Probabilistic CBA

Impact Assessment and Project Appraisal, volume 22, number 3, September 2004, pages 205–220, Beech Tree Publishing, 10 Watford Close, Guildford, Surrey GU1 2EP, UK

Probabilistic CBA Applying a cost–benefit analysis model to the Three Gorges project in China Risako Morimoto and Chris Hope

The world’s largest hydro project is now under construction in China. It is controversial as large environmental and social impacts are anticipated, although it is supposed to control the region’s severe floods, to generate 18.2 GW of hydropower, and to improve river navigation. This study employs a quantitative approach to bring the major economic, environmental and social impacts of this massive project together. This comprehensive probabilistic cost–benefit analysis (CBA) model takes into account the project uncertainty, and its empirical application tries to deliver more robust and justifiable results than those produced by the more usual deterministic CBAs or multi-criteria analyses. This allows the distribution of the net present value to be calculated, and the most significant impacts to be identified. The mean and the 95th percentile of the cumulative net present value at a 5% discount rate estimated by the model will be positive, while the 5th percentile will be negative. Keywords: CBA; China; dam; energy; sustainable development

Dr Risako Morimoto is a research associate, and Dr Chris Hope a senior lecturer in operational research, Judge Institute of Management, University of Cambridge, Trumpington Street, Cambridge, CB2 1AG, UK; Tel: + 44 1223 339700; Fax: + 44 1223 339701; E-mail: [email protected]. The authors would like to thank Mr Tang Jie (World Bank) and the staff from China Three Gorges Project Corporation (CTGPC) who provided us with some useful data and information as well as arranging the tour of the dam site and the visit to the research institute of Yangtze sturgeon. Apart from this, all data and information have been obtained from our own independent study.

Impact Assessment and Project Appraisal September 2004

T

HE MOST ENVIRONMENTALLY controversial project in China today is the construction of the world’s largest dam, the Three Gorges project (TGP) on the Yangtze River in western Hubei province. It is currently under construction and is expected to be completed in 2009. This large-scale project involves about 20,000 workers of whom 40% are women, according to the China Yangtze Three Gorges Project Development Corporation (CTGPC). The dam site is surrounded by scenic landscape with beautiful gorges and many important cultural heritage sites. The area to be submerged is fertile and densely populated. There are numerous studies on the economic, environmental and social impacts of this gigantic project. The Canadian Yangtze Joint Venture published its feasibility report in 1988; CTGPC also published a full environmental impact assessment in 1995; Qigang (1998) surveyed affected people using questionnaires and follow-up interviews to examine people’s awareness and their expectation of changes in their life after this massive scale of resettlement (see also Guojie, 1998; Ren, 1998); Qing (1998b) interviewed the director of the National History Museum of China to investigate the present situation of archaeological sites around the dam site and the budget for preservation work (see also ChildsJohnson and Sullivan, 1998). Hui (1993) studied water pollution in the reservoir area, as there are many factories along the river; Leopold (1996) examined the effects of reservoir sedimentation problems; Bing (1998) considered the possibility of the Three Gorges dam being a target for military attack based on past experiences around the world. There are so many studies that it is difficult to cover them all, but those listed above are the major 1461-5517/04/030205-16 US$08.00  IAIA 2004

205

Probabilistic CBA

ones. This research assesses the project again using a different approach from the previous studies. It employs a quantitative approach to bring the major economic, environmental and social impacts together. The model has been adapted and developed from that used for Sri Lanka in a previous paper (Morimoto and Hope, 2004). This comprehensive probabilistic cost–benefit analysis (CBA) model takes into account the project uncertainty, and its empirical application tries to give more robust and justifiable results than the ones that a more usual deterministic CBA or multi-criteria analysis (MCA) produce. The next section discusses the main issues of the TGP, followed by a detailed explanation on methodology and the presentation of the results including a full sensitivity analysis. The probabilistic analysis allows the distribution of the net present value (NPV) to be calculated, and the most influential impacts to be identified. The final section concludes the study by considering the significance of the results.

Three Gorges project China’s rapid economic growth has resulted in both dramatic improvements in living standards and serious damage to its environment. About 80% of electricity in China is generated by thermal means (more than 70% is coal; the rest is gas and crude oil), 19% is hydro and 1% nuclear (Zeng and Song, 1998).1 Such a high dominance of coal use in China creates serious environmental problems, such as air, land and water pollution.2 . Burning coal produces greenhouse gases, and the emission levels of CO2 in China are increasing continuously. According to Chinese officials, electricity supply would have to increase by 20–30% to eliminate present power shortages; the economic cost of these power shortages is very high (Wu and Li, 1995). The installed hydro capacity in China is less than 15% of its exploitable potential (Zeng and Song, 1998). The Chinese Government is intending to obtain 40% of its power from clean hydroelectric sources3 (WCD, 2000b). As a result of an economic slow down and demand reductions from closures of inefficient state-owned industrial units, China’s electric power industry experienced oversupply in 1998–1999 (EIA, 2001). However, this was only a short-term event and growth in electricity consumption is projected at 5.5% per year until 2020 (EIA, 2001). More electricity is required to boost the economy, including still underdeveloped regions, to meet expected future economic growth, and to meet future increases in household electricity consumption as a result of increased use of electric appliances resulting from improvements in standard of living. Currently, the electricity transmission capacity is limited. However, once this distribution problem is

206

solved, a huge increase in electricity consumption in remote areas will be expected.4

History of the dam Dr Sun Yat-Sen, the forerunner of China’s democratic reform, first proposed to dam the Three Gorges in 1919. The plan of the project has been interrupted over the years by war, the Cultural Revolution, economic troubles and other events. In 1989, the 70m high Gazhouba Dam (40km downstream of the TGP) opened, after nearly 20 years of construction, in response to the increasing demand for electric power in the region. In August 1996, two major projects to improve transportation in the Three Gorges area, the Xiling Bridge and the airport, were also completed and went into service. Construction of the TGP is divided into three stages. The river was blocked at the end of Stage I (1993–1997), then the hydro power station started operating at the end of Stage II (1998–2003), and the project will be completed at the end of Stage III (2004–2009). Appendix IV presents photographs of the progress and impacts of the Three Gorges project.

Benefits of the dam The current dam is intended to provide one tenth of China’s existing energy needs, to raise flood control capacity from the present ten-year frequency to 100year frequency, and to improve navigation along the river (CTGPC, 1995).5 To generate the same amount of electricity, 50 million tons per year of coal, 25 million tons per year of crude oil or 18 nuclearpower plants would be required (Thurston, 1996). The Yangtze River has produced some of China’s worst natural flood disasters. For example, in 1954, 30,000 people were killed (Fung, 1999). In Hubei province alone, the 1998 flood resulted in total economic losses of US$3.6 billion. Agricultural production in this region accounts for about 22% of gross domestic product (GDP) and the flood inundated about 1.7 million hectares of crops (Saywell, 1998). Thus, avoiding these losses must also be taken into consideration in the creation of a realistic model of the project.

Negative impacts associated with the dam Opposition around the world continues to argue that the project will be a social and environmental disaster, having seen the tragedies of large dams elsewhere in the world (for instance, Aswan High dam in Egypt) (Barlow, 1999). One of the major concerns is that the dam would affect a massive area of this densely populated region. The world’s third largest river, the Yangtze, carves its route through the


Probabilistic CBA

The project would displace huge numbers of people and submerge vast areas of fertile farmland; sabotage and earthquake are possible; water quality will probably deteriorate, affecting fisheries; a great tourist attraction and antiquities will be lost; sedimentation is likely to reduce power generation

mountains of southwestern China, springing from the glacial mountains of northern Tibet. It heads North East to surge through a spectacular 200km stretch of deep, narrow canyons known collectively as the Three Gorges. From there, the river widens and meanders across southern China’s vast fertile plains to the East China Sea at Shanghai. The Yangtze River valley is China’s agricultural and industrial heartland. It presently supports one third of the country’s population, produces 40% of the nation’s grain, 70% of its rice, and 40% of China’s total industrial output (Morrish, 1997). The project would displace about 1.98 million people, submerge 100,000 hectares of this fertile farmland, 13 cities, 160 towns, 1,352 villages, 1,500 factories (see Topping, 1996; Goldstein, 1998; Ren, 1998; Childs-Johnson and Sullivan, 1998; Caufield, 1997). There is also a possibility of dam failures and earthquake (Thurston, 1996).6 Large dams can be a prime military target; as was seen in the case of Kajaki dam in southern Afghanistan being bombed by the American air force on the 1 November 2001. Biodiversity is also a great concern. The river possesses 300 species of fish (of which one third are endemic) and its annual aquatic production output accounts for 50% of the whole nation’s gross output (CTGPC, 1995). However, there is a fear of depriving downstream fisheries as a result of a decline in water quality after constructing the dam (Thirston, 1996; ASCE, 1997). The change in habitat may benefit some species but most likely adversely affect others given that different fish species have often quite specific ecological niches (CTGPC, 1995). According to Topping (1996), there are over 3000 factories and mines in the reservoir area; they currently produce 10 billion tons of waste annually containing 50 different toxins. Topping (1996) argues that, if we assume the waste-water level remains unchanged, the waste content will increase 11 times in some areas because of the dam. The reservoir area is a great tourist attraction, with its scenic natural landscapes as well as cultural relics and heritage, all carrying a high aesthetic value. The cultural value of the area is also significant. About


12,000 cultural antiquities and 16 archaeological sites will be submerged as a result of the project (Ren, 1998). Construction of the dam will also result in the destruction of an important link in understanding the birth and early development of China’s ancient civilization. Only a limited number of archaeological remains can be removed or replicated for posterity because of lack of time and funds (CTGPC, 1995). Finally, there is a risk associated with the generation capability of the dam. The water of the Yangtze River carries the fifth largest sediment discharge in the world, most of which is conveyed during floods.7 The Three Gorges area has been intensively cultivated so that soil erosion has become a very serious problem because of a loss of a forest cover (CTGPC, 1995). Sedimentation problems are likely to reduce power generation capacity.8

CBA model There are several project evaluation methods to choose from, each of which has different characteristics. CBA is a project-based approach focusing on net benefits. The most fundamental task of analysis is to define alternative options and quantify their impacts on the objectives established for national energy planning (Munasinghe and Meier, 1993). MCA evaluates multiple objectives simultaneously, which can provide decision-makers with additional information to NPV. Least cost analysis focuses on the entire power system and seeks to minimize total system costs. Hydropower projects generally involve streams of costs and benefits that span many years. Calculating NPV is a relatively easy way to examine the profitability of the project, and it provides policy makers with useful information for their decision-making process. However, in reality, often only certain engineering or economic costs and benefits are included in the analysis. Hence, this paper focuses on the widely used CBA analysis in a broader context, trying to incorporate environmental and social issues into the economic analysis. The model is aimed to lie somewhere in between the highly practical MCA (see for example, Hope and Palmer, 2001) and the highly theoretical CBA (for example, those described in text books on CBA, such as Zeerbe and Dively (1994), Layard and Glaister (1994), Brent (1996), Boardman et al (2001)), with the aim of giving robust and highly justifiable results. This paper presents a CBA model slightly modified from the one applied to Sri Lanka in Morimoto and Hope (2004).9 It is expanded from the previous form by adding some extra variables to cover the differences in the projects. The new model is applied to China’s massive Three Gorges dam. The TGP is selected in this paper, as this is currently the dominant option, which is supposed to produce the

207

Probabilistic CBA

greatest benefit, and indeed it is already under construction. The second dominant alternative scheme is to build a thermal power station. This is justifiable, as about 80% of electricity in China is currently generated by thermal (Zeng and Song, 1998). Other possible alternative schemes are: building a series of dams that generate equivalent amounts of electricity in the less populated upper reach area that has more active water volume whose construction is technically less challenging than the TGP; simply not building a dam; and building nuclear power plants (Fang et al, 1988). The costs and benefits are calculated by determining what might have happened in the absence of the project under construction. The basic specification of the model is given in Morimoto and Hope (2004). The new equations are described in Appendix I and some of the key equations presented in Morimoto and Hope (2004) are presented in Appendix II. Some key variables are: PG is non-environmental cost savings on incremental power; EG is a willingness-to-pay for incremental power; and CP is the cost of environmental damage saved as a result of building TGP instead of thermal plants. The variables EG and CP are multiplied by the factor P and (1–P) respectively, where P is the proportion of time that alternative power generation is not available. This is because the clean power benefit (the variable CP) is obtained only when the alternative power generation is available, and EG is only obtained when the alternative generation is not available. This process avoids double counting. The five new variables added to the model are shown in Table 1: NI (navigation improvement); FC

Table 1. New variables added to the model

Variable

Description

FC (flood control)

The monetary value of flood control benefits. Impact of sedimentation is also considered.

NI (navigation improvement)

The benefit from navigation improvements is expressed by reduction in shipping costs. Changes in transportation costs and impacts of sedimentation are also considered.

DE (downstream effects)

The cost of dealing with downstream pollution caused by the dam construction.

FI (impacts on downstream fishery)

Expressed by a decline in fish catches.

AS (value of the lost archaeological sites)

Approximated by the proportion of the budget for the preservation of cultural antiquities under the international standards, and the budget to rescue those cultural sites.

208

(flood control benefit); FI (negative impact on downstream fishery); DE (mitigation cost of downstream water pollution) and AS (value of lost archaeological sites10). The evaluation method of CP (clean power) is also slightly modified to use China-specific data, estimated by the Battelle Memorial Institute, for externality costs for SO2 emissions based on a number of domestic studies and the marginal costs of CO2 mitigation. The size of the dam is massive, unlike the previous case study, therefore an expression of the possibility of dam collapse as a result of special circumstances (earthquake, technical failures and being a military target) is also added to the evaluation of the variable AC (accident cost). Descriptions of the main parameters in the model and their details are presented in Table 2 and 3 respectively. The rest of the parameters are found in Appendix III. Some of the data used in the analysis may not be very accurate or precise. However, this is inevitable, as many variables are not readily quantifiable and some data have a limited availability because of the project complexity.11 Hence, the data are given as ranges representing our best attempts to extract as much up-to-date information as possible from various sources. All the parameters in the model are assumed to follow either triangular or Beta distributions, and assigned a minimum, most likely and maximum value. Then, 10,000 Monte Carlo simulations are run to generate an expected NPV. Repeated runs of the model obtain a probability distribution of possible outcomes, which is a more defensible procedure than just using single values for inputs that are in reality not well known. In comparison with the Sri Lankan dam, this project is much bigger in size, and is far more complicated in terms of anticipated negative environmental and social issues. Thus, this paper introduces further modifications to allow for premature decommissioning. Table 2. Main input parameters

Parameter

Units

Description

P0

Initial proportion of time during which an alternative power generation technology is not available

ϕ

Parameter that describes the rate of decrease in P over time

EO

Yuan/MWh

A

Initial expected increase in economic output as a result of increased power supply Annual rate of decline in power generation as a result of sedimentation

GC

GW

Generation capacity

AL

billion Yuan

Archaeological loss


Probabilistic CBA Table 3. Main parameter values and descriptionsa

Parameter P0

Minimum value

Most likely value

Maximum value

0.17b

0.2c

1d

e

ϕ

f

0

EOh (Yuan/MWh)

0.018

3200

A

12500

0.001

j

0.005

0.03 k

l

m

18.2n

GC (GW)

12.6

AL (billion Yuan)

1.5 o

Notes:

7800 i

0.085g

16.4

4.9 p

33 q

a Other parameters are presented in Appendix III. PERT distribution (a special form of beta distribution) is used for all the parameters except EO, which is described by a triangular distribution b About 40 million rural households out of 232 million have no access to electricity (40/232=0.17) (US Department of Energy, 1996) c The State development and planning Commission estimated that about 20% of China’s area suffered power shortages in 1998 (South China Morning Post, 10 December 1998, “Obstacles block path to reform of industry”; see also Sinton and Fridley (2000) China’s total energy requirement is projected to increase continuously in the next 20 years (US Department of Energy, 1996) According to WCD (2000a), coal cannot be considered as an alternative to hydropower as a source of peaking power because of its inefficiency. Gas turbines could be an alternative, but would offer fewer benefits. Hence, assume that alternative techniques would not be available in a feasible period f Electricity generated by coal- and gas-fired thermal power plants increased by 1.8% in 1998 (China Energy Efficiency Information Bulletin, March 1999) g Electricity generated by coal- and gas-fired thermal power plants increased by 8.5% in 1995 (China Energy Efficiency Information Bulletin, March 1997) h In China, each kWh of power shortage results in a loss of economic output of US$0.38–1.5 (MOF, 1990) i The average annual active storage loss rate as a result of sedimentation is 0.1% (WCD, 2000a, Figure 2.14, page 65) j It is estimated that, after 100 years, 50% of the reservoir will be filled because of sedimentation. Thus, the annual rate will be 50/100 = 0.005 (Ryder and Barber, 1993) k The maximum annual rate of loss of active storage as a result of sedimentation in WCD Cross-Check Survey: a relatively high rate is set in order to challenge project optimism (WCD, 2000a) l The energy output of Victoria Dam in Sri Lanka is about 31% lower than the planned figure (WCD, 2000a) m The WCD Cross-Check Survey shows that over half of the projects in the sample generate power less than the planned figure: the most likely case is 10% below the target (WCD, 2000a) n The current planned installed capacity (CEB, 1994) o The estimated cost of the necessary excavations in the proposed reservoir area is US$180 million (Topping, 1995) p The estimated cost of salvage work for the TGP area is US$590 million (The Associated Press News, 29 January 1995) q The maximum value for TCC above is 660 billion Yuan. The budget for the preservation of historical relics and cultural antiquities should be about 3–5% of the total construction cost according to the international standards (Qing, 1998a, chapter 9). This approximation is used as a proxy because of lack of other appropriate evaluation methods, thus 660 billion Yuan×5%=33 billion Yuan

Mean present value results There are three large positive impacts, EG (economic growth), PG (power generation), and CP (clean power), and three large negative impacts, CC (construction cost), AS (value of lost archaeological sites),

and RE (resettlement cost) as shown in Table 4. Figure 1 gives a visual presentation of the mean present values by year; the areas under the lines in the figure are the cumulative NPVs in Table 4. The scale of resettlement for the TGP is extremely large compared to most dams, since the area is densely populated.

Table 4. Cumulative mean net present value for the 14n variables at t=100

Benefits

US$ billion

Costs

US$ billion

PVCC

Construction

50

PVEG

Economic growth

82

PVPG

Power generation

31

PVAS

Archaeological loss

15

PVCP

Clean power

17

PVRE

Resettlement

12

PVFC

Flood control

5

PVOM

O&M

5

PVNI

Navigation improvement

3

PVAC

Accident

3

PVDE

Downstream effect

3

PVFI

Fishery loss

0.7

PVLT

Tourism loss

0.4

PVIN

Land inundation loss

0.2

Source: CBA model runs


209

Probabilistic CBA 6

$billion

4

PVEG 2 PVPG

PVFC

PVCP 0 0

10 PVRE

PVAC

-2

-4

20

PVAS

30

40 PVRE

PVOM

50

60 PVDE PVIN

70 PVFI

80

90

100 Year

PVLT

PVCC

-6

-8

Figure 1. Mean present values for the fourteen variables in the model by year Source: CBA model runs

Ranges of the six most influential variables According to Table 4 and Figure 1, the variables PG, EG, CP, AS, CC, and RE are likely to have the largest impact on the NPV.12 Figure 2 plots the 90% ranges of these six most influential variables over time and these are values for the year in question, not present values. The variable PG seems to have a long-lasting positive impact. The magnitude of the impact of EG is much higher than the other variables, though it ceases more rapidly, mainly because alternative power generation techniques would gradually become more readily available, and also the country’s energy scarcity is likely to reduce as time passes. The variable CP gradually moves downward as the level of electricity generation is reduced as a result of sedimentation.13 The variable AS initially shows a downward trend and then stabilizes. Although both capital costs and resettlement costs are huge, they are required only during the construction period.

electricity, ongoing costly operation and maintenance cost, and the loss of important archaeological sites. The likelihood of a negative NPV for the pro ject is 11%. This low level of risk appears to be a result of relatively low environmental and social costs compared to the proposed benefits of the project. However, the sensitivity of the results to the huge uncertainty of the project should be noted. The model includes a premature closure option, which means that, if the ongoing costs outweigh the benefits, the dam is closed down rather than keep making an increasing loss. In practice for the TGP, the possibility of premature closure does not affect the result, even at the 5th percentile, since the possible recoverable costs (after premature closure) of OM (organization and method cost), AC (accident cost), DE (downstream effect), and FI (fishery loss) are not very significant. Therefore, premature decommissioning would be highly unlikely to take place.

Cumulative net present value The evolution of the 5th percentile, mean, and the 95th percentile of the cumulative NPV with a 5% discount rate are shown in Figure 3. The final values are US$ –14, 51, 159 billion respectively. The cumulative NPV is initially negative because of the large construction and resettlement costs. However, a rapid upward movement follows as a result of increased clean electricity sale and stimulated economic growth. The benefit never outweighs the cost for the 5th percentile of the cumulative NPV, because of the reduction in electricity generation resulting from sedimentation problems, low prices for

210

The low level of risk of the project appears to be a result of relatively low environmental and social costs compared to the proposed benefits of the project, but the sensitivity of the results to the huge uncertainty of the project should be noted


Probabilistic CBA $ billion

9

$ billion

30 95%

8

25

7 6

20

5

Mean

15

4

95%

3

10

2

5%

Mean

5

1

5%

0 0

10

20

30

40

50

60

70

80

90

0

100

0

Year

PG (power generation) 2.5

10

20

30

40

50 Year

60

70

80

90

100

EG (economic growth) $ billion 0

$ billion

Year 10

20

30

40

50

60

70

80

90

100

95%

2

5%

-0.5

1.5

Mean

-1

5%

-1.5

Mean

1 0.5

-2

0 0

10

20

30

40

50

60

70

80

90

100

95% -2.5

Year

CP (clean poverty) 0 -1

AS (archaeological loss)

$ billion

$ b illio n 0

Year 10

20

30

40

50

60

70

80

90

100

Year 10

20

30

40

50

60

70

80

90

-0 . 5

-2 -3

-1

-4 -5 -6

5% Mean

M ean

-7 -8

5%

-1 . 5

-2

95%

95%

-9

-2 . 5

CC (construction cost)

RE (resettlement cost)

Figure 2. Range of values of the six most influential variables by year Source: CBA model runs

Sensitivity analysis Table 5 shows that the following input parameters have the most significant impact on the cumulative NPV: P0 (initial proportion of time during which alternative power generation is not available), ϕ (parameter that describes the annual rate of decrease in P (proportion of time during which alternative power generation is not available)), EO (initial expected increase in economic output resulting from increased power supply), a (annual rate of decline in power generation because of sedimentation), GC (power generation capacity), and AL (archaeological loss). The sensitivity of each parameter has the correct sign and is therefore consistent with the model. Although this massive project demands a huge construction cost, it seems not to have one of the largest impacts on the NPV. This may be because a fairly Impact Assessment and Project Appraisal September 2004

Table 5. Statistically significant parameters

Parameter

Student b coefficient

P0

+ 0.58

ϕ

– 0.54

EO

+ 0.39

A

– 0.19

GC

+ 0.15

AL

– 0.14

Note:

Source:

The input parameter values are regressed against the output (NPV). The student b coefficient is a coefficient calculated for each input parameter in the regression equation CBA model runs

211

100

Probabilistic CBA 200

$billion

95%

150 100 Mean 50 5%

0 0

10

20

30

40

-50

50 Year

60

70

80

90

100

-100

Figure 3. Range of cumulative NPV with a 5% discount rate by year Source: CBA model runs

CBA to this controversial project is challenging and novel, as it assesses whether the project, involving such complex environmental and social problems, is worth building, and would contribute to China’s sustainable development. Quantification of the project outcome gives a clear picture and helps policy makers in their decision-making process. Five new variables have been added, namely FC (flood control), NI (navigation improvement), DE (downstream effect), FI (impacts on downstream fishery), and AS (value of lost archaeological sites). The variables with the largest contribution towards the cumulative NPV were PG (power generation), EG (economic growth), CC (construction cost), CP (clean power), RE (resettlement cost), and AS (value of lost archaeological sites). The sensitivity analysis has shown that the parameters P0 (initial proportion of time during which an alternative power generation is not available) and ϕ (parameter that describes the annual rate of decrease in P (proportion of time during which an alternative power generation is not available)) have particularly significant impacts on NPV.

small range of data, 470–660 billion Yuan (US$57– 80), is used, as it is thought that this parameter is fairly well known. Several variable discount rates were used to measure the effect of different rates of pure time preference.14 Figure 4 shows that the choice of pure time preference rate has a strong effect on the value of the cumulative NPV. The fixed discount rate of 5% used for the basic analysis lies in between a pure time preference rate of 1% and 2%. The mean value of the cumulative NPV becomes negative when the pure rate of time preference is approximately above 5%. Therefore, this analysis clearly indicates that the cumulative NPV is sensitive to the choice of discount rates.

Conclusion This paper has illustrated the use of a CBA model of hydropower projects, slightly modified from the one developed in Morimoto and Hope (2004), using China’s massive Three Gorges Project as a case study. The application of the comprehensive probabilistic 350

$billion

300 250 200 150 100 50 95% 0 -50

0

Mean 6

3

5%

%

Figure 4. The cumulative NPV against the pure rate of time reference Source: CBA model runs

212


Probabilistic CBA

The possibility of premature decommissioning has also been introduced in this study, although it did not give a noticeable change to the outcome for the TGP. The improvements of the results could be seen only if sedimentation problems of the TGP were assumed to be even larger than expected, when premature decommissioning would take place. The mean and the 95th percentile of the cumulative NPV at the 5% discount rate at year 100 were positive, whereas the 5th percentile was negative. It is difficult to compare these findings with those from past studies on the TGP, as most of the previous studies have focused only on each impact of the dam and have employed mainly a qualitative approach. However, the finding in this paper supports the view that the social and environmental implications of the project are not negligible, despite the fact that the benefit of power generation is expected to be large. Furthermore, the results obtained by the model need to be treated carefully, as they are sensitive to the valuation methods, the choice of discount rates, and the large project uncertainty. Theoretically, positive NPV implies that gainers should compensate

losers, as some people obtain more benefit than others. However, such distributive effects have not been investigated in this study. Given the large size and geographic scale of the project, investing those effects would be useful in future research. Finally, we have tried to use the most representative data are readily available, to obtain the best possible results. However, it is admitted that data collection was one of the most difficult parts of this research, since there are so many parameters included in the model, and the project is not easy to scrutinize. Wider ranges are used for the particularly uncertain figures. The strength of the model is its simplicity: if the figures used are considered not to be sufficiently accurate or representative, they can easily be replaced and the model run again to obtain better results. For example, the paper shows that the variable AL (archaeological loss) has a significant impact on NPV according to the sensitivity analysis. Therefore, further efforts to search out better estimates for this parameter would be worthwhile, to improve the accuracy of the result.

Appendix I. Additional equations to the previous model (Morimoto and Hope, 2004) FIt

= FB*(1–b*t)

for t>Tc, GYuan/year

where FI FB b Tc NIt TRt

CL2 C1

= benefit from flood control in billion Yuan/year = mean annual benefits from flood control in billion Yuan = annual rate of decline in flood control benefit due to sedimentation = construction period. = [SC*(1–c*t)]*e*TRt = TR0 for t=0 = (1+g)TRt–1

for t≥TS, GYuan/year, for t>0,

C2 QEt CM1 CM2 FIt

= annual decreased SO2 level because of a decrease in coal use as a result of the TGP in Gtons/year = annual decreased CO2 level because of a decrease in coal use as a result of the TGP in tons/GWh/year = annual decreased SO2 level because of a decrease in coal use as a result of the TGP in tons/GWh/yr = quantity of electricity generated in GWh/year = benefits of CO2 reduction in Yuan per ton of CO2 = benefits of SO2 reduction in Yuan per ton of SO2. = (FP*t)/Tc = FP

Yuan/ton

for t≤Tc for t>Tc,

where

where NI

TR+ TS TR0 g

= benefit from navigation improvement in billion Yuan/year (it starts functioning in TS) = annual shipping capacity in Gtons/year = annual rate of reduction in shipping costs = annual rate of decline in navigation control benefit as a result of sedimentation = shipping costs in Yuan/ton = time when hydropower starts operating = shipping costs at t=0 in Yuan/ton = annual rate of change in transportation costs.

CPt

= (1–Pt)*

CLi Pt

= (Ci*QEt)/109 = P0*exp[–ϕt]

SC E c

∑

ϕ

DEt

=0 = ADE

for t≤Tc for t>Tc

GYuan/year

where ADE = annual investment costs to deal with downstream effects in billion Yuan/year (clean-up costs and water pollution mitigation costs).

GYuan/year

= AL*t/Tc = AL

for t≤Tc for t>Tc

GYuan/year

where Gtons/year

where

P0

= annual lost profits from fishery in billion Yuan/year.

ASt [CLi*CMi]

i =1, 2

CP CL1

FP

= benefit from replacing coal use = annual decreased CO2 level because of a decrease in coal use as a result of the TGP in Gtons/yr = initial proportion of time during which an alternative power generation technology is unavailable = parameter that explains the rate of decrease in P over time


AS AL

= value of lost archaeological sites in billion Yuan = archaeological loss in billion Yuan.

Because of the scale of the dam and the geographical characteristics of its location, significant economic damage and deaths/injuries would be expected if the dam collapsed as a result of technical failures, being a target of terrorism, or earthquake. Hence,

(continued)

213

Probabilistic CBA

Appendix I (continued) ACt

=

∑

ACCjt+DMt

GYuan/year,

DCCt = π*TCC = 0 otherwise,

GYuan/year GYuan/year

where

j =1, 2

ACC1t = VD*[DCt+DOMt+DDt] ACC2t = VM*[INCt+IOMt+IDt] where

ACC1t = estimated annual costs of deaths as a result of accidents in billion Yuan ACC2t = estimated annual costs of injuries as a result of accidents in billion Yuan DMt = estimated annual damage costs as a result of special events (costs of economic loss) in billion Yuan VD = value estimate for deaths in million Yuan/death VM = value estimate for injuries in million Yuan/illness DCt = annual number of deaths during construction INCt = annual number of injuries during construction DOMt = annual number of deaths during O&M IOMt = annual number of injuries during O&M DDt = annual number of deaths as a result of dam technical failures/terrorism/earthquake IDt = annual number of injuries as a result of dam technical failures/terrorism/earthquake. Special events, such as dam technical failures, terrorism, and earthquake, are very rare, so that each estimated accident cost will be multiplied by probability of occurrence of these events. DDt

= (GC*DCR′′)*(P+P′+P′′) =0

for t>Tc for t≤Tc

GYuan/year

where DCR′′ = annual number of deaths as a result of special circumstances in deaths/GW/year P = probability of occurrence of technical failures per year P′ = probability of occurrence of terrorism per year P′′ = probability of occurrence of earthquake per year IDt

= (GC*MCR′′)*(P+P′+P′′) =0

for t>Tc for t≤Tc,

GYuan/year

DCCt = costs of decommissioning the TGP in billion Yuan/year, π = proportion of decommissioning costs in construction cost, TDC = time of decommissioning: if policy makers choose premature closure as the cost outweighs the benefit, TDC = time of premature closure. If the costs start outweighing the benefits, it would be better to close down the dam prematurely. Policy makers have an option to close down if TBt–(OMt+DEt+FIt+ACt)+DCCt–DCCt+1Tc, GYuan/year

if TBt–(OMt+DEt+FIt+ACt)+DCCt–DCCt+1Tc+Tcl, GYuan/year where

where ECL

if t=TDC

= economic loss in billion Yuan/year.

Decommissioning costs are currently difficult to predict because of uncertainty surrounding the various parameters affecting the costs and the limited practical experiences with decommissioning (WCD, 2000a). Decommissioning costs vary from project to project, though they are usually large.

TB = total benefits TC = total costs, dt = discount rate (fixed/variable), NPVT = net present value at time T Tc = construction period, Tcl = time to close down the dam (time until which policy makers allow NPV to be negative after construction).

Appendix II. Equations for the other four most influential variables (CP, AS in Appendix I) = [(QEt*103)*PEt]/109 = PE0 for t=0 = (1+f)*PEt–1 for t>0, QEt = [(t–TS)/(Tc–TS)]*AQE*[1–a(t–TS)] = (1–a)*QEt–1 for t>Tc =0 for t