paper review

ej_2017_faehn_hagem_lindholt_maeland_rosendahl_demand_vs_supply_side_climate_policies_norway.pdf

Home >Business & Finance homework help >Economics homework help >paper review

a Corresponding author. Statistics Norway, Research Department. E-mail: [email protected], Tel: ( + 47)95867999. b Statistics Norway, Research Department. E-mail: [email protected]. c Statistics Norway, Research Department. E-mail: [email protected]. d Statistics Norway, Research Department. E-mail: [email protected]. e Norwegian University of Life Sciences, School of Economics and Business. E-mail: [email protected].

Climate Policies in a Fossil Fuel Producing Country: Demand versus Supply Side Policies

Taran Fæhn,a Cathrine Hagem,b Lars Lindholt,c Ståle Mæland,d and Knut Einar Rosendahle

ABSTRACT

In absence of joint global climate action, several jurisdictions unilaterally restrict their domestic demand for fossil fuels. Another policy option for fossil fuel pro- ducing countries, not much analysed, is to reduce own supply of fossil fuels. We explore analytically and numerically how domestic demand and supply side pol- icies affect global emissions, contingent on market behaviour. Next, in the case of Norway, we find the cost-effective combination of the two types of policies. Our numerical results indicate that given a care for global emissions, and a desire for domestic action, about 2/3 of the emission reductions should come through supply side measures.

Keywords: Climate policies, Carbon leakages, Oil extraction, Supply side climate policies, Demand side climate policies

https://doi.org/10.5547/01956574.38.1.tfae

1. INTRODUCTION

In the context of a global climate agreement, a cap on fossil fuel consumption would have the same effects on global emissions as a cap on fossil fuel extraction, as consumption must equal extraction at the global level. If fossil fuel markets were efficient, the global costs of reducing emissions would also be the same. In this first-best situation, demand and supply side policies coincide with respect to efficiency. However, with limited participation in a climate agreement, or with unilateral action by a single country or coalition of countries, demand side versus supply side policies matters. Many jurisdictions show willingness to reduce CO2-emissions by restricting do- mestic demand for fossil fuels. Domestic supply side policies are less frequently discussed, let alone pursued.

The purpose of this paper is to deduce the cost-effective combination of the two types of policies, given a target for a country’s (or coalition’s) contribution to global CO2 abatement. The result hinges critically on how domestic demand side and supply side policies affect global emis- sions through international markets. We explore analytically and numerically how the optimal domestic climate policies depend on market behaviour in the fossil fuel markets, the emissions from extraction, and the costs of downscaling domestic fossil fuel demand and supply.

Domestic policy measures that reduce fossil fuel demand lead to lower international energy prices, and may also reduce the competitiveness of domestic firms in the world markets for energy-

78 / The Energy Journal

1. In practical policy, the domestic action to meet a given global target will have to concur with other existing climate ambitions and commitments. For instance, Norway has demand side commitments in the EU Emissions Trading System and in the Kyoto agreements. Chapter 3 explains how these are accounted for in the computations. Note that to the extent that these commitments are not met by the domestic actions studied here, their fulfillment may imply extra costs. However, Norway has already shown willingness to do more than simply complying with international commitments, e.g., through over-fulfilling the Kyoto obligations in 2008–12, financing technology transfer and engaging in rainforest preservation in Reducing Emissions from Deforestation and Forest Degradation (REDD + ). This is an initiative supported by the UN- REDD program and several other international organizations; see http://www.un-redd.org/aboutredd.

intensive goods. Both effects cause so-called carbon leakages, i.e. increased consumption of and emissions from fossil fuels among free-riders; see, among others, Markusen et al. (1993; 1995), Rauscher (1997), and Böhringer et al. (2010). Leakages occur also through supply side policies, i.e., policies that reduce fossil fuel extraction; see Erickson and Lazarus (2014). Such supply side leakages result from increased supply by countries outside a climate coalition as international fuel prices rise. Harstad (2012) shows that supply side leakages can be completely avoided if the coa- lition buys marginal foreign fossil fuel deposits and conserves them. This renders the non-coalition’s supply curve locally inelastic. Although this is a promising result, buying deposits may face several practical problems such as asymmetric information, contract incompleteness, and bargaining fail- ures. In our paper, we focus on the trade-off between domestic demand and supply side measures. We, thus, reserve a given unilateral contribution to global abatement to domestic action; the options of purchasing foreign fossil fuel deposits or international emission quotas are excluded.

Our case in the numerical analysis is Norway, which has an ambitious target for domestic demand side measures for 2020, but has so far not considered using supply side measures. The Norwegian lack of focus on supply side policies has been questioned by media, analysts and NGOs at home, and has also attracted international attention; see, e.g., The Economist (2009), Science Nordic (2013), Forager Funds (2014) and Greenpeace (2014). While the country accounts for around 2 per cent of global oil production, it contributes to less than 0.3 per cent of global oil consumption (BP, 2013). The global combustion of fossil fuels extracted in Norway leads to CO2 emissions that are about ten times higher than total emissions of CO2 within Norway. Even though leakages are likely to be larger with supply side measures than demand side measures, we conclude that it is cost-effective for Norway to let most of the contribution to global emission reductions be achieved through supply side measures. In our benchmark scenario, only one third of a given global reduction should be realised through demand side measures; the remaining two thirds should come through supply side measures, that is, by reducing oil extraction.1

Previous literature on optimal (second-best) climate policy in the presence of carbon leak- ages through the international fuel markets has derived the optimal combination of producer and consumer taxes in a climate coalition, given a target for global emission reductions. Hoel (1994) models an aggregate fossil fuel market, and derives analytical expressions for optimal tax levels. Golombek et al. (1995) extend Hoel’s analysis by modelling three fossil fuel markets (oil, coal and gas) and provide a numerical illustration of optimal producer and consumer taxation for a coalition of OECD countries, given competitive fossil fuel markets. They find that the optimal producer tax of oil should be negative, due to terms-of-trade effects dominating the leakage effects (OECD is a net importer of oil). Hagem (1994) compares numerically the costs of pure demand side policy with pure supply side policy in the case of Norway, given a target for its contribution to global emission reductions in 2000. The calculations assume competitive fuel markets and conclude that it would be less costly to reduce oil production than to introduce uniform taxes on fossil fuel consumption.

Climate Policies in a Fossil Fuel Producing Country / 79

Our paper contributes to the theoretical literature by analysing how differences in emissions from fossil fuel extraction across countries affect the relative performance of demand side policies versus supply side policies. Furthermore, it supplements previous numerical analyses of demand versus supply sides policies in several ways: First, we analyse the impact of various non-competitive oil market assumptions. Second, we take into account emissions due to extraction of fossil fuels and, particularly, the differences in emission intensity across countries. Third, we incorporate the fact that both production costs and emission intensities are relatively high in the decline phase of an oil field – here we use detailed cost information from Norwegian oil fields. Fourth, previous estimates on cost and emission effects are outdated. In our updating of the information base we have included a review of the empirical literature on relevant price elasticities in order to assess likely carbon leakage rates on the demand as well as the supply side. The robustness of our cal- culations is checked with thorough sensitivity analyses.

Assumptions regarding supply and demand elasticities, as well as the competitive envi- ronment on the fuel markets, are decisive for our results on the optimal distribution of demand versus supply side policies. There is a large literature on OPEC behaviour (see e.g. Griffin, 1985; Alhajji and Huettner, 2000; Smith, 2005; Hansen and Lindholt, 2008). Although the conclusions from this literature are rather mixed, one conclusion is that OPEC does not behave as a competitive producer. In our main case we model OPEC as a strategic player that seeks to maximize its income from annual oil production, while other producers are price takers. To check the robustness of our results, we also consider the competitive case, along with situations where OPEC has price or production targets.

As fossil fuels are non-renewable resources, there are important dynamic properties of the market that our static analysis does not capture. A fossil fuel producer’s optimization behaviour implies finding an extraction path that maximizes the present value of the resource, which depend on the expected, future price path (Hotelling, 1931). If producers expect a gradual tightening of climate policies, they may accelerate their extraction; see Sinn (2008) for a discussion of this “green paradox”. Thus, leaving out dynamic considerations may have implications for the results. On the other hand, Venables (2011) shows that although decreasing prices may speed up production on existing fields, this is offset by their postponing effect on field openings; see also Österle (2012) for a similar study. Furthermore, the government can control the available cumulative production through their production licencing. Hoel (2013) considers supply side policies and argues that conserving the marginal, most costly resources reduces both total and immediate resource extrac- tion. These studies show the relevance of analysing fossil fuel policies in a static framework as ours even if some intertemporal reallocation is ignored. We restrict our carbon leakage considera- tions to those stemming from the fossil fuel markets, disregarding carbon leakages through the market for energy-intensive goods. These leakages can be mitigated or completely abolished by compensation schemes for exposed industries (e.g. free allocation of permits) or by border tax adjustments (Böhringer et al., 2012a, and Hoel, 1996). We therefore ignore this channel of carbon leakages.

2. THEORETICAL ANALYSIS

2.1 Unilateral Climate Policy

We consider a fossil fuel producing and consuming home country that aims to contribute to a certain reduction in global greenhouse gas (GHG) emissions (Ā), through a combination of domestic demand side and supply side policies. The country’s aggregate benefits from domestic

80 / The Energy Journal

consumption of fossil fuels are given by , where yo, yc and yg denote domestic consump-B(y ,y ,y )o c g tion of oil, coal, and gas, respectively. Without loss of generality, all fuels i = o,c,g are measured in units of CO2. We assume that the benefit function is increasing in each of the fuels.

Furthermore, let denote the home country’s aggregate cost of producing fossil fuelc (x )i i i, where xi denotes home production of this fuel. We assume that the cost functions are increasing and strictly convex. Fossil fuels are traded in international markets at prices Po, Pc and Pg. To simplify the analytical derivations, we treat domestic consumption (yi) and production (xi) as ex- ogenous variables, set by the domestic regulator. In the numerical analysis we derive the optimal consumer and producer taxes, given profit maximizing domestic producers and welfare maximizing domestic consumers.

The objective for the regulator is to maximize welfare (W), subject to the global contri- bution target, Ā, where W is utility of consuming fossil fuels net of production and net import costs:

MaxW = B(y ,y ,y ) – c (x ) – P ⋅ (y – x ) (1)∑ ∑o c g i i i i i y ,x i = o,c,g i = o,c,gi i

st.

0 ¯E ≤ E – A,

where E is global emissions and is the global emissions in absence of the unilateral, domestic0E policies.

Domestic policies affect the world market energy prices and, thereby, the global emissions (carbon leakages). We will proceed by deriving the functions for the world market prices and the impact on global emission of unilateral policies in a partial fossil fuel market model. In the following section we disregard emissions in the fossil fuel extraction processes, but return to this in Section 2.3. In section 2.4 we drive the conditions for an optimal climate policy, given that the regulator takes into account the carbon leakages effect of domestic policies.

2.2 Global Emissions from Demand and Supply Side Measures

Let capital letters denote foreign production and consumption of the three fossil fuels (Xi and Yi, i = o,c,g). As all fuels are measured in units of CO2, total global emissions from combustion of fossil fuels, , must equal global fossil fuel production, which again must equal global con-Ẽ sumption:

˜x + X = E = y + Y . (2)∑ ∑ ∑ ∑i i i i i = o,c,g i = o,c,g i = o,c,g i = o,c,g

We assume that foreign consumers are price takers, where demand for each fuel is a

function of all energy prices ( , where for and for ). For ∂D� ∂D�i i

Y = D (P ,P ,P ) � 0 i = j � 0 i ≠ ji i o c g ∂P ∂Pj j

each fuel market, foreign production must equal foreign consumption plus net import from the home country:

X = D (P ,P ,P ) + y – x , i = o,c,g. (3)i i o c g i i

Climate Policies in a Fossil Fuel Producing Country / 81

2. In Fæhn et al. (2013), Appendix A, we derive the equilibrium price functions given that the dominant oil producer a) operates as a competitive price taker, b) keeps the oil price constant, and c) keeps its production constant.

3. The leakage rate is positive and less than one ( ) when the following three conditions hold for each of theD0 � L � 1i fuels: 1) Increased net demand of one of the fuels leads to higher prices of all fossil fuels, 2) An increase in the price reduces the sum of demand of all fuels, measured in carbon content, and 3) Higher net demand increases total production of fossil fuels from abroad, measured in carbon content (see Golombek et al., 1995).

We further assume competitive behaviour by foreign coal and gas producers. Their aggre- gate supply functions are given by:

∂Si X = S (P ), � 0, i = c,g . (4)i i i

∂Pi

The oil market is characterised by a dominant producer (OPEC) with a competitive fringe (Non-OPEC):

X = Z + S (P ), (5)o o o

where Z is output from the dominant oil producer, and is aggregate supply from the com-S (P )o o petitive fringe. From (3)–(5), we write the equilibrium fuel prices as functions of net import from the home country and supply of oil from the dominant oil producer:

P = P (y – x – Z,y – x ,y – x ), i = o,c,g. (6)i i o o c c g g

Our default assumption is that the dominant oil producer maximizes net income. However, we also consider other objective functions in the numerical analysis.2

If the dominant oil producer seeks to maximize net income, Z is found from:

Max [P ⋅ Z – C(Z)] , (7)o z

where is the production cost. The first order condition is given by:C(Z)

P + P� ⋅ Z – C� (Z) = 0 . (8)o oZ

From (6) and (8), we can write all prices as functions of net import from the home country:

P = f (y – x ,y – x ,y – x ), i = o,c,g . (9)i i o o c c g g

Equation (2) can thus be written as

˜x + X = E = y + D [f ( ⋅ ),f ( ⋅ ),f ( ⋅ )] . (10)∑ ∑ ∑ ∑i i i i o c g i = o,c,g i = o,c,g i = o,c,g i = o,c,g

As international fossil fuel prices are functions of net import from the home country, domestic climate policies will affect emissions abroad. We define the marginal demand side CO2 leakage rate of fuel i, denoted , as the increase in consumption abroad (measured in carbonDLi units) following from a unit decrease in domestic consumption of fuel i:3

82 / The Energy Journal

∂ Y∑ i ∂fj = o,c,g kDL = = – D� . (11)∑ ∑i jk– ∂y ∂(y – x )j = o,c,g k = o,c,gi i i

We define marginal supply side leakage rate of fuel i ( ) as the increase in total fossilSLi fuel production abroad (measured in CO2 units) following from a unit decrease in domestic pro- duction of fuel i. As total consumption must equal total production, and yi and xi are exogenous, we find that:

∂ X ∂⎣ D( ⋅ ) – x + y ⎦∑ ∑ ∑ ∑j i i j = o,c,g i = o,c,g i = o,c,g i = o,c,gSL = – = – = (12)i

∂x ∂xi i

∂fk D1 + D� = 1 – L .∑ ∑ jk i ∂(y – x )j = o,c,g k = o,c,g i i

Hence, we can express the marginal impact on total emissions of domestic climate policies as functions of the demand side carbon leakage rate:

DẼ� = 1 – L ,y ii (13) DẼ� = L .x ii

We see from (13) that demand side policies are more (less) effective in terms of global emission reduction than supply side policies when the demand side leakage rate is smaller (bigger) than 0.5 ( for ). We also notice that If both domestic consumptionD˜ ˜ ˜ ˜E� – E� � 0 L � 0.5 E� + E� = 1.y x i y xi i i i and domestic production decrease by one unit, there is no impact on fossil fuel prices, and the final global impact is one unit less emitted.

So far we have disregarded emissions due to extraction of fossil fuels. Fossil fuels are used as input factors in the extraction process, and emission intensities vary quite a lot across sources. Hence, the global impact of domestic policies should be adjusted accordingly.

2.3 Including Emissions from Fossil Fuel Extraction

Let E denote total emissions (fossil fuel consumption including emissions from extraction):

˜E = E + α (x ) + β (X ), (14)∑ ∑i i i i i = o,c,g i = o,c,g

where and are emissions as functions of extraction of fossil fuel i in the home countryα (x ) β (X )i i i i and abroad, respectively. We find (see Appendix A):

∂ β (X ))∑ j j j = o,c,g D˜E� = E� + = 1 – L + B , (15)y y i ii i ∂yi

∂ β (X )∑ j j j = o,c,g D˜E� = E� + α� + = L + α� – B ,x x x i x ii i i i∂xi

Climate Policies in a Fossil Fuel Producing Country / 83

4. See Golombek et al, 1995, Appendix A.

where expresses the increase in emissions from extractions following from increased fossil fuelBi production abroad due to a unit increase in domestic consumption, and is given by:

D DB ≡ β� (1 – l ) – β� ⋅ l , h = o,c,g i = o,c,g . (16)∑i X ii X hii hi h ≠ i

where is the demand side leakage rate for fuel j (increased consumption of fuel j abroad due toDlji reduced consumption of fuel i at home):

∂fkDl = – D� , j = o,c,g i = o,c,g. (17)∑ji jk ∂(y – x )k = o,c,g i i

We cannot in general sign , as it depends on the signs of , and the magnitudes ofDB li ji .β�Xj

Under perfect competition, an increase in any will increase all prices.4 With increas-y – xi i ing supply functions, this means that all will increase. In this paper, the oil market is not perfectlyXi competitive. However, also under such market structures it is likely that the supply of all fossil fuels from abroad increases when domestic demand for one of the fuels increases. If this is the case, we must have that: , and , and is positive and increasing in all ofD D(1 – l ) � 1 l � 0 , i ≠ h Bii hi i the , j = o,c,g. Hence, we see from (15) that demand side policies become more effective inβ�Xj reducing global emissions the larger emission intensities abroad, whereas supply side policies be- come more effective the smaller and the larger domestic emission intensity ( ). A cut inβ� α�X xj i domestic demand drives down the fossil fuel prices and decreases production abroad, and thereby also emissions from extraction. The opposite occurs under supply side policies. Reduced domestic production leads to increased emissions from extraction abroad, but less emission from domestic extraction.

From (15) we also see that If both domestic consumption and productionE� + E� = 1 + α� .y x xi i i decrease by one unit, there are no carbon leakages, but as domestic fuel production causes emissions from extraction, global emissions decrease by more than one unit.

2.4 Optimal Unilateral Climate Policy

The objective for the regulator is to maximize welfare (W), subject to the global contri- bution target and the price functions. That is, the regulator solves (1), where is given by (9) andPi

E = y + D [f ( ⋅ ),f ( ⋅ ),f ( ⋅ )] + α (x ) + β (X ). (18)∑ ∑ ∑ ∑i i o c g i i i i i = o,c,g i = o,c,g i = o,c,g i = o,c,g

From the first-order conditions for this maximization problem, we find that:

∂fki B� = P + (y – x ) + kE� (19)∑y i k k yi i∂(y – x )k = o,c,g i i

84 / The Energy Journal

5. We abstract from the technical challenges of separating oil and gas extraction, but return to this issue in Section 3.3.2.

∂fki c� (x ) = P – (y – x ) + kE� (20)∑i i i k k xi∂(y – x )k = o,c,g i i

k is the shadow cost of the emission constraint. ( and are the marginal effects onE� E�y xi i global emissions of increased consumption and production of fuel i in the home country, respec-

tively, see (15)). is the terms-of-trade effect. If the country is a net exporter ∂fki (y – x )∑ k k

∂(y – x )k = o,c,g i i of a fuel, a higher price improves terms of trade. Hence, the terms-of-trade effects for a fuel exporter will tend to favour supply side policies, i.e., to reduce production rather than consumption. Note that this effect occurs also in the absence of climate policy. In the following we will disregard terms-of-trade effects, as the price changes and consequently the welfare impacts of this can be considered minor for the small home country relative to the other terms in (19)–(20). Small price changes do not imply, however, that global emission effects of these price changes can be ignored – consumption effects abroad may well be of the same order of magnitude as consumption effects in the home country; as analysed in the previous sections. From (19)–(20), we then find:

B� – P P – c�(x )y i i i ii = = k. (21) E� E�y xi i

Hence, optimal climate policy implies that the marginal cost of global emission reductions

through domestic demand side policy should equal the marginal cost of global emission B� – Py ii� �E�yi

reductions through domestic supply side policy across all fuels. Given that domestic P – c�(x )i i i� �E�xi

consumers and producers are price takers and maximize their net benefit and profit, it is shown in Golombek et al. (1995) that the optimal outcome can be achieved by introducing fuel-specific consumer taxes, , and producer taxes, .c pt = kE� t = kE�i y i xi i

3. NUMERICAL ANALYSIS

We now turn to the comparison of demand and supply side policies in the case of Norway. Our focus is on the year 2020, as the Norwegian government has specific climate goals for that year (see below). In Section 3.1 we estimate marginal costs of Norwegian unilateral reductions in fossil fuel demand and supply. This means quantifying and , respectively; see Eq.B� – P P – c�(x )y i i i ii (21). Demand side abatement is assessed by means of a computable general equilibrium (CGE) model for Norway, which is simulated for the year 2020 (Fæhn and Isaksen, 2015). Supply side measures are quantified by identifying representative, marginal cuts in Norwegian oil production based on historic data for the period 2009–2011. We then discuss the relevance of these data for the year 2020. Norway is also a significant producer of gas, accounting for around 3 per cent of global gas production (BP, 2013). Gas is, however, a fossil fuel with relatively low emissions and with larger substitutability against the high-emitting coal. Hence, it is not clear whether reduced Norwegian gas extraction would decrease or increase global emissions, and we do not consider this supply side option in our analysis.5

Climate Policies in a Fossil Fuel Producing Country / 85

6. The same simulated scenario is used in Climate Cure 2020 (2010), the report of an officially appointed commission tasked with preparing the ground for evaluating Norway’s climate policy.

7. This is consistent with the assumption made by e.g., Tol (2014).

In Section 3.2 we analyse the effects on global emissions by exploiting a partial model of the global fossil fuel market effects, where we also take into account emissions from extraction of fossil fuels. The model is calibrated based on historic data (2011), and again the relevance for the year 2020 is briefly discussed. These computations will provide the values of the denominators in Eq. (21), and . In Section 3.3 we combine the findings in the two preceding sections to deriveE� E�y xi i the optimal combination of demand and supply side policies for Norway as expressed in Eq. (21).

3.1 Unilateral Climate Policy

3.1.1 Demand side policies

The Norwegian parliament has announced high ambitions for its contribution to global (demand side) emission reductions, corresponding to a 30 per cent reduction from Norwegian 1990 emissions by 2020. Moreover, it has emphasised that the lion’s share of the reductions is to result from domestic action. To obtain a marginal cost function for demand side measures in Norway, we use Statistics Norway’s technology-rich CGE model for the Norwegian economy, MSG-TECH (see Fæhn and Isaksen, 2015). We simulate costs of uniform emissions pricing given different demand side abatement levels. The effects are measured from a reference scenario that incorporates climate policies already implemented, approved, or promised for the years up to 2020. From 2008, this includes the participation in the EU ETS.6

Since we assume that the demand side abatement aims to contribute to global emission reductions, we only consider emissions pricing in sectors outside the EU ETS. With the cap on total emissions in the EU ETS, additional cuts in Norwegian ETS sectors will merely displace emissions to ETS-regulated installations in other European countries.

Based on a number of simulations, we find a marginal cost curve for Norwegian demand side measures as expressed by Eq. (22) and depicted in Figure 1. (See the numerator of the first fraction of Eq. (21)). For all the simulated emission targets, all abatement takes place as reduced oil consumption, mostly within the transport sector. These measures are a mixture of investments in new vehicle technologies, substituting public for private transport and reducing transportation demand.

2B� – P = – 2.5A + 86.6A + 23.4 . (22)y o D Do

AD denotes the level of domestic emission reductions (measured in million tonnes of CO2). As Figure 1 suggests, the marginal abatement cost is increasing and fairly linear.7 The interpretation is that you will have to resort to yet more expensive cuts as you add to the abatement ambitions, but the cost increase between the last CO2 unit abated and the next cost-efficient marginal measure is fairly constant. The abatement cost at the margin is a result of a vast number of simultaneous responses in the large-scale CGE model that depend on a variety of elasticities and other parameters.

3.1.2 Supply side policies

The costs of supply side measures in our static framework are the forgone profits by not extracting the oil, corresponding to ; see the numerator of the second fraction of Eq. (21).P – c�(x )o o o

86 / The Energy Journal

Figure 1: Marginal Costs of Foregone Fossil Fuel Consumption in Norway

8. This is not explained by the averaging of the oil price; even if we use the actual yearly oil prices some of the fields come out with negative profit in some of the years. They still produced all the years.

We single out oil fields which can be characterised as marginal, in the sense that terminating extraction involves small profit losses per unit. Oil fields in the decline phase generally have higher costs than fields in the plateau phase. Explanations are that marginal operating costs, including energy input, are increasing as remaining oil in the reservoir declines. In addition, IOR (Improved Oil Recovery) activities to prolong the lifetime of maturing fields can involve new costly invest- ments, implying that the profit losses of not undertaking an IOR project may be modest (though not always). Typically, these fields also have higher emission intensity. Unfortunately, we have limited information about IOR costs (see below).

For the years 2009–2011 we have singled out nine Norwegian fields where oil constituted a major part of total petroleum production (several were pure oil fields). In addition, these fields were in, or close to, the decline phase. We have field data from Statistics Norway on production volumes and variable costs, costs that would not accrue if oil production were terminated. Based on these data we have constructed a marginal production cost curve; see Appendix B.

To calculate marginal forgone profits by reduced oil production, we apply the average oil price over the period 2009–2011 (USD 84.5 per barrel of Brent Blend), and subtract the marginal production costs. The results can be considered as the marginal costs of forgone oil extraction in Norway, and are shown in Figure 2. The resulting curve of marginal forgone profits is concave, since the marginal production cost curve has the usual convex shape. Note that we disregard the fact that reduced supply of Norwegian oil would increase the oil price marginally, slightly increasing the revenue of the remaining Norwegian oil export; see the sensitivity analysis. The supply side marginal cost curve, where AS is reduced extraction measured in million tonnes of CO2, is:

2P – c�(x ) = – 0.7A + 19.6A – 6.1 . (23)o o o S S

We see that it is actually profitable to reduce a small amount (0.3 Mt of CO2) irrespective of climate benefits, due to high production costs of some of the smaller fields.8

Climate Policies in a Fossil Fuel Producing Country / 87

Figure 2: Marginal Costs of Foregone Oil Extraction in Norway

9. An oil price of USD 84.5 per barrel, and a break-even price of USD 60 per barrel, implies a cost of USD 24.5 per barrel forgone oil production, corresponding to USD 58 per tonne CO2. An average production of 1.4 million Sm

3 over the period 2016–2028 leads to 3.7 million tonnes of CO2 per year when it is combusted.

In our study we are interested in abatement options in the near future such as 2020. Thus, the relevant question is to what extent the marginal cost function depicted in Figure 2 is represen- tative for coming years. Several of the fields we have studied for the years 2009–2011 will stop producing before 2020. On the other hand, some fields that are now in their plateau phase will be in their decline phase around 2020, suggesting that their costs per unit production will increase. It is difficult to know whether the net effect of these considerations will push the cost curve in Figure 2 up or down.

However, there are several reasons why we may have underestimated the total costs of production, i.e., overestimated the costs of reducing production. First, we do not have specific information about the costs of IOR projects, which are often projects with limited profits per unit of extraction. Second, we have only considered advanced termination of maturing fields. A cost- effective downscaling of oil production may also imply that some fields with limited profitability are not developed at all.

To help assessing the relevance of Figure 2 for the year 2020, we have also gathered information on an oil field named Ivar Aasen that is decided to be developed. Here we have access to information about both expected annual development and operating costs, as well as production (The Norwegian Oil Company, 2012). Based on the reported data we calculate a break-even oil price of USD 60 per barrel for this field, using a discount rate of 6 per cent which the oil company uses. The forgone profits of not developing this field are comparable to the data behind Figure 2.9

In addition, Rystad (2013) points to several Norwegian (undeveloped) oil fields with break-even

88 / The Energy Journal

prices above USD 72 per barrel. These observations support our belief that the costs of reducing oil production are lower than what we assume in our analysis.

The oil price around 2020 may be different from what it was in 2009–2011. The steep decline in the oil price lately illustrates this quite clearly. If the oil price in 2020 becomes lower than expected, then forgone profits of reduced oil extraction will also be lower if extraction costs are unchanged (and vice versa if the oil price becomes higher than expected). However, extraction costs have tended to be positively correlated with the oil price (see e.g., Osmundsen et al. (2015), who find that rig rates in the Gulf of Mexico increase significantly with oil and gas prices), meaning that the effect of a different oil price on forgone profits could be moderated. In addition, a lower (higher) oil price could entail lower (higher) oil production in the reference case (i.e., in a situation without supply side policies), at least if the lower (higher) oil price is expected. We are interested in the marginal costs of reducing oil production compared to the reference case. Hence, it is not certain that a lower (higher) oil price will lead to lower (higher) costs of supply side measures if the oil price change is expected by the Norwegian oil producers.

To sum up, although the uncertainties are rather large, it seems more likely that the marginal costs of supply side measures around 2020 lie below than above the curve shown in Figure 2.

3.2 Numerical Analysis of Global Fossil Fuel Markets

3.2.1 The partial fossil fuel market model

Based on the exposition in Section 2, we construct a simple numerical model that makes it easy to identify and adjust the basic assumptions driving the results (the model equations are specified in Appendix C). The main drivers are i) price responsiveness on the demand side (in- cluding substitution effects between oil and other fossil fuels), ii) price responsiveness of Non- OPEC supply, iii) OPEC’s response, and iv) differences in emission intensity in oil extraction. We consider iso-elastic demand functions (i.e., with constant direct and cross price elasticities), iso- elastic supply functions for competitive fossil fuel producers, and constant unit production costs for OPEC (when behaving as a dominant producer). As we are focusing on a permanent cut in oil supply as a potential supply side measure, we are mostly interested in the long-run effects in the market, i.e., we consider long-run elasticities. Finally, we model fixed emission intensities in oil extraction, but these should be interpreted as emission intensities of marginal production. Appendix D contains a detailed discussion of the main drivers, in particular a review of existing demand and supply elasticity estimates from the literature. Here we only present the assumptions of our bench- mark case, which are motivated in Appendix D.

Oil price increases may reduce oil consumption in various ways. Oil consumers may reduce their total energy use, or they may switch to other energy goods such as coal, gas, or renewables. Switching to other energy goods requires that there are viable alternatives, which will vary across sectors. Reducing total energy use may either involve reduced use of energy services (e.g., driving fewer miles, producing/consuming less energy-intensive products), or using more energy-efficient vehicles (or transport modes), capital, or equipment. In the long run, higher prices may also stimulate the development of more oil-efficient technologies. In principle, long-run price elasticities should capture all these effects. Based on the literature review, we apply a direct price elasticity of – 0.5 in the long run, and cross-price elasticities for coal and gas of 0.08. However, we report the effects of other estimates as well.

Higher prices of oil increase the profitability of oil exploration, new fields developments, and IOR projects. An increase in the price of oil will mostly affect extraction of so-called marginal

Climate Policies in a Fossil Fuel Producing Country / 89

10. The market in 2020 will likely deviate somewhat from the market outcome in 2011. However, this deviation has marginal impacts on our results.

11. Note that the calibrated unit cost for OPEC is increasing in the absolute value of the residual demand elasticity. When the residual demand is more elastic, OPEC is less interested in cutting supply to increase the oil price, and hence unit costs must be higher to obtain a reference case consistent with base year data.

12. OGP (2012) reports both emissions and production data for seven regions of the world. We deduct emissions and production from the Middle East and half of those from Europe (i.e., Norway), and calculate the emission intensity for the remaining regions, which we then assume is representative for Non-OPEC.

resources, such as exploration and field development in ultra-deep waters, developments of smaller fields and unconventional oil, and IOR projects. Higher oil prices may also lead to improved technologies in the long run, similarly to oil-efficiency improvements on the demand side. Based on the literature review, we use a supply elasticity of 0.5 for Non-OPEC. This implies that oil demand and Non-OPEC supply are equally price elastic. However, due to substitution between oil and other fossil fuels, the fossil fuel demand elasticity (with respect to the oil price, and measured in CO2 units) becomes around – 0.4.

As discussed in Section 2, our default assumption in our benchmark case is that OPEC behaves as a dominant producer. The unit production cost of OPEC is calibrated so that our reference simulation is consistent with the market outcome of the year 2011.10 In our benchmark case, the unit marginal production cost of OPEC turns out to be 45 per cent of the oil price in 2011, which is within the range of production costs reported by IHS CERA for OPEC countries (see e.g. Figure 3.9 in Ministry of Petroleum and Energy, 2011).11 When we model OPEC as a competitive producer, we assume the same supply elasticity as for Non-OPEC.

Although the lion’s share of CO2-emissions from oil use takes place as the oil is combusted, emissions from oil extraction have to be counted, as well. According to OGP (2012), the average GHG emissions per unit production worldwide in 2011 were 159 tonnes CO2e (CO2 equivalents) per 1,000 toe hydrocarbon produced. The figure for the Middle East is only 51 tonnes CO2e, but the coverage is less comprehensive for this region – hence the real average could potentially be higher.

The European figure is 84 tonnes CO2e. OGP (2012) does not report figures for Norway, but based on data from Statistics Norway we calculate the average Norwegian emission intensity in 2011 to be 60 tonnes CO2e per 1,000 toe . For the rest of Non-OPEC we make a rough calculation based on the OGP (2012) figures for the Middle East, Europe and the world, arriving at around 200 tonnes CO2e.

The average figures reported above will typically deviate from the marginal change in emissions of increased or reduced oil production. Reduced oil production in Norway could e.g., involve reduced IOR activity or advanced termination of a field. In both these cases, energy use per unit extraction will tend to be higher than average; see Fæhn et al. (2013), Appendix C. The same could be true for reduced oil exploration or field developments, at least in aggregate, as the marginal areas or fields will tend to be less profitable, which often means that more costly energy is needed per unit production.

Similarly, increased supply from other Non-OPEC producers could imply higher-than- average emission intensities. For instance, Canadian oil sands are considered relatively costly and thus marginal resources, with average emission intensities around three times the world average. When it comes to OPEC supply, however, increased production may come from increased extraction of developed fields in countries like Saudi Arabia and, thus to a lesser extent involve higher emission intensities.

90 / The Energy Journal

13. The marginal emission intensities could in fact be even higher – at least some fields have even higher emission intensities. However, it is difficult to know if the most emission-intensive fields are the marginal fields, both in Norway and elsewhere. Anyway, as the results in Table 1 indicate, assuming even higher emission intensities in production would not alter our results substantially, as emissions related to production are much lower than emissions from combustion.

14. It could be argued that the emission intensity of Norwegian oil extraction should be set to zero, as these emissions are regulated by the EU ETS, which has a cap on overall emissions (cf. the discussion of ETS sectors in Section 3.1.1). As seen in the following section, however, these emissions are of less importance.

Our benchmark case assumption is that marginal emission intensities are 50 per cent above the reported average figures above.13 For Norway and (other) Non-OPEC countries this is related to the marginal supply most likely being more emission-intensive than average supply. For OPEC, the increase is partly related to less comprehensive reporting and reliance on Middle East figures (see above) and partly to marginal supply possibly being more emission-intensive than average supply. Thus, we set the emission intensities in Norway, OPEC and Non-OPEC equal to respectively 90, 76 and 300 tonnes CO2e per 1,000 toe.

14 For comparison, emissions from consuming (i.e., combusting) 1,000 toe of oil is about 3,070 tonnes of CO2. Although of minor importance here, we also account for emissions from extracting other fossil fuels, and set emission intensities for coal and gas equal to the Non-OPEC emission intensity reported above (i.e., 300 tonnes of CO2e per 1,000 toe).

3.2.2 Effects on global emissions of demand and supply side policies

We first report the simulation results of exogenously reducing Norwegian oil extraction or consumption by one unit of CO2. We are interested in the net effects on global emissions, i.e., the denominators and in Eq. (21). As shown in Section 2, the sum of and should equalE� E� E� E�x y x yi i i i one plus , i.e., the emissions from domestic extraction (relative to emissions from consumption).α�xi

Table 1 displays the net global emission reductions when OPEC acts as either a dominant or a competitive producer. The table also shows the various components of the emission reductions. Note that the leakage rate defined in Section 2 is equal to minus the sum of “Oil market leakage”DLo and “Coal/gas market leakage” under “Demand side” policy (and also equal to the sum of the three first components under “Supply side” policy).

We first notice that leakage through the oil market is around 50 per cent for both demand side and supply side leakage. This is certainly the case if OPEC acts competitively, and follows straightforwardly from the assumption of equal (absolute values of) supply and demand elasticities. If OPEC acts as a dominant producer, it is optimal for the producer group to adjust its supply slightly more to changes in Norwegian supply or demand compared to in the competitive case, but the difference is not big: Supply side leakage through the oil market is 55 per cent, compared to 45 per cent for demand side leakage.

Next, we see from Table 1 that overall market leakage is substantially lower under demand side policy than under supply side policy, whether OPEC behaves competitively or as a dominant producer. This is due to substitution between oil and other fossil fuels, which obviously goes in different direction depending on whether the oil price drops (demand side) or increases (supply side). When oil demand abroad increases (decreases) due to reduced Norwegian oil consumption (extraction), coal and gas consumption is somewhat reduced (increased). This effect alone accounts for almost 10 per cent of the gross emission reduction.

Finally, the importance of emissions from fossil fuel extraction is modest, accounting for respectively 3.7 per cent and 6.1 per cent of the net global emission effect of reduced Norwegian

Climate Policies in a Fossil Fuel Producing Country / 91

Table 1: Net Global Emission Reduction from Reduced Norwegian Oil Extraction or Consumption by One Unit of CO2. Benchmark Case

15. The Norwegian government and parliament have high ambitions for domestic abatement by 2020. An agreement (The Climate Settlement) signed in 2008, and reinforced in 2012, by the parliamentary majority states that 2/3 of the emission reduction from 1990 to 2020 will take place within own borders. When assuming that Norway fulfils this domestic target through abating Norwegian non EU-ETS emissions, a 5Mt cut in global emissions in 2020 is in line with what can be achieved when the leakages are accounted for; see Fæhn et al. (2013).

extraction and consumption of oil. The effects are highest for demand side policy, as under supply side policy increased emissions from oil extraction outside Norway are modified by reduced emis- sions from oil extraction in Norway.

Obviously, net emission reductions are sensitive to a number of assumptions such as price elasticities and OPEC behaviour. Hence, in Section 3.3.2 below we present a detailed sensitivity analysis. We now use the findings in Table 1 to analyse the optimal balancing of demand and supply side policies, focusing on the case with OPEC as a dominant producer.

3.3 Optimal Balancing of Demand and Supply Side Policies

3.3.1 The cost-effective solution

By combining the demand side and supply side cost curves expressed in eqs. (22) and (23) with the quantifications of their net effects on global emissions, and , derived above, we canE� E�x yi i find the optimal composition of domestic action; see Eq. (21). Eq. (21) expresses that the marginal cost of global emission reduction in optimum is equal for demand side and supply side action. In our benchmark case with OPEC as a dominant producer the net global emission reduction of demand side policies, A, is estimated to 67.6 per cent of domestic abatement, AD (see Table 1). The corre- sponding estimate for supply side policies is 35.3 per cent of AS. As seen from Eq. (21), the marginal cost of global emissions reduction is equal to Eq. (22) divided by 67.6 per cent for demand side policies and to Eq. (23) divided by 35.3 for supply side policies.

We exploit these relations to depict the optimum combination of demand side and supply side policies for a global contribution target, Ā(see Eq. (1)). We pick a target of 5 Mt of CO2 by 2020 (equal to 1.9 million Sm3), which corresponds to about 10 per cent of Norway’s present domestic greenhouse gas emissions.15 Moreover, In Figure 3 we show a bath tub diagram with

92 / The Energy Journal

Figure 3: Combining Leakage-adjusted Demand and Supply Side Marginal Cost Curves

16. 1USD/tonne CO2 = 0.42 USD/barrel. 1 tonne CO2 = 0.317 toe.

length equal to Mt global emission reduction, and where the marginal costs of global emissionĀ = 5 reduction through demand (supply) side measures are shown from left to right (right to left). The intersection point between the two curves shows the optimal combination of demand and supply side measures. We notice that 3.3 Mt CO2, or about 2/3 of the global contribution target, should be met through reduced oil extraction. The remaining 1.7 Mt is met by demand side policies. The corresponding marginal costs of reducing global CO2 emissions through this combination are 336 USD per tonne16. The figure also reflects that if the global reductions were to be met through demand side measures alone, costs would more than double.

Implementing this optimal combination of demand and supply side measures would mean that domestic CO2 emissions should be reduced by 2.5 Mt of CO2 . Given our benchmark case, this domestic abatement is necessary to obtain a global reduction of 1.7 Mt CO2 since global achieve- ment is only 67.6 per cent of the domestic abatement. The domestic CO2 reduction can be obtained, e.g., by a domestic CO2 tax on non-EU ETS emissions of 228 USD per tonne CO2 (cf. Figure 1). This tax rate corresponds to the marginal cost of reducing global CO2 emissions in optimum (336 USD per tonne) multiplied by the same share as explained above (67.6 per cent). Almost 90 per cent of the demand measures that are profitable to carry out relates to transportation, of which reduced private transport accounts for 20 per cent and transition to more climate friendly vehicles accounts for the rest.

In the optimum, Norwegian oil extraction should be reduced by 3.5 million Sm3 (one standard cubic meter, Sm3, equals 6.29 barrels), which is 3.1 per cent of total Norwegian oil pro- duction in 2012. 3.5 million Sm3 contains 9.2 Mt CO2, however, with benchmark assumptions only

Climate Policies in a Fossil Fuel Producing Country / 93

35.3 per cent of this, or 3.3 Mt, is the resulting global emission reduction in the optimal equilibrium. This reduction can be achieved in different ways, e.g., through a production tax on Norwegian oil extraction. The optimal marginal cost of reducing global CO2 emissions estimated above (336 USD per tonne) corresponds to a production tax of USD 50 per barrel, i.e., 60 per cent of the crude oil price assumed. This tax is found by multiplying the marginal cost by the net effect on global emissions of reduced Norwegian extraction (0.353) and by the CO2 content of a barrel of oil (0.42). As mentioned in Section 3.1.2, the break-even price of the Ivar Aasen oil field, which can be characterised as relatively profitable, is around USD 60 per barrel.

Below we discuss the pros and cons of implementing a production tax. Here we want to emphasise that a production tax of around USD 50 per barrel could potentially lead to a much bigger reduction in oil extraction than the 3.1 per cent calculated above. The reason is that, as underlined in Section 3.1.2, we most likely overestimate the costs of reducing oil extraction.

3.3.2 Sensitivity analysis

There are other uncertainties in our calculations, too, especially the effects in the fossil fuels markets of reduced oil extraction or consumption in Norway. As our brief literature review in Appendix D shows, both supply and demand elasticities vary a lot across empirical studies, with most estimates ranging between 0.1 and 1 in absolute value. For our purpose, the relative difference between supply and demand elasticities is the most important. OPEC’s behaviour is also somewhat uncertain. In Table 2 we present a number of sensitivity analyses where we adjust assumptions from our benchmark case. The global contribution target is held fixed at 5 Mt CO2.

We notice from Table 2 that assuming competitive behaviour by OPEC gives more or less the same results as above – this is not surprising given the results in Table 1. Besides that, we see that whether demand side policies or supply side policies are most effective in reducing global emissions depends quite a lot on what we assume about the oil market. If we think that OPEC keeps its supply fixed, or if demand is twice as elastic as supply, cuts in oil extraction are even more effective in reducing global emissions, and 90 per cent of the global emission reductions results from supply side measures. Nevertheless, the optimal production tax does not change much. The global emission effects of reduced oil extraction ( ) are increased, shifting the supply sideE�xi curve in Figure 3 downwards. Likewise, the demand side curve in Figure 3 shifts upwards. Still, the intersection point drops down, meaning that the shadow cost of the emission constraint (cf.k Eqs.(19)–(20)) declines. However, the optimal tax on oil extraction is proportional to (cf. SectionE�xi 2), which has increased. The domestic CO2 price drops quite substantially, though, due to a com- bination of lower and lower .k E�yi

If we think that supply is twice as elastic as demand, cuts in oil extraction is less effective and the share of emission reductions resulting from supply side measures drop to 25 per cent. Again, we see that the optimal production tax is less affected, while the domestic CO2 price has increased quite a lot. If supply is even more elastic relative to demand, supply side measures become even less attractive. The literature review in Appendix D does not suggest so, but the large variation in estimated elasticities implies that such a case cannot be ruled out. Related to this, if OPEC for some reason chooses to keep the oil price fixed, reduced oil extraction by the home country gives insignificant global emission reduction, as supply is merely shifted to other producers and the only effect will arise from the small differences in emissions from extraction at the margin. The optimal choice will be the conventional one of only doing demand side policies; see Table 2.

As mentioned, we disregard the fact that reduced supply of Norwegian oil would increase the oil price marginally. This would reduce the costs of supply side measures by roughly 10 per

94 / The Energy Journal

Table 2: Sensitivity Analysis. Effects of Reducing Norwegian Extraction or Consumption of Oil by One Unit of CO2

cent in the cost-effective solution depicted in Figure 3 and slightly increase the costs of demand side measures. If we have overestimated the costs of reduced oil extraction, we should undertake even more supply side measures than suggested by Figure 3. Moreover, the optimal domestic CO2 price and the optimal production tax for Norwegian oil extraction should then be reduced. For instance, if we scale down the supply side cost curve by 50 per cent, 83 per cent of total abatement should be caused by supply side measures, with the optimal domestic CO2 price and production tax being 126 USD per tonne CO2 and 28 USD per barrel; see Table 2.

On the other hand, we have ignored the challenges of separating oil and gas extraction, which may suggest that we have underestimated the forgone profits of reduced oil extraction. However, the share of gas in total oil and gas production for the nine fields studied above was merely 5 per cent. Moreover, for 8 of the 13 fields currently under development on the Norwegian shelf, more than 90 per cent of recoverable reserves are oil (Ministry of Petroleum and Energy, 2013). Hence, this may be of limited importance.

The higher the oil price, the less profitable it is to restrict extraction from a given oil field as the profit margin is higher. As explained in Section 3.1.2, however, a higher oil price does not necessarily mean that supply side measures become more costly if the higher price is expected by the oil producers, as more expensive resources will then be extracted in the reference case. The costs of supply side measures then depend on the profit margin of these more expensive resources. Most likely, the marginal unit available for supply side policies will still have a profit near zero and, thus, be a low-cost supply side measure. Anyway, it is very unlikely that it is cost effective to

Climate Policies in a Fossil Fuel Producing Country / 95

17. The Norwegian CO2 tax is differentiated across fuels and sectors. The highest tax level in non-ETS sectors is on petrol, at 393 NOK per tonne CO2 in 2013 (67 USD), cf. Ministry of Finance (2012).

rely only on demand side measures. Given the benchmark case estimates of and , it is optimalE� E�x yi i to implement some supply side measures as long as the net revenue of the least profitable oil extraction is less than 116 USD per barrel. That is, even if the oil price should unexpectedly approach 200 USD per barrel, some supply side measures could be optimal.

3.3.3 Policy alternatives and discussion

So far, we have taken for granted that the Norwegian government will impose sufficiently strong measures to reach its global contribution target,Ā. A reasonable first step towards this goal could be to implement supply side policies comparable to the demand side policies already in place in Norway. The current CO2 tax imposed on Norwegian non-ETS sectors is 67 USD per tonne CO2.

17 Using the benchmark case value of (i.e., 0.676 – see Table 2), this translates into aE�yi shadow price of global emission reductions ( ) of 98 USD (66/0.676, cf. Eq. (21)), which furtherk translates into a corresponding production tax of 14 USD per barrel (98 ⋅ 0.353 ⋅ 0.42; cf. section 3.3.1 and Eq. (21)). That is, supplementing a domestic CO2 price of 66 USD per tonne CO2 in non- ETS sectors with an oil production tax of 14 USD per barrel would imply a cost-effective combi- nation of demand and supply side climate policies. Naturally, the global target,Ā, would not be reached with these moderate measures – global emissions would decline by a little more than one million tonnes. The main conclusion that the lion share – about 2/3 – of the global contribution left for domestic actions should be supply side measures in optimum still holds.

As explained above, the derived marginal costs of emission reductions translate into a shadow price on oil production equal to 50 USD per barrel (in the benchmark case). This shadow price can in principle be implemented through a corresponding production tax on all oil production in Norway. However, implementing such a large tax overnight is not without drawbacks. First, we have already noted above that we may have overestimated the costs of reducing oil extraction. As a thought experiment, assume that half of Norwegian oil production becomes unprofitable with the indicated tax level, and that the forgone profits amount to on average 25 USD per barrel, i.e., half of the tax. Using the production level of 2012, total costs would then be 17 billion USD, compared to 1.1 billion USD in the benchmark solution. Although this thought experiment may be somewhat extreme, it illustrates that there is a substantial downside risk by implementing such a large pro- duction tax for such a big sector.

Second, Norwegian authorities have, for good reasons, been cautious about changing the taxation rules, at least for already developed fields. Implementing additional taxes could be seen as changing the rules of the game, increasing the risk of doing business on the Norwegian continental shelf. Hence, it is easier to make a case for imposing a large production tax on extraction from undeveloped fields, unexplored areas and even developed fields requiring upgrading through IOR projects, than on sanctioned extraction from developed fields.

An alternative supply side policy, e.g., combined with a more limited production tax, could be to have a more restrictive practise when it comes to opening new areas for oil exploration. At least it seems reasonable to take a global perspective similar to the one in this paper when under- taking impact assessments of opening new areas for exploration.

4. CONCLUDING REMARKS

The conventional way of implementing policies to reduce CO2 emissions is through the demand side, that is, introducing measures or instruments to reduce the consumption of fossil fuels.

96 / The Energy Journal

In a closed market such as the global economy, demand and supply side measures may be equivalent. This is not the case, however, when only one or a group of countries implement climate policies. Demand and supply side measures will then have different effects, depending, in particular, on the price responsiveness on the demand and supply side of the market.

In this paper we have derived analytical expressions for the optimal combination of demand and supply side policies for a fossil fuel producing and consuming country that has a fixed target for its contribution to reducing global emissions. We have also accounted for emissions from the extraction of fossil fuels, which comes in addition to emissions from the use (i.e., combustion) of the fuels.

Based on this analytical framework, we have analysed the optimal combination of demand and supply side climate policies for a small oil producing country, Norway, using data for domestic abatement costs and forgone profits for Norwegian oil production, as well as a transparent model of international fossil fuel markets. We find that a majority of measures should be implemented on the supply side, that is, by reducing Norwegian extraction of oil. In our benchmark case the optimal combination of demand and supply side measures involves annual cuts in Norwegian oil extraction of around 3.5 million Sm3 (around 3 per cent of current Norwegian oil production), and annual domestic reductions in CO2 emissions of 2.5 million tonnes of CO2 (almost 5 per cent of current Norwegian CO2 emissions). In contrast, the Norwegian Government suggests using demand side measures, only. We find that such a strategy more than doubles the costs. As our numerical ap- proaches are designed for long run assessments, transaction costs of reallocating resources to the new equilibrium are not included. These can be significant, given our focus on the year 2020. However, it should be noted that transaction costs are relevant for both demand and supply side strategies. Despite a number of uncertainties in our calculations, the conclusion that a majority of the Norwegian contributions to global abatement should be taken through supply side measures, seems quite robust. With our benchmark assumptions on market behaviour, 2/3 of the global emis- sion reductions should come through supply side measures. In particular, irrespective of the ambition level set in terms of global abatement contributions, the optimal domestic action will have a lion share of supply side measures.

The optimal policy combination is, at least in principle, a tax per tonne domestic CO2 emissions and a tax per barrel of domestic oil extraction. The tax levels we derive in our benchmark case are high, driven e.g. by the high costs of reducing Norwegian emissions from sectors that are not regulated by the EU ETS. Implementing such high taxes overnight is not without drawbacks, especially on the supply side, and we have discussed alternative ways of implementing cuts in Norwegian oil extraction.

Although our numerical case has been Norway, the question of demand side versus supply side climate policies is relevant for all fossil fuel producing countries. For instance, the EU has introduced a number of demand side measures such as the EU ETS, but no supply side measures yet. Reduced coal extraction is a potential climate policy measure, given that it leads to higher international coal prices and lower coal consumption outside the EU. Thus, an interesting follow- up study could be to examine the optimal combination of demand and supply side measures in the EU or other countries with unilateral climate policies.

ACKNOWLEDGMENTS

We are grateful to Torbjørn Hægeland, Ådne Cappelen, and anonymous referees for com- ments on an earlier draft. We would also like to thank the research centre CREE (Centre for Research

Climate Policies in a Fossil Fuel Producing Country / 97

on Environmentally friendly Energy) for funding. The authors Fæhn, Hagem, Lindholt and Rosen- dahl are associated with the CREE-centre.

REFERENCES

Alhajji, A. and D. Huettner (2000). “OPEC and World Crude Oil Markets From 1973 to 1994: Cartel, Oligopoly, or

Competitive?” The Energy Journal 21(3): 31–60. http://dx.doi.org/10.5547/ISSN0195-6574-EJ-Vol21-No3-2.

Askari, H. and N. Krichene (2010). “An oil demand and supply model incorporating monetary policy.” Energy 35: 2013–

2021. http://dx.doi.org/10.1016/j.energy.2010.01.017.

Böhringer, C., C. Fischer, and K.E. Rosendahl (2010). “The Global Effects of Subglobal Climate Policies.” The B.E. Journal

of Economic Analysis&Policy 10(2), (Symposium) Article 13. http://dx.doi.org/10.2139/ssrn.1694104.

Böhringer, C., B. Bye, T. Fæhn and K. E. Rosendahl (2012a). “Alternative designs for tariffs on embodied carbon – a global

cost-effectiveness analysis.” Energy Economics 34 (Suppl. 2): 143–153. http://dx.doi.org/10.1016/j.eneco.2012.08.020.

Böhringer, C., J.C. Carbone, and T.F. Rutherford (2012b). “Unilateral climate policy design: Efficiency and equity impli-

cations of alternative instruments to reduce carbon leakage.” Energy Economics 34 (Suppl. 2): 208–217. http://dx.doi.org/

10.1016/j.eneco.2012.09.011.

BP (2013). “BP Statistical Review of World Energy 2013.” London: BP.

Brons, M., P. Nijkamp, E, Pels and P. Rietveld (2008). “A meta-analysis of the price elasticity of gasoline demand. A SUR

approach.” Energy Economics 30(5): 2105–2122. http://dx.doi.org/10.1016/j.eneco.2007.08.004.

Climate Cure 2020 (2010). “Tiltak og virkemidler for å nå norske klimamål mot 2020.“ Klima- og forurensningsdirektoratet,

Norges vassdrags- og energidirektorat, Oljedirektoratet, Statistisk sentralbyrå, Statens vegvesen. Rapport TA2590.

Dahl, C. (2006). “Survey of Econometric Energy Demand Elasticities; Progress Report.” August 2006, Unpublished report.

Dahl, C. (2012). “Measuring global gasoline and diesel price and income elasticities.” Energy Policy 41: 2–13. http://

dx.doi.org/10.1016/j.enpol.2010.11.055.

Dahl, C. and T.E. Duggan (1998). “Survey of Price Elasticities from Economic Exploration Models of US Oil and Gas

Supply.” Journal of Energy Finance & Development 3(2): 129–169. http://dx.doi.org/10.1016/S1085-7443(99)80072-6.

Ellis, J. (2010). “The Effects of Fossil-Fuel Subsidy Reform: A review of modelling and empirical studies.” International

Institute for Sustainable Development (IISD), March 2010. http://ssrn.com/abstract = 1572397.LISHED ONLINE: 10

AUGUST 2014 ⎪ http://dx.doi.org/10.2139/ssrn.1572397. Erickson, P. and M. Lazarus (2014). “Impact of the Keystone XL pipeline on global oil markets and greenhouse gas

emissions.” Nature Climate Change 4: 778–781. DOI: 10.1038/NCLIMATE2335. http://dx.doi.org/10.1038/ncli-

mate2335.

Farzin, Y.H. (2001). “The impact of oil prices on additions to US proven reserves.” Resource and Energy Economics

23(3):271–291. http://dx.doi.org/10.1016/S0928-7655(01)00040-9.

Forager Funds (2013). “Norwegian Hypocrisy on Global Warming.” https://www.foragerfunds.com/bristlemouth/norwegian-

hypocrisy-global-warming.

Fournier, J. I. Koske, I. Wanner and V. Zipperer (2013). “The Price of Oil – Will it Start Rising Again?” OECD Economics

Department Working Papers, No. 1031, OECD Publishing. http://dx.doi.org/10.1787/5k49q186vxnp-en.

Fæhn, T and E. Isaksen (2016). “Diffusion of climate technologies in the presence of commitment problems.” The Energy

Journal. 37(2): 155–180. http://dx.doi.org/10.5547/01956574.37.2.tfae.

Fæhn, T., C. Hagem, L. Lindholt, S. Mæland and K. E. Rosendahl (2013). “Climate policies in a fossil fuel producing

country- demand versus supply side policies.” Discussion paper 747, Statistics Norway.

Golombek, R., C. Hagem and M. Hoel (1995). “Efficient incomplete international climate agreements.” Resource and Energy

Economics 17(1): 25–46. http://dx.doi.org/10.1016/0928-7655(94)00022-C.

Greenpeace (2014). “Frozen in Time: Coal Mining at the Heart of Climate Change.” http://www.greenpeace.org/africa/en/

News/Blog/frozen-in-time-coal-mining-at-the-heart-of-cl/blog/37006/.

Griffin, J.M. (1985). “OPEC behavior: a test of alternative hypotheses.” American Economic Review 75: 954–963.

Hagem, C. (1994). “Cost-effective climate policy in a small country.” Energy Journal 15(4): 119–139. http://dx.doi.org/

10.5547/ISSN0195-6574-EJ-Vol15-No4-6.

Hansen, P.V. and L. Lindholt (2008). “The Market Power of OPEC 1973–2001.” Applied Economics 40(22): 2939–2959.

http://dx.doi.org/10.1080/00036840600972480.

Harstad, B. (2012). “Buy Coal! A Case for Supply side Environmental Policy.” Journal of Political Economy 120(1): 77–

115. http://dx.doi.org/10.1086/665405.

98 / The Energy Journal

Hoel, M. (1994). “Efficient Climate Policy in the Presence of Free Riders.” Journal of Environmental Economics and Management 27(3): 259–274. http://dx.doi.org/10.1006/jeem.1994.1038.

Hoel, M. (1996). “Should a carbon tax be differentiated across sectors?” Journal of Public Economics 59(1): 17–32. http:// dx.doi.org/10.1016/0047-2727(94)01490-6.

Hoel, M. (2013). “Supply side climate policy and the green Paradox.” CREE Working Paper 2/2013. Hotelling, H. (1931). “The economics of exhaustible resources.” The Journal of Political Economy 39: 137–175. http://

dx.doi.org/10.1086/254195. International Energy Agency (IEA) (2007). “World Energy Outlook 2006.” International Energy Agency, Paris. Krichene, N. (2002). “World crude oil and natural gas: a demand and supply model.” Energy Economics 24: 557–576.

http://dx.doi.org/10.1016/S0140-9883(02)00061-0. Markusen, J. R., E. R. Morey and N. D. Olewiler (1993). “Environmental Policy when Market Structure and Plant Locations

Are Endogenous.” Journal of Environmental Economics and Management 24(1): 69–86. http://dx.doi.org/10.1006/ jeem.1993.1005.

Markusen, J. R., E. R. Morey and N. Olewiler (1995). “Competition in regional environmental policies when plant locations are endogenous.” Journal of Public Economics 56(1): 55–77. http://dx.doi.org/10.1016/0047-2727(94)01419-O.

Ministry of Finance (2012). “Skatter, avgifter og toll 2013.” http://www.regjeringen.no/nb/dep/fin/ dok/ regpubl/ prop/2012- 2013/prop-1-ls-20122013/11/9/2.html?id = 702766.

Ministry of Petroleum and Energy (2011). “An industry for the future – Norway´s petroleum activities.” Meld. St. 28 (2010– 2011) Report to the Storting (white paper). http://www.regjeringen.no/upload/OED/Petroleumsmeldingen_2011/Overset- telse/2011-06_White-paper-on-petro-activities.pdf.

Ministry of Petroleum and Energy (2013). “Facts 2013 (Faktaheftet 2013).” The Norwegian Petroleum Sector. http://npd.no/ Global/Engelsk/3-Publications/Facts/Facts2013/FACTS_2013.pdf.

Mohn, K. and P. Osmundsen (2008). “Exploration economics in a regulated petroleum province: the case of the Norwegian Continental Shelf.” Energy Economics 30: 303–320. http://dx.doi.org/10.1016/j.eneco.2006.10.011.

OGP (2012). “Environmental performance indicators. 2011 data.” Report No. 2011e, October 2012, International Associ- ation of Oil & Gas Producers. http://www.ogp.org.uk/publications/environment-committee/environmental-performance- indicators/environmental-performance-indicators-2011-data/.

Osmundsen, P., K.E. Rosendahl and T. Skjerpen (2015). ”Understanding rig rate formation in the Gulf of Mexico.” Forth- coming in Energy Economics. DOI: 10.1016/j.eneco.2015.03.001. http://dx.doi.org/10.1016/j.eneco.2015.03.001.

Österle, I. (2012). “Fossil Fuel Extraction and Climate Policy: A Review of the Green Paradox with Endogenous Resource Exploration.” FEEM Working Paper Series 13.2012. http://dx.doi.org/10.2139/ssrn.2030090.

Ramcharran, H. (2002). “Oil production responses to price changes: an empirical application of the competitive model to OPEC and non-OPEC countries.” Energy Economics 24: 97–106. http://dx.doi.org/10.1016/S0140-9883(01)00091-3.

Rauscher, M. (1997). International Trade, Factor Movements, and the Environment. Oxford: Oxford University Press. Ringlund, G. B., K. E. Rosendahl and T. Skjerpen (2008). “Does oilrig activity react to oil price changes? An empirical

investigation.” Energy Economics 30: 371–96. http://dx.doi.org/10.1016/j.eneco.2007.06.002. Rystad (2013). “Petroleum production under the two degree scenario (2DS).” Report for The Ministry of the Environ-

ment.http://www.regjeringen.no/pages/38425303/2013_Rys- tad_Energy_Climate_report_NorwegianMinistry_of_the_environment.pdf.

Science Nordic (2013). “Economists call Norway’s climate policy hypocritical.” http://sciencenordic.com/economists-call- norways-climate-policy-hypocritical

Sinn, H.-W. (2008). “Public policies against global warming: a supply side approach.” International Tax and Public Finance 15(4): 360–394. http://dx.doi.org/10.1007/s10797-008-9082-z.

Smith. J. (2005). “Inscrutable OPEC? Behavioral tests of the cartel hypothesis.” The Energy Journal 26(1): 51–82. http:// dx.doi.org/10.5547/ISSN0195-6574-EJ-Vol26-No1-3.

The Economist (2009). “Norway and the environment - Binge and purge.” http://www.economist.com/node/12970769. The Norwegian Oil Company (2012). “Konsekvensutredning for Ivar Aasen.” (Impact study for Ivar Aasen (in Norwegian). Tol, R.S.J. (2014). “Ambiguity Reduction by Objective Model Selection, with an Application to the Costs of the EU 2030

Climate Targets.” Energies 7: 6886–6896. http://dx.doi.org/10.3390/en7116886. Venables, A. J. (2011). “Depletion and development: Natural resource supply with endogenous field opening.” CESifo

working paper: Resource and Environment Economics, No. 3608. http://hdl.handle.net/10419/52473. World Bank (2008). “Climate Change and the World Bank Group: Phase I: An Evaluation of World Bank Win-Win Energy

Policy Reforms.” World Bank: Washington, DC.

Climate Policies in a Fossil Fuel Producing Country / 99

APPENDIX A: DERIVATION OF EQUATIONS (15)

APPENDIX B: MARGINAL PRODUCTION COSTS OF OIL

Figure B1: Marginal costs and corresponding production

Source: Statistics Norway

The marginal cost curve is based on production and cost data of nine Norwegian oil fields 2009–2011. The fields are Glitne, Gungne, Gyda, Jotun, Norne, Sygna, Ula, Varg and Veslefrikk.

100 / The Energy Journal

The production volume for these fields is 5.3 per cent of total Norwegian liquids production in 2012. Hence, we have 27 observations of costs and production. We arrange our observations ac- cording to cost level and estimate by means of OLS a marginal production cost curve representing an average year within the time span. For more details, see Fæhn et al. (2013).

APPENDIX C: SPECIFICATION OF THE FOSSIL FUEL MARKET MODEL

Foreign demand Xj for fossil fuel j is given by:

d d djo jg jcX = D (P ) (P ) (P )j j0 o g c

where Pi is the price of fuel i, dji is the direct price elasticity if i = j and cross price elasticity if i ≠ j, and Dj0 is a constant. Foreign competitive supply Yj of fossil fuel j is given by:

γjY = S (P )j j0 j

where γj is the supply elasticity, and Sj0 is a constant. Market balance of fossil fuel j is given by:

Y + Z + y = X + xj j j j j

where yj and xj denote exogenous domestic production and consumption of fuel j, respectively, and Zj denotes OPEC’s oil production (i.e., Zj = 0 for gas and coal).

When OPEC acts as a dominant producer, it maximizes profits (c denotes OPEC’s unit costs):

Max P = (P – c)Zo o

given the demand, Non-OPEC supply and market balance equations above. Global emissions EC from consumption are fuel specific:

CE = je (X + x )∑C j j j

where is the fuel-specific emission intensity. Global emissions EP from production are fuel and Cej

region specific:

PX Px PZE = j(e X + e x + e Z )∑P j j j j j j

where , and are the fuel-specific emission intensities of foreign competitive supply,PX Px PZe e ej j j domestic supply and OPEC production (of oil). Global emissions E are then:

E = E + EC P

Climate Policies in a Fossil Fuel Producing Country / 101

18. The database contains studies dating from the 1970’s up until today. The standard deviation for the long-run gasoline elasticity is 6.37! See http://dahl.mines.edu/courses/dahl/dedd/.

19. Dahl refers to them as long-run elasticities, but notes that dynamic models, estimating both short- and long-term elasticities, tend to find long-term elasticities 50–100 per cent above the elasticities found in static studies.

20. It is reasonable to assume that price elasticities for crude oil are lower than for oil products, as oil products are higher priced than crude oil (Fournier et al., 2013). At least this is the case if the markup, i.e., the difference between the product and the crude oil price, is independent of the crude oil price itself. Own estimations suggest that the markup and the crude oil price is somewhat correlated, but a 1 per cent increase in the crude oil price will in general increase the product price by less than 1 per cent.

21. The model has been used in e.g. Böhringer et al. (2010, 2012a,b). The GTAP database is available at www.gtap.org. 22. Obviously, from the same simulations we can derive the implicit direct price elasticity for oil, which turns out to be

– 0.45 for crude oil, i.e., quite close to the estimate of the benchmark case.

APPENDIX D: PRICE RESPONSIVENESS

Price Responsiveness on the Demand Side

There is a large empirical literature on direct price elasticities. However, the estimation results vary quite substantially. Using a meta-analysis of 43 primary studies of gasoline demand from different countries, Brons et al. (2008) find a mean long-run price elasticity of – 0.84. How- ever, all the primary studies were published before the year 2000. Carol Dahl has developed a large database with inter alia 247 studies of gasoline and diesel demand studies from around the world. According to her summary statistics, the median long-run price elasticities are – 0.55 and – 0.33 for gasoline and diesel, respectively.18 In Dahl (2012) she presents an analysis based on the static studies in her database, reporting median elasticities of – 0.34 and – 0.16 for gasoline and diesel, respectively. These may be interpreted as intermediate elasticities, i.e., between short- and long-run elasticities.19 Ellis (2010) reviews empirical literature on price elasticities, and refers e.g., to studies by the World Bank (2008) and the IEA (2007). Whereas the World Bank estimates long-run price elasticities for gasoline and diesel at – 0.61 and – 0.67, respectively, the IEA estimates the long- run price elasticity for crude oil demand at – 0.15. Fournier et al. (2013) estimate the average medium- to long-run price elasticity in OECD and BRIICS (Brazil, Russia, India, Indonesia, China and South Africa) countries to be around – 0.2.20 Askari and Krichene (2010) find very low elas- ticities: Their estimates of long-run demand and supply elasticities are both around 0.01 in absolute value.

Most empirical studies of oil demand focus on gasoline and diesel demand. As stated by Ellis (2010), demand tends to be less elastic in the transport sector than in other sectors due to fewer viable alternatives. This is confirmed by an unpublished survey by Dahl (2006), based on the database referred to above, reporting a mean long-run elasticity for fuel oil at – 0.9.

As is evident, a consensus estimate of the long-run price elasticity of oil demand is difficult to nail down. As mentioned in Section 3.2.1, we use – 0.5 as our benchmark case estimate. However, we use other estimates, as well in the sensitivity simulations.

Whereas estimates of direct price elasticities vary quite a lot, estimates of cross-price elasticities are rarely reported (none of the studies mentioned above do so). Instead, we will rely on simulations on a large-scale CGE model building on the extensively used GTAP database and using benchmark GTAP parameters for crucial elasticities in production and consumption of goods and services.21 By simulating an exogenous increase in the crude oil price, we find that global consumption of coal and gas (measured in CO2) increases by respectively 0.10 and 0.09 units for every unit reduction in oil consumption.22 This corresponds to cross-price elasticities of around 0.08 for both fuels, which we use as our benchmark case estimates.

102 / The Energy Journal

Price responsiveness of Non-OPEC supply

As opposed to oil demand price elasticities, there exist rather few empirical studies of oil supply price elasticities. This is also pointed out by Fournier et al. (2013), who set the price elasticity of supply equal to the (absolute value of the) estimated demand elasticity ( – 0.2) in their simulations. Above we referred to a study by Askari and Krichene (2010), who estimates long-run demand and supply elasticities around 0.01 in absolute value. In an earlier study, Krichene (2002) reports a long-run supply elasticity of 0.1 for the period 1973–1999. Importantly, however, all these studies consider world supply of oil, not Non-OPEC supply. Ramcharran (2002) finds an average price elasticity of 0.11 for Non-OPEC over the period 1973–1997. In a study of OPEC behaviour, Alhajji and Huettner (2000) find support for a model where Saudi Arabia acts as a dominant producer – with this specification oil supply price elasticity from the rest of the world (Non-OPEC + OPEC minus Saudi Arabia) is found to be 0.20. In a similar study, Hansen and Lindholt (2008) find a long-run supply elasticity of 0.38 for the period 1974–2001.

Empirical studies that focus on oil drilling tend to find higher price elasticities. For instance, Ringlund et al. (2008) find an average long-run elasticity of 0.99 for oilrig activity in Non-OPEC, with elasticities ranging between 0.51 and 1.86 in different regions. Dahl and Duggan (1998) find elasticities for oil exploration in the U.S. above one, whereas Mohn and Osmundsen (2008) find a long-run elasticity of 0.41 for exploration drilling in the Norwegian Continental Shelf. Farzin (2001) finds even lower elasticities for reserve additions of known fields in the U.S. (0.16 in the long run).

Again, it is difficult to pin down the exact price elasticity of Non-OPEC supply. As a benchmark case estimate, we will use 0.5, i.e., the same absolute value as for the demand price.