Flexible mandates for investment in new technology

Abstract

Environmental regulators often seek to promote forefront technology for new investments; however, technology mandates are suspected of raising cost and delaying investment. We examine investment choices under an inflexible (traditional) emissions rate performance standard for new sources. We compare the inflexible standard with a flexible one that imposes an alternative compliance payment (surcharge) for emissions in excess of the standard. A third policy allows the surcharge revenue to fund later retrofits. Analytical results indicate that increasing flexibility leads to earlier introduction of new technology, lower aggregate emissions and higher profits. We test this using multi-stage stochastic optimization for introduction of carbon capture and storage, with uncertain future natural gas and emissions allowance prices. Under perfect foresight, the analytical predictions hold. With uncertainty these predictions hold most often, but we find exceptions. In some cases investments are delayed to enable the decision maker to discover additional information.

Appendix: Technology choice and timing in the simulations

Table 4 illustrates the technology choice and timing of investments for the base case and the three versions of technology policy, for the twelve scenarios, both under perfect foresight and uncertainty. Table 5 provides a mapping from numeric labels to technologies. For additional supporting information see Patino-Echeverri et al. (2012).

Table 4 Year of investment and technology installed with ACP that produces investment in CCS at the same time as the inflexible standard

Table 5 Cost and performance of investment alternatives

Three options for new investment use solid coal: sub-critical (sub), supercritical (super) and ultra-supercritical (ultra). In addition we investigate integrated gasification combined cycle coal-fired (IGCC) and natural gas combined cycle (NGCC). Each of these would satisfy standards for new sources for emissions of $\text{ SO }_{2},\,\text{ NO }_{x}$, mercury and particulates. Each could come with or without CCS, or with retrofit CCS sometime after initial construction, which results in 10 different generation technologies but a total of 15 different investment alternatives with associated costs and emissions as shown in Table 4.

1.1 Appendix 1: effects of an inflexible performance standard

For each scenario under the base case and the inflexible performance standard, Table 4 presents the year when the first installation comes online Y1, the technology first installed T1, the year when a second installation (new plant or retrofit) comes online Y2, and the technology installed in the second installation T2.

When there is perfect foresight about the future natural gas price and climate policy, the introduction of an inflexible performance standard on new investments leads to delays in investment for all scenarios except where natural gas prices are high and climate policy is stringent (high150), where technology policy is not expected to have an effect because the initial investment includes CCS in the absence of the standard. When natural gas prices are at mid- level, and there is no federal climate policy (mid0), new investment never occurs.

When there is uncertainty about future natural gas prices and climate policy, compared to no technology policy, an inflexible performance standard causes investments to be delayed in 9 scenarios, unaffected in 1 scenario, and sped up in two scenarios. The acceleration of investment for the scenarios with high natural gas prices and stronger climate policy (high100 and high150) shows that under uncertainty the intuition that an inflexible standard would never speed up investment does not hold. In the case of scenario high150 the inflexible standard reduces uncertainty. For the scenarios with high natural gas prices, the only two power generation technologies that an investor would consider for an initial investment are an ultra-supercritical coal plant for the cases with no or weak climate policy (high0 or high50), or an IGCC plant for the mid or strong climate policy (high100 or high150). An NGCC plant is not competitive due to the high prices of the fuel, and the alternative of not installing any plant is not competitive because for these scenarios the expected electricity prices after 2020 are sufficiently high to motivate investment. Under no technology policy, it is optimal to wait one more year to have certainty about the climate policy scenario and decide whether an ultra-supercritical (without CCS) or an IGCC with CCS should be installed. Under an inflexible performance standard, there is no value of waiting one more year because the choice of an ultra-supercritical without CCS is not available.

1.2 Appendix 2: effects of a flexible standard

For the flexible standard we present the minimum surcharge level that produces investment in CCS at the same time or before the inflexible standard, which is denoted by $\beta ^{*}$. For this $\beta ^{*}$, we present the corresponding year of installation $Y^{*}$ and CCS technology installed $T^{*}$. For some scenarios $T^{*}$ is different than the CCS technology installed for the first time under the inflexible standard.

The value of $\beta ^{*}$ refers to the surcharge in $ per ton that must be paid for each ton in excess of the emissions standard. We assume $\beta ^{*}$ increases every year at the same rate of discount used by the investor in the expected NPV calculations.

In each scenario, with perfect foresight, we identify a value of $\beta ^{*}$ between $0 and $20 per ton of $\text{ CO }_{2}$ that leads to investment in CCS at least as soon as under the inflexible performance standard. An increase in the value of the surcharge moves forward the time at which CCS technology is built. When natural gas prices are at mid- level and there is no federal climate policy (mid0) the inflexible standard causes investment to be indefinitely postponed (beyond 2052). In this case the surcharge value needed to replicate this result under a flexible performance standard is $0. Under perfect foresight and under the most strict federal climate policy, when natural gas prices are high (high150), no technology policy is necessary to get the IGCC with CCS to come on line, so the surcharge value is $0. For any other case under perfect foresight there is always a surcharge level $\beta ^{*}$, for which investment in CCS will happen at the same time or before as under the inflexible standard.

The change in the timing and choice of generation technology can have an important effect on cumulative emissions. In three cases investment in newer technology, albeit absent CCS, comes years earlier than under the inflexible policy (low0, mid100, mid150). For the scenario with low natural gas prices and no federal climate policy (low0), a surcharge of $3 produces investment in IGCC with CCS the same year that it occurs under the inflexible NSPS policy. However, in the flexible case NGCC without CCS appears several years earlier and is subsequently replaced. For the scenarios with mid-level natural gas prices and mid- and stringent- level federal climate policy (mid100, mid150) a surcharge of $6 yields investment in a CCS retrofit for NGCC, subsequently replaced by IGCC with CCS, instead of the IGCC with CCS initially chosen under the inflexible standard. For these scenarios there is no surcharge value that would cause identical investment as under the inflexible standard.

When there is uncertainty about the future natural gas prices and federal climate policy, there are three scenarios (low100, high100, and high150) for which there is no surcharge value in the range of $\beta ^{*}\le 20$ that yields installation of CCS at the same time or before than the inflexible standard. In these scenarios and for the set of values we examine, installation happens one or more years later than under the inflexible standard. For the scenario with mid-level natural gas prices with no climate policy (mid0), and the scenario with mid-level natural gas prices and stringent climate policy (mid150), no surcharge is needed to yield identical investment (e.g. $\beta ^{*}=0$). For the remaining scenarios there is a surcharge level that yields an identical investment to the one produced by the inflexible standard.

1.3 Appendix 3: effects of a flexible standard with escrow account

The escrow fund comprised of accumulated emission surcharge payments provides a source of funds that can be used to subsidize the cost of retrofitting a facility with CCS in the future. Thus the policy is most effective when the flexible policy by itself does not lead to CCS being installed with the initial investment. We have assumed the flexible performance standard with escrow fund operates with 3 rules that help destroy incentives for delaying CCS investment in the hopes for lower capital costs. The first rule specifies that the funds accumulated in the escrow account do not gain any interest. This makes delaying in CCS costly since the surcharge payment accumulates in an escrow fund that loses value with time. The second rule specifies that the maximum amount of funds withdrawn from the escrow cannot exceed the capital costs of the CCS investment (be it a CCS retrofit or a new plant with CCS included). This discourages accumulating funds in the escrow that exceed the capital costs of the needed CCS investment. The third rule specifies that funds from the escrow account can be withdrawn only once. This means that any funds not used for the first CCS investment (a retrofit or a new plant) are lost, which discourages accumulation of funds in the escrow and accelerates investment.

Under perfect foresight the $\beta ^{*}$ values for the flexible standard with escrow are lower than without escrow for the 6 scenarios with climate policy and low or mid-natural gas prices (scenarios low50, low100, low150, mid50, mid100, mid150), and the same for the scenarios with no climate policy or high natural gas prices. For two of the scenarios (mid50 and mid100) the lower $\beta ^{*}$ required to obtain CCS also causes earlier investment. Under low50 the flexible standard requires a $\beta ^{*}$ of $13 to cause an investment in NGCC plant subsequently retrofit in year 2023, while the standard with escrow requires a $\beta ^{*}$ of only $7 to produce earlier investment in NGCC and a CCS retrofit in 2023. In the mid50 scenario the flexible standard requires a $\beta ^{*}$ of $9 to produce the installation of an IGCC plant with CCS in year 2030, while the standard with escrow requires a $\beta ^{*}$ of $6 to cause an investment in NGCC subsequently retrofit in year 2025, and then replaced by IGCC with CCS. These results are consistent with a hypothesis suggesting that the introduction of an escrow account should not delay the timing of investment in CCS.

With uncertainty, the timing and choice of investments is identical to the flexible standard and there is no change in the $\beta ^{*}$values, and as expected the fund does not delay investment in new generation or CCS.

1.4 Appendix 4: ACP values required for an effective flexible standard

For a flexible policy to be effective the ACP must be low enough to allow installation of uncontrolled facilities when the capital costs of CCS are too high, but high enough to motivate a CCS retrofit at the same time or earlier than when a CCS plant would be installed under an inflexible technology policy.

An investment in a new uncontrolled plant will occur when the emissions costs of such plant (i.e. paying for a carbon tax + ACP) are lower than the capital and O&M costs of a CCS plant¹⁷:

$$\begin{aligned} \sum _{t=\tau }^{\tau +T} {d^{(t-\tau )}v\left( {e o+\left( {e-s} \right) \beta } \right) } \le vc_\tau ^{pc} +\sum _{t=\tau }^{\tau +T} {d^{(t-\tau )}v\left( {m^{pc}+e^{pc}o} \right) } \end{aligned}$$

(6)

Assuming a discount rate of zero (i.e. $d$=1), no carbon tax (i.e. $o$=0), and emissions from the CCS plant $e^{pc}$ being less than $s$we find the lower bound of the ACP as:

$$\begin{aligned} \beta \le \frac{c_\tau ^{pc} }{T(e-s)}+\frac{m^{pc}}{(e-s)} \end{aligned}$$

(7)

Note that since $\left( {e-s}\right)$ represents the amount of emissions abated with CCS, Eq. 7 indicates that at the time of installation of the uncontrolled plant $t=\tau$ the ACP must be less than the per unit cost of emissions abatement (capital + O&M) from a CCS plant.

Similarly, an upper bound for the ACP is given by the per unit capital and O&M cost of abatement of a retrofitted plant at a time $t=\upsilon$:

$$\begin{aligned} \beta \ge \frac{\left( {1+z} \right) c_\upsilon ^{pc}}{T(e-s)}+\frac{m^{pc}}{(e-s)} \end{aligned}$$

(8)

In conclusion if the ACP is in the range indicated by 7 and 8 for $\tau ^{\mathrm{bsln}}\le \tau \le \upsilon \le \tau ^{\mathrm{std}}$ the flexible standard will be effective. In the case where the flexible standard is complementing a carbon tax policy, the ACP value needs to be reduced by the carbon tax $o$, so that the new threshold is:

$$\begin{aligned} \frac{c_\tau ^{pc} }{T(e-s)}+\frac{m^{pc}}{(e-s)}\ge \beta +o\ge \frac{\left( {1+z} \right) c_\upsilon ^{pc}}{T(e-s)}+\frac{m^{pc}}{(e-s)} \end{aligned}$$

(9)

1.5 Appendix 5: effects of the retrofit penalty

As discussed in Sect. 3.2, the effectiveness of a flexible new source performance standard depends on the existence of a pollution control technology without prohibitive cost penalties for retrofit installations and the expectation that the capital costs of this technology will gradually come down. Under the flexible policy mechanism, investors will find that delaying the CCS installation is profitable if there is the expectation that sometime in the future the gains from postponing the investment to take advantage of reductions in capital costs are higher than the extra cost of a retrofit installation.

In this section we find a threshold for the retrofit penalty $z$ above which the flexibility of the emissions standard becomes irrelevant.

An investment in CCS (at the same time as the initial investment in the plant or as a retrofit) will occur if and only if the net present value of the reduction in emissions costs over the planning horizon exceeds the capital and O&M costs of the CCS equipment. If we assumed that the CCS retrofit extends the lifetime of the plant, then a retrofit will occur when Eq. (10) is satisfied¹⁸:

$$\begin{aligned} \sum _{t=\upsilon }^{\upsilon +T} {d^{(t-\upsilon )}v\left( {e o+\left( {e-s} \right) \beta } \right) } \ge v\left( {1+z} \right) c_\upsilon ^{pc} +\sum _{t=\upsilon }^{\upsilon +T} {d^{(t-\upsilon )}v\left( {m^{pc}+e^{pc}o} \right) } \end{aligned}$$

(10)

Under this assumption, retrofit will occur at or after a year $\upnu >\uptau$ such that:

$$\begin{aligned} \left( {1+z} \right) c_\upsilon ^{pc} \le c_\tau ^{pc} \end{aligned}$$

(11)

If capital costs of CCS decline by a factor $l$such that the capital costs at a time t are given by:

$$\begin{aligned} c_t^{pc} =\left( {1-l} \right) ^{t}c_{0}^{pc} \end{aligned}$$

(12)

Then for a retrofit to occur at year $\nu$ after an initial installation of the base plant at year $\tau$, the retrofit penalty factor $z$ must be less than the cumulative reduction in capital costs due to learning by doing:

$$\begin{aligned} z\le \left( {1-l} \right) ^{\tau -\upsilon }-1 \end{aligned}$$

(13)

Table 5 shows the maximum value of the retrofit penalty factor for different combinations of the annual percentage reduction in CCS capital costs, and the number of years between the initial plant installation and the retrofit:

Note that if the lifetime of the retrofitted plant is determined by the age of the uncontrolled plant as stated in Eq. (2), then the upper bounds of the retrofit penalty are lower than those presented in Table 6, since investors will have less time to recover the capital costs of the CCS investment. However, in the simulation analysis we assume that the retrofit extends the lifetime of the CCS plants as in Eq. 6.

Table 6 Maximum value of the retrofit penalty factor $z$( %)

As discussed in the paper, the simulation assumes that current capital costs of CCS are twice what they will be in year 12 of the simulation. This implies an annual learning rate $l$ of 5.6 percent. If we assumed that emissions costs are constant (i.e. $o$ and $b$ are constant as suggested by Eq. 6) and there are no other environmental benefits or costs (due to emissions costs of other pollutants such as $\text{ SO }_{2},\,\text{ NO }_{x}$, and mercury) then from Table 6 it can be observed that as long as there are 3 or more years between the initial plant installation and the retrofit, penalties of 20 % as assumed for the Pulverized Coal plants are within the bounds of penalties that allow the Flexible Emissions Standard to have a positive effect. Similarly, for the IGCC, a CCS retrofit penalty of 30 % requires that there are at least 5 years between the base plant installation and the retrofit. Note however than in the simulation there are other factors such as economic benefits from reduced compliance costs for other air pollutants, affecting the profitability of initial investments and retrofits.

1.6 Appendix 6: cost of CO$_{2}$ emissions abatement

Table 7 shows the cost of $\text{ CO }_{2}$ emission abatement under the three technology policies considered. The cost is found by dividing the reduction in the net present value of the change in investors’ profits by the reduction in cumulative $\text{ CO }_{2}$ emissions over 43 years. Under perfect foresight, $\text{ CO }_{2}$ emissions under the inflexible standard increase and profits decrease for the first 8 scenarios implying that compared to the baseline (i.e. no technology policy) the inflexible policy increases emissions while reducing investors’ profits. In contrast, the flexible policies in general reduce emissions at a cost that range from near $0 to $46/ton of $\text{ CO }_{2}$.¹⁹ For the mid100 scenario the flexible policy has higher emissions and lower profits than under no technology policy, but profits are higher and emissions are lower than under the inflexible standard. Since the ACP for the flexible policies is set to achieve CCS installation at the same time or earlier than under the inflexible standard, it is unsurprising that the outcomes of these flexible policies are not always superior to those with no technology policy.