Document(s) 8 of 15

Introduction

The electricity sector plays a central role in the South African economy as the supplier of a key input to the industrial, mining and commercial sectors, as an employer and as a service provider for households.¹ This role is likely to increase in the foreseeable future as the country’s growth and development objectives translate into greater economic output and improved service levels for low-income households. Cheap electricity is widely viewed as an important basis for achieving these objectives. In this context it is pertinent to ask: How serious are the negative environmental impacts caused by electricity generation? What is their significance in economic terms? And what would it mean if those impacts were to be accounted for in the regulatory and pricing regimes? These are some of the questions addressed in this chapter.

To begin with, a brief overview is provided of South Africa’s electricity supply industry. Next, reference is made briefly to the theory of externalities and methodologies for their valuation—some understanding of this is necessary for the subsequent analysis. The third section of this chapter summarises briefly the results of major international attempts to account for electricity sector externalities. In the fourth section, the main externalities in South Africa’s electricity sector are identified, focusing especially on the coal-based subsector, but including aspects of the nuclear-based electricity industry. Thereafter, economic valuations of key externalities are made, and these results are compared to current price levels. The chapter ends by considering the limitations of the analysis, as well as several important policy implications.

¹ This chapter is based on a study undertaken as part of the Industrial Strategy Project, published as a book entitled Counting the social costs: electricity and externalities in South Africa (Van Horen 1996a).

An overview of South Africa’s electricity supply industry

South Africa’s electricity supply industry comprises three main subsectors, corresponding to their functional activities: the generation of electricity, its transmission from power stations through a high-voltage national network and the distribution of electricity from the transmission network to end-consumers. While the focus of this chapter is primarily on the generation sector, it should be noted that there is an important new stakeholder in the electricity supply industry, namely, the National Electricity Regulator, which was established by government in early-1995 on the recommendation of the National Electrification Forum.² While the principal concern of the National Electricity Regulator is presently with matters related to the electrification programme, such as the issuing of licences to distributors and overseeing the rationalisation of the distribution industry, this is unlikely to be the case forever. As is the case in many countries with large electricity supply industries, it can be expected that the National Electricity Regulator’s mandate will be widened to include also the governance of the environmental performance of the industry.

The key player in the electricity generation sector is Eskom, the national electricity utility. In 1994, Eskom generated 96% of all electricity in South Africa, with the balance being produced by some local authorities with their own power stations (such as Johannesburg, Port Elizabeth and Cape Town) and by private concerns producing electricity for their own consumption (Eskom 1995a: 54). Most of the municipal power stations are used as a backup to the Eskom supply, and have not been fully operational for some time, with the exception of those in Johannesburg, which have ready access to cheap coal (Steyn 1994: 7).³ Given the dominance of Eskom in the electricity generation sector, therefore, the remainder of this chapter focuses on its own power stations. This is not to say that the environmental impacts of local authorities’ power stations are less significant than those of Eskom—if anything, the opposite is true given that their stations are generally older, less efficient and located in more densely populated areas. Nonetheless, their size in relation to Eskom’s capacity does not warrant further attention here.

As at the end of 1994, Eskom had a total of 19 power stations in commission, with a total nominal capacity of 37 840 megawatts (MW) (Eskom 1995a: 56). A breakdown of this capacity by fuel source is shown in Table 1.

It is evident from the table that 95% of total electricity capacity is based on nonrenewable resources: primarily coal, but including also nuclear and small amounts of gas. The coal and nuclear power stations provide the bulk of the base electricity load, while the pumped storage schemes and gas turbines are used to meet electricity demand during peak times and in cases of emergency. Pumped storage schemes are net consumers of electricity, which is used during off-peak hours to pump water to storage reservoirs, and then allowed to run down during peak hours when electricity demand is especially high.

2 The National Electrification Forum was a body representing all key stakeholders in the electricity industry, established in 1993 to formulate policies for an accelerated electrification programme for South Africa. It was disbanded in early 1995, having reached agreement on some issues, but since it could make decisions only by consensus, its ability to resolve important policy questions was constrained.

3 The Cape Town electricity department recommissioned its Athlone Power station in 1995 for the purpose of meeting peak demand.

TABLE 1 Breakdown of Eskom’s generation capacity as at 31 December 1994 (Eskom 1995a: 56)

Of the total capacity of 37 840 MW, some 4 531 MW, or 12%, was mothballed as at the end of 1994 (ibid: 56), because of the excess capacity of the Eskom system—these were generally the older and less efficient stations. Peak demand in 1994, which occurred on the night of 26 July, was 24 798 MW (ibid: 2), reflecting the large amount of excess capacity, even with the standard reserve margins commonly employed by utilities. In 1995, however, the peak demand on the system came much closer to supply capacity.

The dominance of coal in the production of electricity is even more significant, with coal-fired power stations accounting for 92% of all electricity generated in 1994—this higher percentage, relative to their share of total capacity, reflects the high level of utilisation of coal plants which are used for base-load. The Koeberg nuclear power station has supplied a relatively constant 5% to 6% of Eskom’s electricity. While the contribution of the two pumped storage schemes has been fairly constant at around only 1%, the amount of electricity generated by the hydroelectric schemes dropped considerably in 1992 and 1993; this was due to the drought which affected the flow rate of the Orange River.

Eskom operated nine of its coal power stations during 1994; their location and that of Koeberg nuclear power station is shown in Figure 1.

Eskom in an international context

Eskom is one of the largest electricity utilities in the world. As shown in Table 2, it ranks among the top five utilities, measured on the basis either of capacity or of sales. Interestingly, Eskom is one of only two utilities from non-Organisation for Economic Cooperation and Development countries, the other being South Korea’s Korea Electric Power Company. This makes South Africa the country with the lowest per capita GDP of the countries shown in the table (World Bank 1993). Eskom is, therefore, an unusually strong utility among developing countries.

Another important point of comparison between Eskom and international electricity utilities is their relative price levels. Here, Eskom ranks among the cheapest producers in the world.⁴ At the end of 1993, its industrial electricity tariffs were the lowest of a basket of industrialised countries: Japan (whose average price was over three times higher than Eskom’s), Germany, the UK, the USA, France, Canada, New Zealand and Sweden (Eskom 1995a: 9). This comparison is all the more striking when the resource bases of some of those countries are taken into account: in particular, several of the listed countries are predominantly hydro-electricity based, which is generally regarded as one of the cheapest sources of electricity (when compared to coal, gas, nuclear, etc.). This comparison, therefore, begs the question: why is South Africa’s electricity so cheap in relative terms?

FIGURE 1 Location of Eskom’s operational coal and nuclear power stations, 1994

Trends in Eskom’s electricity prices

A notable development in relation to Eskom’s electricity prices is its ongoing commitment to reduce the real price of electricity on the back of internal efficiency gains. Eskom announced its “price compact” in 1991, in terms of which it undertook to decrease the real price of electricity by 20% over the five-year period 1992 to 1996. This came on top of a 14% reduction in real price which had already been achieved from 1987 to 1991 (Eskom 1992: 5). In 1994, it made a further commitment, in terms of its so-called “RDP commitments” to reduce the real price by 15% over the period 1995 to 2000 (Eskom 1995a: 9). Until 1994, all of these targets had been met and, therefore, 1994 prices were approximately 76% of 1987 levels in real terms. There is little reason to believe that the utility will not meet its latest commitment and, if this is the case, the average electricity price in 2000 will be approximately 60%, in real terms, of the 1987 price level (Van Horen 1996b). This is a dramatic decrease by any standards.

⁴ This comparison excludes utilities whose tariffs are too low to recover their costs.

TABLE 2 The largest electricity utilities in the world in 1993 (Eskom 1995a: 59)⁵

Eskom suggests that these price reductions will be achieved through ongoing productivity improvements, reduced operating expenditure and cost containment. This commitment is linked to the goal of becoming “the world’s lowest-cost producer of electricity” (Eskom 1995a: 3). It is evident that other factors underlie this decrease in electricity prices, notably the decrease in the number of employees, from 66 000 in 1985 to 40 000 in 1994 (ibid: 55). In addition, its reduced exposure to finance charges has contributed significantly to its financial health, with the debt-equity ratio declining from 3,0:1 in 1985 to its 1994 level of 1,7:1 (ibid: 10). The utility’s stated intention is to reduce this ratio to parity by 1998. This is achievable given that its levels of capital expenditure have declined considerably in real terms since the mid-1980s because of the situation of over-capacity in which Eskom found itself (Van Horen 1996b).

It is in this context that the question again arises whether South Africa’s electricity prices, which are declining in real terms, adequately reflect environmental costs. Before investigating international and South African experience with external costs, the following section reviews briefly the economic theory of externalities.

⁵ All data as at 31 December 1993, except for TEPCO (31 March 1994), ENEL (31 December 1992), RWE (30 June 1993) and TVA (30 September 1993).

Theoretical and methodological issues

There is a large body of economic literature which deals with the theory of externalities and market failure. It is important to understand this theory because of the complex methodological questions which arise in attempts to apply it. This section therefore briefly reviews the theory of externalities.⁶

A central concept in the economic analysis of environmental impacts is that of an externality. An externality or, “external effect”, can be either positive or negative, although policy is most frequently concerned with the latter because of the implied welfare loss. In its earlier usage, the term was sometimes defined so broadly as to include most sources of market failure (Mishan 1971:6, Baumol and Oates 1975:16), although in its more contemporary usage it generally refers to a situation where two conditions are met:

• activity by one economic agent causes a change in the utility or welfare of another agent, (more formally, the first entity’s production function includes variables whose values are determined by the second entity); and
  
  
• this change in welfare is not compensated or appropriated (Baumol and Oates 1975: 17, Pearce and Turner 1990: 61).

An externality, therefore, arises in situations such as where a productive facility causes the emission of pollutants or waste products which, in turn, impact upon human health or environmental elements which have value for humans (such as agricultural crops), where the costs of those impacts are not captured in the market relationship between the producer and its customers, and those who bear the costs are not compensated in any way. It is apparent that this scenario applies to the production of electricity in so far as it results in environmental and health impacts, the costs of which are not accounted for in the utility’s costing or pricing structures.

Another way of explaining the idea of externalities is by reference to the divergence between private and social costs (Pearce and Turner 1990: 66). Private costs are those costs which are borne by the producer of the good—in Eskom’s case, these comprise the costs of the factors of production: coal, enriched uranium, labour, capital and so on. Social costs go further than this to include the full costs of producing or consuming a commodity, and may be borne not only by the producer but also by other groups in society at large. The difference between private and social costs, then, represents the external cost or the externality which is borne by society at large.

The principles of external costs are illustrated graphically in Figure 2. Any individual producer faces a horizontal marginal revenue curve (shown by MR) equivalent to the price of the commodity, and a marginal private cost curve (shown by MPC). If the producer seeks to maximise its surplus, it will clearly produce at the point where its marginal revenues and costs are equal—point B, that is, at a level of output equal to Q₁. With the assumption that it seeks to maximise profits, it makes little sense to deviate to either side of that point. The marginal external cost at any given level of output is given by the vertical difference between the marginal social cost (MSC) and MPC lines, and so the total external cost at the individual’s optimum (Q₁) is equal to the triangle OBD.

⁶ One of the classic texts dealing with externalities is Baumol and Oates (1975). For a more accessible text see Pearce and Turner (1990).

While this may be optimal from the individual producer’s perspective, it is not ideal for society as a whole. The socially optimum level of output will be at a lower point, that is, point A, where the marginal benefit equals the marginal social cost, equivalent to a level of output of Q₂. At this point, the benefits to society are exactly equal to the costs to society of producing the commodity, and so, if the objective is to maximise social welfare, rather than individuals’ surpluses, then it does not make sense to deviate to either side of Q₂.

FIGURE 2 Illustrative marginal revenue (MR), marginal private cost (MPC) and marginal social cost (MSC) curves in a competitive market

It is worth noting that in terms of this formulation, which is derived from basic microeconomics, the economic optimum will still be associated with some level of pollution, the external costs of which are shown by the triangle OAC in the figure. This is an important point of departure in economic analysis, namely, that there is an optimal level of pollution, which is often above zero, and that this depends upon the interplay between costs and benefits. In an environmental analysis, by contrast, the optimum level of pollution is usually at or close to zero, corresponding to very low levels of economic activity, depending not on costs and benefits, but upon ecological processes. This highlights the frequently differing approaches of economists and environmentalists in analysing pollution and related issues.

The theory as outlined above, therefore, demonstrates that externalities constitute an important source of market failure. If the above conceptual representation bears any relation to reality, then it can be expected that the existence of externalities may lead to several outcomes which are less than ideal. Firstly, resources could be allocated inefficiently due to the decision of producers to produce a higher level of output than is economically ideal. Secondly, the burdens of the external costs are seldom spread equitably across society and often fall on social groups which are least able to afford them. Thirdly, a higher rate of productive activity usually translates into more rapid consumption of resources, including non-renewable ones, and this undermines goals of environmental sustainability. Collectively, the effect of externalities is, therefore, in conflict with the goals of economic efficiency, social equity and environmental sustainability—the three pillars upon which “sustainable development” is based.

Externality valuation: methodological issues

Two sets of methodological issues are relevant to the measurement of externalities in practice: the overall framework for identifying externalities, and the methods used to value them in economic terms.

Given that there is a wide range of environmental impacts arising from the generation of electricity, it is necessary to adopt a systematic approach to their identification and evaluation. The method used in this study, which has been used in the majority of international externalities studies, is the impact pathway or damage function approach. This approach is illustrated conceptually in Figure 3.

The damage function approach entails the identification and quantification of environmental and other damages arising at each stage in the fuel cycle: from the extraction of raw materials (such as coal or uranium), to their transport and processing, to their consumption in power stations, to the impacts of waste products arising in the electricity generation process, and their impacts on human health and amenity, and on the physical and natural environments.

This approach corresponds with the real-world steps in the fuel cycle. The damage function approach is generally regarded as the preferred approach to assessing environmental externalities in practice (Rowe et al 1995:2). It is important to note, however, that the damage function approach is also subject to several weaknesses and limitations:

• the approach is highly data-intensive, since information is required about each step in the impact pathway;
  
  
• professional judgements are required about the most appropriate data to use, since there are often conflicting views in the externalities literature, and the result can be sensitive to these judgements;
  
  
• if there are omissions, inaccuracies or biases in the data, these can be compounded throughout the assessment chain, thereby limiting the usefulness of the end results; and
  
  
• the damage function approach is fairly complex and draws upon a number of disciplines, which can render such studies inaccessible to the wider audience which may be interested in their results.

While the data requirements of a comprehensive impact pathway assessment are formidable, there is a relatively large body of relevant information which makes it possible to employ the approach in South Africa.

A further important characteristic of the damage function approach is that it allows for the analysis to proceed as far down the impact chain as the decision-maker chooses. Thus, for instance, if there is uncertainty or disagreement over the valuation of human health effects or crop damages from air pollution, it is possible to use the information about physical impacts as the basis for decision-making; in other words, it is entirely possible to stop short of the economic valuation of externalities if the decision-maker chooses. Moreover, provided these input assumptions are made explicit, it is possible for users to perform sensitivity analyses or enter alternative values for particular variables.

FIGURE 3 The damage function approach for power station emissions (after Rowe et al 1995:3)

In order to make estimates of the value of the externalities identified in this chapter, it is important to describe briefly the valuation approach which has been used in respect of changes in human morbidity and mortality.

A number of valuation approaches exist to value morbidity and mortality. Dealing first with morbidity, two broad methods can be used: first, those based on individual preferences, that is, willingness to pay for environmental and health improvements, or willingness to accept compensation for deterioration and, second, those methods based on resource or opportunity costs (Freeman 1993:343). There is a considerable body of literature around these methods, which is summarised briefly in Van Horen (1996a). For present purposes, it can be noted that the second approach, sometimes also called the “cost of illness” approach is used in this study, for the main reasons that willingness to pay and willingness to accept compensation studies have not been undertaken yet in South Africa and, in any event, these approaches are more complex and controversial in the context of a developing country such as South Africa. Thus, the valuation of health effects generally includes actual expenditure on health care (both public and private), transport costs, medication and foregone income, such as lost time at work. This data has been derived from published medical data wherever possible, otherwise from estimates by public health practitioners.

Turning to the valuation of premature death, the complexities and controversies are significant. Not least of all is the ethical problem which arises in reducing human life to a finite monetary value, and the implications this holds for policy-making. The economics literature frequently makes a distinction between the valuation of an individual’s life and the value individuals or society place on the risk or probability of early mortality (Freeman 1993, Pearce et al 1991:5).

Another, perhaps more compelling, justification for undertaking valuation is that in many respects this is done already by individuals and society implicitly in many of their activities, but often without making the trade-offs explicit. This is acutely evident, for example, in the decision over the allocation of public resources to primary health care services versus high-level tertiary care such as heart transplants. To assign some monetary value to human life merely makes transparent or explicit whatever judgements are being made. Furthermore, provided the values are used in a decision-making context which seeks to balance the full range of interests as best as possible, the use of monetary values for early death can serve an important strategic purpose: for example, by highlighting the losses suffered by society due to inadequate supplies of potable water and sanitation services, a case can be made, perhaps more strongly, for investment in improved service levels.

As in the case of morbidity effects, there are a number of methods which may be adopted to value premature death. One is the “human capital” approach: essentially, this entails valuing a lost life at the discounted value of future income which that person might have been expected to generate. Most simply, average GDP would be used as a proxy for that person’s income: this was done, for instance, in a study by Dutkiewicz and de Villiers (1993). Problems soon arise, however, particularly if there is any differentiation between social classes, age groups, males and females, employed and unemployed, and so on—the implications of the differing values which result are especially problematic for policy and resource allocation decisions where “equity” is a goal: in most societies, adult males of about 25 years of age will have the highest “value” (Freeman 1993:324). Furthermore, this approach is highly sensitive to the choice of a discount rate: for example, for a male child between one and four years of age in the US in 1987, at a discount rate of 2,5%, its human capital value would have been $761 000, compared to only $60 000 at a discount rate of 10% (ibid: 325).

The valuation approach which is generally preferred in the literature entails the use of individual preference approaches: not so much a person’s willingness to pay to avoid death (which would probably be their entire wealth) or willingness to accept compensation for death (which would probably be an infinite amount), but the valuation of a changed probability of death. Such decisions are made on a daily basis: for example, in paying a higher price for a ticket with an airline or bus service which is considered safer than the alternatives.

The choice of values of premature death for South Africa is made difficult by two factors: first, there have been no studies of this nature in this country from which values can be derived; second, there are sharp inequalities in the distribution of income and wealth, which presents problems (from an equity perspective) if differential valuations are used for different income groups—as would apply if the human capital approach was adopted.

Taking both of these issues into account, this study uses a consistent valuation set for premature death across the entire population in the region. The valuation approach adopted here has drawn on international studies of revealed preference, adjusted to take account of differential levels of income. At least two major international externality valuation exercises have undertaken their own reviews of the literature and, on that basis, selected a range of values for premature deaths. These estimates, which are based on revealed preference approaches, are shown in Table 3. The study by Rowe et al (1993,1994) was undertaken for New York state, USA, and drew upon North American valuation studies; likewise, the study by ETSU (1995) estimated values for the European Union.

Simply applying these valuations to South African conditions would be problematic from a theoretical perspective, since individual valuations of the risk of death must, by definition, take some account of income levels. Assuming these valuations vary in direct proportion to income, an adjustment can be made to the North American and European values to reflect average South African income levels. These adjustment factors are also shown in Table 3.

The figures in the bottom right cell of Table 3 represent the average of the adjusted valuations for the American and European studies in the last row of the table. If these numbers are rounded to the nearest R100 000, the following valuations for premature deaths are derived:

• low estimate:R700 000;
  
  
• central estimate: Rl 000 000; and
  
  
• high estimate Rl 400 000.

These are the values which have been used in this study for purposes of attaching an economic value to premature mortality. The final consideration for present purposes concerns the possible differentiation of valuations depending on variables such as age, gender, race and location. While the use of different valuations might yield results which more accurately reflect the influence of these variables on the economic standing of individuals, this is not done in this study, for the following reasons:

• the empirical basis for introducing a wider range of valuations for different social groups, is weak. The only method would involve adjusting the above estimates for average incomes of each group being identified (in the same way as this was done in Table 3). However, this would introduce the same problems as are attributable to the human capital valuation approach—particularly the ethical and equity considerations arising in policy decisions encompassing different social groups;⁹
  
  

• the modelling of health outcomes does not permit differentiation between, for example, employed and unemployed victims of pollution. Thus it would not be possible to apply differentiated valuation sets for the range of externalities being considered in this thesis; and
  
  

• the marginal increase in “accuracy” which might be achieved through further disaggregation of the above valuation sets, would require a disproportionately large increase in data collection and modelling sophistication.

TABLE 3 Valuations of premature deaths used in international studies and in this study

Externality valuation: The international experience

Distinguishing features of the environmental costs produced by externality studies undertaken to date are the numerical discrepancies and lack of consistency in their results. This uncertainty is sometimes used by sceptics to discount altogether the validity of such attempts. This, however, is an unconstructive and uncritical response since many of the differences can be accounted for by the varying technical and environmental conditions pertaining to the studies, as well as to methodological differences. The estimates of external environmental costs produced by nine significant studies are summarised in Table 4.¹⁰

⁷ Calculated as follows: GDP for 1992 of R238 711 million (SAIRR 1995: 380), divided by population estimate for 1992 of 38,8 million (1994 estimate of 40,7 million, reduced by annual growth rate of 2,44%) (ibid: 5).

⁸ The US and EU valuations are divided by the income adjustment factor to give South African valuations. The following exchange rates are used: $1 = R3,66, ecu 1 = R4,78.

⁹ Because such comparisons are not being made on an international scale in this study (for example, between South African externalities and those elsewhere), there is no contradiction in adjusting international valuations to reflect different income levels in South Africa, as per Table 3.

For purposes of this study, the most important fuel cycles are coal and, secondarily, nuclear; nonetheless, it is interesting to compare the external costs of these cycles with others, which becomes especially relevant when making decisions about new investments with various resource options. As would be expected, renewable options fare much better than fossil fuel and nuclear fuel cycles, with large relative differences in external costs. Of all fuel cycles, coal had the consistently highest external cost, with the exception of nuclear in Hohmeyer’s study and oil in Pearce’s study.

Upon closer inspection of externalities in the coal cycle, it is apparent that the studies can be grouped into two categories, based on the order of magnitude of their results. The first group comprises the three earlier studies—Hohmeyer, Ottinger and Pearce—which produced external cost estimates in the range of 7,3 c/kWh to 33,1 c/kWh. The second group of studies, which were all conducted post-1992, produced lower estimates, in the range of 0,21 c/kWh to 9,52 c/kWh. In analysing the differences in these studies, the following main factors may be identified (partly drawn from Lee 1995:2-3):

• Methodology. While all the studies summarised in the table used the damage cost approach, they used different methods to obtain data on components in the impact pathway. The earlier studies generally used other studies’ estimates of pollution emissions and impacts, and multiplied these by economic values to calculate damage costs. The second group of studies, on the other hand, either used more complex and specific methods to collect data on pollution emissions, dispersion and impacts, such as atmospheric models and dose-response functions, or they used lower valuations. In practice, the effect has been that the earlier studies used higher estimates of emissions, concentrations and impacts.
  
  
• Emission factors. The earlier studies’ emissions factors (measured in tonnes of pollutant per gigawatt-hour of electricity generated) were considerably higher than in recent studies, sometimes by a factor of 10. This is partly due to technical differences in the plants which were addressed, in that more recent studies have selected newer plants which have better environmental performance in general and, in some specific cases, are fitted with desulphurisation and other control equipment.
  
  
• Sulphate and nitrate aerosols. The older studies contain different assumptions about emissions of SO₂ (from which damaging sulphate aerosols are formed) and about their dispersion in the atmosphere, which lead to higher external cost valuations. SO₂-related externalities accounted for 60% of total externalities in Ottinger et al’s work, and 75% in Hohmeyer’s study, both of which are considerably higher than the more recent studies.
  
  
• Climate change damages. In earlier studies, fairly high values were attributed to damages caused by climate change (impacts include, for example, sea level rise, increased drought and climatic extremes), whereas in more recent studies, analysts have argued

¹⁰For more details on each of these studies, see Van Horen (1996a).

TABLE 4 Fuel cycle external costs (in cents per kilowatt-hour—c/kWh) estimated by various studies (adapted from Lee 1995: 5; Friedrich and Voss 1993:121, Dutkiewicz and de Villiers 1993:41, Schleisner et al 1995)

Notes

¹All amounts in 1994 SA c/kWh, converted from 1994 US cents at a rate of R3,66/$l, and from 1993 German mark at R1,98/DM1. Where relevant, amounts in rand have been adjusted using a 10% annual inflation factor.

² + denotes an external benefit.

³Estimates for a “new” and “old” plant respectively.

⁴High estimate reflects risk averse valuation of health impacts of nuclear disaster.

⁵ Estimates are for 4 000 full load hours per year, and include external costs of back-up system.

⁶Estimates originally in 1989 rand; annual adjustment of 10% used.

⁷ Estimates for plants in the rural south-west and south-east USA respectively.

⁸Estimates for retrofit on existing dams in Kentucky, and diversion project in Washington State, respectively.

⁹All but 0,02 c/kWh was due to the aesthetic value of a waterfall, which was estimated in a contingent valuation study.

¹⁰Estimates using a 3% and 0% discount rate respectively.

¹¹The first estimate was for a site at West Burton, UK; the second was for Laufen, Germany.

¹²Estimates for various sites in the UK.

¹³Estimates are for a rural site in New York State.

¹⁴For a boiling water reactor, rather than pressurised water reactor.

¹⁵Estimates are for a “conventional coal-fired plant” defined as a 350 MW plant with desulphurisation and de- NO equipment.

that there is too much scientific uncertainty about the impacts of climate change (without questioning its likelihood of occurring) to make meaningful estimates of damage costs. Recent studies have, in turn, been criticised for understating the likely scale of those impacts by avoiding their valuation (Ottinger 1995:4).

If any conclusions are to be drawn from these international studies, then the first would be that external costs can be significant in absolute terms, as well as in relative terms, when comparing alternative fuel cycles. International experience, therefore, points to the need to investigate externalities in South Africa’s power sector. Secondly, it is important not to take external costs simply at face value; rather, they need to be evaluated in the specific contexts insimply applied uncritically to the South African situation, but that local circumstances be texts in which they were calculated. Thus, it is important that the values summarised above are not simply applied uncritically to the South African situation, but that local circumstances be taken into account as far as possible. Finally, it is important that assumptions, methodologies and limitations are made explicit in order that the results of the valuation exercise can be appraised in the appropriate context.

Externalities in South Africa’s fuel cycle

Given time and information constraints, it was not possible in this study to evaluate externalities at every step in the chain; moreover, beyond a certain point, there are diminishing returns from widening the scope of the study to include less serious environmental and other impacts.

A degree of judgement is required in making decisions regarding which externalities are potentially significant and which are not. Consequently, a classification system has been used, which makes explicit the criteria used in determining the scope of the quantification exercise. Environmental impacts were classified according to the following three categories:

• Class One impacts. These are environmental impacts which are potentially serious, and for which sufficient information exists to permit an estimate of their economic value.
  
  
• Class Two impacts. These are impacts which are potentially serious, but for which there is insufficient data to permit an economic assessment of external costs within an acceptable range of certainty.
  
  
• Class Three impacts, these are impacts which, on balance of evidence and probability, are not likely to be highly material in relation to other impacts and, therefore, no attempt is made to quantify them in economic terms; alternatively, the environmental costs of these impacts have already been substantially internalised.

Environmental impacts arise at most stages in the fuel cycles. These impacts are subject to varying levels of management and attempts to ameliorate them; in other words, some of these impacts are already fully or partially internalised into the pricing structure of electricity. In considering the coal and nuclear fuel cycles, 10 potentially important impacts have been considered, of which five have been categorised as Class One impacts for present purposes. These are summarised in Table 5.

Numerous other impacts arise from coal-fired generation: for example, the health impacts of electromagnetic fields around high-voltage transmission lines, the loss of productive land above underground mines due to subsidence and the aesthetic impacts of large power stations and transmission lines in rural areas. While any number of such impacts may have significant environmental or social impacts in their specific or local contexts, they have not been considered in any detail in this study, either because they are not significant in aggregate on a national scale, or because the costs of those impacts have already been substantially internalised.

TABLE 5 Summary of potentially significant environmental impacts and their classification in this study

Space constraints preclude any discussion of impacts other than those in the Class One category. For information about the state of knowledge regarding Class Two and Class Three impacts, reference can be made to Van Horen (1996a). The sections which follow describe the nature and extent of externalities for the five Class One impacts noted above.

Occupational health effects in coalmining

Workers in coal mines are exposed to a number of risks. These include rockfalls, methane explosions, transport accidents and accidents in the handling of materials, which may result in immediate injury or death. A second category of occupational risks results from prolonged exposure of workers to air pollution resulting from mining activities: although the illnesses which occur are often significant, insufficient data exists regarding their scale and so this has been classified as a Class Two externality.

The Leon Commission of Enquiry reported in mid-1995 on the state of health and safety on South Africa’s gold and coalmines. The general tone of the commission’s report was sharply critical of the health and safety management in the country’s gold and coalmines, and highlighted the high morbidity and mortality rates compared to other major mining countries. This was more especially the case in the gold mines, which generally operate at much deeper levels than coal mines.¹¹

Although there is no publicly available information on the rate of injury and death on coalmines supplying Eskom, this has been estimated with reference to data reported by the Leon Commission. The details of this calculation are described more fully in Van Horen (1996a), but the essence is that accident rates attributable to coalmines supplying Eskom are probably lower than industry averages, because the proportion of coal derived from open-cast mines (which have a lower accident rate than underground mines) is higher for Eskom than for the industry as a whole.

The calculated fatality and injury rates for coal mines serving Eskom’s power stations are that an average of 0,30 fatalities and 1,68 injuries occur for every million tonnes of coal mined for purposes of power generation. Based on average coal consumption (from Eskom 1995a:54), this translates into 0,156 fatalities and 0,874 injuries per thousand GWh of electricity produced. If this rate is appliced to 1994 electricity production of 149 443 GWh, then a total of 23 workers would have died in coalmines supplying Eskom in that year, and a further 131 would have been injured.

The nature of injuries sustained by coalminers varies widely, from relatively minor injuries to permanently disabling ones. Their costs include the costs of medical treatment and the opportunity costs of not working. These have been estimated based on discussions with the Government Mining Engineer.

Based on the above estimates and using the valuation data described earlier, the valuation estimates are shown in Table 6.

TABLE 6 Valuation estimates for mortality and morbidity on coal mines, 1994, in total and in c/kWh

As noted above, an important category of occupational morbidity has not been included in the above valuation estimates, namely, the occurrence of respiratory and other diseases resulting from prolonged exposure of coalminers to high levels of dust and other airborne particulate matter. The effect of this omission will be to understate actual externality values.

Water consumption in power generation

Coal-powered electricity generation accounts for about 3% of total water consumption in South Africa (Roome 1995). Water is an integral part of the thermal power generation process, being used not only (as steam) to drive the turbines which generate electricity, but also to cool down the steam in large cooling towers. Most of Eskom’s power stations utilise conventional wet-cooling processes, although two—Kendal and Marimba—use dry-cooling processes which were pioneered by Eskom.¹² Average (net) water consumption in 1994 for all of Eskom’s coal power stations was 1,43 litres per kWh generated; the two dry-cooled power stations consumed just over 10% of this (Fraser 1995). Although water prices varied widely from one power station to another, the average in 1994 was R0,66 per m3.

¹¹ Ironically, the release of the Leon Commission report coincided with a serious accident at the Vaal Reefs Gold Mine, in which 104 miners were killed when their lift-cage dropped hundreds of metres down a mine shaft.

Water costs represent a small fraction of Eskom’s total operating costs: in the region of 1,8% of direct operating costs (excluding depreciation and finance charges) (calculated from Eskom 1995a:41). This, however, understates the importance of water as an input into the electricity generation process—it is an essential raw material, underlined by the fact that Eskom was directly involved in the construction or financing of several dams, long before other water consumers demanded water in those areas.

As in the case of coal, Eskom has benefited from water prices which have been low and stable over time. There is a wide range of pricing contracts in place with respect to Eskom’s water purchases, each dependent upon the specific source of supply. In general, Eskom purchases its water from the Department of Water Affairs and Forestry and pays on the basis of the historic costs of supplying that water, as distinct from the marginal cost of supplying water. Since the cost of supplying water is dependent primarily on the capital costs of constructing the necessary infrastructure, the use of historic costs for water pricing leads to low prices, especially where the capital infrastructure was constructed some time ago. Lethabo power station, for instance, draws its water from the Vaal River which, was dammed in the early 1900s, hence its water price (R0,12 per m3 in 1994) was very low.

A review of the national pricing policy for bulk water supply was commenced in 1995, with a view to developing a policy which provides adequate signals to water consumers regarding the economic cost of water (Department of Water Affairs and Forestry 1995). By the end of 1995, no firm estimates had been made regarding marginal costing scenarios for water supply, although several values have been presented. One benchmark which is informing the policy debate is the cost of supplying water to the Highveld from the Lesotho Highlands Water Scheme. Again, some uncertainty exists regarding the economic cost of that water, especially because the feasibility studies are several years old; however, a value of Rl,50 per m3 has commonly been quoted (Roome 1995). Some estimates of the economic cost of supplying additional water on the Highveld are higher, at about R3,00 per m3, although there has been little analysis to underpin those estimates (ibid).

For purposes of this study, the central estimate of the economic value of water supplies to the power stations is Rl,50 per m3, with low and high values of Rl,20 and Rl,80 respectively. The external costs implicit in these water prices, expressed in terms total rand and c/kWh of electricity generated, are shown in Table 7.

¹² In these power stations, the cooling towers utilise massive fans to blow air onto hot water pipes to cool them down.

TABLE 7 Valuation estimates for water consumption external effects

Air pollution from power generation: health impacts

The evaluation of the health impacts from power station emissions is one of the more complex but important externalities to be considered. Quantification of health impacts requires information for four steps in the impact pathway:

• the quantities of pollution emitted by power stations;
  
  
• the dispersion and ultimate deposition of those pollutants;
  
  
• the responsiveness of human health to various exposures (doses) of pollution; and
  
  
• the valuation of increased morbidity and mortality.

Space limitations do not allow for data for each of these steps to be described in full here (refer to Van Horen 1996a for details); instead, only pertinent considerations are outlined.

For purpose of quantifying externalities in this study, a computer modelling tool called EXMOD was used. EXMOD follows the damage function approach as outlined above. Data regarding each of the above four categories was collected and used to model the actual damages from air pollution emissions. EXMOD was developed over the period 1993 to early-1995 for the New York State environmental externalities cost study, with support from the Empire State Electric Energy Research Corporation, the New York State Energy Research and Development Authority, the New York Department of Public Service and the Electric Power Research Institute. The aim of that study was to develop a user-friendly damage function tool with which impacts of new or relicensed electricity supply and demand management options could be evaluated. The work in the New York study was summarised in four comprehensive volumes (Rowe et al 1993,1994, Bernow et al 1995a, 1995b),¹³ as well as a shorter paper (Rowe et al 1995). Following a review of international externalities studies, and particularly of models which could be of potential use in the South African context, EXMOD was investigated and selected for use in this study.

With regard to the quantity of pollution emitted, data on actual emissions for 1994 was used for this study. It is worth noting briefly that the coal used in South Africa’s power stations is generally of poor quality, since the highest grade coals are exported. The average sulphur content of coals used by Eskom is relatively low at around 1 %, whereas ash content is high, ranging from 21% at Arnot power station to 39% at Lethabo. The latter also has the lowest calorific value (energy content), at 15,2 megajoules per kilogram of coal. It is significant to note that with such a low energy content, “coal” could not be used in any conventional commercial or domestic process.¹⁴ Thus, electricity is being generated from a product which would otherwise have little or no economic value.

¹³All of these reports will be published in a two volume book by Oceana Press, New York State in 1996.

The negative side of this is that, all other things being equal, particulate emissions and ash production are relatively high. As a consequence, Eskom’s pollution control policy is focused mainly on the reduction of particulate emissions. All of its operational coal power stations utilise electrostatic precipitators, which operate with an efficiency of around 90 to 99,7% (Tilley and Keir 1994), although it should be noted that it is the finest particles (that is, with the smallest diameter) which present the greatest potential health hazard, since it is these which are small enough to be respirable.¹⁵ In addition to electrostatic precipitators, Eskom has installed bag filters on a trial basis at some of its power stations. Where these are used successfully in combination with electrostatic precipitators, efficiencies of 99,99% are possible (Hanson 1992).

Furthermore, Eskom’s newer power stations have tall chimney stacks so that emissions penetrate the inversion layer and are released into the upper atmosphere. This inversion layer is an important feature of atmospheric conditions in the Mpumalanga Highveld because it inhibits the dispersal of ground-level or low-level emissions, especially during winter months (Tyson et al 1988). All of Eskom’s power stations have relatively tall chimney stacks of 200 metres or more, except Arnot (193 metres) and Hendrina (110 metres).

A final point to note in relation to Eskom’s air pollution policy is that it has decided that desulphurisation and denitrification technologies are not warranted. Therefore, no active measures are taken to reduce the emissions of sulphur dioxide or nitrogen oxides. This is an issue which has received much attention in recent years, although there has been no systematic investigation of the costs and benefits of respective pollution control options.

Total emissions from the nine operational coal power stations in 1994 were as follows: total suspended particulates, 122,42 kilotonnes (equivalent to 0,84 kg/MWh); for sulphur dioxide, 1 166,7 kilotonnes (7,88 kg/MWh); and for nitrogen oxides, 960,9 kilotonnes (6,49 kg/MWh) (Eskom Generation Group 1995).

It is important to note that this data excludes pollution originating from the ash dumps at power stations. Although Eskom has an extensive programme of ash management, it produces an enormous quantity of ash: some 22 million tonnes in 1994, of which about 3% was reused for cement or brick-making (Eskom 1995b:23). This translates into a continuous production rate of 42 tonnes of ash per minute.

The second step in the damage function approach concerns what happens to those pollutants after they are emitted, that is, how they are dispersed in the atmosphere and where they are deposited. For this to be assessed, information was assembled regarding physical emission characteristics, such as the height of chimney stacks and the speed, volume and temperature of flue gas emissions, as well as atmospheric conditions, including wind patterns (derived from long-term data), mixing heights and atmospheric stability.

¹⁴ Indeed, this coal cannot even be ignited with a blowtorch!

¹⁵ Particulate matter with a diameter of 10 mm (microns) or less is usually regarded as being in the respirable range (hence the label PM10).

A reasonably significant amount of air quality research has been undertaken by Eskom in the Mpumalanga Highveld since the mid-1980s. This cannot be reviewed in detail here (refer to Van Horen 1996a for this), save to mention that the air quality data collected suggests that the pollutants which exceed or approach health guidelines most frequently are ozone and sulphur dioxide. Of less concern are concentrations of particulates and nitrogen oxides, which are reported to be well within government guidelines.

Dose-response relationships constitute the third step in the impact pathway, namely, the link between ambient pollution exposures and health outcomes: in this case, respiratory illnesses. The human health effects of pollution exposures have been widely studied in a number of countries in response to a range of environmental conditions. In South Africa, there have been relatively few studies of the health effects of air pollution, and although a handful of studies have attempted to find correlations between environmental quality—mainly particulate concentrations—and health outcomes—mainly respiratory illnesses (see, for example, Terblanche et al 1992,1993)—there have been no studies which have quantified the dose-response function for pollution exposures.¹⁶

Reference has, therefore, been made to the international literature to identify dose-response functions which might be applicable to South Africa. For purposes of the project in New York State, USA, which developed the EXMOD model, extensive reviews were undertaken of epidemiological and bio-medical literature in order to derive dose-response functions which could be used with a satisfactory degree of certainty (Rowe et al 1994: chapter V). A strict set of criteria were applied in the selection of epidemiological studies for that purpose.¹⁷ On the basis of a large number of studies which followed similar methodologies and were reasonably comparable, dose-response functions were derived for a range of air pollutants.

Again, space constraints do not allow for these relationships to be described in full here, but to illustrate the nature of these relationships it has been estimated from epidemiological studies that one in 3,3 million people will die for every 1mg per m3 increase in ozone concentrations. Put differently, if 40 million South Africans are exposed to an additional 1mg per m3 of ozone pollution, the central estimate implied by this risk factor is that 12 people will die prematurely each year. Dose-response functions were used for several health outcomes, such as asthma attacks, acute and chronic bronchitis, visits to doctors and hospitals for respiratory symptoms, mortality, hospital admissions for respiratory ailments and days away from work due to illness. In addition, for each dose-response function, low, central and high values have been used to reflect uncertainty.

There is an unavoidable measure of uncertainty in applying dose-response relationship data which was derived in North America to South Africa. Clearly, environmental and health characteristics differ in these two environments. At the most basic level, human physiology and responses to environmental conditions do not differ according to national boundaries. A factor such as lower nutritional status, for instance, means that South Africans may have lower resistance to environmental hazards and, therefore, be more susceptible to illness (Terblanche 1995). The bias which this factor introduces means that the selected dose-response relationships would tend to understate the actual health outcomes in South Africa. Other factors would also affect these functions, and so the overall direction of bias is not clear.

¹⁶ See Van Horen (1994) or Lerer (1995) for a review and analysis of South African and international studies at the energy-health interface.

¹⁷ These are described in detail in Rowe et al (1994: V-3 to V-4).

Another, more technical, uncertainty concerns assumptions about threshold levels of pollution. Although environmental standards are traditionally based on an assumption that there is a level of pollution below which health effects can be safely assumed not to exist, recent epidemiological evidence suggests that such thresholds do not exist. Consequently, this study has used a zero threshold level.¹⁸

The valuation of health effects resulting from air pollution exposures is the final step in the impact pathway analysis. Valuation was undertaken using the EXMOD modelling tool, with data on the following items:

• Eskom’s nine coal power stations, including their fuel types and composition;
  
  
• demographic data for each magisterial district in South Africa from 1991 census data, including total population, age distribution, land area, geographical coordinates and the average altitude of the district;
  
  
• aggregated demographic data for the following neighbouring countries: Lesotho, Swaziland, Mozambique, Zimbabwe, Botswana and Namibia;
  
  
• long-term (10 to 20-year average) surface wind data for 15 monitoring stations in South Africa (and two in Namibia);
  
  
• emissions data for each of Eskom’s nine operational power stations for the main air pollutants: particulates, sulphur dioxide and nitrogen oxides;
  
  
• dose-response data for the health outcomes described earlier; and
  
  
• valuation data for health outcomes, based on opportunity costs of health effects.

The EXMOD model incorporates three air dispersion models, covering short-range, medium-range and long-range transport of pollutants.¹⁹ Consequently, the model was run nine times, corresponding to data sets for each of the nine operational coal power stations, and using the above data categories.

The results of the model computations include both the physical health outcomes (number of cases of asthma attack, chronic bronchitis, and so on) and the economic costs of those effects. The incidence of certain of these health outcomes is relatively high: notably, asthma attacks, respiratory symptom days and days of restricted activity. Also significant is the number of mortalities which are expected to occur each year: 174 in the central estimate (EXMOD computations). This is not insignificant, even in relation to the high population on the Highveld within a few hundred kilometres of the power stations.

¹⁸ See Van Horen (1996a) for a more detailed discussion of this assumption.

¹⁹ There is inherent uncertainty in using these dispersion models on the Highveld because of its particular atmospheric conditions, which differ from those elsewhere. However, there are no other models available specific to Highveld conditions.

The results of applying opportunity cost valuation data to the physical health outcomes are shown in Table 8. This table shows the range of estimates, in cents per kWh, of morbidity and mortality resulting from emissions of sulphur dioxide, nitrogen oxides and particulate matter.

The valuation results for the nine coal power stations are of a similar order of magnitude: the lowest damages arise from Matimba power station with a central estimate of 0,23 c/kWh, due to the fact that it is located in a less-densely populated area in Northern Province; the highest damages arise from Kriel and Hendrina, probably because their nitrogen oxide and particulate emission factors, respectively, were relatively high (based on 1994 emissions).²⁰ Based on actual electricity generated during 1994, the weighted average cost of damages from air pollution was calculated as 0,54 c/kWh for the central estimate, with a low of 0,39 c/kWh and a high of 0,67 c/kWh. It should be noted, once again, that this excludes the Class Two damages which have not been quantified, such as damages resulting from acidic deposition or impaired visibility. Thus, there is certainty over the direction of at least one source of bias in these estimates, namely, that the estimates will understate some external effects.

TABLE 8 Valuation of morbidity and mortality from power station emissions in c/kWh, 1994 (EXMOD computations)

²⁰ Nitrogen oxide emissions impact upon human health more particularly through the formation of ozone.

Greenhouse gas emissions from power generation

Electricity generation, where it is based on coal power, is unavoidably a significant source²¹ of greenhouse gas emissions. The principal greenhouse gas emissions are carbon dioxide (CO₂), methane, chlorofluorocarbons and nitrous oxides, the first two of which are most significant in South Africa. South Africa was responsible for about 1,2% of global greenhouse gas emissions in 1988, making it the 18th largest source in the world, and one of the largest sources on a per capita basis (Van Horen and Simmonds 1995). It was also the largest source of greenhouse gases in Africa, accounting for 15% of the continent’s CO₂ emissions.

Significant quantities of carbon dioxide are emitted by the electricity generation industry, and smaller amounts of methane during coalmining. Eskom is the single largest source of greenhouse gases in South Africa which, by its calculations, amounted to some 142,9 million tonnes of carbon dioxide in 1994 (Eskom 1995b).²²

There is a vast body of international literature which has sprung up around the climate change phenomenon, addressing both its physical and political-economic dimensions. It is impossible to summarise all aspects of this issue here. For present purposes, it is important to note that there is considerable uncertainty around climate change in various of its dimensions. One of the key uncertainties is around the potential impacts of climate change on specific subregions, such as southern Africa. This uncertainty, exacerbated by the extremely long time periods over which it might occur,²³ makes it extremely difficult to make assessments of the economic and social costs of possible climate change. These may include, for example, the costs of possible increased drought in the future, and more frequent occurrences of extreme weather events (storms, floods and droughts). Several attempts have been made internationally to estimate the range of damage costs which might result from climate change (for example, Cline 1992, Fankhauser 1992, Nordhaus 1993): not surprisingly, these have produced very different estimates and have attracted their share of criticism. This uncertainty presents special difficulties for the present study.

What is certain, however, is that the climate change issue will not disappear from the international political economy in the near future. Given the prominent role of South Africa among developing countries, it is essential that the issue is not ignored.

²¹ Briefly, greenhouse gases are relevant in so far as they are widely believed to enhance the naturally occurring greenhouse effect—in terms of which greenhouse gases increase the ability of the earth’s atmosphere to retain warmth. The balance of scientific opinion suggests that continued emission of greenhouse gases at present rates will lead to global climate change with variable, but often negative, consequences in many regions of the world. For more details on recent developments in the international climate change debate, see Rowlands (1995a), and on the South African energy sector’s contribution, see Van Horen and Simmonds (1995).

²² The power sector is also indirectly responsible for the emission of methane from coal mines; however, these are not included in the present analysis for lack of data.

²³ With concomitant importance attached to the selection of a discount rate when economic effects are being considered.

One of the most recent economic studies of climate change damages has been done by working group III of the Intergovernmental Panel on Climate Change (IPCC), the body of leading social and natural scientists informing the international negotiation process. The analysis undertaken by this group has suggested that the annual costs of global warming will be in the region of 1,5% to 2% of gross world product by the time CO₂ concentrations reach double their natural levels—somewhere around 2050 or 2060 on the basis of current trends.²⁴

By relating global damage costs to current emissions of greenhouse gases, it is possible to estimate a cost per tonne of greenhouse gas or, more particularly, CO₂ emissions. Corresponding with the high level of uncertainty over future damage costs, there is an equally wide range of “per ton” damage cost scenarios. The IPCC, in its 1995 second assessment report, “does not endorse any particular range of values for the marginal damage of CO₂ emissions”, but instead referred to published estimates which fall in the range of $5 to $125 per tonne (IPCC 1995). One such estimate, by Fankhauser and Pearce (1993), both of whom were centrally involved in this aspect of the IPCC working group III’s work, amounted to 14 pound per tonne of CO₂, equivalent to R80 or $22 per tonne (reported in Pearce 1995: 31).²⁵ This value was used as the central estimate in this study. For the low estimate, the lowest value referred to in the IPCC report will be used: $5 per tonne, equivalent to R18 per tonne. The choice of a high value is made more complicated by the wide range of estimates which have been published. For present purposes, the high estimate of damages has been taken as being 50% higher than the central estimate, that is, R120 per tonne (or nearly $33 per tonne). On this basis, the estimates of damages from South African power station emissions of greenhouse gases are shown in Table 9.

TABLE 9 Valuation estimates for CO₂ -induced climate change damages

Fiscal externalities in the nuclear industry

A final category of externalities which arises in the South African electricity generation industry are those subsidies which have flowed to the nuclear industry from public funds. This may be termed a “fiscal externality”—a fiscal transfer payment (for example, a subsidy) which is made to the nuclear industry and which is not reflected in the price of nuclear electricity (Lockwood 1992). While this is not an environmental externality per se, it is a potentially material externality in South Africa which could be significant in relation to the price of electricity, and so it is included in the scope of this study.

²⁴ Notably, this analysis has been criticised by some economists as being far too conservative, with alternative estimates of global costs being in the range of 12% to 130% of gross world product by the year 2050 (Meyer and Cooper 1995).

²⁵ An exchange rate of R5,68 to the UK pound is used.

Unlike the coal-fired electricity sector, which has historically received little or no financial subsidy from public funds, the local nuclear industry has enjoyed a very privileged position in this respect. This is similar to the nuclear industry in the UK and elsewhere (Lockwood 1992).

The nuclear industry has received an average of 69,3% of the Department of Mineral and Energy Affairs annual parliamentary grant over the period 1971-72 to 1995-96 (calculated from Auf der Heyde 1993). The total allocation over this period was R8 528 million in nominal terms; when these amounts are adjusted to real 1995 rand by the producer price index (from Central Statistical Service 1995), the total allocation amounts to some R21 753 million.

Thus, considerable resources have been directed to the local nuclear industry. These have been directed to three main categories of expenditure: capital expenditure, operating expenditure and servicing and repayment of loans (ibid:7). Not all of these amounts should be attributed to the nuclear generation industry, since non-electricity aspects of the industry have also benefited from state subsidies: notably the nuclear bomb programme, research and development in non-electric areas, and the production of medical isotopes. Unfortunately there is no publicly available information on the allocation of the subsidy to these various sectors; consequently, calculations and assumptions have been made on the basis of available information. To be conservative, only those costs which are known to be related to electricity generation have been included in the analysis; thus these represent a minimum estimate of the fiscal subsidy to the industry and probably underestimate actual costs. The details of this calculation are provided in Van Horen (1996a). The estimated portion of the total allocation which can be attributed to the electricity industry is R12 298 million (in 1995 rand) for the period 1971-72 to 1995-96, or 57% of the total grant.

In order to calculate an average external cost, three scenarios have been used for purposes of spreading the fiscal subsidy over the lifetime of the assets to which it relates:

• low estimate: the subsidy ceases from 1996-97 onwards and the power station operates at current capacity until 2023;
  
  
• central estimate: the subsidy is phased out from its current level to zero by the year 2000,²⁶ while operations remain at their current level; and
  
  
• high estimate: support for the industry terminates and production of electricity ceases at the end of 1996.

Table 10 summarises the external costs for each of these three scenarios. Electricity generation figures are based on actual production since the power station was commissioned, and assumed output for the remainder of its 40-year life at 1995 levels.

²⁶ Thus, the allocation to the electricity component of the industry is assumed to be 57% of the annual allocations: R489,2 million for 1996-97, R366,9 million for 1997-98, R244,6 million for 1998-99, and R122,3 million in 1999-2000 (all in 1995 rand).

TABLE 10 Valuation of fiscal externalities in the nuclear industry

It should also be noted that the central scenario is consistent with the Atomic Energy Corporation’s so-called “2000 plus” business plan. However, the trend in the 1995-96 and 1996-97 budget allocations does not appear to be in line with this plan; in fact, the subsidy has not been reduced by as much as envisaged, which means the central estimate may understate the amount of the fiscal externality.

Summary of results and limitations

The values for the five Class One externalities are summarised in Table 11. It should be noted that the valuations in the table are not additive, since they do not have a common base: they are expressed in relation to the amount of coal and nuclear electricity generated, respectively.

TABLE 11 Summary of valuation results for Class One externalities in c/kWh, 1994

Notes

¹The external costs of coalminers’ morbidity (chronic and acute illness) have not been quantified because insufficient information exists regarding their pollution exposures.

²The external costs of health impacts caused by pollution originating from ash dumps on the power stations have not been quantified because insufficient information exists regarding the quantity of emissions and their dispersal.

Summarising the Class One externalities in Table 11, totals can be derived for the coal and nuclear cycles, and for an average of both. These results are shown in Table 12. The weighted average external cost takes into account the relative proportions of coal and nuclear electricity generated by Eskom.

TABLE 12 Summary of externality valuations for coal and nuclear cycles in c/kWh, 1994

The central estimate for all Class One externalities included in this study is, therefore, 8,11 c/kWh, with a lower bound of 2,29 c/kWh and an upper estimate of 12,34 c/kWh. It is evident from the above tables that the external costs of the nuclear cycle are high on a per unit basis. Taking only the externalities in the coal fuel cycle, the resultant cost is 8,41 c/kWh for the central estimate, 2,23 c/kWh for the low estimate and 12,45 c/kWh for the high estimate. Furthermore, if damages attributable to greenhouse gas emissions are removed altogether, and fiscal externalities in the nuclear cycle are also ignored, the central estimate is 0,69 c/kWh, with low and high ranges of 0,49 c/kWh and 0,86 c/kWh respectively.

These results can be placed into context by comparing them with current electricity price levels. The relative significance of the externalities obviously varies depending on the choice of a benchmark tariff; for present purposes, Eskom’s weighted average tariff for all consumers is used. Table 13 makes this comparison, with a breakdown for each of the five Class One externalities.

TABLE 13 Average Class One externalities as a percent of Eskom’s average tariffs, 1994

Because nuclear electricity accounted for just over 6% of the coal-nuclear total, its relative impact on electricity prices becomes much smaller than when compared to nuclear-generated electricity alone. Much more significant in this comparison is the impact of assumed damage costs from climate change where these are anything other than negligible on a per tonne basis. Of the environmental externalities experienced within the country and its neighbours in the relatively short-term, health impacts of air emissions are most significant, representing between 3% and 5% of current electricity prices.

It is worth making a brief comparison of the results of this analysis with the results of externality studies undertaken elsewhere—as summarised in Table 3. The valuations in this study fall somewhere between the two main sets of externality values derived internationally—perhaps closer to the lower, more recent bottom-up studies in the second half of Table 3. This is consistent with expectations: on the one hand, it would be expected that the external costs in South Africa might be higher than those in Europe or North America where emissions standards and environmental controls are generally much stricter; on the other hand, this study excluded many externalities for which there was insufficient data, which would be expected to result in lower valuation results here. In conclusion, the results of the present study, albeit subject to a number of limitations which are outlined below, appear plausible on the basis of international experience.

Limitations and weaknesses of the study

It is important in a quantitative analysis such as this to stress that the calculated figures are as good only as the input assumptions and information from which they are derived. By its nature, any externalities study is subject to a number of limitations and weaknesses, which means that the economic values should not be taken simply at face value. Rather, they should be analysed in full awareness of the limitations of the study, of which there are three main categories: omissions, uncertainties and biases.

Firstly, in the case of omissions, Class Two and Class Three impacts were not quantified in the present study. Class Two impacts included the following:

• chronic and acute illnesses experienced by workers on coalmines supplying Eskom;
  
  
• impacts of air and water pollution emitted by coalmines supplying the power stations;
  
  
• impacts on human health of air pollution originating from coal power stations’ ash dumps;
  
  
• impacts of coal power station emissions and resultant acidic deposition, in terms of human health, damages to crops, forests, water supplies and other physical assets in the Mpumalanga Highveld and neighbouring regions;
  
  
• impacts of coal power station emissions on visibility conditions, particularly in the Mpumalanga Highveld;
  
  
• impacts of coal power station emissions into watercourses; and
  
  
• impacts of nuclear power stations on environmental quality and human health.

It is possible that the economic value of some of these externalities will be significant and they, therefore, warrant further investigation. In addition to these issues, it is possible that some of the externalities which were accorded a Class Three rating in this study could have significant economic values. Ideally, all of these effects should be subjected to a more comprehensive analysis.

With respect to uncertainties in the present analysis, Table 11 summarised the level of uncertainty for each of the Class One impacts, at a very broad level. The most significant areas of uncertainty include the following:

• in the case of injuries and deaths occurring in the coalmining sector, data for the coalmining industry as a whole had to be apportioned between the main consumers, since data was not available specifically for the coalmines supplying Eskom. The effect of this may have been either to understate or overstate the results;
  
  
• there is uncertainty over the applicability of dose-response functions derived in North America to South African populations. No epidemiological studies have yet derived these relationships for South Africa;
  
  
• with regard to the valuation of health impacts of air pollution, there is a moderate level of uncertainty regarding the atmospheric modelling approach used. The EXMOD model used a Gaussian plume type of dispersion model to approximate the dispersion of emissions from power station chimneys, whereas actual conditions on the Highveld are not especially well-represented by this kind of model (Turner 1995). In the absence of any atmospheric model designed specifically for South African conditions, however, this level of uncertainty is unavoidable;
  
  
• there is a high level of uncertainty regarding the future global impacts of anthropogenic greenhouse gas emissions in the economic, social and environmental spheres; and
  
  
• there has been very little previous analysis of the economic value of environmental and health issues, from the pricing of water to the value of human health and mortality and, consequently, there is a high level of uncertainty in this respect.

These uncertainties have been accommodated, to an extent, by utilising a range of estimates rather than a single one so as to reflect the inherent uncertainty. In most of the cases above, there would be benefits to undertaking further investigation to narrow the range of uncertainty.

It is important to be explicit about the effect of these omissions and uncertainties on the economic valuations which have been reported in this study, that is, whether the direction of the resultant bias will be to understate or overstate the externality valuations. A summary of these potential biases is shown in Table 14.

TABLE 14 Summary of potential biases in this study due to omissions and uncertainties

The overall effect of these biases is difficult to assess since any one of them (for example, greenhouse gas impacts) could be large enough to more than offset all others. Nonetheless, it is important to note that there is one definite source of bias, namely, the omission of Class Two and Class Three impacts, which would cause estimates to understate actual impacts. In all cases in this study the chosen route has been to err on the side of understating external effects rather than the opposite. Thus, there is a fairly high level of confidence that the range of quantified externalities does not overstate the minimum value of externalities in the South African power sector.

Implications for South Africa’s development policy

In effect, the analysis in this chapter represents a first (and by no means a conclusive) attempt to value the external costs of electricity generation. This is useful for several purposes. Firstly, it represents a baseline against which comparisons can be made in future. If, for instance, environmental policies are introduced with a view to reducing air pollution concentrations, progress can be evaluated in the future in economic terms by comparing them with the 1994 baseline described here.

Secondly, the valuation of external costs is a necessary component of any cost-benefit analyses of abatement options which may be considered. Thus, for instance, an assessment of the relative costs and benefits of utilising technologies with lower sulphur emissions (such as scrubbers or fluidised bed combustion), should include, on the one hand, the costs of those technologies and, on the other hand, the benefits which they will bring about. Their environmental benefits will, simplistically, be equal to the avoided external costs. To the extent that this study estimated the value of these externalities, it provides one of the basic building blocks for such a cost-benefit analysis.

Thirdly, quantification of external costs is an essential component of integrated resource planning. Integrated resource planning is a planning approach in which all energy supply and demand options are evaluated with a view to making the resource choice which is optimal from the view of society as a whole.²⁷ Typically, integrated resource planning assessments compare energy supply investments (such as building another coal power station, or importing hydroelectricity) with demand-side management options and seek to make the comparison on an equal footing. Since the private costs and external costs of these options vary widely, it is important to include all costs in such comparisons. Therefore, when Eskom makes its next decision about how to satisfy the country’s electricity needs, it should do this on the basis of full cost comparisons, using a comparable methodology to that in this study.

²⁷ Integrated resource planning is widely used in North America, where electricity regulators stipulate that utilities have to undertake an integrated resource planning assessment of all the options they face for meeting demand. These assessments typically include quantification of external costs.

²⁸The philosophy of China and certain newly industrialised countries in South-East Asia is sometimes characterised in this way.

Eskom’s vision is to “provide the world’s lowest-cost electricity for growth and prosperity” (Eskom 1995: i). The results of this study suggest that this vision, by itself, might not best serve the country’s interests. To illustrate, it would be quite possible for South Africa to be the world’s cheapest electricity producer if it paid no regard whatsoever to environmental conditions and polluted the environment freely.²⁸ In essence, such a situation would amount to a subsidy being granted to the industry by the environment—in both its natural and social dimensions. This kind of subsidy is just as unsustainable as the large fiscal subsidies which many utilities in Latin America and Africa received from their governments, notably during the 1970s and 1980s.

It is obvious that Eskom does not act in single-minded pursuit of low-cost electricity, without regard for the environmental consequences of its actions. It has invested in technologies to reduce its impact on its surroundings (for example, electrostatic precipitators), and it incurs ongoing expenditure on the management of its impacts. It reported in its first environmental report that approximately R135 million had been spent on environmental management and research during 1994 (Eskom 1995b: 35). Over half of this was on air quality management in the generation division (Roos 1995).

The relevant question, therefore, relates to the degree to which environmental considerations are taken on board in the pursuit of cheap electricity. Clearly, there are external costs associated with producing electricity, and “artificially” low prices provide a powerful but dangerous signal to electricity consumers. It is not difficult to imagine a scenario in which major electricity-intensive industries establish themselves in South Africa on the basis of low electricity prices (all other things being equal), but that some years down the line the environmental costs become too large to ignore. At that point, the clean-up costs and control costs could be significant and would probably translate into an unavoidable price shock. Thus, Eskom’s original vision—to be the lowest-cost producer in the world—would be severely compromised, with potentially adverse economic effects. It is important, therefore, that Eskom’s vision be tempered by its stewardship responsibilities to the natural and social environments.

This study has focused almost exclusively on the supply-side—in other words, on the generation of electricity. While this focus has been deliberate, it is important to flag the demand-side of the industry and, in particular, the possible benefits from implementing energy efficiency and demand-side management policies. Clearly, policy responses aimed at internalising or managing externalities will bring about upward pressure on electricity prices. If, however, energy efficiency interventions are adopted at the same time, the potential exists to offset any such price increases and, indeed, sustain current downward trends.

Internationally, electricity utilities have invested in energy efficiency and demand-side management programmes over the past decade or more, partly in response to public and regulatory pressures to reduce the environmental costs of generating electricity from coal and other fossil fuels. In South Africa, there has been relatively little attention paid to the demand side of the equation, partly as a result of the overcapacity situation in Eskom’s coal power stations. Eskom has recently established a residential demand-side management programme, although it remains a relatively low investment priority for the organisation. Nonetheless, indications are that considerable energy savings can be achieved, at low or negative cost, through the adoption of well-proven technologies and practices. For example, considerable savings can be made in the household sector by improving the thermal performance of low-cost houses which are being connected to the electricity grid. While this is not the place to discuss the large number of energy efficiency and demand-side management options which can be readily implemented, it is important that policy interventions dealing with externalities are part of an integrated approach which balances cost pressures on the supply-side with potential gains on the demand-side.

Policy implications related to specific external effects

This chapter has included quantification of five external effects, and policy responses related to each are outlined briefly below.

Injuries and mortalities in the coalmines

The safety performance of South African mines in general, including coalmines, was heavily criticised by the Leon Commission. It was the commission’s view that the number of injuries and fatalities was “unacceptably high” (1995:16).

It is not necessary to repeat here the range of recommendations made by the Leon Commission, nor to describe actions taken recently by government. The relevant point for present purposes is to note that more attention to worker health and safety in the coalmines, whether this is driven by stricter regulations, union demands or management initiatives, should have the effect of reducing accident rates. To the extent that this results in incremental costs being incurred by the industry, it is likely that these will be passed on to consumers, meaning ultimately that there would be upward pressure on coal and electricity prices. In economic terms, this is entirely correct: consumers should be aware of, and should pay for, the full implications of their consumption behaviour.

The pricing of water consumed in power stations

The country’s bulk water pricing policy is currently under review (Department of Water Affairs and Forestry 1995), and it is not the intention of this study to pre-empt the results of that process. Nonetheless, it is incontrovertible that the historical cost of supplying water to the power stations bears little relation to the current economic value of that water. As such, Eskom is not being given the correct price signals regarding its consumption patterns and, as a result, its technology choices do not necessarily take account of water scarcity. Only two of its coal power stations, Kendal and Matimba, employ dry cooling technologies.

The appropriate policy is for bulk water supply authorities—generally the Department of Water Affairs and Forestry—to move away from historic cost pricing towards economic pricing of water. This applies both to new supply facilities and to ongoing water consumption in existing power stations. With respect to the former, marginal cost pricing should be implicit in the planning and feasibility studies for any new power stations which are under consideration. In the case of existing facilities, a phased introduction of economic prices should be agreed upon in advance, so that unexpected price effects can be minimised as far as possible. The increase of estimated economic water prices over current prices amounts to about 127% in the central estimate: the doubling of water prices would have a noticeable effect on electricity prices, and so it is important that new pricing policies are introduced in a well-planned and phased manner. It is important to note, however, that the main objective, ultimately, is for prices to be set at a sufficiently high level to reflect the scarcity of water, and so there is a limit to the concessions which can or should be made for water consumers. As a consumer which is guaranteed the most secure water supplies, even in times of drought, it is correct that Eskom (and its customers) should pay for that privilege.

Health effects of air pollution emissions

The analysis of the external costs of air pollution emissions on human health suggested that these costs are in the region of 0,54 c/kWh (range: 0,39 to 0,67 c/kWh). The central estimate represents approximately 4% of average electricity tariffs in 1994. Significantly, this estimate excludes a large number of externalities which may have significant costs, notably, the long-term effects of acidification on buildings, crops, forests and other objects at ground level.

There are numerous strategies which can be employed to address these external effects, including the setting of emissions limits, the specification of control technologies to be used, the introduction of pollution taxes or tradable emissions permits and self-regulation by polluters. It is beyond the scope of this study to analyse all these options and to propose suitable policies. Nonetheless, it is possible to make explicit some key implications of the analysis.

First, the order of magnitude of the health costs which currently occur are not insignificant and may merit consideration of abatement measures. Taking as a point of departure the current stock of coal power stations, the range of technical options for reducing gaseous and particulate emissions is relatively narrow. Commonly discussed in this regard are flue gas scrubbers, which could be retrofitted to existing power stations. There has not been any thorough public investigation to date of the economics of this technology option. Eskom’s position on this has been that the costs of retrofitting its power stations with scrubbers are prohibitively high, at anything from 33% to 49% of prices (King and Rodseth 1993: 20), and that with costs of this order of magnitude, it is more effective to invest in electrification in highly polluted townships (Lennon and Turner 1992:5).

Clearly, there are constraints on capital and investment resources which mean that investment decisions need to weigh up all alternatives, but it may be misleading to present air pollution abatement options as a direct trade-off between power station controls and electrification. Experience in recent years has shown that urban electrification does not lead to significant pollution reduction, and much less rural electrification (Eberhard and Van Horen 1995:166). The national electrification programme is certainly not motivated by environmental concerns.

In assessing whether pollution abatement technologies are justifiable in economic terms, two sides of the equation need to be considered. The first is the cost of abatement: Eskom’s analysis suggests that this would lead to increases in electricity prices in the region of 30% to 50%. This estimate is somewhat higher than experience elsewhere suggests: for example, Petrie et al (1992: 434) reported international experience of this cost penalty was in the region of 10 to 15%—therefore, taking this wider range, the cost penalty will be in the region of 10 to 50%.29 The second part of the equation is the benefit of abatement: this will consist primarily of the avoided environmental costs. The health costs quantified in this study represent 3% to 5% of electricity prices but, as noted several times already, this excludes several other impacts of air pollution. It could be expected that these costs would be reduced by up to 90% depending on the abatement technologies adopted; in addition, abatement benefits would also be felt in other areas not quantified here, such as reduced costs of acidification. It appears, therefore, from the orders of magnitude mentioned here, that the costs of abatement options do not necessarily outweigh the benefits by a large margin.

A range of technological options exist to reduce air pollution emissions: flue gas scrubbers are the most well-known of these, but are not necessarily the best option from an economic or environmental perspective. A more promising option is fluidised bed combustion: although this is a relatively young technology, life cycle cost analyses suggest that its performance is better than flue gas desulphurisation in most respects (Diekmann and Notten 1995). Possibly of great significance in South Africa’s dry conditions is that scrubbers require large quantities of water for ongoing operations, and in the context of increasing water scarcity and higher water prices this represents a serious constraint.

With regard to fiscal instruments such as pollution taxes or tradable emissions permits, it would make little sense from an economic or environmental perspective to impose such instruments only on Eskom’s power stations, without including other major sources of pollution. For such instruments to be adopted, an integrated policy is required in which all major pollution sources are included. Thus, in the absence of such a framework, this is not a feasible policy option at present.

Emission of greenhouse gases

The international governance of greenhouse gas emissions and mitigation of their effects is a rapidly evolving arena. The current situation is that South Africa, if and when it ratifies the Framework Convention on Climate Change, will not face any specific emission reduction targets or obligations.30 Thus, there is no immediate necessity for South Africa to introduce greenhouse gas mitigation measures at its own expense. A range of international funding sources—for example, the Global Environment Facility—exist to assist developing countries achieve reductions in their greenhouse gas emissions.

While it is clear from the international climate change debate that there is no question, for the present, of introducing a carbon tax or similar externality “adder” on South Africa’s carbon dioxide emissions, this does not mean, however, that South Africa can afford to ignore the climate change issue. South Africa was the 18th largest source of greenhouse gas emissions in 1988, and one of the largest on a percapita basis (Van Horen and Simmonds 1995). The energy sector is the single largest source of greenhouse gas in South Africa; therefore, it has some responsibility to consider the effect of emissions. Also, it is possible that middle-income developing countries such as South Africa will face stricter commitments at some point in the not too distant future (Rowlands 1995b). If this is the case, it would be imprudent to ignore the relative impacts of different electricity supply options in terms of their climate change implications.

²⁹Clearly, variations in the base cost of electricity (the denominator) will influence the amount of the percentage increase.

³⁰The future tense is used because, as at the beginning of 1996, South Africa had signed, but not yet ratified the FCCC.

The quantification of externalities made it abundantly clear that the potential scale of damages is very large, and so the relative cost of more COz-intensive energy sources will be considerably higher than other alternatives. This factor could be highly consequential when planning and investment decisions are taken around the country’s next bulk supply option, given that coal is much more carbon-intensive than hydroelectricity or gas.

Subsidies to the nuclear industry

The analysis of the fiscal subsidy to the nuclear industry showed that this amount averaged from 3,3 to 11,3 c/kWh, depending on the assumptions made about the future of the industry. Even in the most optimistic scenario, the subsidy represents about one-third of the current average price of electricity. On this basis, it is clear that nuclear electricity has historically been heavily subsidised, and will continue to be until the end of the life of Koeberg power station.

The externality calculations in this study reflect average external costs. From an economic point of view, an analysis of future policy options should be concerned with the marginal costs of various alternatives. Thus, the subsidy to the industry over the past 25 years is a “sunk cost”: it has been expended and cannot be recovered. Policy analyses regarding the nuclear industry should, therefore, concern themselves with the cost-effectiveness of future investments and expenditure from public resources, and with the marginal environmental impact of future operations.

Thus the central question for policy-making purposes is whether the advantages of continuing to operate the country’s nuclear facilities outweigh their disadvantages. Recent investigations of Koeberg’s operating performance have raised serious questions about the reliability and effectiveness of its operations, thus underlining the need for a systematic and comprehensive investigation into the economics of the facility (Thomas 1996). An analysis of this question should be comprehensive and take into account all the future economic, environmental, social and other impacts related to the various options. In economic terms, all of the private and external costs should be included in the calculation: the latter include, for instance, society’s valuation of risks of environmental hazards, catastrophes and so on. The absolute scale of subsidies which have historically been granted to the industry appear to be so large as to make it highly unlikely that they can be justified on economic grounds, especially when there are so many other sources of cheap electricity in the region. Of course, factors other than economically quantifiable ones will also enter the decision.

Conclusion

Electricity prices play an important role in a resource-intensive economy such as South Africa’s. Equally, any increase in prices which may come about as a result of shifting the burden of environmental costs from society at large to electricity producers and consumers, could have significant economic effects. At a microeconomic level, the price elasticity of demand (responsiveness to price changes) is relatively low in the short-term, for the principal reason that it is usually not easy to switch from electricity to an alternative energy source, particularly for large consumers.

In the longer-term, however, price levels may play a more significant role at the microeconomic level. This probably depends largely on the nature of the consumer, since electricity is a small input cost for most commercial, industrial and high-income domestic consumers and, so, they are more likely to absorb (and, where possible, pass on) price increases. For large consumers, however, even small price changes can have a major effect on their competitiveness: for example, gold mines and aluminium producers.

At a macroeconomic level, electricity prices have an important effect on GDP, inflation and employment, since electricity is an intermediate good which affects most sectors of the economy. In the short-term, any price increases can be expected to have a negative effect on these variables (Gibson and Van Seventer 1995).

In the longer-term, however, it is less clear what the effects of higher electricity prices are on macroeconomic performance. Two of the most successful economies in the world, those of Japan and Germany, were built up after the devastation of World War II, with energy prices among the highest in the world. This is not to suggest that their successful economic performance was a result simply of high energy prices—clearly it was not—but to point out that cheap energy prices will not necessarily be in South Africa’s best interests in the long-term, especially if these lock the country into a capital-intensive and resource-intensive development path as opposed to a higher value-added route.

References

Auf der Heyde, T., The South African nuclear industry: history and prospects, Energy for Development Research Centre, University of Cape Town, 1993.

Bernow, S., Rowe, R., White, D., Bailly, K. and Goldstein, J., New York state environmental externalities cost study, report 3: user and reference manual, Tellus Institute and RCG/Hagler, Bailly, Inc., Empire State Electric Energy Research Corporation, Albany, 1995a.

Bernow, S., Rowe, R., White, D., Bailly, K. and Goldstein, J., New York state environmental externalities cost study, report 4: case studies, Tellus Institute and RCG/Hagler, Bailly, Inc., Empire State Electric Energy Research Corporation, Albany, 1995b.

Baumol, W.J. and Oates, WE., The theory of environmental policy, Cambridge University Press, Cambridge, 1975.

Cline, W.R., The economics of global warming, Institute for International Economics, Washington, 1992.

Department of Water Affairs and Forestry, “Bulk water tariffs in South Africa: a possible new approach?”, paper presented at a conference on water conservation, Department of Water Affairs and Forestry, Pretoria, 1995.

Diekmann, U. and Notten, P, “Life cycle assessment: a tool for technology choices in the South African coal industry”, Department of Chemical Engineering, University of Cape Town, 1995.

Dutkiewicz, R.K. and de Villiers, M.G., “Social cost of electricity production”, Engineering Research, report for the National Energy Council, Pretoria, 1993.

Eberhard, A. and Van Horen, C, Poverty and power: energy and the South African state, Pluto Press, London, and UCT Press, Cape Town, 1995.

Eskom, Annual report 1991, Eskom, Johannesburg, 1992.

Eskom, Annual report 1994, Eskom, Johannesburg, 1995a.

Eskom, Eskom environmental report 1994, Eskom, Johannesburg, 1995b.

ETSU, “Externalities of fuel cycles: ExternE project summary report”, report number 1, prepared for the Commission of European Communities, Harwell, UK, 1995.

Fankhauser, S., “Global warming damage costs: some monetary estimates”, working paper GEC 92-29, Centre for Social and Economic Research on the Global Environment, London, 1992.

Fankhauser, S., Valuing climate change: the economics of the greenhouse, Earthscan, London, 1995.

Fraser, M., personal communication, water management section, Eskom, Johannesburg, 1995.

Freeman, A.M., The measurement of environmental and resource values: theory and methods, Resources for the Future, Washington, 1993.

Friedrich, R. and Voss, A., “External costs of electricity generation”, Energy Policy Vol. 21. No. 2, 1993.

Gibson, B. and Van Seventer, D.E., ”The macroeconomic and environmental implications of green trade restrictions on South Africa, draft paper, July 1995.

Hanson, R., personal communication, generation group, Eskom, Johannesburg, 1992.

Hohmeyer, O., Social costs of energy consumption, Springer Verlag, Berlin, 1988.

Intergovernmental Panel on Climate Change, “Summary for policy-makers: second assessment report”, working group III draft report, 1995.

King, H.A. and Rodseth, K.L., “Gaseous detoxification technologies”, TPR/P93/083, Eskom TRI, Johannesburg, 1993.

Lee, R., “Externalities studies: why are the numbers different?”, paper presented at the Third International Workshop on Externality Costs, Ladenburg, Germany, 27-30 May 1995.

Lennon, S.J. and Turner, C.R., “Air quality impacts in South Africa: addressing common misconceptions”, Journal of Energy Research and Development in Southern Africa, May 1992.

Leon Commission, Commission of inquiry into safety and health in the mining industry, Department of Mineral and Energy Affairs, Pretoria, 1995.

Lerer, L., “Review of the bio-medical literature relating to energy and health”, Medical Research Council, Cape Town, 1995.

Lockwood, B, “The social costs of electricity generation”, report GEC 92-09, Centre for Social and Economic Research on the Global Environment, University of East Anglia and University College London, 1992.

Meyer, A. and Cooper, T, A recalculation of the social costs of climate change, Global Commons Institute, London, 1995.

Mishan, E.J., “The postwar literature on externalities: an interpretative essay”, Journal of economic literature Vol. IX No. 1, 1971.

Nordhaus, W.D., “Reflections on the economics of climate change policy”, Journal of economic perspectives Vol. 7, No. 4, 1993.

Oak Ridge National Laboratory and Resources for the Future, “Estimating fuel cycle externalities: analytical methods and issues”, Report No. 2, McGraw-Hill/Utility Data Institute, Washington, 1994a.

Oak Ridge National Laboratory and Resources for the Future, “Estimating externalities of coal fuel cycles”, Report No. 3, McGraw-Hill/Utility Data Institute, Washington, 1994b.

Oak Ridge National Laboratory and Resources for the Future, “Estimating externalities of natural gas fuel cycles”, Report No. 4, McGraw-Hill/Utility Data Institute, Washington, 1995a.

Oak Ridge National Laboratory and Resources for the Future, “Estimating externalities of oil fuel cycles”, Report No. 5, McGraw-Hill/Utility Data Institute, Washington, 1995b.

Oak Ridge National Laboratory and Resources for the Future, “Estimating externalities of hydro uel cycles”, Report No. 6, McGraw-Hill/Utility Data Institute, Washington, 1995c.

Oak Ridge National Laboratory and Resources for the Future, “Estimating externalities of biomass fuel cycles”, Report No. 7, McGraw-Hill/Utility Data Institute, Washington, 1995d.

Oak Ridge National Laboratory and Resources for the Future, “Estimating externalities of nuclear fuel cycles”, Report No. 8, McGraw-Hill/Utility Data Institute, Washington, 1995e.

Ottinger, R.L., “Have recent externality studies rendered environmental externality valuation irrelevant?”, paper presented at the Third International Workshop on Externality Costs, Ladenburg, Germany, 27-30 May 1995.

Ottinger, R.L., Wooley, D.R., Robinson, N.A., Hodas, D.R. and Babb, S., Environmental costs of electricity, Oceana Publications Inc., New York, 1991.

Pearce, D.W., “The development of externality adders in the United Kingdom”, workshop on the external costs of energy, Brussels, 30-31 January 1995.

Pearce, D.W and Turner, R.K., Economics of natural resources and the environment, Harvester Wheatsheaf, Hemel Hempstead, UK, 1990.

Pearce, D., Barbier, E., Markandya, A., Barrett, S., Turner, R.K. and Swanson, T, Blueprint 2: greening the world economy, Earthscan, London, 1991.

Pearce, D.W, Bann, C. and Georgiou, S., The social costs of fuel cycles, HMSO, London, 1992.

Petrie, J.G., Burns, Y.M. and Bray, W, “Air pollution”, in Fuggle, R.F. and Rabie, M.A. (eds), Environmental management in South Africa, Juta and Co., Kenwyn, 1992.

Roome, J., “Water pricing and management”, World Bank, presentation at a conference on water conservation, Department of Water Affairs and Forestry, Pretoria, 1995.

Roos, G. personal communications, Eskom TRI, Johannesburg, 1995.

Rowe, R., Lang, C, Bird, L., Callaway, M., Chestnut, L., Eldridge, M., Latimer, D., Lipton, J. and Rae, D, “New York state environmental externalities cost study, report 1: externalities screening and recommendations”, RCG/Hagler, Bailly, Inc., Empire State Electric Energy Research Corporation, Albany, 1993.

Rowe, R.D., Bernow, S.S., Bird, L.A., Callaway, J.M., Chestnut, L.G., Eldridge, M.M., Lang, C.M., Latimer, D.A., Murdoch, J.C., Ostro, B.D., Patterson, A.K., Rae, D.A., White, D.E., “New York state environmental externalities cost study, report 2: methodology”, RCG/Hagler, Bailly, Inc., Empire State Electric Energy Research Corporation, Albany, 1994.

Rowe, R.D., Chestnut, L.G., Lang, CM., Bernow, S.S. and White, D.E., “The New York environmental externalities cost study: summary of approach and results” workshop on the external costs of energy, Brussels, 30-31 January 1995.

Rowlands, I.R., “The climate change negotiations: Berlin and beyond”, discussion paper 17, Centre for the Study of Global Governance, London School of Economics, 1995a.

Rowlands, I.R., “South Africa and global climate change”, draft paper, London School of Economics, 1995b.

Schleisner, L., Meyer, H.J. and Mothorst, P.E., “Assessment of environmental external effects in the production of energy”, workshop on the external costs of energy, Brussels, 30-31 January 1995.

South African Institute of Race Relations, Race relations survey 1994/95, SAIRR, Johannesburg, 1995.

Steyn, G., “Restructuring the South African electricity supply industry”, paper 14a, Energy Policy Research and Training Project, Energy for Development Research Centre, University of Cape Town, 1994.

Terblanche, R, personal communication, CSIR Environmental Services, Pretoria, 1995.

Terblanche, E, Nel, C.M.E. and Opperman, L., “Health and safety aspects of domestic fuels”, report to the National Energy Council, Medical Research Council, Pretoria, 1992.

Terblanche, P, Nel, C.M.E. and Opperman, L., “Health and safety aspects of domestic fuels: phase 2”, report to the Department of Mineral and Energy Affairs, Pretoria, 1993.

Thomas, S. The operating performance of Koeberg nuclear power plant, Energy and Development Research Centre, University of Cape Town, 1996.

Tilley, H.A. and Keir, J., ”1993 KED database, report no. TRR/P94/127, Eskom TRI, Johannesburg, 1994.

Turner, C.R., personal communications, Eskom TRI, Johannesburg, 1995.

Tyson, P.D., Kruger, EJ. and Louw, C.W., “Atmospheric pollution and its implications in the Eastern Transvaal Highveld”, South African National Scientific Programmes, report no. 150, Pretoria, 1988.

Van Horen, C., “Household energy and environment”, Paper 16, South African Energy Policy Research and Training Project, Energy for Development Research Centre, University of Cape Town, 1994.

Van Horen, C., Counting the social costs: electricity and externalities in South Africa, Elan Press and UCT Press, Cape Town, 1996a.

Van Horen, C., “Eskom, its finances and the national electrification programme”, Development Southern Africa, in press, 1996b.

Van Horen, C. and Simmonds, G., “Greenhouse gas abatement in South Africa’s energy sector”, paper presented at the regional workshop on greenhouse gas mitigation for African countries, Arusha, Tanzania, 28-30 August 1995.

World Bank, World development report 1992, Oxford University Press, Oxford, 1993.

World Bank, World development report 1994: infrastructure for development, Oxford University Press, Oxford, 1995.

Document(s) 8 of 15