA believes that when completely implemented, the CPP will result in electricity generation (and therefore demand from consumers) that is 8% below its own 2030 base case, and 14% below what EIA believes will occur based on current policies. In case you wondered, EPA grudgingly acknowledges in one of its Technical Support documents that EIA’s AEO Reference Case already includes at least some of the same programs driving EPA’s assumed “efficiency” gains.

How credible are EPA’s numbers? History shows that improving efficiency is not the same as reducing demand. Since 1949 (the earliest year for which EIA has records) there has been literally not even one 10-year period (let alone a 17-year period) over which electricity demand in the final year was lower than the first year. There were only 7 individual years where electricity demand was lower than the year before, in most cases by very little, and growth always resumed in the subsequent years. This is because (1) electricity demand is fundamentally a reflection of economic growth, and (2) improvements in energy efficiency often reduce the marginal cost of energy consumption, which, in turn, increases energy demand in many applications.

The chart below shows the trends and comparison to GDP growth over the same periods. Even the 17-year period through 2013, which included the severe 2008/9 recession and slow subsequent recovery, saw electricity demand grow by over 22%. Since 2000, the average 17-year growth rate is 43%.

Screen Shot 2015-08-18 at 11.34.57 AM

The CPP envisages electricity generation being basically the same in 2030 as it was in 2013, with cumulative growth of just 1.6 % over those 17 years. EIA assumes growth of 16% over that same period. So we have at minimum, a very large potential demand shortfall v. generation problem.

So what happens if it turns out that EPA’s assumed electricity demand number is too low? Under a truly rate-based system, there would be no problem as long as the megawatts were generated from a system where the composite rate met the target emission rates per MWhr. But there are a host of practical and internal-to-CPP barriers erected to make it almost impossible for states to adopt a rate-based system that would average out their megawatts and their emissions to meet the targets.

Instead, states are forced / “encouraged” to adopt the mass-based system which sets an absolute cap on emissions. The economics alone will likely drive states in this direction given that that EPA estimates that a “rate-based” system will cost an annualized $8.4 billion by 2030 while a “mass-based” system will cost but $5.1 billion [RIA, p. ES-9]. And in such a mass-based system, states will have only one short-term option: trading emission allowances with other states to meet their targets.

This might be reasonable were the assumed demand reductions not so large. But the EPA assumes–and has built into the state CPP caps–that we can do without one-seventh (EIA’s estimate) or, at minimum, one twelfth (EPA’s Base Case) of the generation required in 2030 under business-as-usual. But if the CPP demand assumptions are wrong–and the data quoted above suggest there is every reason to believe that they are optimistic–we could be in for some very expensive surprises, especially if they are wrong across the board and few states have excess allowances to sell.

In the longer term (2-3 years minimum) states could encourage providers to build new renewable capacity that (somehow) will overcome intermittency and meet base-load demand. Or they could add new gas-fired generation, which may be feasible under the new source provisions of the CPP (see below) but would mean missing the CPP emission targets, perhaps by a lot.

So what will happen? EPA claims in its Final Rule on page 349 that “the emission limitations reflective of the BSER [Best System of Emission Reduction] are achievable even if aggregate generation is not reduced through demand-side EE [Energy Efficiency].” That is correct: states could either mandate enough renewables to completely cover the demand (a potentially enormous amount of new capacity), or some additional renewables and gas-fired generation for back-up. Alas, this runs counter to EPA’s declared policies of avoiding leakage to new sources and not encouraging additional natural gas-fired capacity.

In fact, in order to avoid “leakage” to new sources, EPA is requiring states that adopt a mass-based cap to also adopt one of two measures to prevent this from happening.

  • Include new sources in the cap and allocate to all generators an additional 2.5% allowances, thus effectively capping new generation at that level; or
  • Set aside a number of allowances to be allocated to renewable providers and then (presumably) sold to new fossil fuel providers, which also effectively caps new gas generation.

Putting aside whether EPA can legally cap new gas-fired capacity in the context of a rule dealing with existing sources, we will see if either would provide an attractive or viable basis for a long-term investment like a power plant.

It is a considerable achievement for a single set of assumptions in the CPP to have such an impact on the feasibility of the plan and its potential economic and environmental consequences. We’d be a lot more comfortable if the demand number was not “exogenous” (in the technical vernacular), to the modeling process; i.e., input as an assumption, with the suspicion that it is there to make the emission reduction numbers work. We look forward to seeing the electricity industry’s reactions.