   A. Trading System

These comments focus on the structure of the emission trading system envisioned by EPA, including in particular (1) the classification of states into two distinct groups with no trading allowed between groups (Section VI of the preamble); (2) the determination of variability limits (also Section VI); (3) the allocation of allowances (Section VII); (4) the assurance provisions (also Section VII); and (5) the use of banked NOx allowances from the CAIR program (Section IX).  Attention is focused on these areas because the reviewers following up on these issues also believe that the scope and stringency of the rule are appropriate in terms of their economic impact; indeed, if anything, based on the analysis EPA has provided, a benefit-cost perspective would support greater reductions in emissions than EPA is requiring (The more stringent scenario has a cost estimate of $2 billion more, with additional benefits of $20 to $50 billion).  This makes it seem as though the more stringent scenario would be the preferred one.  Moreover, these reviewers believe that the determination of state-level budgets, which was a crucial issue in the court decision regarding CAIR, is both analytically rigorous and responsive to the court's concerns in North Carolina.  Indeed, EPA is to be commended for the terrific job it has done to produce a rule with benefits that vastly outweigh the costs, in a manner that is responsive to a difficult and disappointing court decision.  It is the soundness of the overall rule that provides the opportunity to focus more narrowly on critiquing and potentially improving the design of the market mechanism that EPA envisions to implement it.

These comments are presented according to the order that the issues appear in the preamble. The overarching theme is that the limitations on trading imposed by EPA, although well intended, are likely to greatly impair the program's ability to achieve emissions cost-effectively.  Much of the problem can be traced to the design of the assurance mechanism, and in particular the way that it would impose penalties on individual units or groups of units for exceeding an essentially arbitrary limit on emissions.  

1. EPA should reconsider its decision to prohibit trading between Groups 1 and 2.

As in the proposal, the final rule would define two distinct groups of states with respect to required SO2 emissions reductions.  Group 1 states are those for which EPA calculates allowable emissions budgets on the basis of a marginal cost threshold of $2300 per ton of SO2 emissions; Group 2 states are those for which the modeled contribution to downwind air quality problems (nonattainment or interference with maintenance) is eliminated at a marginal control cost of $500/ton.  Under the trading system envisioned by EPA, EGUs in Group 1 and Group 2 states would be issued distinct kinds of emission allowances; units in Group 1 states would not be allowed to use Group 2 allowances for compliance, or vice versa.

EPA explains its approach on p. 245 of the Preamble by arguing that "to allow Group 1 or Group 2 allowances to be used interchangeably ... would be to allow the shifting of reductions from areas where they are needed to eliminate significant contribution to areas where they are not needed to eliminate significant contribution."  But the same logic would apply to any interstate trading.  Consider a hypothetical example in which units in state A (in the aggregate) have surplus allowances to sell to units in state B, and units in state B (in the aggregate) have demand for those allowances.  Trading between the states would shift reductions from state B to state A.  By definition, this results in shifting reductions from "areas where they are needed to eliminate significant contribution to areas where they are not needed"  -  even if the two states are in the same group.  That is because EPA constructs state budgets by calculating the emissions left after subtracting the reductions needed to eliminate a state's significant contribution.  As a result, by definition the reductions in state B that end up not occurring are needed to eliminate significant contribution, and the excess reductions in A that result from the trade are not.  Hence EPA's main rationale for not allowing intergroup trading fails because it holds equally well for intragroup trading, which EPA allows.

EPA's rationale is unconvincing for a second reason (noted by the commenters): any restrictions on intergroup trading are entirely unnecessary given the assurance mechanism that EPA is also putting in place.  The assurance mechanism ensures that the rule makes measurable progress toward eliminating the significant contribution of upwind states to downwind nonattainment problems.  With it in place, no further limitation on trading is necessary.  In response, EPA argues that "allowing for trading between the two groups ... would increase risk of a state exceeding its variability limit."  But again, this argument is unpersuasive, for the same reason as noted above: it applies equally well to trading between any two states, whether or not they are in the same group.

The determination of which states are sellers and buyers of permits depends on the marginal abatement costs, not the level of stringency, though those can be correlated.  Has the EPA determined that Group 2 states are more likely to be sellers of permits to Group 1 states that are other Group 1 states?

Meanwhile, there are clear drawbacks to prohibiting intergroup trading.  First, maintaining two separate markets will increase the complexity of administering the program  -  an issue that EPA emphasizes as one of its primary concerns in designing the program elsewhere in the Preamble (e.g. in the discussion of the assurance provisions).  Perhaps more importantly, preventing intergroup trading will inhibit trading and reduce market liquidity, limiting the cost-effectiveness of the market mechanism.  As discussed below, the assurance mechanism already threatens to interfere with the functioning of the market.  But at least in that case there is a clear need for the assurance mechanism as a response to the court's concerns in North Carolina.  In contrast, there appears to be no good reason for limiting trading among groups of states.

It is worth noting that this is an easy change to make: it would require nothing in the way of new modeling and little work beyond rewriting the Preamble.

   > EPA should reconsider its prohibition of intergroup trading in light of these concerns.
   > If EPA chooses to continue to prohibit intergroup trading, can it provide any data or modeling to support its concern that allowing intergroup trading will put greater pressure on the assurance mechanism or otherwise undermine the performance of the rule?

2. EPA should consider raising the variability limits to 15% or 20% of the state budgets.  At a minimum, more analysis is needed to (i) analyze the historic variability of emissions (not conditional on emissions rates) and (ii) consider how the choice of 10% interacts with the assurance provisions to translate into limits on emissions at the level of individual units or designated representatives.

EPA elects to use 10 percent as the variability limit in almost all cases, on the grounds that "[m]ost states' historic variability in year-to-year emissions of the covered pollutants ... were within 10 percent of their average annual emissions" (p. 261).  This variability limit then determine the absolute upper bound for allowable emissions for each state. However, EPA's choice of 10 percent appears to be an unnecessarily tight constraint, for two reasons.

(i)  First, EPA states that the variability limit is based on the variation in emissions "holding emission rates constant"  -  in other words, the variation in heat input but not in total emissions.  While EPA is right to note that heat input may vary as a result of factors outside of a unit's control, it is also true that a unit's emissions rate may vary as a result of such factors.  For example, coal varies widely in sulfur content  -  even for coal of otherwise similar characteristics from similar regions.  It would not be surprising if the sulfur content of the coal burned in many units varied by at least 5 percent in a given year.  As another example, scrubbers may break down, or be taken down for repairs, etc.  It is not hard to imagine that relatively short scrubber outages could result in variations of at least 5 percent; given that wet scrubbers typically reduce 95 percent of SO2 from the flue gas, taking a scrubber offline for a month would increase emissions by roughly 8 percent of uncontrolled emissions (and thus a much larger percentage of controlled emissions).

The point is that absolute emissions at an individual unit may vary from year to year around a long-term average, not only because of variation in heat input but also because of unanticipated and hard-to-control variation in emissions rates.  In considering only variation in heat input, EPA is missing an important source of unit-level variation.  On the reasonable assumption that variation in heat input and variation in emissions rates are uncorrelated (or at least not negatively correlated), EPA's failure to consider variation in emissions rates will result in an understatement of the true variability in unit-level emissions.

In a trading system, of course, these variations are inconsequential, because they end up as "noise" in unit-level emissions that can be covered by allowance purchases.  However, the proposed assurance mechanism would impose penalties at the level of individual DRs for exceeding strict upper bounds on emissions, defined by using EPA's variability limits.  Calculating those variability limits on the basis of variation in heat input only, as EPA does, will result in unnecessarily strict limits on compliance flexibility and a corresponding loss in cost-effectiveness.

   > Has EPA done an analysis to calculate the variability of absolute emissions (rather than the variability of heat input)? 

(ii) Second, in discussing its selection of a variability limit, EPA only considers historical variation at the state level.  In doing so, EPA seems to have overlooked the fact that the variability limit will effectively apply at the level of the designated representative (DR)  -  not at the level of the state.  That is because the assurance mechanism penalizes units (aggregated to the DR level) for emissions in excess of their initial allocation plus their share of the state's variability limit.  As a result, almost all units (at the DR level) face a hard constraint on their emissions equal to 10 percent above their allocation.

This could severely undermine the cost-effectiveness of the program, since it sharply limits the ability of some units/DRs with relatively high marginal abatement costs to comply by emitting relatively large amounts and buying allowances to cover those emissions.

Nowhere, however, does EPA do any analysis on variability at the DR level.  As EPA acknowledges in the context of discussing variability in individual states versus regions, variability is typically lower at higher levels of aggregation.  The same applies in reverse: variability is likely to be greater at lower levels of aggregation, e.g. at the level of designated representatives relative to states.  Thus even if state-level historical variability is less than 10 percent of annual emissions, variability may be greater than 10 percent at the level of individual units (or DRs).

Moreover, EPA's discussion suggests that the benefits from imposing a 10 percent variability limit rather than a 15 or 20 percent limit are negligible.  According to EPA, modeling of different variability limits "suggests that the air quality impacts are small when all upwind states linked to a particular receptor monitor increase their SO2 emissions to any of the variability levels (5, 10, 15, or 20 percent)."  This statement seems to suggest that EPA could fairly easily justify 20% variability limits.

   > Has EPA done an analysis of emissions variability at the unit and/or designated representative level?  Can it share that analysis with interagency staff?
   > In light of the minimal air quality benefits but potentially adverse impacts on cost-effectiveness, EPA should reconsider its choice of a 10 percent variability limit for most state-pollutant pairs, and instead consider using a higher limit (15 or 20 percent).  At a minimum, EPA should provide a more explicit discussion of its choice in light of the proposed assurance mechanism.

3.  EPA should reconsider its decisions for allocating allowances to new and retired units

Although EPA's basic allocation methodology appears sound, the treatment of units that are entering and exiting the program raises some potential concerns.

(i) For new units, EPA envisions a complicated two-step methodology that is poorly explained in the rule (pp. 365).  As a result, it is hard to tell what the impact of the allocation will be.  However, it appears that the consequence of EPA's approach  -  in the absence of a shortage of allowances for new units  -  will be that units in their first year of operation are guaranteed to receive exactly as many allowances as they need to cover their emissions.  In other words, they will face an effective marginal price of zero.  (This is because EPA proposes that "a unit's new unit set-aside allocation initially equals that unit's emissions for the control period ... in the preceding year" (p. 371).)  Doesn't this provide an underincentive for new units to control emissions?  The same issue also arises for the allocations to new units in subsequent control periods, which the Preamble also explains poorly.  

This discussion raises a number of questions:
   > To the extent that EPA bases new unit allocations in a given period on a unit's emissions in the prior control period, how can EPA avoid effectively eroding the incentive for new units to reduce their emissions? 
   > To the extent that units receive more allowances in future periods as a result of increased emissions in a current period, won't this create an incentive to overemit in early periods?  
   > Is there an alternative, more neutral approach that EPA could use instead?  
   > Can EPA provide a clearer description of the new unit allocation rule and a fuller explanation of the resulting incentive impacts?

(ii) For existing units, in the final rule EPA has revised its provisions regarding units that cease operation.  At proposal, EPA proposed continuing to allocate allowances to units for six years after they ceased operation.  In the final rule, EPA plans to reduce the window of time to four years.  In explaining this change, EPA does not provide any discussion of the implication for incentives for retiring coal plants.  By reducing the allocation of allowances to units that retire, EPA is increasing the incentive to keep those old plants operating  -  to the detriment of public health and other policy considerations. 
 
   > Can EPA explain its reasoning in greater detail and specifically address the impact on incentives to continue to operate aging coal plants rather than retiring them?

Finally, when will the unit-by-unit allocations be provided?  When will the public and states see them?

4.  EPA should invest significant time in reevaluating and reconsidering the design of the assurance mechanism.

EPA faces constraints as a result of the court's ruling in North Carolina.  However, it appears the agency has not fully appreciated the extent to which its proposed assurance mechanism may limit trading.  While agency staff have correctly pointed out concerns with alternatives, they have not fully considered concerns with their proposed approach.  Moreover, the assurance mechanism is at the heart of the trading mechanism used to implement the Transport Rule, and its performance will have significant implications for future market-based mechanisms.  

The primary problem with the assurance mechanism is that it will impose very tight and essentially arbitrary hard ceilings on emissions at the level of designated representatives.  No DR will be able to emit more than 10 percent of its allocation without risking penalty.  Because the allocations are essentially arbitrary (connected to past heat input) they do not bear any relation to actual emissions.  That is okay from the perspective of allocation per se, but EPA does not discuss how the allocation decision interacts with the assurance mechanism.

   > Has EPA done any analysis of how large the designated representatives are likely to be?  On a call, EPA mentioned how many there might be per state; but this is a different question, namely what is the distribution of the number of units (and the diversity of fuel sources/emissions rates) in designated representatives?
   > How do the allocations at the DR level compare to baseline emissions?  
   > Has EPA modeled the ability of individual units or DRs to comply with the allocation-plus-10-percent ceiling, using the allocations in the final rule?

As a result of this tight constraint, it seems likely that the assurance mechanism will severely limit trading.  It is hard to imagine, for example, that a regulated utility would choose to comply by leaving its emissions uncontrolled and covering those emissions with purchased allowances, because that would leave it exposed to the risk of higher-than-expected emissions and a steep penalty.

   > Has EPA considered the incentives facing regulated generators, including the likely treatment by PUCs of emission allowances, penalties, fuel-switching costs and so on?  Has EPA considered the incentives facing regulated generators, including the likely treatment by PUCs of emission allowances, penalties, fuel-switching costs and so on?
   > One potential response to these concerns might be that other concurrent EPA regulations are likely to end up requiring significant investment in advanced pollution control equipment at existing units anyway.  Has EPA modeled the performance of the assurance mechanism taking other rules into account, such as MATS? 

A second problem that EPA does not appear to have recognized is that the proposed assurance mechanism depends very heavily on the allocation mechanism used.  It is designed specifically with EPA's proposed allocation rule in mind, since the penalties for individual units (or DRs) are based on their allocation plus the variability limit. 

   > How does EPA propose to implement the assurance mechanism under alternative allocation schemes that states might enact as part of their SIPs, such as allowance auctions?

Finally, EPA appears to have not fully appreciated how the assurance mechanism interacts with other enforcement provisions.  EPA points out that a violation of the assurance mechanism will not constitute a violation of the Clean Air Act  -  presumably by way of reassuring people that the consequences of unexpectedly exceeding the emissions limit will be limited to a higher marginal price of emissions.  However, EPA also specifies that "failing to hold sufficient allowances to meet the allowances surrender requirement [under the assurance mechanism] will be a violation of the regulations and the CAA" (p. 386).  Effectively this means that entities will need to purchase and hold excess allowances in preparation for the potential eventuality that the assurance mechanism might be triggered and that their emissions might exceed their DR-level variability limit.  In principle, an entity might wait until after a control period or even just before the "true-up" date to see whether extra allowances were needed; but in practice many entities have expressed a preference for managing their allowance holdings in a more continuous fashion, i.e., ensuring that they are accumulating allowances in line with their emissions.  In that case, the threat of very steep penalties under the CAA may appear to be a real one, and entities contemplating compliance strategies that would put them at risk of triggering the assurance mechanism may act as if they face an effective marginal cost that is much greater than three times the allowance price.

   > Has EPA considered how entities are likely to comply with the assurance mechanism given the prospect of full CAA enforcement for failing to hold sufficient allowances?

Again, the common theme is that EPA does not appear to have carefully thought through how the components of its scheme interact.  While each of the components (the variability limits, the allocation rule, the assurance mechanism) may appear reasonable in isolation, in combination they appear very likely to stifle the development of an emissions market and undermine the scheme's cost-effectiveness. 
The complexities are compounded in regulated electricity markets, where compliance expenditures can be added to utilities' rate bases and earn a rate of return, whereas penalty payments most likely cannot.

   > What has EPA done in the way of "worst-case" modeling to capture what might happen if the assurance mechanism effectively shuts down emission trading markets?  Can EPA compare the cost of that scenario to the cost of other scenarios, such as one that only allowed intrastate trading (but in which intrastate trading was robust)?

During this review, we have discussed the possibility of the alternative mechanism proposed by commenters, with a parallel assurance market and interstate emissions trading market.  EPA has chosen to reject this approach on two grounds: complexity and market power.  The complexity argument is unpersuasive, for several reasons.  First, the complexity could be reduced simply by allowing trading between Groups 1 and 2, as recommended above.  That would cut the number of extra markets by 25 percent.  Second, EPA appears to be understating the complexity of its proposed approach.  Implementing the assurance mechanism in the final rule will require EPA to calculate allocations (or some equivalent benchmark in states that choose to use other allocation rules  -  see discussion above) at the DR level, compare those to actual emissions levels, and calculate the penalty.  How is this easier and less complex than running a separate market?

Finally, EPA appears to be overstating the additional complexity of having separate assurance markets.  Given automation and computing power, the costs of actually administering the markets are likely to be small.  Moreover, although there may be startup costs associated with introducing an unfamiliar assurance market to utilities, those are fixed initial costs rather than ongoing ones.  It is worth noting that EPA considered in its proposal having intrastate markets, and this would be no more complex to administer than that approach.

   > Does EPA have actual experience or data it can point to support its concern that there will be large administrative costs associated with multiple markets?  Can EPA explain in greater detail why its proposed approach is less complex?

The market power argument appears to hold more weight, but is worth digging into some more.

   > Can EPA share its analysis of market power? Was this in one of the TSDs?
   > How do the states in which market power would be a concern line up with the states that have regulated electric power sectors?  Has EPA considered the ability of state regulators to address and mitigate market power considerations?
   > In considering its intrastate approach, did EPA consider any design features that might limit market power, that might be applicable to the case of assurance markets?

In general, more careful thought needs to be given to alternatives, including the two-track market idea.  In addition to its elegance and structural simplicity (which may or may not be offset by administrative complexity), the assurance market has three clear and important advantages.  First, it would offer transparency and predictability for market participants, by providing a clear market-driven price signal of the likelihood that the state variability limits would be reached in a given year.  Second, it could provide an ironclad assurance that state variability limits will be met, or alternatively could be tweaked for more flexibility if desired (e.g., by allowing states to sell additional assurance allowances at a set price, such as three times the market allowance price).  Third, it can work independently of the particular allocation mechanism that a state uses to distribute emission allowances.

While the market power concern is worth considering, one can imagine addressing it through fairly simple design changes.  For example, freely allocating the assurance allowances to individual units (e.g., on the basis of historical heat input) could help mitigate market power, since all units will have the option to hold onto their assurance allowances.  Or, if (as suggested above) states were allowed to sell additional assurance allowances at a set price, the market power problem would be largely mitigated.

These comments are not meant to overemphasize the assurance market mechanism, which may not be the best approach.  However, it appears EPA has done a much better job critically evaluating alternatives than it has done in looking squarely at the problems with its own approach.  More work is needed in this area.

5.  EPA should consider allowing NOx allowances to be carried over into the Transport Rule program.

Several reviewers believe that EPA appears to have strong grounds (given the North Carolina decision) for not accepting SO2 allowances into the Transport Rule program (although we are still getting interagency comment, and on this point in particular we may want to have further conversations about this issue).  However, its rationale is not as strong for NOx allowances.  EPA defends its decision on the basis of three arguments: (1) the bank of CAIR NOx allowances is too large to be added to the state budgets calculated in the Transport Rule; (2) allocating unit-level Transport Rule allowances on the basis of banked CAIR allowances (e.g. by allowing banked allowance to be exchanged for Transport Rule allowances) would invite unacceptable legal risk by linking Transport Rule allocations indirectly to the fuel-factors approach used under CAIR and rejected by the court; and (3) allocating Transport Rule allowances on the basis of the CAIR bank would raise technical difficulties because of the time needed to determine final CAIR allowance holdings.

On the first of these arguments, it would be useful for EPA to share data on the provenance of the existing bank of CAIR NOx allowances, to help evaluate the argument that these represent "excess" emissions that should not be allowed.  On interagency calls, EPA has said that a significant fraction of the outstanding allowances were excess or bonus allowances of some sort, but we haven't seen hard data on that.  To the extent that they were simply carry-overs from the earlier NOx SIP Call, they could still represent legitimate early reductions.  
   > Can EPA share data on the existing bank of NOx allowances and how they were generated?

On the second argument, the legal risk does not seem particularly acute given that the current holders of allowances are not necessarily the same as the entities to whom the allowances were originally allocated.  
   > Can EPA provide data on that point, at least in the aggregate?

Moreover, even if the current banks largely reflect the original allocations, the court's ruling in North Carolina applied strictly to the question of how to allocate allowances among states, not among units within a state.  Couldn't EPA allow for some (perhaps limited) exchange of banked NOx allowances without disturbing the state-level budgets calculated under the Transport Rule?

In addition, has EPA considered reviewing approved SIPs as a proxy for acceptable allowance banks?  If approved SIPs do not include fuel adjustment factors, they should be carried over.

The third argument is important, but there might be a simple solution.  What if EPA assigned unit-level allocations under the FIP for the 2012 control period, and then provided that existing banked NOx allowances could be exchanged for allowances in subsequent years?  In other words, why can't EPA simply suspend the use of banked allowances for a year, rather than making them worthless?


The bottom line is that the principle of continuity in an emissions trading program should be of paramount importance, and EPA should seek to respect it unless there are very strong legal reasons preventing them from doing so.  That appears to be the case for SO2 allowances, but not for NOx allowances.  The simple fact that there are a large number of banked NOx allowances is not grounds for eliminating them  -  to the contrary, it is something EPA should celebrate, since it represents early reductions made under previous programs.  The existing rule gives lip service to the importance of continuity but honors it only in the breach.

Further, sunsetting the existing Title IV SO2 and CAIR banks in 2011 may likely set a bad precedent.  Assigning no value to existing Title IV and CAIR emission banks because of their lack of future use will increase uncertainty in the value of any cap‐and trade market from this Transport Rule or any other future rule, including any future CO2 cap‐and‐trade market. This rule has the potential to adversely affect all future cap and trade markets. EPA should give serious consideration to some method for pre‐2012 allowance use. 
This may also lead to perverse incentives and uncertain emissions for 2010.   With the elimination of Title IV and CAIR allowances after 2011 and the complete loss of economic value to 2010 and 2011 allowances and their associated banks, the draft rule appears to cause the opposite of early emission reductions. Has EPA considered the perverse incentive created by a no carryover requirement, i.e., that holders of allowances may have an incentive to "dump" emissions before allowances expire? 
6. EPA should consider the implication for other banked permits
Declining to carryover all CAIR allowances would also affect the Title IV allowance market (effectively rendering them valueless).  Given this impact, EPA might still lawfully base the Transport pool of allowances on existing volume of banked Title IV allowances.  This approach would be similar to that of the Compliance Supplement Pool (CSP) used in early NOx emissions programs transitions.  The reference to the Title IV bank needn't be a full replication if the agency is concerned about the court's decision regarding the limits of Sec. 110(a)(2) authority in relation to Title IV markets.  In addition, EPA has received public comment suggesting this approach.  What is EPA's response to this approach?

7.  Further interagency discussion of other trading mechanisms

The RIA discusses alternatives: one more stringent and one less stringent.  But it does not discuss alternative trading mechanisms.  Further, in order to know the benefits of allowing limited interstate trading, relative to the costs (especially the risk of there being no trading at all), we need to know how much the marginal abatement cost varies across states and within states.  Has EPA made any attempt to estimate that and is it reflected in the rule or the RIA?
We would like greater discussion of the following:
   1. Only intrastate trading allowed.  This might have two disadvantages:
         a. Market power in smaller states.  But market power leads to an inefficiently low number of trades, which would be preferred relative to zero trades.
         b. Numerous markets: one for each pollutant in every state.  The complexity inherent in a multiplicity of markets would be offset by the simplicity of each one of the markets.  Within markets, there would be no penalties, no uncertainty over what other utilities in the state do, no risk of exceeding assurances, etc.  The preamble claims (p.280) that intrastate trading would be more resource intensive for sources, but it is not clear why that is true.  The rule also claims that the intrastate only option would be less transparent.  Again, that seems unclear. 
   2. Some version of the dual-track trading program mentioned during one of our phone calls.  There would be state-specific permits and (fewer) interstate permits.  Each ton of emissions would require one of each type of permit.  No state could exceed its assurance level, because of the fixed number of state-specific permits.  But utilities in states with low compliance costs could sell interstate permits to utilities in states emitting more (though still less than the assurance level).  The market price of the state-specific permit would be zero in states selling the interstate permits, and positive in states buying the interstate permits.

   A. FIP/SIP Structure
Many states requested that EPA put out a rule that will allow for states to submit SIPs and manage their own program as soon as possible.  There was concern that EPA's timeline with Phase I starting in January, 2012, and Phase II starting in January, 2014, does not allow time for states to develop and submit SIPs to EPA.  This in turn will mean that they will be subject to a FIP for a number of years, which may lead to higher costs/burden for both the state agencies and the regulated entities.  This may also carry a legal risk.  
   > Has EPA considered other options for allowing great state control of this program, outside of what has been outlined in the draft final rule? 
   > For instance, has EPA considered eliminating Phase I and instead starting with Phase II? 

As one commenter suggested, EPA cited the need for a Phase I requirement in order to assure there were sufficient allowances banked by Phase II, but they requested that EPA skip Phase I and instead add a CSP for Phase II.  

   B. Impact of CAIR 
Could EPA provide an analysis where you utilize the emissions reductions due to the CAIR rule in projecting 2012 design values?  Although this rule replaces CAIR, we are interested in seeing whether CAIR reductions have already alleviated some of the attainment and maintenance issues identified in this final rule.    Further, has EPA considered utilizing CAIR outcomes (perhaps in conjunction with approved SIPs) as a more flexible approach to determining inclusion of transport states?

   C. Electricity Price Modeling
The RIA examines the effects of the rule on electricity prices.  What assumptions about trading are built into that estimate?  For example, is the assumption that all market-clearing trades take place so long as they don't exceed state variability levels?  If so, is that realistic?  If the complexities of the assurances and penalties result in there being no trading, what would be the effect on electricity prices?

   D. Timing/Transition
It is unclear if states and affected facilities will be prepared for a January 1, 2012 start date, especially given other changes that EPA is making in the draft final rule.  For instance, modeling results used in the final rule are substantially different than those in the original August 2, 2010 Proposed Rule and subsequent notices.  Six (6) States are being dropped from the proposed rule; Texas is being added; 3 States have their SO2 Group status change; and the sheer magnitude of change to the budgets of all of the states results in a significantly different rule than originally proposed. 
In the sixteen (16) States where the EPA reduced SO2 emission budgets for 2012, the reduction in the State's trading budgets are dramatic ‐ ranging from 2% to 69% and averaging 26%. 
Similar issues are evident with the 2014 SO2 State trading budgets. Budget reductions range from 7% to 72% with an average of 26% in the 19 affected states. The Tennessee 2014 SO2 budget has been reduced by 41%. 
For Annual NOx budgets, sixteen (16) States face budget reductions of 2 to 40% with an average of 18% for the 2012 budget. In 2014, seventeen (17) States face budget reductions of 4% to 50% with an average of 23%. In Tennessee the annual 2014 NOx budget has been reduced by 32%. 
EPA has not provided details that would show what site‐specific controls/measures the agency's modeling indicates would be necessary to eliminate a State's significant contribution. 
Further, accelerating the date the assurance provision becomes effective from 2014 (in the proposed rule) to 2012 (latest interagency draft), greatly changes compliance planning for 2012 and 2013. Such a substantial change occurring six month prior to the effectiveness of the assurance provision leaves sources with few options to respond in a cost‐effective manner, increasing the likelihood of disrupting system reliability if it becomes necessary to achieve compliance through derates and/or idling.  Can EPA explain why they are accelerating the assurance provision effective date? 
 Units may have to be idled to meet allocation caps until controls are installed thereby increasing costs and decreasing reliability.
EPA control installation schedules do not account for accessibility to construction sites at existing plants. EPA's examples of rapid control equipment installations were at sites with good access around the units allowing rapid construction of new controls. These are atypical of most installations. 
Controls needed to achieve reductions under the TR rule would be very site specific due to varying unit designs, fuels burned, and generation requirements (capacity factor, etc.) Accordingly, site specific detailed designs would be necessary, increasing the time necessary to achieve the reductions. The EPA schedule doesn't allow adequate time for project planning, scope development and detailed design.  EPA received comments on this issue, and others related to the estimated construction time in its modeling.  Further information can be provided on this issue, for the SCR and FGD lead time assumptions.  
   > Did EPA make any changes to its estimated construction times? 
   > How do longer equipment installation times affect the achievement of the current limits?  Has EPA done any sensitivity analysis on these assumptions?  Was this part of a reliability assessment?

   E. Technology Concerns
EPA estimates that one third (3 GW) of the SO2 reduction technology installed to meet the Transport Rule are projected to be dry sorbent injection (DSI). Nearly all of the 5.9 GW of FGD retrofits are comprised of 12 units at 7 plants.  We believe that the number of FGDs that may be necessary to meet the TR reductions may have been underestimated. It is unlikely that reliance on lower sulfur coals and on dispatch of lower‐emitting generation units alone can achieve these reductions, setting aside the technical problems involved with switching to lower sulfur fuels. 
DSI adoption is assumed to occur at an SO2 price of $2,300/ton.  FGD assumed to be built at $1,600/ton.  This set of assumptions is at odds with those made in the utility MACT rule, where DSI was viewed as a less expensive control option than FGD.  Please clarify.
Utilities must plan controls to meet all anticipated regulatory requirements including the proposed Utility MACT. To this end, the TR proposal's reliance on DSI is unwarranted. It would not be prudent to install a technology (DSI) to meet the reductions required under the Transport rule if a different technology (FGD) may be necessary to meet the standards of the Utility MACT. Required controls for MACT are anticipated to be more complex and require longer construction cycles than what EPA is anticipating for the Transport Rule. Also the proposed MACT is still subject to change. For example, since the proposed rule was published in the Federal Register on May 3, 2011, EPA has announced a change to the mercury limit for existing sources by 20%. Moreover, the alkali injection assumption for DSI technology may not work with high sulfur coal. Likewise, there are concerns about water issues from sodium when Trona is used for duct sorbent injection for SO2 control. Even if these unresolved issues with DSI technology did not exist, the use of this technology carries the risk that it may become a stranded investment in light of the stricter standards in the later MACT rule. 
Finally, For dry sorbent injection, what type of system or sorbent is actually used in the IPM modeling? For the LNB/OFA improvements, how many units (GW capacity) are installing new low-NOx burners and how many units are installing existing combustion control improvement?  What has EPA assumed for improvements, i.e., the type of improvements and NOx reduction levels? 
We suggest EPA should increase their prediction of the number of FGD's that may be required and included the cost and schedule increase in their estimates of the rule impacts. The 2014 deadline in the TR rule should be dovetailed with the compliance deadline in the MACT rule, consistent with the statutory standard to install controls as expeditiously as practicable. Further, the assurance provision should become effective on this deadline for meeting the MACT standards.

   F. Reliability Concerns
EPA suggests shifts in generation where controls cannot be installed in time without considering transmission constraints.  For instance, such constraints exist at several TVA plants and are particularly severe at TVA's Shawnee plant (Kentucky) where more unit retirements may be required than predicted by EPA. 
Recommendation: EPA should include the time required for transmission upgrades and/or construction of replacement transmission lines in their schedule for compliance and capture the associated costs.

   G. Regional Haze
Please clarify relationship of this rule to the Regional Haze Rule.  Since EPA determined that the emissions reductions under the original CAIR would achieve greater emissions reductions than the Best Available Retrofit Technology requirements of the regional haze rule, regional haze SIPs submitted by many eastern states have cited EPA's determination rather than performing source specific BART analyses for electric generating units.   As requested by commenters, EPA should consider demonstrating/determining that emissions reductions under the proposed transport rule will also be better than BART for both SO2 and NOx or states will need to revise their regional haze SIPs to include source specific BART analyses.  It is not clear why EPA is not making that finding in this draft final rule.
   H. Future Transport Rule
EGUs account for only 14% of nationwide NOx. Why is EPA planning to do another Transport rule for these sources in the future when there may be more cost-effective NOx tons to be reduced?  
