Green Expectations: Current Effects of Anticipated Carbon Pricing

51 Pages·2013·1.33 MB·English

Green Expectations:
Current Effects of Anticipated Carbon Pricing∗
Derek Lemoine
Department of Economics, University of Arizona
McClelland Hall 401, 1130 E Helen St, Tucson, AZ, 85721-0108, USA
[email protected]
University of Arizona Working Paper 13-09
June 2013
First version: April 2013
Forward-looking markets respond not just to current policies but also to expectations of
future policies. The “green paradox” literature suggests that a future environmental policy
increases emissions today, potentially even rendering the policy ineffective. However, the
theory’s relevance has been an open question in the absence of empirical confirmation and
measurement. I generalize the green paradox to a market with imperfectly substitutable
commodities in order to measure it in U.S. energy markets. An event study confirms the
primary prediction: coal futures produced positive excess returns of around 2% upon the
unexpected collapse of the U.S. Senate’s 2010 climate effort. If coal producers had expected
the proposed legislation to become law, their resulting production increase would have made
the future emission constraint act like a current subsidy of up to $12 per ton of carbon
dioxide. The response of natural gas prices and the theoretical analysis together suggest
that while the legislative process was indeed increasing U.S. emissions prior to 2013, passing
the legislation would have eventually decreased cumulative emissions.
JEL: H23, Q41, Q58
Keywords: green paradox, anticipation, hotelling, energy, futures, commodities, emissions,
climate, leakage
∗IamgratefulforcommentsfromseminarparticipantsatArizona,ArizonaState,HECMontr´eal,Toronto,
UC Berkeley, UC Davis, and Yale. Hoa Nguyen provided research assistance. This work was supported by
the University of Arizona’s Renewable Energy Network, with special thanks to Ardeth Barnhart and Stan
Reynolds. I also thank Keisuke Hirano and Ashley Langer for helpful discussions.
Lemoine Green Expectations June 2013
In 2009, the U.S. House of Representatives passed a bill to cap carbon dioxide emissions.
TheSenatesoonbegandraftingcomplementarylegislation. Theemissioncapwastobeginin
2013, four years after the House bill’s passage, twelve years after Congress first voted on cap-
and-trade legislation, and twenty-five years after attention-grabbing Congressional hearings.
Energymarketshadampleforewarning. The“greenparadox”literaturesuggeststhatenergy
producers increase near-term extraction when they anticipate future adoption of a climate
policy. In many models, the increase in near-term emissions completely offsets the eventual
environmental benefits of climate legislation. However, the existence and magnitude of green
paradox effects are both open questions. For instance, van der Werf and Di Maria (2012)
conclude their recent review by lamenting that “the most striking void in this literature is
an empirical assessment of the green paradox, without which it is hard (if not impossible) to
provide even order of magnitude estimates of green paradox effects.” I use an event study of
the weekend collapse of the U.S. Senate’s climate effort to establish the existence of a green
paradox in U.S. energy markets and to bound its magnitude from below.
I find that if coal producers had anticipated that the proposed climate policy would
actually pass, they would have increased coal production as if offered an emission subsidy of
between $1 and $12 per ton of carbon dioxide. This current subsidy stands in stark contrast
to the tax of between $5 and $65 per ton required to internalize the climate externality
(Greenstone et al., 2013). The current situation of not pricing carbon emissions for a bit
longer is generating more contemporary emissions than simply not pricing carbon emissions
at all. Further, the increase in coal production during the unsuccessful legislative process
distortedpricestoasimilardegreeasdidrailroads’marketpower(BusseandKeohane,2007)
and utility regulation (Cicala, 2012). Green paradox effects are economically significant, and
they have probably been distorting energy markets for years.
Theoretical models of green paradox effects disagree about whether intertemporal emis-
sion leakage completely offsets the future benefits of climate policy. First, in Hotelling (1931)
models of physical exhaustion, producers have a finite endowment of homogeneous energy
resources to allocate over time. An anticipated climate policy increases current emissions
but has no effect on cumulative emissions, which are completely determined by the physical
endowment (e.g., Sinn, 2008, 2012; Di Maria et al., 2012). By tilting the emission profile
towards the present, the future policy actually increases the present value of environmental
damages. In contrast, Heal (1976) models endogenize total resource use by making marginal
cost increase in cumulative extraction. Commodity producers no longer decide merely how
to allocate a given resource over time but also decide how much to extract in total. The
future carbon price reduces planned future extraction, which also reduces the incentive to
conserve low-cost resources for the future. In this setting, the anticipated climate policy
decreases cumulative emissions because it decreases cumulative extraction. The policy’s net
effect on the present value of damages depends on the importance of the early increase in
emissions relative to the ultimate decrease in emissions (e.g., Gerlagh, 2011; Hoel, 2012).
The U.S. Senate’s climate legislation would have primarily regulated electricity markets,
1
Lemoine Green Expectations June 2013
which are dominated by generation from coal and natural gas. Physical exhaustion is not a
prime concern for these resources.1 However, previous models which endogenize cumulative
extraction have only considered markets with a single resource. When there are multiple
resources, substitution effects complicate the standard green paradox story. The future
policy taxes both commodities, but it taxes the higher-emission commodity (coal) more
than the lower-emission commodity (natural gas). The future policy therefore could increase
future natural gas consumption and reverse the direction of that commodity’s anticipation
effect. My theoretical analysis generates a testable prediction for coal markets and links
natural gas market outcomes to electric power sector emission outcomes. First, I predict
that coal consumption increases in anticipation of an emission price. Second, I show that
if the policy proposal also increases pre-implementation consumption of natural gas, then
it eventually succeeds in decreasing cumulative emissions. While green paradox effects do
undercut climate policy, they do not render it completely ineffective as long as natural gas
and coal markets move in the same direction.
I find that coal prices responded to the collapse of the Senate’s climate effort with a
statistically significant jump in the theoretically predicted direction. Further, the estimated
response of natural gas prices has the same sign as the response of coal prices. Therefore,
while energy producers were increasing emissions in anticipation of the proposed cap-and-
trade program (manifesting a “weak” green paradox), the program’s implementation would
have decreased emissions after 2013 and in total (avoiding a “strong” green paradox). While
not discussing a climate bill at all would have generated fewer contemporary emissions, the
climate would nonetheless have eventually benefited from the bill’s passage. However, the
bill did not pass. The additional emissions from its discussion in 2009–2010 were therefore
neveroffsetbyacarbonprice. Whilepassingthelegislation wouldhavedecreasedcumulative
emissions, the unsuccessful legislative process actually increased cumulative emissions.
The anticipation effect at the heart of the green paradox is a more general feature of pub-
lic economics. Anticipating future changes in taxes on investment, income, or consumption
can induce smoothing behavior or offsetting behavior (Hall, 1971; Branson et al., 1986; Judd,
1985; Auerbach, 1989; Yang, 2005; Mertens and Ravn, 2011). These anticipation effects can
strongly affect the excess burden of taxation (Judd, 1987). The Hall (1971) consumption
tax has particularly similar effects to a proposed emission tax: the future change in the tax
causes a jolt to real flows which an anticipation effect offsets by shifting earlier real flows
in the opposite direction.2 The empirical literature on tax anticipation has obtained mixed
1Indeed, Jones et al. (2013) question the relevance of the green paradox on the basis of the common
assumption of exhaustibility: “Perhaps most fundamentally, however, whether fossil fuels are best modeled
as exhaustible is questionable: empirically, the evolution of resource prices is not well-described by simple
Hotelling-typemodels;andstocks—especiallyofcoal—aresolargethattherelevanceofexhaustionismoot.”
2Inmodelsofrationaladdiction,presentconsumptionincreasesthemarginalutilityfromfutureconsump-
tion. Anticipated taxes therefore reduce present and future consumption. These effects have been identified
in, for instance, cigarette markets (e.g., Becker et al., 1994; Gruber and K¨oszegi, 2001). In our setting,
present production increases the marginal cost of future production. Anticipated taxes therefore increase
2
Lemoine Green Expectations June 2013
results that depend on how one constructs the timing of tax changes (Mertens and Ravn,
2012; Perotti, 2012). Identifying anticipation effects in commodity markets is more straight-
forward for two reasons: these markets probably lack liquidity constraints that could mask
the effect for individuals or households, and proposals for emission pricing are not themselves
responding to commodity market movements as, for instance, personal taxation responds to
aggregate output or consumption. My results indicate that hypothesized anticipation effects
need not make overly strong demands of market efficiency.
I use an event that occurred among the parties developing climate legislation to isolate
an exogenous shift in the probability of regulation. Narrative evidence, contemporary news
accounts, and prediction market trades all indicate that expectations of regulation shifted
when Senator Lindsey Graham walked out of his climate collaboration over the weekend
before its scheduled unveiling. I measure the strength of green paradox effects by estimating
how futures markets responded to Graham’s decision. Event studies typically examine stock
market returns. In particular, event studies of energy policy have used changes in equity
prices to learn about the cost of regulation and about the distributional consequences of
cap-and-trade programs (Lange and Linn, 2008; Linn, 2010; Bushnell et al., 2013). Futures
markets instead teach us about the consequences for energy consumption and emissions.
In contrast to equity market outcomes, these consequences are largely independent of how
regulators choose to allocate emission permits in a cap-and-trade program.
I begin by analyzing how commodity spot prices, futures prices, and emissions respond
to new information about future emission policies. The theoretical model highlights the in-
terplay between higher- and lower-emission commodities and generates predictions relevant
to proposed climate legislation. Section 2 combines several lines of evidence to argue that
Senator Graham’s weekend withdrawal from his climate bill conveyed unanticipated infor-
mation that altered expectations. Section 3 introduces the estimation framework by which
I identify the event’s effect on futures markets. Section 4 demonstrates the anticipation
effect in coal futures across a range of specifications. Section 5 demonstrates that other days
with similar prediction market movements also saw sizable increases in coal prices. It then
evaluates the magnitude of the event’s effect on coal prices and uses the effect on natural gas
prices to learn about the existence of a green paradox in emissions. Section 6 concludes with
the policy implications of finding that markets are distorted by the suggestion of regulation.
1 Current effects of anticipated taxation
Consider a two-period model with two commodities, indexed by H and L. A representative
consumerobtainsutilityU(qH,qL)fromconsumingquantitiesqH andqL attimet,whereutil-
t t t t
ity is increasing, quadratic, twice-differentiable, and strictly concave. The two commodities
present production and decrease future production, though substitution between commodities can reverse
these effects.
3
Lemoine Green Expectations June 2013
are at least partially substitutable: U < 0, where subscripts indicate partial derivatives.3
HL
Consumingthehigher-emissioncommodityH generatesemissionsatrateeH, andconsuming
the lower-emission commodity L generates emissions at rate eL, with eH ≥ eL ≥ 0. Aggre-
gate time t emissions are E ≡ eHqH + eLqL. A two-commodity market roughly matches
t t t
the empirical setting: the two major sources of dispatchable generation in U.S. electricity
markets are coal (the higher-emission product) and natural gas (the lower-emission product).
The emissions are externalities, with the usual first-best policy pricing them in each
period at their marginal damage. We consider the case where emissions are not priced in the
first period, but the regulator imposes a linear emission tax τ in the second period. There are
twoequivalentwaysofinterpretingthissetting: thegovernmentdoesnotsucceedinadopting
the pollution policy until the second period, or the government adopts the policy in the first
period but delays its implementation to allow regulators and firms time to prepare. Firms
anticipate the second-period tax when selecting their first-period production schedules.
Firms are identical price-takers, allowing us to analyze them via a representative firm.
Production cost for product type k is an increasing, quadratic, twice-differentiable, strictly
(cid:0) (cid:1)
convex function of cumulative production: Ck Qk . Let Qk indicate the cumulative quan-
t
(cid:0) (cid:1) (cid:0) (cid:1)
tity produced prior to time t. Then the cost of time t extraction is C qk +Qk −C Qk .
t t t
Increasing either current or previous extraction increases current marginal cost by the same
amount.4 Neither coal nor natural gas markets face a clear threat of exhaustion, but pro-
ducers in both markets have steadily moved towards less accessible reserves: coal mining has
progressed from surface seams to mountaintop removal, while gas companies have developed
techniques to access ever deeper reservoirs and “unconventional” shale reserves.
The representative firm maximizes the present value of profits, with discount factor β ∈
(0,1] and Q normalized to zero:
1
(cid:88) (cid:8) (cid:0) (cid:1) (cid:2) (cid:0) (cid:0) (cid:1) (cid:0) (cid:1)(cid:1) (cid:3)(cid:9)
max pkqk −Ck qk +β pkqk − Ck qk +qk −Ck qk −τekqk , (1)
1 1 1 2 2 1 2 1 2
{qH,qL,qH,qL}
1 1 2 2 k∈{H,L}
3Other work on extraction profiles with multiple commodities has assumed physical exhaustion and/or
perfectsubstitutability (e.g.,Chakravortyet al.,2008; Smuldersand van derWerf,2008). Michielsen (2011)
develops a green paradox model with one exhaustible resource and two imperfectly substitutable, inex-
haustible resources. Because these two backstops are each produced at constant marginal cost, their supply
is not intertemporally linked. Di Maria et al. (2012) consider implementation lags when there are two
perfectly substitutable, exhaustible resources. I also find that first-period extraction of the high-emission
commodity usually increases in the emission tax (as with their “abundance effect”), but its cumulative ex-
traction nonetheless usually decreases in the present setting. I also describe cases in which extraction of the
low-emission resource declines in the first period (similar to their “ordering effect”), but in the present set-
ting,thischangeneednotbefullyoffsetbyincreasedextractioninthesecondperiodandisevencompatible
with declining second-period extraction.
4The crucial feature of the cost function is that second-period marginal cost increases in first-period
extraction, as is common in Heal (1976) models. Two papers take other approaches to endogenizing total
resource use. Fischer and Salant (2012) posit a series of exhaustible resource pools, each with a different
marginal cost of extraction. Smulders et al. (2012) generate a green paradox via households’ consumption
and savings decisions in a general equilibrium setting where energy and capital are complements.
4
Lemoine Green Expectations June 2013
where subscripts index time and superscripts index commodities. The firm’s optimal quan-
tities solve the following four first-order conditions:
(cid:0) (cid:1) (cid:2) (cid:0) (cid:1) (cid:0) (cid:1)(cid:3)
pk =Ck(cid:48) qk +β Ck(cid:48) qk +qk −Ck(cid:48) qk , for k ∈ {H,L}, (2)
1 1 1 2 1
(cid:0) (cid:1)
pk =Ck(cid:48) qk +qk +τek, for k ∈ {H,L}, (3)
2 1 2
where primes indicate derivatives. The firm equates marginal revenue to a comprehensive
measure of marginal cost. In both periods, marginal cost includes the marginal extrac-
tion cost. In the first period, it also includes the (discounted) additional cost imposed by
extracting from more costly reserves in the second period. In the second period, it also
includes emission tax payments. Quantity decisions are linked over time by the dependence
of marginal extraction cost on cumulative extraction.
Substituting among the four equations and recognizing that equilibrium prices are equal
to marginal utility, we obtain:
(cid:0) (cid:1) (cid:0) (cid:1) (cid:0) (cid:1)
U qH,qL +βτeH =βU qH,qL +(1−β)CH(cid:48) qH , and
H 1 1 H 2 2 1
(cid:0) (cid:1) (cid:0) (cid:1) (cid:0) (cid:1)
U qH,qL +βτeL =βU qH,qL +(1−β)CL(cid:48) qL ,
L 1 1 L 2 2 1
where subscripts on the utility function again indicate partial derivatives. These equations
characterize firms’ intertemporal decisions. The left-hand side measures the marginal benefit
of extracting in period 1 instead of period 2: it includes the revenue from selling the com-
modity in period 1 and the gain from not paying the tax in period 2. The right-hand side
measures the marginal cost of extracting in period 1 instead of period 2: the firm forgoes
revenue by not selling in period 2 and no longer discounts extraction costs. At an optimum,
the firm is indifferent between extracting in these two periods. The anticipated tax provides
an incentive to extract more of the resource earlier, and that incentive is stronger for more
emission-intensive commodities.
1.1 Effects on coal markets
Begin by considering a system with only one commodity. The solid lines in the left-hand
plot of Figure 1 show demand and supply in a model in which second-period extraction
costs are independent of first-period extraction or, equivalently, in which the firm is myopic
(β = 0). The firm’s profit-maximizing quantity (point A ) occurs at the intersection of
1
these curves. When extraction costs instead depend on cumulative extraction and the firm
is forward-looking, the first period’s supply curve is shifted inward (dashed line) because
additional extraction increases costs in the second period. These intertemporal linkages
reduce first-period extraction (to point B ).
1
Now consider the effect of introducing an emission tax in the second period. The right-
hand plot in Figure 1 shows second-period supply and demand. Introducing the tax reduces
the second period’s equilibrium quantity from B to C(cid:48). This is the tax’s direct effect.
2 2
5
Lemoine Green Expectations June 2013
Figure 1: When costs depend on cumulative extraction, first-period marginal cost (dashed
line) includes the effect on second-period costs, which reduces the equilibrium quantity to B
1
from the myopic equilibrium A . Imposing a second-period tax reduces the second-period
1
market-clearing quantity from B to C and increases the first-period quantity from B to
2 2 1
C .
1
Because the tax reduces the marginal benefit of conserving resources for the second period,
the firm has a smaller opportunity cost to exploiting first-period reserves. Equivalently,
when second-period extraction decreases, additional first-period extraction increases second-
period costs by a smaller amount. The first-period supply curve therefore shifts out (dotted
line) part of the way towards the myopic curve. Raising the second-period tax increases
first-period extraction from point B to C , but extraction remains lower than it would have
1 1
been in a world with only one period. Greater first-period extraction shifts supply inward in
the second period (dotted line), which further reduces the equilibrium quantity to C . The
2
increase in first-period extraction and the further decrease in second-period extraction are
anticipation effects.
In a system with both a more emission-intensive commodity (coal) and a less emission-
intensive commodity (natural gas), the effect of a marginally greater tax depends on the
relative emission intensity of the two commodities. Figure 2 summarizes the appendix’s
formal results regarding how the higher-emission commodity responds to the tax. In the
region on the left, coal is relatively dirty and its first- and second-period quantities respond
as in the system with only a single commodity. In the three regions on the right, coal is
not much dirtier than natural gas. Now coal’s quantity responses can be reversed from the
single-commodity setting due to interactions with natural gas markets.
These rightmost regions exist only if marginal utility responds strongly to coal consump-
tion while neither marginal utility nor marginal extraction cost respond strongly to natural
6
Lemoine Green Expectations June 2013
Figure 2: The response of the equilibrium quantity of coal to the tax, as proved in the
appendix. The region marked with an asterisk is guaranteed to exist. The existence of the
other regions depends on the curvature of the utility and cost functions.
gas consumption.5 Both gas and coal might have a similar price elasticity of demand in
electricity markets, but natural gas also has major markets in relatively price-inelastic res-
idential and industrial applications. Further, natural gas probably has a more curved cost
function due to the smaller size of each reserve, and its emission intensity is around half that
of coal. Therefore, supply, demand, and emission considerations suggest that the leftmost
region of Figure 2 is the only region of interest in our empirical application. In that case,
a greater tax has a qualitatively similar effect on coal consumption in both the one- and
two-commodity settings.
1.2 Effects on natural gas markets
Natural gas markets more plausibly deviate from the single-commodity intuition. Three
effects determine the sign of the change in second-period natural gas consumption. First,
the tax wedge always acts to reduce consumption (direct effect), where the per-unit tax is
smaller in the natural gas market than in the coal market due to its lower emission intensity.
Second, changes in first-period natural gas extraction via anticipation effects affect second-
period supply. For instance, when first-period extraction increases in anticipation of reduced
second-period extraction, the corresponding reduction in second-period supply further re-
duces second-period extraction. Third, the reduction in coal’s second-period consumption
5When marginal utility and marginal cost are flat for product L, its quantity responds strongly to small
changes in its per-unit tax. This strong change increases the value of additional product H (i.e., it shifts
out residual demand for H). When residual demand for H is sufficiently inelastic, the tax’s direct effect is
small. Therefore, when both products have similar emission intensities (and so similar per-unit taxes), the
greater residual demand for H can overturn the direct effect of the tax.
7
Lemoine Green Expectations June 2013
Figure 3: The response of the equilibrium quantity of natural gas to the tax, as proved in
the appendix. The region marked with an asterisk is guaranteed to exist. The existence of
the other regions depends on the curvature of the utility and cost functions.
increases the residual demand for natural gas, which tends to increase second-period con-
sumption of natural gas (substitution effect). The net effect is ambiguous because the direct
effect always opposes the substitution effect. The change in first-period extraction is also
ambiguous. While first-period substitution effects always decrease residual demand for nat-
ural gas, the first-period change in supply from anticipation effects is ambiguous because
the second-period change in extraction is ambiguous. Because of the stronger substitution
effects and smaller direct effects in gas markets, each period’s natural gas consumption could
plausibly increase or decrease in response to a larger second-period tax.
Figure 3 orders the possible outcomes based on the formal analysis in the appendix.
Towards the right, natural gas is nearly as dirty as coal. Substitution effects are weak
because the two commodities are taxed at nearly the same level, and natural gas markets
therefore respond to a greater tax as they would in a single-commodity setting. Towards the
left, natural gas is very clean relative to coal. Substitution effects are so strong that they
reverse its quantity responses in all periods and in total: because the emission tax imposes
a much greater cost on each unit of coal, second-period consumption shifts strongly towards
natural gas and first-period consumption of natural gas decreases in anticipation of this
greater second-period demand. As U shrinks towards zero, substitution effects become
HL
smaller and the rightmost region grows to fill nearly the whole area. As U increases in
HL
magnitude, substitution effects become stronger and all vertical lines shift to the right.
The crucial part of Figure 3 is the ordering of the middle regions, where the response of
natural gas to the tax is the same as in the single-commodity system over some timeframes
butdifferentoverothers. Inparticular, notethatwhenmovingfromrighttoleft(i.e., making
naturalgascleanerrelativetocoalandtherebyincreasingthestrengthofsubstitutioneffects),
thefirst-periodquantityresponseswitchesitssignbeforethesecond-periodquantityresponse
8
Lemoine Green Expectations June 2013
Figure 4: When a marginally greater second-period tax increases the second period’s equilib-
rium quantity of natural gas (from X to Y ), the first period’s anticipation and substitution
2 2
effects both work to decrease the first period’s equilibrium quantity (from X to Y ).
1 1
does. While it might seem that there are four plausible combinations of first- and second-
period quantity responses, only three are actually possible. The analytic results rule out
any case where a marginally greater tax increases natural gas consumption in both periods.
If we observe a greater tax increasing first-period natural gas consumption, then we know it
decreases natural gas consumption in the second period and in total.
Figure 4 provides the intuition for why a greater tax cannot increase natural gas con-
sumption in both periods. Assume that second-period consumption of natural gas increases
in the tax. Recall that three effects determine second-period consumption: the tax wedge
tendsto reduceit(direct effect), the greaterresidualdemand due todecreased coal consump-
tion tends to increase it (substitution effect), and the altered supply curve due to changes
in first-period extraction (anticipation effect) has an ambiguous effect. The anticipation
effect increases first-period extraction (and shifts second-period supply inward) when the
height of the tax wedge is greater than the outward shift in residual demand at the original
market-clearing quantity, and it decreases first-period extraction otherwise. The anticipa-
tion effect reinforces the net effect of the other two. Under our assumption of an increase
in second-period consumption (e.g., from X to Y ), the residual demand curve must shift
2 2
out (to the dashed line in the right-hand plot) by a large amount relative to the tax, and
the anticipation effect must decrease first-period extraction while shifting out second-period
supply (to the dotted line).
Now consider the determinants of first-period consumption. Second-period consump-
tion of natural gas increases when substitution effects make future sales more profitable, in
9