Social Choice, Uncertainty About External Costs and Trade-Off Between Intergenerational Environmental Impacts.

Ecological Economics 57 (2006) 282 305 www.elsevier.
com/locate/ecolecon
ANALYSIS
Social choice, uncertainty about external costs and trade-off between intergenerational environmental impacts: The emblematic case of gas-based energy supply decentralization
Francesco Gull * `
Iefe-Institute for Energy and Environmental Economics and Policy, Universita Bocconi, Viale Filippetti 9, 20122 Milano, Italy ` Received 9 January 2004; received in revised form 4 April 2005; accepted 13 April 2005 Available online 24 June 2005
Abstract The performance of the small natural gas-fired power technologies has improved remarkably over the last decade. This has aroused the interest of operators, regulators and legislators in natural gas-fired distributed generation (gas-fired DG), namely, the integrated or stand-alone use of small, modular gas-fired power generation close to the point of consumption as an alternative to large power generation and electricity transport over long distances. Gas-fired DG can provide an important benefit from the environmental point of view. Customer proximity, in fact, greatly increases the potential for combined heat and power generation, involving energy saving and reduced greenhouse gas (GHG) emissions. Unfortunately this kind of decentralized supply also determines higher non-GHG emissions (mainly NOx, compared to the best available central power technology) which occur in urban areas (high populated) instead of extra-urban areas (where large power plants are generally located). It is therefore difficult to make a reliable evaluation of gas-fired DG environmental benefits without comparing centralized and decentralized models in terms of external costs, that is without an analysis which allows us to compare the extent of global and localregional impacts in terms of monetary damage. If, on the one hand, this underlines the (potential) importance of the methods adopted to assess the economic value of environmental externalities (even for policy decisions that are binary, i.e. the choice between different energy technologies), on the other, it raises the crucial question of the uncertainty about the economic estimates. This article aims at demonstrating that the uncertainty about external costs, even if large, does not undermine the possibility of verifying whether gas-fired DG is preferable (or not) to centralized supply. The paper compares centralized and decentralized models in terms of the external environmental costs which are calculated by using the results of the available studies in this field (in particular the results of the dissemination process of the so-called ExternE project, one of the most recent and accurate methodologies, and the results of a meta-analysis, with regard to the marginal cost of GHG emissions). The uncertainty about external costs is substantial but not so large that it is not possible to say anything about the environmental raking of alternative technology solutions involving trade-off between the impacts of
* Tel.: +39 258363824; fax: +39 258363890. E-mail address: francesco.gulli@uni-bocconi.it. 0921-8009/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.ecolecon.2005.04.014
F. Gull / Ecological Economics 57 (2006) 282305 `
283
different pollutants (or between different kinds of impacts). The literature on external costs provides several studies accounting for a large part of uncertainty by means of appropriate statistical and sensitive analysis. By using and elaborating these results, the analysis described in this paper seems to support the conclusion that centralized supply, and especially the completely electric solution (based on the reversible electric heat pump), is still preferable to natural gas-fired CHP distributed generation. This is not a definitive conclusion but, we hope, a useful (scientific based) contribution for policy decisions under the state of the art. In fact, this result has an interesting policy implication. It suggests unless questioning the current enthusiasm on natural gasfired CHP distributed generation deployment (e.g. the European Commission is indeed advocating DG as a contribution to GHG emission reduction) and helps us to reflect upon gas-fired DG supporting environmental policies which focus on the reduction of GHG emissions and totally disregard the possible trade-off between the impacts of global and local-regional pollutants. Unless one denies the rationality attributed to making tradeoffs, on the basis of ethical limits of economic valuations. Even in this case, however, cost-benefit analysis seems to be legitimate and a necessary step of the public discourse. We think that the results of this paper are emblematic, from this point of view. D 2005 Elsevier B.V. All rights reserved.
Keywords: Distributed generation; External costs; Uncertainty; Cost-benefit analysis; Environmental policy
1. Introduction The performance of the small natural gas-fired power technologies1 (reciprocating engine and gas turbine) has improved remarkably over the last decade. This has aroused the interest of operators, regulators and legislators in natural gas-fired distributed generation (gas-fired DG), namely, the integrated or stand-alone use of small, modular gas-fired power generation close to the point of consumption as an alternative to large power generation and electricity transport over long distances (centralized supply). Gas-fired DG involves (internal) costs higher (on average) than those of centralized supply but can provide important benefits from the environmental point of view.2 Customer proximity, in fact, has two important advantages. On the one hand, it greatly increases the potential for combined heat and power generation3 (CHP or cogeneration) and, on the other,
1 Ackermann et al. (2001) consider distributed generation an electric power source connected directly to the distribution network or on the customer site of the meter. They suggest the following categories: micro DG (1 Watt b 5 kW); small DG (5 kW b 5 MW); medium DG (5 MW b 50 MW); large DG (50 MW b 300 MW). In this article, we refer to conventional small gas-fired technologies. Therefore we do not take into consideration the fuel cells. 2 As regards DG benefits and the relationship between technological change and market organization, see Arthur D. Little Inc. (1998) and Pfeifenberger et al. (1997). 3 The high costs of transporting heat even over short distances make large-scale cogeneration unattractive.
it avoids electricity transmission losses. Therefore, compared to centralized supply, natural gas-fired CHP distributed generation can provide energy saving and, consequently, can involve reduced greenhouse gas (GHG) emissions. Nevertheless, the realization that gas-fired DG could provide energy saving and lower GHG emissions does not mean that decentralization is undoubtedly preferable to large power generation from the environmental point of view, for two reasons. First, despite the higher overall energy efficiency, gas-fired DG technologies might involve higher nonGHG emissions (compared to the best available central power technology). Second, there are considerable differences between centralized and decentralized technologies in terms of the impact of non-GHG emissions (SOx, NOx, particulate, etc.). These differences might be due to micro-localization effects. Unlike large power plants (high stack and extra-urban location), gas-fired distributed technologies have low stacks and are generally located in densely populated urban areas. Because of low stacks (emissions at extremely low altitudes), pollutant atmospheric dilution could be lower so that the increases in pollutant concentration close to the plant could be higher than those of a large power plant. Due to location, these high increases in pollutant concentration occur in highly populated areas and seriously damage human health. These combined effects might cause an environmen-
284
tal impact (per unit of pollutant emitted) much higher than that of a large power plant. In other words, compared to centralized supply, gas-fired DG energy saving not necessarily implies lower non-GHG emissions, and lower non-GHG emissions not necessarily imply lower localregional environmental impact. In consequence, it is impossible to make a reliable evaluation of gas-fired DG environmental benefits without comparing centralized and decentralized models in terms of external costs, that is without an analysis which allows us to measure the extent of global and localregional effects in terms of monetary damage. If, on the one hand, this underlines the importance of the methods adopted to assess the economic value of environmental externalities, on the other, it raises the crucial question of the uncertainty about the economic estimates of some important impact categories. The uncertainty about external cost estimates is one of the main reasons many authors do not recommend using cost-benefit analysis to support environmental policies.4 These authors argue that the high uncertainty about external costs, especially the economic value of global warming, negatively affects the significance of comparing different technological and organizational solutions. The overall external costs of a particular technology or kind of energy supply might be in fact higher or lower than another one depending, for example, on the value assigned to the marginal cost of the GHG emissions. This value can vary enormously depending on statistical uncertainty (e.g. uncertainty about physicalchemical phenomena, values of some economic parameters, etc.) and politicalethical uncertainty which affects the choice of the discount rate, whose value is crucially important when we deal with far-distant future (such as in the case of global warming effect). Nevertheless, when policy decisions are binary (e.g. the choice between two different power technologies), uncertainty is not generally considered very important (Rabl and Spadaro, 1999; Krewitt, 2002). In most comparisons between power plants the situation is, in fact, quite simple because there is a positive correlation between GHG and non-GHG
4 Among these authors we must mention Stirling (1997) who proposes some interesting arguments for supporting his critics.
emissions (and between the different non-GHG emissions). For instance, coal and oil plants emit much more CO2 than gas-fired plants, but they also emit much more SOx, NOx and particulate. In consequence, whatever the uncertainty about external costs gas-fired plants performs better than other fossil fuel plants, from environmental (and perhaps social) point of view. Rather, we could argue that the environmental ranking of these technologies could be easily deduced without estimating external costs.5 Instead, for the reasons illustrated above, this simple reasoning can not be certainly applied to the comparison between central power supply and gasfired distributed generation (and, in general, fossil fuel fired DG), even if we consider the sole environmental performance.6 The analysis of the gas-fired distributed generation, therefore, looks very interesting from both methodological and policy points of view. On the methodological side, it underlines the (potential) importance (and usefulness) of the cost-benefit analysis even for policy decisions that are binary. On the policy side, it might provide useful indications for legitimating or questioning the current enthusiasm on gas-fired DG development. Concerning policy implications, however, it is also necessary to highlight that a part of the economic literature questions the validity of the cost-benefit analysis, on the basis of the ethical limits of methods to assess external costs. The theoretical support for externality valuation exercises is drawn from the neoclassical welfare economics literature. Within this strand of research, there are a number of valuation methods in use, but ultimately they all aim at discovering peoples preferences expressed as willingness to
The comparison between fossil fuel plants and renewable technologies (wind, solar, mini-hydro) leads to similar conclusions. However, the problem of uncertainty might arise when we compare the different technologies in terms of overall social costs (internal plus external costs), for example, the internal (market) costs of the renewable plants are generally higher than those of the fossil fuel plants. In consequence, the overall performance of these technologies might depend on the value assigned to the external costs. 6 The framework becomes even more complex when we take into account internal (market) costs. The internal cost of large power supply is lower (on average) than that of distributed generation (Gull, 2003). In consequence, the bsocial competitivenessQ of DG ` depends on the environmental performance. This emphasizes the importance of the uncertainty about external costs.
5
285
pay (WTP) for environmental goods and services. Soderholm and Sundqvist (2003) argue that bthe behavioral and ethical foundations of environmental valuation, as applied to the valuation of external effects, are likely to be too restrictive for serving as the sole basis for social choice. [. . . . .] Also, since various power sources give rise to different types of externalitiessome likely to be less amenable to social cost pricing than othersthe choice between different technologies becomes more complex than is implied by the welfare economics literatureQ. Moreover, Spash (1997) asserts that bevidence is found to support the view that environmentalists choose to operate on a right-based approach which rejects the relative welfare arguments of economics and positively attributes compensation to future generation for environmental damages. [. . . . .] In terms of policy, environmental management on the basis of totaling economic values is liable to be undemocratic because of the systematic exclusion of a section of the general publicQ. In this paper, we do not deal with the problem of the ethical limits of external cost methods. We report this argument in order to highlight that the results of the analysis presented here could be legitimate and necessary but likely not sufficient step of the public choice process. This article aims at demonstrating that the uncertainty about external cost estimates (even if large) does not undermine the possibility of verifying whether gas-fired DG is preferable (or not) to centralized supply, from environmental point of view. The results showed in this paper are based upon a detailed technical analysis of energy flows of the fuel cycles (centralized and decentralized systems). Environmental externalities are assessed by using the results of the dissemination process of the ExternE methodology,7 one of the most ambitious and internationally recognized attempts at coming up with btrueQ external cost estimates for the different power technologies (Krewitt, 2002). We are aware such a methodology could be largely imperfect. Nevertheless
The ExternE project is a major research program launched by the European Commission at the beginning of the 1990s to provide a scientific basis for the quantification of energy related externalities and to give guidance supporting the design of internalization measures (Krewitt, 2002). In this paper we use and elaborate original ExternE estimates (ExternE-1998) and more recent ExternE estimates (NewExternE). http://www.externe.info/.
7
we think that it could provide useful and reliable indications when used to compare technological alternatives and when the uncertainty about value estimations can be internalized into the estimating model. Moreover, since using a specific methodology does not allow us to account for uncertainty about the choice of the model (particularly important in the case of the estimate of the global warming effects), we also use the results of a meta-analysis based on the literature about the marginal damage cost of carbon dioxide emissions. Finally, the analysis focuses on a simulation of a particular territorial context, the case of Italy. However, as we will explain in Section 2, this simulation is particularly significant, so that the results obtained can be generalized. The article is organized in the following way. The second section is divided into two parts. The first part describes the overall approach to the subject and explains the reasons for choosing the technological solutions, the applications and plant locations. The second compares centralized and decentralized supply in terms of primary energy consumption and atmospheric pollutant emissions. The third section describes the methodology adopted to calculate the external costs and benefits and the results obtained. The final results are presented in the form of cumulative probability distributions in order to verify the impact of the uncertainty about external costs. The last section sums up the main results obtained.
2. Assumptions and technical assessment 2.1. Assumptions The aim of this article is to determine the environmental benefit of a new model based on a wide decentralization of the energy supply. Therefore we are interested in analyzing applications, technologies and locations which could be representative of a large potential DG deployment. In order to achieve this it is necessary to formulate the following assumptions and make the following choices. First, we should not analyze specific applications or technologies (i.e. renewable energy sources) which are not representative of large potential DG deployment. We therefore have to deal with fossil fuel plants
286
Fig. 1. Regional environmental impact.
(especially natural gas-fired technologies,8 gas-fired DG) and applications for residential uses (i.e. the largest customer category9). Second, gas-fired DG external costs greatly depend on the geographic position of the plant (which greatly influences the regional impact). This obviously makes impossible to provide a single value of gas-fired DG environmental benefit. Therefore a significant range of variability must be proposed. This means that our estimating model should be as flexible as possible in order to take into account the variability of the geographic location. Since this variability can be very high it would be impossible (and superfluous) to examine all the possible value combinations. Our analysis will therefore focus on the best and worst cases. The effect of the geographic location of the plants (which considerably affects the environmental impact of the fuel cycles) could be considered by finding a
Since DG plants are generally located in urban areas, natural gas is the most preferred fuel because of its low environmental impact compared to other fossil fuels. 9 This kind of application represents a conservative case involving a low ratio of heat and electricity consumption. Industrial customers are more suitable to CHP (higher ratio of heat and electricity consumption). Nevertheless, we do not take into consideration applications that involve power plant size higher than 10 MW (such as the typical industrial applications) since this power size is considered centralized (generation that does not bypass the electricity transport grid).
8
geographic context (i.e. a country) where there are at least two locations which represent extreme environmental situations. In this respect, Italy, the country chosen for this analysis, provides a very useful case study for the following reasons (Fig. 1). First, its geographic configuration, a long latitudinal extension from the centre of Europe to the centre of the Mediterranean Sea, makes it easy to identify two locations with the required characteristics. In fact, if we applied our model in the north (for example, in Milan) and in the south (for example, in Palermo) we would find two opposite situations in terms of regional environmental damage. The case of Milan represents a very high environmental impact since it is located near a region (the centre of Europe) which is densely populated and relatively far from the sea. On the contrary, the case of Palermo represents a very low regional environmental impact since the city is located in a region far from the centre of Europe and surrounded by the sea (since plant pollution is mainly discharged directly into the sea, the potential environmental impact is very low10).
10 Moreover, Milan and Palermo also differ in terms of climate conditions (cold in the first case and mild in the second) which influence CHP costs (the lower the heat needs for domestic heating and/or air conditioning, the higher the CHP costs). We can therefore analyze this effect without introducing an explicit variable in the estimating model.
F. Gull / Ecological Economics 57 (2006) 282305 ` Table 1 Applications: capacity needs and energy consumption
287
2.2. Technical assessment Before assessing gas-fired DG external benefits, it is necessary to compare centralized and decentralized systems in terms of energy efficiency and atmospheric pollutant emissions11 (GHG and non-GHG). We consider six technological solutions: two based on centralized supply (A and B) and four on decentralized supply (C1, C2, D1 and D2). Energy consumption include all customer needs: heat for domestic heating and sanitary uses, cold for air conditioning, and structural electricity needs (lighting, power for domestic appliances, etc.). Heat for space heating and cold for air conditioning depend on plant location (Milan or Palermo) while the structural electricity needs are assumed to be identical (Table 1). The simplified technical schemes of the different solutions are reported in Fig. 2. Centralized systems include two solutions, conventional (A) and advanced (B). Solution A includes a conventional condensing boiler providing heat for space heating and sanitary uses (hot water). A conventional compressing refrigerator supplies cold for air conditioning. Imported electricity (from the utility grid) is assumed to be generated by a combined cycle-gas turbine plant (CCGT) with 56% electrical efficiency, which is the power generating marginal technology.12 This choice is consistent with the need to compare centralized and decentralized models in terms of environmental external marginal costs. Energy losses due to electricity transport and distribution are assumed equal to 6%. Furthermore, we assume that the large power plant uses the selective catalytic reduction (SCR), as method of removing NOx from the exhaust stream. SCR is highly effective and it is typically used in large industrial and electric generating facilities. It makes use of toxic chemicals (e.g. urea, ammonia) and produces solid waste, two features which (aside from its high cost for small-scale
Residential building (Volume: 57,600 cm) Milan Structural electricity need Maximum capacity (kW) Annual consumption (MWh/y) Air conditioning needs Cool maximum capacity (kW) Annual consumption (summer) (MWh/y) Space heating needs Maximum capacity (kW) Annual consumption (winter) (MWh/y) Hot water for sanitary uses Capacity (assumed constant) (kW) Annual consumption (MWh/y) Source: Bruzzi (2002). 270 788 Palermo 270 788
1300 2316
1420 2575
1006 1800
603 561
55 482
55 482
applications) render it impractical for many DG applications. Solution B is based on a reversible heat pump which provides both heat for space heating (in winter) and cold for air conditioning (in summer). Electricity is imported from the utility grid at the same conditions of the previous system. Decentralized systems include four solutions, two are bopenQ (C1 and C2) and two are bisolatedQ (D1 and D2). C1 and C2 are bopenQ solutions. Power plants are sized in order to satisfy the maximum heat demand13 so that they generate power in excess of customer needs (see the Appendix). This excess power is
Generally, CHP plants operate with a peak load boiler to satisfy peak demand. Nevertheless this technical configuration is prevalently used in the case of large CHP steam turbine cycle plants. In the cases of small plants (gas engine and turbine gas), it is preferable to size the plant in order to satisfy the maximum heat demand. In fact, it is important to take into account two problems. First, there is a problem of space availability (residential and service applications). Second, in this analysis, CHP plants provide not only heat for space heating but also cold for space conditioning (generated by an absorbing refrigerator). In this case, generally, plants are sized in order to satisfy the maximum heat needs for the absorbing refrigerator. Furthermore, it is necessary taking into account that the absorbing refrigerator needs large space.
13
11 The impact of atmospheric pollutants represents the relevant part of the total damage. 12 Indeed, gas-fired DG must be compared to the marginal mix of central power. This implies that in our calculations we also consider other marginal power technologies providing energy during the peak hours (large gas turbine). Nevertheless, the weight of these technologies (in terms of energy production) is very low compared to that of the CCGT.
288
Convetional centralized (A)
Inported electricity Compressing refrigerator Natural gas Condensing boiler

Advanced centralized (B)
Strctural electricity need cold for space conditioning (summer) heat for space heating (winter), hot water, etc. Inported electricity Strctural electricity need cold for space conditioning (summer) heat for space heating (winter)
Heat pump
Natural gas
Boiler
Hot water
"Open" DG (C1 and C2)

Exported electricity Power Natural gas "open" DG plant Heat Strctural electricity need Absorbing refrigerator cold for space conditioning (summer) Natural gas
"Isolated" DG (D1 and D2)
Structural electricity need Power "isolated" DG plant Heat Hot water Heat pump cold for space conditioning (summer)
heat for space heating (winter) Hot water
Absorbing refrigerator
heat for space heating (winter)
Fig. 2. Centralized and decentralized systems: simplified technical schemes.
exported to the utility grid and accounted for in terms of avoided fuel consumption of the large power generation (56% electrical efficiency minus the percentage of power transport losses14). Therefore, net primary energy is equal to gas-fired DG fuel consumption minus the avoided fuel consumption. Solution C1 produces combined heat and power by using a gas engine technology. Cold for air conditioning is generated by means of an absorbing refrigerator making use of the bcogeneratedQ heat. Solution C2 has the same configuration as C1 but combined heat and power is produced by a gas turbine instead of a gas engine. The D1 and D2 systems are bisolatedQ since they do not involve importing/exporting electricity from/to the utility grid. Power plants are sized in order to satisfy the maximum customer electricity needs. This implies that the amount of bcogeneratedQ heat is not sufficient to satisfy energy needs for space heating and air conditioning (see the Appendix). We assume that the remaining heat (in winter) and cold (in summer) are supplied by a reversible heat pump using
bcogeneratedQ power. Solution D1 uses a gas engine and solution D2 a gas turbine. Tables 2 and 3 show the pollutant emission rates of the different technologies per unit of electric energy generated and per unit of fuel used, respectively.15 By following the method illustrated in the Appendix (Tables A1 and A2) it is now possible to obtain the primary energy consumption of the different energy supply solutions. Table 4 reports the final results and shows that the primary energy consumption of the decentralized solutions is always lower than that of the conventional centralized solutions (except for the bopenQ gas turbine in Palermo) and depends considerably on technologies and locations. Energy saving is higher in Milan where the climate conditions are more favorable to combined heat and power generation (high ratio of heat and electricity needs) and the reciprocating engine always performs better than the gas turbine. In the best case (bisolatedQ gas
15 Note that there are other GHGs such as CH4 and N2O. The specific contribution of these gases to global warming is higher than that of CO2 but the amount emitted is much lower (of one order of magnitude). In order to take even these gases emitted by fuel cycles into account, the equivalent CO2 greenhouse potential is normally used. For a criticism of this procedure, see Schamaleense (1993).
In the case of bopenQ solutions, only the transport losses (2%) must be accounted for, because this kind of supply does not bypass the electricity distribution network.
14
F. Gull / Ecological Economics 57 (2006) 282305 ` Table 2 Power technologies: electrical efficiency and emissions rate Plant size (kWe) Efficiency % Emission rate (CO2eq) (g/kWh) Uncontrolled gas-fired lean burn gas engine Uncontrolled small gas turbine CCGT (SCR)(2)
(1) (1)
289
(SOx) (g/MWh) 2.5 2.5 1.6
(NOx) (g/MWh) 897.8 469.3 24.9
(PM10) (g/MWh) 12.3 32.4 16.6
700 700 400,000

(2)
40 30 56
452.8 611.0 320.8
Per unit of electricity generated; SCR: selective catalytic reduction. Sources: Regulatory Assistance Project (2001).
engine in Milan), energy saving can reach almost 30%. Table 5, where the amounts of pollutant emitted are reported, allows us to identify the core of this analysis. As we can see, on the one hand, distributed generation reduces CO2 emissions (except for the bopenQ gas turbine in Palermo), compared to the conventional centralized supply, but, on the other hand, it greatly increases NOx emissions (whereas, in the case of the other local and regional pollutants, the differences are low, except for the particulate in the case of bopenQ gas engine16). Thus, on the basis of the technical assessment, we can not answer the crucial question of this paper: what is the real environmental benefit of gas-fired DG? In order to answer this question we need to compare the different kinds of impact. This means that we need to calculate the monetary impact of the different pollutant emissions.
the natural ecosystem, materials, monuments, etc. Such damage is referred to as external costs since it is not reflected in the market price of energy. The damage is due to various agents including atmospheric and non-atmospheric pollutants, accidents and occupational diseases, noise, etc. Our analysis considers only the effects of the principal atmospheric pollutants (which represent the significant part of the total damage). External costs can be permanent since they can appear many years after the plant lifetime and/or persist over time. For some categories of impacts this behavior is negligible (i.e. local and regional pollutants) while for others (i.e. global warming) it is rather considerable so that the time damage horizon can be much higher than the plant lifetime. These introductory remarks are meant to explain why we will distinguish localregional pollutants (essentially SOx, NOx and particulate) and global warming (due to GHG emissions).
3. The economic valuation 3.1. Methodology Energy production and consumption cause damage to a wide range of receptors, including human health,
16 The bopenQ solutions export power to the grid. They generate power exceeding customer needs. This exported power is accounted for in terms of avoided primary energy consumption of the central power (and avoided pollutant emissions). Since particle emission of the gas engine DG technology is lower than that of the marginal central power (see Table 3), this implies that the overall particle emission is negative (particle emission of DG minus particle emission of avoided central power). Obviously this effect is not important in the case of bisolatedQ DG because this does not imply excess power exported to the electricity grid.
Table 3 Emission rate per unit of primary energy Emissions(1) CO2eq (g/kWh) Gas-fired boiler Gas-fired engine Gas turbine Large power plant (combined cycle) 188.7 181.1 183.3 179.7 SOx 1.7 1.0 0.8 0.9 PM10 (g/MWh) 0.0 4.9 9.8 9.2 NOx 32.9 359.1 140.7 13.9
(1) Excluding emissions from the upstream stages (extraction, transportation, etc.) of the natural gas fuel cycle. Sources: elaboration on data reported in Table 2. For the case of the gas boiler, European Commission (1999d).
290
Table 4 Primary energy consumption and energy saving North (Milan) Natural gas consumption (MWh/y) Centralized Solutions Conventional (A) Electrical heat Pump (B) B/A Decentralized solutions bOpenQ solutions Gas engine (C1) C1/A Gas turbine (C2) C2/A bIsolatedQ solutions Gas engine (D1) D1/A Gas turbine (D2) D2/A 2535 438 Net electricity imported from the grid (MWh/y) 1414 1959 Primary energy (MWh/y) 5363 4356 0.81 South (Palermo) Natural gas consumption (MWh/y) 1158 438 Net electricity imported from the grid (MWh/y) 1545 1651 Primary energy (MWh/y) 4248 3740 0.88
9710 9416
3096 2316
3977 0.74 5127 0.96 3902 0.73 4413 0.82
7808 7849
2336 1800
3482 0.82 4515 1.06 3384 0.80 3891 0.91
3902 4413
0 0
3384 3891
0 0
Source: Appendix (Tables A1 and A2).
The total external cost EXTC can be expressed as (for centralized and bisolatedQ and bopenQ decentralized supplies, respectively) EXTCcentr X
z
EXTCopen DG
X
z
aDG QzDG z acentr Qzcentrav z

P P
X
z
1c
acentr Qzcentr z aDG QzDG z

P
1a 1b
EXTCisolated DG
X
z
where z is a generic pollutant and Qz is the discounted amount P z pollutant emitted during the of P plant operating. QzDG and Qzcentr are the discounted amount of pollutant emitted of decentralized and
Table 5 Total emissions of atmospheric pollutants(1) North Italy (Milan) CO2 (t/y) Centralized solutions Conventional (A) Electrical heat pump (B) Decentralized solutions bOpenQ solutions Gas engine (C1) Gas turbine (C2) bIsolatedQ solutions Gas engine (D1) Gas turbine (D2)
(1)
South Italy (Palermo) PM10 (kg/y) 26.0 36.0 NOx (kg/y) 122.7 68.9 CO2 (t/y) 773.7 676.0 SOx (kg/y) 4.7 3.7 PM10 (kg/y) 28.4 30.4 NOx (kg/y) 81.0 60.3
SOx (kg/y) 6.9 4.7
986.6 786.7
728.2 955.2 706.7 808.9
4.6 3.7 3.9 3.5
5.2 52.8 19.1 43.2
3407.2 1265.2 1401.2 620.9
636.7 839.7 612.9 713.2
3.9 3.3 3.4 3.1
1.5 46.3 16.6 38.1
2743.7 1058.0 1215.2 547.5
Excluding emissions from the upstream stages (extraction, transportation, etc.) of the natural gas fuel cycle.
291
centralized systems, respectively. Qzcentrav is the discounted amount of avoided pollutant emissions of the large power generation (CCGT with 56% electrical efficiency17). Finally a z is the unit external cost (per unit of pollutant emitted). Given Eqs. (1a) (1b) (1c) and according to the methodological approach, gas-fired DG benefits (EXTB) can be expressed as EXTBDG 1 EXTCDG EXTCcentr 2
In order to calculate the a z coefficients, we use the ExternE methodology briefly described below. ExternE18 follows the bottom-up approach and is based on a step-by-step procedure. In fact, it requires dealing with cascade phases: 1) determining the emissions for each stage of the fuel cycle (from the production of primary input to the output production); 2) simulating the dispersion of the pollutants both on a local and regional scale; 3) identifying all the receptors; 4) calculating the impact (by applying the bdose responseQ functions); 5) where possible, making an economic evaluation of such an impact. The pollutants taken into consideration are solid, liquid and gaseous residues. The main impact is due to the emissions of CO2, SOx, NOx and particulate (PM10). The damage taken into consideration includes the effects on public health, agriculture, forests (acid rain), the ecosystem in general, materials (deterioration of buildings and monuments) and the damage related to global warming due to GHG emissions. Doseresponse functions provide the marginal damage caused by increment of concentration due to plant emissions. They can be linear, non-linear and with a threshold. bWillingness to payQ or bto acceptQ is the
standard measure of the value in environmental economics adopted throughout the ExternE project. The analysis conducted in the ExternE project shows that the uncertainties about the external cost assessment are significant, especially as regards global warming estimations. Part of this uncertainty depends on socialpolitical choices which mainly reflect on the discount rate choice (another important ethicalpolitical choice concerns the value of statistical life), while the remaining part is statistical (scientific nature). Statistical uncertainty regards data uncertainty (e.g. slope of a doseresponse function, cost of a day of restricted activity, deposition velocity of a pollutant, etc.) and model uncertainty (assumptions about causal links between a pollutant and a health impact, assumptions about form of a dose response function, choice of models for atmospheric dispersion and chemistry, models of simulating the temperature increase and its profile over time due to GHG emissions, time horizon, etc.). In order to isolate the problem of political uncertainty (the role of discounting in assessing external costs), which mainly refers to global warming estimations, our analysis deals separately with the impacts of localregional pollutants (LR), mostly due to SOx, NOx and PM10 emissions, and the global effects (GW), due to GHG emissions. 3.1.1. Local and regional pollutants (LR) Political uncertainty (the choice of discount rates), which is crucial in the case of the global warming, is less important in the case of the impact of local regional pollutants. Most localregional impacts of air pollution are, in fact, fairly immediate and discounting is not significant.19 In consequence, the choice of discount rate has a low (negligible) impact on a z . In order to evaluate the external costs coefficients we have used the results of the ExternE dissemination process (the application of the same methodology to different plants and locations across Europe). These results show that there are two effects which determine
The main exceptions are chronic health impacts and cancers, which are important in the so-called chronic mortality from particulate. For the sake of simplicity, in our analysis, we disregard this effect because of the very low emissions of particulate. See European Commission (1999e).
19
Minus the energy losses due to electricity transport and distribution (6% for bisolatedQ systems and 2% for bopenQ systems). 18 ExternE is one of the most recent and reliable methodologies in this field. It was developed by a group of leading European and US research centres with the financial support of the EC (DG XII). Valette (1995) briefly describes the methodological approach and reports the results when it was first adopted. Krewitt (2002) provides a more recent analysis of the results of the ExternE project. He gives examples on how external costs in spite of remaining uncertainties are successfully used to support environmental policy. In this paper we use and elaborate the original ExternE estimates (ExternE-1998) and more recent ExternE estimates (newExternE). See also http://www.externe.info/.
17
292
significant differences in external cost coefficients (being equal the source of pollution) and are important in the case of natural gas-fired distributed generation: the macro-localization of the plant (in our case, north or south of Italy) and its micro-localization (urban or extra-urban location). Gas-fired DG plants are generally located, in fact, in urban areas while large power plants in extra-urban areas. Regarding macro-localization effects (north of Italy versus south of Italy, described in Fig. 2) we have used the results of the ExternE dissemination process in Italy20 (the application of the same methodology to different plants across the country). These results point out how external costs per unit of pollutant emitted are, in the north of Italy (Milan), higher than the average Italian value (+ 38% for SOx; +59% for PM10 and + 77% for NOx) and, in the south of Italy (Palermo), lower than the average value ( 34% for SOx; 24% for PM10 and 39% for NOx). With regard to micro-localization effect (urban versus extra-urban locations) we have used the results of the external costs of transport,21 which provide a comparison between urban and extra-urban locations. Urban areas involve an environmental impact higher than that of the extra-urban areas. In fact, the high increases in pollutant concentration (close to the source of pollution) occur in highly populated areas and seriously damage human health. With the macrolocalization being equal, these effects cause an environmental impact (per unit of pollutant emitted) of natural gas-fired distributed generation (which is generally located in an urban area) higher than that of a large power plant (which is generally located in an extra-urban area). The influence of the local population density (urban areas) is considerable in the case of PM10 (urban cost is 150% higher than the extra-urban cost) and SOx (urban cost is 37% higher than extraurban cost) but it is much lower (in some cases nonexistent) in the case of NOx.22 In this respect,
20 21
we have to observe, however, that applying the results of transport to power generation might not be a correct approach. Indeed, transport emissions occur at the ground level so that the effect of pollutant atmospheric dilution (dispersion) is very low (much higher pollutant concentrations close to the source of pollution), compared to that of a power plant with a high stack. Consequently, environmental impact of transport (mainly health impact) is higher (being equal the amount of pollutant emitted) and gas-fired DG micro-localization effect might be overestimated. Nevertheless, it is important to bear in mind that the stack of a small power plant in an urban area is much lower (around 20 m) than that of a large power plant (100150 m). Thus, even gas-fired DG emissions occur at a relatively low quota. ExternE simulations of SOx impact23 for different emission quotas (stack height) demonstrate that the difference between the ground level (transport) and the 20 m level (gasfired DG) is almost negligible24 (+ 5 H 10%). Therefore, using the results of the transport to evaluate micro-localization effects of gas-fired DG can be acceptable. The introduced approximation will have only a marginal impact on the overall results, also because of the extremely low emissions of particulate and SOx (see below). Finally, in its original formulation, ExternE (ExternE-1998) assumed nitrate aerosols to be as harmful as primary PM10 particles.25 Indeed, this is an important source of uncertainty. In fact, this is a kind of bprecautionaryQ approach because there was no empirical evidence (at that time) of nitrate aerosols damages on human health. However, more recent epidemiological studies have reduced the uncertainty about the impact of nitrate aerosols (EPA, 2004) and the new version of the ExternE Project provides more reliable results on this topic. The current approach of ExternE (Rabl, 2005a) assumes that nitrate are half as toxic as PM10 and PM and O3 as independent
23 See European Commission (1999a) and Rabl and Spadaro (1999). 24 This difference refers to locations near urban areas. Note, however, that for a source of pollution in an agglomerate (such as Paris) this effect is much greater (see Rabl and Spadaro, 1999). 25 See European Commission (1999a) and Funk and Rabl (1999).
See Frigerio et al. (1995) and European Commission (1999c). See European Commission (1997). 22 Most impact of NOx is due to nitrate aerosols, which form in air (via chemical transformation) far to the source of pollution. Therefore the share of local range of the total damage costs is only small and is almost independent on the local population density.
F. Gull / Ecological Economics 57 (2006) 282305 ` Table 6 External costs per unit of pollutant (ExternE) Sectors
293
pollutant (i.e. their impacts are additive). A group of internationally renowned epidemiologists and toxicologists (USA and EU) were asked to provide comments and recommendations on this assumption (among others). Their basic conclusion was that the approach of ExternE is valid (Rabl, 2005a). Taking into account these improvements of the ExternE approach, we have used the new concentrationresponse function for the human health impact of nitrate.26 By using these data, we have obtained the values illustrated in Table 6. In this table the median of the statistical distribution and the corresponding geometric standard deviation are reported. Gas-fired DG and gas-fired boiler values have been calculated by applying the coefficients previously described for each pollutant to the Italian average value, corresponding to the large power plant. For example, in the case of SOx in Milan, we have multiplied 7.5, the external cost coefficient of the large power plant, by 1.89 = 1.38d 1.37. The following conclusions can be drawn from Table 6. First, the effects of local and regional pollutants largely depend on plant geographic localization. All other conditions being equal, external costs are much lower in Palermo than in Milan. This is due to the micro- and macro-localization effects (thus also confirming our preliminary intuition about the choice of geographic locations). With regard to uncertainty, ExternE proposes a methodology which simplifies the treatment. In fact, the expression for the total damage is shown to be largely multiplicative, even though it involves a sum over receptors at different sites. For example, air pollution effect on health is: Damage Dpollution concentrationd population d dose response function d economic valuation Since the central limit theorem implies that the error distribution for multiplicative processes is likely to be approximately lognormal, one may be able to bypass the need for a detailed and tedious Monte
Local and regional effects (a z ) (Median) (SOx) USD/kg (PM10) USD/kg (NOx)a USD/kg 9.1 3.7 5.5 35
Gas-fired boiler and gas-fired DG technologies Milan 14.2 35.8 Palermo 6.8 17.2 Large power plant (Combined cycle) Italy 7.5 9.0 Geometric standard 35 35 deviation (r g)
a
Including the damage via ozone (1.8 USD/kg).
26 See also Rabl (2002). The new estimates are lower (approximately 0.60.7 times) than those of ExternE-1998.
Carlo calculation. In fact, following Rabl and Spadaro (1999), it may not be necessary to worry about details of the probability distributions, because they wash out in the final result thanks to the central limit theorem. Thus, it suffices to specify geometric mean tl gb and geometric standard deviations tr gb, or equivalently, multiplicative confidence intervals about the geometric mean (which is usually close to the median): tl g / r g, l g d r gb for 2 2 approximately 68% and, tl g / r g , l g d r g b for approximately 95%. Thus, to the extent that the distribution of the results is lognormal, the geometric mean equals the median and the geometric standard deviation has a simple interpretation in terms of multiplicative confidence intervals around the median. For the reporting of uncertainties, the ExternE project chooses a simplified format, in terms of uncertainty labels. These labels are: A= high confidence, corresponding to r g = 2.54; B = medium confidence, corresponding to r g = 46; and C = low confidence, corresponding to r g = 612. A variant of this presentation is to indicate the order of magnitude of the range (between upper and lower limits of the confidence intervals). Then label A corresponds to approximately 1 order of magnitude, label B to approximately 1.5 orders of magnitude, and label C to approximately 2 orders of magnitude, respectively, between upper and lower limits of 68% confidence interval. The ExternE project typically recommends for NOx, SOx and PM damages low and medium confidence intervals (labels A and B). More specifically, Rabl and Spadaro (1999) analyze several types of air pollution damage and show that
294
the geometric standard deviation r g is in the range 3 5 (Table 6). 3.1.2. Global warming It is apparently easier to assess the external cost coefficients of global warming. The marginal damage (per unit of pollutant emitted) does not, in fact, depend on technology and plant localization. Therefore, there is no need to analyze the ExternE dissemination process since it is sufficient to utilize the general estimations proposed by ExternE which recommends the use of a bcentral estimateQ. Nevertheless, given the high uncertainty mentioned earlier (and that the impact assessment is cut off after 100 years), the recommendation of a bcentral estimateQ is at least badly misleading (Krewitt, 2002). Fortunately, ExternE-1998 also reports the statistical distributions of the marginal costs of CO2 emissions. These distributions have been obtained by means of a series of simulations, each for a particular value of the discount rate, and are well fitted by lognormal functions.27 Concerning the choice of the discount rate, ExternE-1998 recommends using a discount rate ranging from 1% to 3%. This range can be derived from the concept of social time preference (Markandya and Pearce, 1991) which depends on the rate of pure individual time preference or impatience, on the growth rate of real consumption per capita, and on the elasticity of the marginal utility of consumption.28 ExternE argues that, for intergenerational damage, individual time preference is irrelevant, and therefore a discount rate equal to the per capita growth rate is appropriate
These simulations are based on the use of the Montecarlo method and the FUND (Climate Framework for Uncertainty, Negotiation and Distribution) model. For a description of this model, see European Commission (1999a,b). Concerning FUND model, Tol (2003) shows that all relevant current variables of FUND have finite variances. However, Tol also recognizes that bthere is a small chance that climate reverses economic growth in some regions. In that case, the discount rate becomes negative and the net marginal benefits of greenhouse gas emissions reduction become very large. So large, that its variance is unbounded. Nevertheless, Tol argues that, even in this case, bcost-benefit analysis is a valid toolQ. 28 The equation is STP= ITP + WU where STP is the social time preference, ITP the rate of individual time preference, W the growth rate of real consumption per capita and U the elasticity of the marginal utility of consumption. Regarding the debate over the choice of discount rate and the approach adopted by ExternE, see European Commision (1999a,b).
27
(European Commission, 1999b; Rabl, 1996). In the IPCC scenarios29 (used by ExternE) the per capita growth rate is between 1% and 3%. It is also important to point out that, although ExternE refers to constant discounts rates, it recognizes that it is important to bear in mind that these rates can vary when phenomena are analyzed on a very long time scale (European Commission, 1999b). Effectively, several models have been proposed in order to simulate discount rate variability over long time. These models can be divided into two categories. The first category includes contributions of authors arguing that a discount rate for intergenerational effects should be defined by taking into account the perspective of future generations. For example, Rabl (1996) argues that market interest rates can only be taken to the extent that market exists (3040 years). Beyond this threshold, there is no inconsistency in lowering the interest rate. Other authors (e.g. Bayer and Cansier, 1996) have developed models of overlapping generations. These models include several generations and calculate costs of climate change by estimating the discounted value of investments into climate protection when benefits go beyond the life expectancy of the current generation. The second category includes contributions of the literature which links the variability of the discount rate over time to the uncertainty about the future value of the discount rate. In fact, according to Weitzman (1998), the uncertainty about future discount rates provides a strong generic rationale for using certainty-equivalent discount rates which decrease over time, from the best current estimation, presumably based on the observable market values, down to the lowest imaginable rates for the far distant future. In order to quantify this uncertainty, Weitzman (2001) carried out an extensive e-mail survey of more than 2000 economists who were asked to express their bprofessionally considered gut feelingQ about the appropriate value to adopt in assessing environmental projects. In this way, uncertainty represents a current lack of consensus about the correct discount rate for all future time periods.
ExternE uses the IPCC (Intergovernmental Panel of Climate Change) IS92a and 92d scenarios. For a description of such scenarios, see Pepper et al. (1992).
29

0,07 0,06
295
probability density
r = 0.03 0,05 0,04 0,03 0,02 0,01 0 0 20 40 60 80 100 120 140 160 180 200 r = 0.02 r = 0.01
USD per tonne of carbon dioxide

Fig. 3. Uncertainty about the marginal costs of carbon dioxide (ExternE-1998 range). Source: European Commission (1999b).
Newell and Pizer (2003), however, adopted a different approach. They assume that there is no lack of consensus about the current value of the discount rate, based on the market rates, but that it varies over time. They also assume that the historic trends of the interest rates indicate possible future trends. This is one of the most significant aspects that set their work apart from earlier contributions on this same issue. The authors use the market rates to quantify the uncertainty and persistence of the future discount rates. In particular, they find significant empirical evidence showing that the historic rates are, in fact, uncertain and persistent.30 Using this evidence to forecast future performance, they find that the certainty-equivalent discount rate gradually decreases from 4% to 1% over a period of 400 years. In particular, they show that under random walk assumption the certainty equivalent rate falls from 4% to 2% after 100 years (this is the time-horizon used by ExternE to evaluate the effects of global warming31). Note that this value is within the range recommended by ExternE. In order to be fully consistent with the other steps of the analysis conducted in this paper, we will adopt the range of constant discount rates proposed by ExternE-1998 (13%). Initially, we will use the cenFor example, the series of data shows that there has been a decrease in the discount rate going back at least 200 years (Newell and Pizer, 2003). 31 It is important to underline that the time horizon of ExternE is considered inappropriate for the assessment of external costs because of the long term nature of the expected effects (Krewitt, 2002).
30
tral value of this range (2%). Later, when we will implement the sensitivity analysis, the extreme values of the range will be adopted. Fig. 3 shows the lognormal distributions of ExternE-1998 for 1%, 2% and 3% discount rates. In order to consider the effects of the uncertainty and evaluate the environmental costs and benefits of the decentralized supply, these distributions will be directly used in the estimating model. Nevertheless, this is not sufficient yet. Using a specific methodology does not allow us to take into account the uncertainty due to the choice of model and basic assumptions (in particular, the time horizon) and adopting the results of the new version of ExternE, which uses a longer time horizon, does not resolve the problem. For this reason, we will use (together with the original analysis based on the ExternE-1998, which is more conservative than the new version of ExternE32) a meta-analysis, that is a joint probability density function formed by combining the one hundred and three estimates of 28 studies (carried out from 1991 to 2003) on the marginal damage of carbon dioxide emissions (Tol, 2005). This approach allows us to account for the large uncertainty due to the choice of models and related assumptions. It can be interpreted as plausible results, given the state of the art. Fig. 4 shows the probability distribution (cumulative) arising from the meta-analysis.
The new version of ExternE recommends (Rabl, 2005b) a central value of 19 o / tCO2 while ExternE-1998 recommended+29o / tCO2.
32
296
1,1 1 0,9 0,8
3% All 1% 0%
probability
0,7 0,6 0,5 0,4 0,3 0,2 0,1 0 -8 -4 0 4 8 12
16
20
24
28
32
USD per tonne of carbon dioxide

Fig. 4. Composite cumulative density function (meta-analysis), author weighted, for all studies and for those studies that use a 3%, 1% and 0% discount rate. Source: Tol (2005). Note: in the original paper values are reported in USD/tC.
3.2. Results Given the estimates of the monetary damages due to local and regional pollutants (reported in Table 6) and the probability distributions of global monetary damage reported in Fig. 3 (ExternE-1998) and Fig. 4 (meta-analysis), we can now calculate the external gas-fired DG benefits in terms of cumulative distributions. Figs. 5 and 6 show the results obtained by applying ExternE-1998, for the global warming, and the new dose response functions for local and regional pollutants (NewExternE). We report the ratio of costs (gasfired DG costs divided by centralized costs) on the horizontal axis (the benefit is one minus the value of this indicator). For the moment, since we are prevalently interested in showing the effect of global impact uncertainty we do not take into account the statistical uncertainty about the impact of localregional pollutants. The influence of this uncertainty will be accounted for later, when we will attempt to evaluate the overall range of environmental benefit of natural gas-fired CHP distributed generation. As regards the bisolatedQ solutions, the following conclusions can be drawn from Fig. 5. The estimate dispersions are considerably lower than those of global warming cost estimations themselves (the curves are almost vertical), so that gas-fired DG social cost-
benefit can be clearly identified. In the case of gas engines in Milan, and compared to the conventional centralized system, there is roughly a 50% probability that gas-fired DG could provide benefits but this probability disappears compared to the heat pump (the advanced centralized system). A similar result (slightly better) emerges from the gas turbine application. In Palermo, the situation is significantly better when we compare gas-fired DG to the conventional system but again gas-fired DG (almost) never provides environmental benefits compared to the heat pump (indeed, there is roughly a 30% probability that gas-fired DG could provides benefits, in the case of gas engine). The results of the bopenQ solutions (which are the most likely system configuration33) are more negative for gas-fired DG. Fig. 6 shows that the environmental benefits are almost completely absent even comparing gas-fired DG to the conventional centralized solution (except for the gas engine in Palermo, compared to the conventional centralized solution). It is important to underline that the macro- (plant geographic locations) and micro-localization (urban versus extra-urban areas) effects (previously described) have only a small impact on these results. These effects mainly regard, in fact, SOx and PM10
33 In fact, bopenQ solutions involve bprivateQ costs lower than those of bisolatedQ solutions.
297
CHP gas engine - Milan

1 0,9 cumulative distribution 0,8 0,7 0,6 0,5 0,4 0,3 0,2 0,1 0 0,7 1 1,3 1,6 1,9 2,2 2,5 2,8 EXTC DG / EXTC CENTR D1 versus B D1 versus A cumulative distribution 1 0,9 0,8 0,7 0,6 0,5 0,4 0,3 0,2 0,1 0 0,7 1
CHP gas turbine - Milan
D2 versus A
D2 versus B
1,3
1,6
1,9
2,2
2,5
2,8
EXTC DG / EXTC CENTR
CHP gas engine - Palermo

1 0,9 cumulative distribution cumulative distribution 0,8 0,7 0,6 0,5 0,4 0,3 0,2 0,1 0 0,7 1 1,3 1,6 1,9 2,2 2,5 2,8 D1 versus B D1 versus A 1 0,9 0,8 0,7 0,6 0,5 0,4 0,3 0,2 0,1 0 0,7 1
CHP gas turbine - Palermo
D2 versus A
D2 versus B
1,3
1,6
1,9
2,2
2,5
2,8
Fig. 5. External cost-benefit of gas-fired DG (bisolatedQ solutions with 2% discount rate): ExternE-1998 (for the global warming) and NewExternE (for the LR pollutants). Note: Values include non-GHG impacts (0.89 mUSD/kWh) and the monetary damage of GHG emissions (14 g/kWh) due the upstream of the natural gas fuel cycle. EXTC DG=external costs of DG; EXTC CENTR=external costs of centralized supply.
whose emissions of distributed generation and large power generation are similar and very low. The relevant part of gas-fired DG low performance is due to the much higher gas-fired DG emissions of NOx, compared to those of the large power plant (recall values reported in Tables 2 and 3). In Table 7 we report the shared analysis of the difference between external costs of gas-fired DG and advanced centralized supply (for the case of gas engine in Milan). As we can see, the low environmental
performance of distributed generation is prevalently due to the higher non-GHG pollutant emissions, localregional effects which counterbalance the positive effect of the lower consumption (energy saving, which mainly involves lower GHG emissions). As pointed out above, the need of accounting for uncertainty about the choice of the model suggests analysing the case in which the probability distribution of the marginal cost of GHG emission is obtained by using a meta-analysis. Fig. 7 shows the results of
298
CHP gas engine - Milan

1,0 0,9 cumulative distribution cumulative distribution 0,8 0,7 0,6 0,5 0,4 0,3 0,2 0,1 0,0 0,7 1,0 1,3 1,6 1,9 2,2 2,5 2,8 D1 versus B D1 versus A 1,0 0,9 0,8 0,7 0,6 0,5 0,4 0,3 0,2 0,1 0,0 0,7 1,0
CHP gas turbine - MIlan

D2 versus A
D2 versus B
1,3
1,6
1,9
2,2
2,5
2,8

1,0 0,9 cumulative distribution 0,8 0,7 0,6 0,5 0,4 0,3 0,2 0,1 0,0 0,7 1,0 1,3 1,6 1,9 2,2 2,5 2,8 D2 versus B D1 versus A cumulative distribution 1,0 0,9 0,8 0,7 0,6 0,5 0,4 0,3 0,2 0,1 0,0 0,7 1,0
D2 versus A
D2 versus B
1,3
1,6
1,9
2,2
2,5
2,8
Fig. 6. External cost-benefit of gas-fired DG (bopenQ solutions with 2% discount rate): ExternE-1998 (for global warming) and NewExternE (for LR pollutants). Note: Values include non-GHG impacts (0.89 mUSD/kWh) and the monetary damage of GHG emissions (14 g/kWh) due the upstream of the natural gas fuel cycle. EXTC DG=external costs of DG; EXTC CENTR=external costs of centralized supply.
this simulation (obtained by utilizing the probability distribution in Fig. 4). We have simulated only the case of the bisolatedQ solution in Palermo (both gas engine and gas turbine) which is the best case for the gas-fired distributed generation (in all the other cases gas-fired DG performance is lower). As we can see, the meta-analysis accounts for much larger uncertainty and, this time, gas-fired DG (almost) never provides benefits (even comparing to the conventional centralized supply). Therefore, the overall conclusion emerging from the previous analysis (based on Ex-
ternE-1998) does not change. Rather, the performance of gas-fired DG appears quite worse. Our analysis does not end here. We still have to verify the robustness of the final results by measuring their sensitivity to the variability of those parameters previously considered fixed. This sensitivity analysis will help us find the overall range of gas-fired DG environmental benefits. The main factors affecting gas-fired DG environmental benefits are the structural climate conditions and the technology-related factors.
F. Gull / Ecological Economics 57 (2006) 282305 ` Table 7 Shared analysis of the difference between external costs: gas engine (D1) versus advanced centralized (B)-Milan Net difference (EXTC D1-EXTC B) (median, 2% discount rate) Share of the different effects
299
Lower fuel consumption of DG Values a (kUSD) Percentage (%) +165.7 100.0 46.1 27.8
Higher emission rates of DG +135.0 +81.5
Micro-localization effect +7.8 +4.7
Macro-localization effect +6.8 +4.1
Combined effects +62.2 +37.5
EXTC D1: external cost DG CHP gas engine in Milan; EXTC B: external costs of the advanced centralized supply. a Excluding external costs of exploration, production and transportation fuel cycle stages.
For the sake of simplicity, we restrict our investigation to the bisolatedQ gas engine, which is the best gas-fired DG solution (compared to the best centralized solution, B) from the environmental benefit point of view (conservative approach). The base case concerns two locations, Milan and Palermo, which represent two extreme situations in terms of environmental impact. Unfortunately, these two locations are complementary in terms of climate conditions. In Milan, where the localregional environmental impact is higher, the climate is colder so that energy saving due to cogeneration is higher (and consequently GHG emissions are lower). In the case of Palermo, the situation is just the opposite. Since the
1,0 0,9 cumulative distribution 0,8 0,7 0,6 0,5 0,4 0,3 0,2 0,1 0,0 0,7 1,0 1,3 1,6 1,9 2,2 2,5 2,8 r= 0.03 r= 0.01
situation is so complementary, it implies a sort of partial compensation between global and localregional effects so that the two locations do not perfectly represent the two extreme situations of climate and geographic locations considered together. Therefore, in order to eliminate this effect, we can switch the climate situations of the two locations by simulating in Milan the climate conditions of Palermo (worst case) and in Palermo the climate conditions of Milan (best case). Technological progress affects gas-fired DG environmental cost-benefit mainly through the electrical efficiency of CHP (combined heat and power) small plants. In order to verify the impact of this factor, we
1,0 0,9 0,8 r= 0.01
cumulative distribution
0,7 0,6 0,5 0,4 0,3 0,2 0,1 0,0 0,7 1,0 1,3 1,6 1,9 2,2 2,5 2,8 r= 0.03
EXTC D1 / EXTC A
EXTC D2 / EXTC B
Fig. 7. External cost-benefit of gas-fired DG (bisolatedQ gas-fired DG in Palermo): meta-analysis (for global warming) and NewExternE (for LR pollutants). Note: Values include non-GHG impacts (0.89 mUSD/kWh) and the monetary damage of GHG emissions (14 g/kWh) due the upstream of the natural gas fuel cycle. EXTC D1=external costs of gas engine CHP DG; EXTC D2=external costs of gas turbine CHP DG; EXTC A=external costs of conventional centralized supply; EXTC B=external costs of advanced (heat pump) centralized supply; r=discount rate.
300
can refer to plausible ranges of variability. Since engineers predict only marginal improvements in the next few years, F 10% seems a plausible range for plant efficiency. We also report the lower boundary in order to take into account current possible overestimation of DG performance. In order to evaluate the overall range of gas-fired DG environmental benefits, we have to analyze the case in which all variables are favorable (i.e. Milan climate in Palermo and high gas-fired DG plant efficiency) and the case in which all the variables are unfavorable to gas-fired DG deployment (i.e. Palermo climate in Milan and low gas-fired DG plant efficiency). Finally, this time, we also consider the uncertainty about the impacts of the local and regional pollutants. As pointed out above, ExternE, in this case, utilizes multiplicative confidence intervals that are much easier to specify than an entire probability distributions. Thus, we internalize this uncertainty by means of the extreme values corresponding to 95% confidence interval. The worst case corresponds to the upper limit 2 of the 95% confidence interval l g d r g . The best case corresponds to the lower limit of the 95% confidence 2 interval l g / r g . We use the central value of the range reported in Table 6 (r g = 4) as estimate of geometric
standard deviation. The geometric means l g approximately equals the median values, reported in Table 6. These two extreme cases identify the best and worst case of gas-fired DG environmental benefits. The results are presented in terms of cumulative distributions. Fig. 8 (ExternE-1998, for the global warming) and Fig. 9 (meta-analysis, for the global warming) allow us to clearly identify whether gasfired CHP distributed generation is preferable (or not) to centralized supply and to be confident about the robustness of the analysis implemented here. In fact, these figures show very similar results, in terms of overall range of gas-fired DG cost-benefit variability, and show that the range of positive values of gas-fired DG benefits (ratio of costs lower than one) is very restricted compared to the range of negative values. In conclusion, this analysis seems to highlight that gas-fired DG is low competitive from the environmental point of view. This result is quite surprising, considering that the supposed environmental benefits generally inform the policies designed to support gasfired DG. This analysis would thus support the idea that such environmental policies are ambiguous since, by focusing on global warming, they disregard the possible trade-off between localregional and global impacts.
1 0,9 0,8 Best case (1% discount rate; 95% lower limit for LR)
0,7 0,6 0,5 0,4 0,3 0,2 0,1 0 0,6 1 1,4 1,8 2,2 2,6 3 3,4 3,8 4,2 4,6 5 Overall range of variability Worst case (3% discount rate; 95% upper limit for LR)
EXTC D1 / EXTC B
Fig. 8. Overall range of gas-fired DG environmental benefits (CHP gas engine): ExternE-1998 (for global warming) and NewExternE (for LR pollutants). Note: EXTC D1=external costs of CHP gas engine; EXTC B=external costs of heat pump; LR=localregional.

1 0,9 0,8 Best case (1% discount rate; 95% lower limit for LR)
301
0,7 0,6 0,5 0,4 0,3 0,2 0,1 0 0,6 1 1,4 1,8 2,2 2,6 3 3,4 3,8 4,2 4,6 5 Overall range of variability Worst case (3% discount rate; 95% upper limit for LR)
EXTC D1 / EXTC B
Fig. 9. Overall range of gas-fired DG environmental benefits (CHP gas engine): meta-analysis (for global warming) and NewExternE (for LR pollutants). Note: EXTC D1=external costs of CHP gas engine; EXTC B=external costs of heat pump; LR=localregional.
However, we have to point out two potential limits of this analysis. First, some models (Tol, 2003) highlight that there is a small chance that uncertainty about climate change could be infinite as well as the expected value of the marginal cost of carbon dioxide emissions. This happens when models include catastrophic impacts whose effects are amplified by the discount rate and perhaps by equity weighting. In this case, it would be difficult to defend the validity of the cost-benefit analysis. Nevertheless, Tol proposes an alternative interpretation, arguing that infinity is present in both the base case and the policy scenario (therefore irrelevant and thus the cost-benefit analysis a valid tool). Moreover, Tol, in his subsequent paper (Tol, 2005), focusing on the practical purposes of the climate change assessment, does not mention this extreme case stating that b. . . people argue that climate change is too uncertain to say anything about the marginal damage costs of carbon dioxide emissions. The uncertainties are indeed substantial, but not as large as the people think. [. . . . .] One can therefore safely say that, for all practical purposes, climate change impacts may be very uncertain but is unlikely that the marginal damage costs of carbon dioxide emissions exceed $50/tC and are likely to be substantially smaller than thatQ.
Second, one can deny the rationality attributed to making tradeoffs, on the basis of the ethical limits of the economic valuations. This is the approach of that part of the literature on ecological economics emphasizing the priority of the right on the good. Following Soderhlom and Sundqvist (2003), bIn many cases, individuals are willing to defend the existence or the well being of the environment apart from any instrumental value it provides. In particular, most people would not be willing to engage in a trade-off discussion regarding events that may lead to disastrous effects (for present or future generations) even though the probability of that disaster is extremely low. Thus, in such cases there simply exists no well-defined private utility function on which to base external cost estimatesQ. These arguments certainly apply to the impacts of global warming. The effects of global warming, in fact, are inherently global, irreversible, long-term and asymmetrically distributed over time. This is in heavy contrast with other emissions from the power sector, whose impacts are more tangible and directly connected to present human (dis)utility. However, using this reasoning for the case of gas-based energy supply decentralization raises the following question. Which should the right approach
302
be when we have to choose between two technologies involving a trade-off between intergenerational environmental impacts? As pointed out earlier, answering this question is not the aim of this analysis. We propose this interrogative only to recall that the results presented here might be not sufficient for a definitive public choice. However, we also think that these results could help policy makers (who push the deployment of gas-fired distributed generation, on the basis of the environmental benefits) to be aware that their enthusiasm (or optimism) might be exaggerated. In this sense, the analysis of gas-based energy supply decentralization helps us to support the idea that the cost-benefit analysis in the field of environmental choice is fully legitimate and represents, in spite of uncertainty and ethical limits, at least a nice (and in some cases necessary) starting point for a further discussion about the content of political and moral debate.
4. Conclusions The high uncertainty about the estimations of external environmental costs is one of the main reasons why the classic cost-benefit analysis is not considered very useful to support environmental policies. By dealing with a very topical case regarding technological change in the energy sector (the deployment of natural gas-fired CHP distributed generation), this article attempts to put in perspective such criticism and could contribute to reassess the normative role of the cost-benefit analysis. The uncertainty about external costs is substantial but not so large that it is not possible to say anything about the environmental raking of alternative technology solutions involving trade-off between the impacts of different pollutants (or between different kinds of impacts). The literature on external costs provides several studies accounting for a large part of uncertainty by means of appropriate statistical and sensitive analysis. By using and elaborating these results, the analysis described in this paper seems to support the conclusion that centralized supply, and especially the completely electric solution (based on the reversible electric heat pump), is still preferable
to natural gas-fired CHP distributed generation. This is not a definitive conclusion but, we hope, a useful (scientific based) contribution for policy decisions under the state of the art. In fact, this result has an interesting policy implication. It suggests unless questioning the current enthusiasm on natural gas-fired CHP distributed generation deployment (e.g. the European Commission is indeed advocating CHP DG as a contribution to GHG emission reduction as well as the recent European directive on the promotion of cogeneration is very favorable to micro-cogeneration and small CHP plants, on the basis of their supposed environmental benefits) and helps us to reflect upon gas-fired DG supporting environmental policies which focus on the reduction of GHG emissions and totally disregard the possible trade-off between global and local regional effects. As a matter of fact, although it is true that thanks to the energy saving from cogeneration, gas-based decentralized supply results in reduced GHG emissions (global effects), it is also true that this benefit is counterbalanced by the higher local and regional impact due to the emission of other atmospheric pollutants (SOx, NOx, and particulate). The fact we have found that there is a trade-off between global and localregional effects shows how important it is to make an economic evaluation of the environmental externalities, even for policy decisions (in the energy sector) that are binary. Therefore, this helps us reflect upon the possible ambiguity of environmental policies which systematically ignore cost-benefit analysis as a valid objective tool. Finally, one can say that the extent of gas-fired DG environmental cost-benefits may be uncertain but is unlikely that gas-based energy supply decentralization is preferable to centralized supply, from this point of view. Rather, gas-fired DG external costs are likely to be substantially higher than those of centralized supply. Therefore, one should be prudent in using the environmental argument in order to legitimate (explicit and generalized) public support to gas-fired DG deployment. Removing technical and economic barriers is welcomed but providing subsidies must be carefully evaluated. We must not conclude without pointing out two limits of this analysis. First, with regard to climate
303
change, some (specific) models do not exclude the possibility that, in the case of extreme scenarios (catastrophic impacts whose effects are amplified by the discount rate and perhaps by equity weighting), uncertainty about climate change could be infinite as well as the expected value of the marginal cost of carbon dioxide emissions. In this case, it would be difficult (but not impossible) to defend the validity of the cost-benefit analysis. Second, in a similar way, one could deny the rationality attributed to making tradeoffs between intergenerational environmental impacts, on the basis of the ethical limits of economic valuations. However, cost-benefit analysis, even in this case, seems to be fully legitimate and represents, in spite of uncertainty (provided that not infinite) and
supposed ethical limits, at least a nice (and necessary) starting point for a further discussion about the content of political and moral debate. We think that the case presented here is emblematic, from this point of view.
Acknowledgments An earlier version of this paper was presented at the bRegulation SeminarsQ of the Department of Applied Economics (DAE) of University of Cambridge. I would like to thank the seminar participants at DAE for their constructive comments. All errors pertain to the author.
Appendix
Table A1. Energy flows and assessment of energy primary consumption (MWh): Milan S: Summer; W: Winter Structural electr. need Heat space heat./con. Heat for hot water Cold for space condit. Ratio TEchp / EEchp Ratio CEar / TEchp Ratio TEhp / EEhp Ratio CEcr / EEcr CB thermal effic. CHP total effic. Central electr. eff. TE from CB CE from AR CE from CR TE low from HP EE for CR EE for HP TE from CHP TE from CHP EE from CHP PE for CB PE for CHP Imported EE Total primary energy Primary energy EEs TEh TEw CE gt g ar COP g cr g cb g chp ge TEcb CEar CEcr TEhp EEcr = CEcr / g cr EEhp = TEhp / COP TEchp = CEar / g ar + TEw TEchp EEchp = TEchp / g t PEcb = TEcb / g cb PEchp = (TEchp + EEchp) / g chp EEimp = EEs + EEcr + EEhp EEchp PE = PEcb + PEchp + EEimp / g e PEa = PEs + PEw A S 394 W 394 B S 394 W 394 1800 241 0 C1 S 394 232 241 2316 1.25 0.90 3.70 90% 50% 241 2316 232 626 3.30 3.70 109% 109% 50% 241 90% 54% 2316 2316 232 626 1800 545 2814 2251 267 2268 219 219 5628 1020 394 1020 939 2097 4082 5837 3579 2066 1836 2400 2013 0 0 0 2041 1633 3329 1924 2041 1180 232 90% 54% 90% 54% 2316 232 90% 54% W 394 1800 241 0 1.25 C2 S 394 232 241 2316 1.73 0.75 W 394 1800 241 0 1.73 D1 S 394 W 394 D2 S 394 W 394
232 1800 232 241 241 241 2316 0 2316
232 1800 232 1800 241 241 241 241 2316 0 2316 0 1.25 0.90 3.70 90% 50% 90% 50% 1.25 3.30 3.70 90% 50% 90% 50% 1.73 0.75 1.73 3.30
90%
50% 50% 2041 241
718 846 1598 1470 232 1123 232 432 397 340 1039 1369 918 831 734 791
894 271 1147 665
1857 1239 1530 786 0 2189 3977 1788 3004 5127 2123
2307 3056 2259 5363 4356
2066 1836 2400 2013 3902 4413
304
Table A2. Energy flows and assessment of energy primary consumption (MWh): Palermo S: Summer; W: Winter Structural electr. need Heat space heat./con. Heat for hot water Cold for space condit. Ratio TEchp / EEchp Ratio CEar / TEchp Ratio TEhp / EEhp Ratio CEcr / EEcr CB thermal effic. CHP total effic. Central electr. eff. TE from CB CE from AR CE from CR TE low from HP EE for CR EE for HP TE from CHP TE from CHP EE from CHP PE for CB PE for CHP Imported EE Total primary energy Primary energy (year) EEs TEh TEw CE gt g ar COP g cr g cb g chp ge TEcb CEar CEcr TEhp EEcr = CEcr / g cr EEhp = TEhp / COP TEchp = CEar / g ar + TEw TEchp EEchp = TEchp / g t PEcb = TEcb / g cb PEchp = (TEchp + EEchp) / g chp EEimp = PEcb + PEimp / g e PE = PEcb + PEchp + EEimp / g e PEa = PEs + PEw A S 394 258 241 2575 W 394 561 241 0 B S 394 258 241 2575 W 394 561 241 0 C1 S 394 258 241 2575 1.25 0.90 5.30 3.40 90% 50% 241 2575 258 757 90% 50% 802 3.40 109% 50% 241 2575 258 757 3.40 109% 50% 241 90% 54% 2575 561 106 3102 802 642 1604 248 1145 3674 802 464 1407 70 1277 1141 551 441 1102 0 1102 258 90% 54% 90% 54% 2575 258 90% 54% 90% 50% 810 1765 258 519 90% 50% 90% 50% 950 1625 258 478 1508 711 411 1247 0 1247 90% 50% W 394 561 241 0 1.25 C2 S 394 258 241 2575 1.73 0.75 W 394 561 241 0 1.73 D1 S 394 258 241 2575 1.25 0.90 W 394 561 241 0 1.25 5.30 3.40 D2 S 394 258 241 2575 1.73 0.75 W 394 561 241 0 1.73 5.30
251 47
91 17
2482 267 891 219 219 6204 1151 2569 4248 394 1679 1151 2521 3740 500 1219 2088 2337 3482
2124 6442 1730 3238 4515
913 2282 0 2282 3384
872 2644 0 2644 3891
A: centralized conventional system, AR: absorbing refrigerator, B: advanced centralized system (HP), C1: bopenQ gas engine, C2: bopenQ gas turbine, B: condensing boiler, CHP: combined heat and power generation, CR: compressing refrigerator, D1: bisolatedQ gas engine, D2: bisolatedQ gas turbine, EE: electrical energy (power), HP: heat pump, PE: primary energy (fuel), TE: thermal energy.
References
Ackermann, T., Andersson, G., Soder, L., 2001. Distributed generation: a definition. Electric Power Systems Research 57 (3), 195 294. Arthur D. Little Inc, 1998. Distributed Generation: Understanding the Economics. Arthur D. Little White Paper, Massachusetts. Bayer, S., Cansier, D., 1996. Methodisch Abgesicherte intergenerationelle Diskountiering am Beispiel des Klimaschutzes. Diskussionspapier der Eberhard-Karls-Universitat Tubingen, cited in European Commission (1999b), pp. 311. Bruzzi, V., 2002. Cogenerazione Distribuita, Pompe di Calore Reversibili e Frigoriferi ad Assorbimento nelle Utenze Civili. Quaderni di Ricerca Iefe, Univesita Bocconi, Milano. `
EPAUS Environmental Protection Agency, 2004. Air Quality Criteria for Particulate Matter (vol. I and II). EPA/600/P-99/ 002 aF-bF, NC, http://www.epa.gov. European Commission, 1997. ExternE: ExternE Project. Publishable report. External Costs of Transport in ExternE. http:// externe.jrc.es/reports.html. European Commission, 1999a. ExternE: Externalities of Fuel Cycles. ExternE Project, Methodology Updated 1998 Report. Office for Official Publications of the European Communities, Luxembourg. European Commission, 1999b. ExternE: Externalities of Energy. Final report, Global Warming. Office for Official Publications of the European Communities, Luxembourg.
F. Gull / Ecological Economics 57 (2006) 282305 ` European Commission, 1999c. ExternE: Externalities of Energy. ExternE Project, National Implementation. Office for Official Publications of the European Communities, Luxembourg. European Commission, 1999d. ExternE: Externalities of Energy. Fuel Cycles for Emerging and End-Use Technologies, Transport and Waste. Office for Official Publications of the European Communities, Luxembourg. European Commission, 1999e. ExternE: Externalities of Energy. ExternE Project, Common Annexes of the ExternE National Implementation Reports. http://externe.jrc.es/reports.html. Frigerio M., Gull F., Lorenzoni A., 1995. External Costs of Fuel ` Cycles. Final Report, ITGCC Fuel Cycle. Report commissioned by the EC, ExternE Project. Funk, K., Rabl, A., 1999. Electric versus conventional vehicles: social costs and benefits in France. Transportation Research. Part D, Transport and Environment 4, 397 411. Gull, F., 2003. Distributed Generation Versus Centralized Supply: a ` Social Cost-Benefit Analysis, Working Papers, vol. 30. The Cambridge-MIT Institute (CMI). Krewitt, W., 2002. External costs of energydo the answers match the questions? Looking back at 10 years of ExternE. Energy Policy 30, 839 848. Markandya, A., Pearce, D.W., 1991. Development, the environment, and the social rate of discount. The World Bank Research Observer, vol. 6, pp. 137 152. Newell, R.G., Pizer, W.A., 2003. Discounting the distant future: how much do uncertain rates increase valuations? Journal of Environmental Economics and Management 46, 52 71. Pepper, W., Leggett, J., Swart, R., Wasson, J., Edmonds, J., Mintzer, I., 1992. Emissions Scenarios for the IPCC. An Update. Assumptions, Methodology and Results, Supplementary Documentation to the 1992 IPCC Report. Pfeifenberger, J.P., Hanser, P.Q., Ammann, P., 1997. Whats the cards for distributed resources? The Energy Journal, 1 16 (DR Special Issue). Rabl, A., 1996. Discounting of long term costs: what would future generations prefer us to do? Ecological Economics 17, 137 145.
305
Rabl, A., 2002. Environmetal benefits of natural gas for buses. Transportation Research. Part D, Transport and Environment 7, 391 405. Rabl, A., 2005a. Health impacts of pollutants. Presentation at Maxima Workshop on Extension of ExternE, Krakow. (http:// maxima.ier.uni-stuttgart.de/). Rabl, A., 2005b. Uncertainty, variability and gaps. Presentation at Maxima Workshop on Extension of ExternE, Krakow. (http:// maxima.ier.uni-stuttgart.de/). Rabl, A., Spadaro, J.V., 1999. Damages and costs of air pollution: an analysis of uncertainties. Environment International 25, 29 46. Regulatory Assistance Project, 2001. Model Regulations for the Output of Specified Air Emissions from Smaller-scale Electric Generation Resources. RAP Report, Maine. Schamaleense, R., 1993. Symposium on global climate change. Journal of Economic Perspectives 7, 3 10. Soderhlom, P., Sundqvist, T., 2003. Pricing environmental externalities in the power sector: ethical limits and implications for social choice. Ecological Economics 46, 333 350. Spash, C.L., 1997. Ethics and environmental attitudes with implications for economic valuation. Journal of Environmental Management 50, 403 416. Stirling, A., 1997. Limits to the value of external costs. Energy Policy 25, 517 540. Tol, R., 2003. Is the uncertainty about climate change too large for expected cost-benefit analysis? Climatic Change 56, 265 289. Tol, R., 2005. The marginal damage costs of carbon dioxide emissions: an assessment of the uncertainties. Energy Policy 33, 2064 2074. Valette, P., 1995. The European Commission ExternE project. Revue de lEnergie 467, 245 265. Weitzman, M.L., 1998. Why the far-distant future should be discounted at its lowest possible rate. Journal of Environmental Economics and Management 36, 201 208. Weitzman, M.L., 2001. Gamma discounting. American Economic Review 91, 260 271.

Social Choice, Uncertainty About External Costs and Trade-Off Between Intergenerational Environmental Impacts.

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Social Choice, Uncertainty About External Costs and Trade-Off Between Intergenerational Environmental Impacts.

Uploaded by

Copyright:

Available Formats

Ecological Economics 57 (2006) 282 305 www.elsevier.

F. Gull / Ecological Economics 57 (2006) 282305 `

F. Gull / Ecological Economics 57 (2006) 282305 `

F. Gull / Ecological Economics 57 (2006) 282305 `

F. Gull / Ecological Economics 57 (2006) 282305 `

Fig. 1. Regional environmental impact.

F. Gull / Ecological Economics 57 (2006) 282305 `

"Open" DG (C1 and C2)

"Isolated" DG (D1 and D2)

heat for space heating (winter) Hot water

heat for space heating (winter)

Fig. 2. Centralized and decentralized systems: simplified technical schemes.

(SOx) (g/MWh) 2.5 2.5 1.6

(NOx) (g/MWh) 897.8 469.3 24.9

(PM10) (g/MWh) 12.3 32.4 16.6

700 700 400,000

452.8 611.0 320.8

F. Gull / Ecological Economics 57 (2006) 282305 `

3977 0.74 5127 0.96 3902 0.73 4413 0.82

3482 0.82 4515 1.06 3384 0.80 3891 0.91

Source: Appendix (Tables A1 and A2).

aDG QzDG z acentr Qzcentrav z

acentr Qzcentr z aDG QzDG z

SOx (kg/y) 6.9 4.7

728.2 955.2 706.7 808.9

4.6 3.7 3.9 3.5

5.2 52.8 19.1 43.2

3407.2 1265.2 1401.2 620.9

636.7 839.7 612.9 713.2

3.9 3.3 3.4 3.1

1.5 46.3 16.6 38.1

2743.7 1058.0 1215.2 547.5

F. Gull / Ecological Economics 57 (2006) 282305 `

F. Gull / Ecological Economics 57 (2006) 282305 `

Including the damage via ozone (1.8 USD/kg).

F. Gull / Ecological Economics 57 (2006) 282305 `

F. Gull / Ecological Economics 57 (2006) 282305 `

USD per tonne of carbon dioxide

F. Gull / Ecological Economics 57 (2006) 282305 `

1,1 1 0,9 0,8

0,7 0,6 0,5 0,4 0,3 0,2 0,1 0 -8 -4 0 4 8 12

USD per tonne of carbon dioxide

F. Gull / Ecological Economics 57 (2006) 282305 `

CHP gas engine - Milan

CHP gas turbine - Milan

EXTC DG / EXTC CENTR

CHP gas engine - Palermo

CHP gas turbine - Palermo

EXTC DG / EXTC CENTR

EXTC DG / EXTC CENTR

F. Gull / Ecological Economics 57 (2006) 282305 `

CHP gas engine - Milan

CHP gas turbine - MIlan

EXTC DG / EXTC CENTR

EXTC DG / EXTC CENTR

CHP gas engine - Palermo

CHP gas turbine - Palermo

EXTC DG / EXTC CENTR

EXTC DG / EXTC CENTR

Higher emission rates of DG +135.0 +81.5

Micro-localization effect +7.8 +4.7

Macro-localization effect +6.8 +4.1

Combined effects +62.2 +37.5