A Critique of "The World Bank’s Experience With Large Dams: A Preliminary Review Of Impacts"

Patrick McCully
Friday, April 11, 1997


Given the importance of the debate over the development effectiveness and social and environmental impacts of large dams, a comprehensive, unbiased and competent review of World Bank lending for large dams is essential. This review must be based on the actual performance and impacts of the projects. The evaluation by the World Bank Operations Evaluation Department (OED), entitled ‘The World Bank’s Experience of Large Dams: A Preliminary Review of Impacts’, fails to fulfil the need for this type of review. The OED review does not assess the actual performance of the projects it covers, is based on flawed methodology and inadequate data, and displays a systematic bias in favour of large dam building. Its conclusions must be rejected as untenable.

The extremely poor quality of the OED review strongly indicates that OED is not a suitable body to entrust with the task of undertaking a comprehensive, unbiased and competent review of World Bank lending for large dams – and also casts doubts on the competence and independence of the work of OED on other areas. For a review of World Bank–funded dams to be seen as credible and unbiased it must be done by a commission of eminent persons independent of the World Bank. The commission must be able to command respect and confidence from all parties involved in the large dams debate and must seek and utilize substantive input from critics of current dam–planning and building practices

Specific reasons for rejecting the conclusions of the OED review include:

  • OED does not compare projected with actual costs and benefits for any of the 50 dams reviewed;
  • for many dams OED has exaggerated both the actual amount of power generated and the economic value of each unit of power produced. OED also includes in its estimates of benefits for individual dams electricity generated at other dams downstream while excluding any of the costs of these downstream plants. OED also neglects the serious adverse impacts of droughts on hydropower production;
  • for dams with irrigation components OED has exaggerated both the areas of cropland receiving irrigation water and the yields from the irrigated areas. OED has not assessed and does not appear to understand the seriousness of the impact of salinization and waterlogging upon irrigated crop yields;
  • projected flood control, water supply and navigation benefits from dams are cited by OED without any review of whether these benefits were realized or whether dams represented least cost methods of achieving them;
  • OED evaluations of the social and environmental costs of the projects are based on seriously incomplete data as are evaluations of the costs of mitigating these impacts;
  • OED adopts an arbitrary amount for the cost of mitigating environmental impacts – $10 million for each of several categories of impacts – for which no justification is given based on the actual costs of the impacts or the cost or effectiveness of their mitigation;
  • OED misrepresents the record of World Bank resettlement operations and makes unsubstantiated claims of "major improvements" in resettlement;
  • assumed dam operation and maintenance (O&M) costs are not based on data on actual O&M costs and no allowance is made for rising O&M costs over time due to ageing equipment and structures;
  • OED fails to deal with the issue of decommissioning dams at the end of their useful lives. Instead, OED wrongly assumes that dams will last indefinitely;
  • OED fails to assess what the environmental impacts of alternatives to any of the specific projects reviewed might have been, nor does it review the accuracy of statements on the costs, effectiveness or impacts of alternatives in project appraisal reports;
  • OED fails to mention demand side management and improved distribution infrastructure as a strategy of reducing growth in urban and agricultural water consumption;
  • OED seriously exaggerates the global warming benefits of dams and fails to note the possible impact on dams of climate change–induced changes in stream and sediment flow patterns;
  • OED’s conclusion that health problems "have in most cases been controlled at a moderate cost" is made in the absence of supporting evidence;
  • OED’s conclusions on the impacts of dams on natural habitats and wild fisheries are made without an informed understanding of the ecological impacts of dams in general or on the overall impacts of any of the specific projects;
  • OED is misleadingly optimistic about the potential of reservoir fisheries to compensate for the loss of wild fisheries. OED gives no figures on predicted and actual yields from reservoir fisheries;
  • OED reviewed very little data on downstream impacts of dams and displays little understanding of the importance of the subject.

A further sign of the seriously inadequate nature of OED’s review process is that the only publicly available product is a 4–page Précis which lists the conclusions of the OED review but contains no details on how these conclusions were reached. None of the sections of the review admitting the poor quality of the data used to reach the conclusions are included in the Précis. In a process of incremental censorship, negative comments on project performance in the "Profiles of Large Dams" background document to the OED report are cut or weakened in the body of the main report, those which survive are emasculated further in the executive summary of the report, and weakened again in the Précis.

One positive aspect of the review process started by OED is that the forthcoming Phase II of the review gives an opportunity for the World Bank to begin in a credible and unbiased way the process of reviewing its financing of large dams.

Until the completion of a redesigned, authoritative independent review and the implementation of any recommended policy reforms, the World Bank should:

  • impose a moratorium on the provision of loans, credits, guarantees and other forms of support for large dams;
  • ensure that the External Affairs Department cease disseminating the conclusions of the review to the press and ensure that the Précis based on the current OED review be withdrawn from the Bank’s Public Information Centre.


The World Bank’s Operations Evaluation Department (OED) undertook this "Preliminary Review" to answer the question "Should the World Bank continue supporting the development of large dams?" (pg.3). The OED’s answer to their question is "a conditional yes". However the seriously flawed methodology and poor quality of data presented in the review in no way support this conclusion – or indeed any other definitive conclusion. This is in particular because most of the quantitative estimates used in the review to evaluate the economic benefit of electricity, irrigation and other services provided by these dams are not based on actual project performance and appear systematically to exaggerate actual project benefits. Attempted evaluations of the social and environmental costs of the projects are based on incomplete and inadequate data, as are evaluations of the costs of mitigating these impacts. OED’s conclusions on environmental impacts and mitigation do not appear to be based on an understanding of the nature of the ecological effects of dams. No attempt was made by OED to compare projected with actual costs and benefits for any of the 50 dams reviewed.

The decision by the OED to restrict is review only to dams over a certain size which have been completed, for which a Project Completion Report has been prepared, and for which the Bank funded the actual dam structure itself (as opposed to essential associated infrastructure such as power and irrigation components) has weakened the effectiveness of the review. The projects reviewed were all appraised before 1987 and most have little information readily available on their social and environmental impacts. Recent controversial projects such as Sardar Sarovar and Yacyretá for which extensive information on social and environmental impacts is available were excluded. Major multi–dam schemes which the Bank supported but on which they did not fund individual dam structures of a sufficient scale to meet the size criteria in the review (such as the Mahaweli project in Sri Lanka) are excluded. OED also failed to review any projects in important countries with large World Bank investments in dams – most notably India.

For these reasons and others, this report and its conclusions should be rejected as untenable.

Given the importance of the large dam controversy for the economic development of World Bank borrowers, a comprehensive, unbiased and competent review of World Bank lending for large dams based on the actual performance and impacts of the projects needs be undertaken. The extremely poor quality of this present review strongly indicates that the World Bank OED cannot be relied upon to undertake a comprehensive, unbiased and competent review of World Bank lending for large dams – and also casts doubts on the competence and independence of the work of OED on other areas where there is less public scrutiny.

The credibility of the World Bank’s ability to assess in an objective manner its own record in the controversial field of dam building is further strained by the fact that the Bank has not made the review publicly available and – despite a promise to the contrary – did not allow NGOs to comment on drafts of the report before its conclusions were circulated to the press and the public. The only information publicly available is a 4–page Précis which lists the conclusions of the review but contains no details on how these conclusions were reached. None of the sections of the review admitting the poor quality of the data used to reach the conclusions are included in the Précis. Furthermore, the conclusions of the report were released to the press before the Précis was available to critics of dam–building practices. It is disturbing that the much–criticized secrecy which surrounds the process of decision–making on dam building is reflected in this review.

One positive aspect of the review process started by the OED, however, is that the second phase of the review – especially the World Bank/IUCN workshop to be hold in Gland, Switzerland on April 10–11, 1997, which will include NGOs and other "representative stakeholders" – gives an opportunity for the World Bank to begin again the process of reviewing its financing of large dams, this time in a credible and unbiased way and including substantive input from critics of current dam–planning and building practices. Such a review must be done by an independent commission comprised of people who can command respect and confidence from all parties involved in the large dams debate.

Such a redesigned review process offers the possibility of restoring trust in the World Bank’s ability to commission objective assessments of controversial lending areas and to reach some common ground in the "bitter and polarized" debate over large dams – a result which would produce substantial benefits for national economies, and for river ecosystems and the countless millions who depend on them.

Until the completion of a redesigned, authoritative independent review and the implementation of any recommended policy reforms, the World Bank should:

  • impose a moratorium on the provision of loans, credits, guarantees and other forms of support for large dams;
  • ensure that the External Affairs Department cease disseminating the conclusions of the review to the press and ensure that the Précis based on the current OED review be withdrawn from the Bank’s Public Information Centre.


For the past decade, large dams have been probably the most intensively criticised of all World Bank investments. This review must be seen as a response to this criticism, and in particular as a response to the June 1994 "Manibeli Declaration", which was endorsed by 326 NGOs and NGO coalitions in 44 countries, and which calls for a moratorium on Bank funding of large dams until a number of conditions can be met. One of these conditions was that the Bank should commission an "independent comprehensive review of Bank–funded large dam projects to establish the actual costs, including direct and indirect economic, environmental and social costs, and the actually realized benefits of each project."

International Rivers was first informed that the OED was undertaking a review of World Bank–funded large dams at the end of 1994. At this time a consultant working for OED requested International Rivers to supply information on dam projects and addresses of relevant NGOs. Through 1995 International Rivers cooperated with and supplied information to the OED team and twice gave written comments on the scope of the review. The OED consultant recognized International Rivers as "an immense source of assistance and information" (letter from Lara Helfer to Shripad Dharmadhikary, July 19, 1995). On November 20, 1995, the Task Manager of the review expressed to International Rivers and the Bank Information Center his "appreciation for the inputs and cooperation which you have already provided" and promised that the OED would "be sending you a copy of the Phase I report as soon as it is ready, for your comments and discussion" (letter from Andres Liebenthal to Patrick McCully and Kay Treakle). However no draft of Phase 1 was ever made available to International Rivers, BIC or any other NGO.

The Bank’s External Affairs Media department began making use of the draft review to attempt to give a positive image of the record of World Bank–funded dams in July, 1996, faxing one journalist a note claiming that "according to a recent World Bank review of 49 large dams financed by the Bank . . . the majority of the projects (31 out of 49) did not experience any seriously negative social or environmental impacts" (the review in fact looked at 50 dams and it is difficult to ascertain from the review which 31 dams External Affairs were referring to). The review itself was circulated internally to the Bank’s Executive Directors and senior management as a "draft report" on September 11. In early 1997 the report (no longer marked draft) together with an accompanying ‘Background Document’ was circulated to the invitees in the April 1997, World Bank/IUCN workshop. The report is marked "restricted distribution" and "Official Use Only". OED has refused to release the review to NGOs who are not on the invitation list for the April workshop (see e.g. electronic mail from Silvana Valle to Alex Wilks, February 11, 1997).

The 4–page OED Précis of the 67–page review (which the Précis terms an "interim report") was released in September 1996. This is the only publicly available version of the review. Due to its brevity and the omission of any detailed explanation of the data on which the conclusions of the review were based, this Précis is more misleading than informative.


From the beginning of the review it is clear that it was undertaken by OED with the pre–determined conclusion that dams by their very nature deliver significant economic benefits, provide "cheap" and "clean" energy, increase food production and control flooding. Examples of this preconceived bias include:

The rationale for the development of large dams is based on the generation of clean power and improved management of water resources . . . (para 1.5)
large dams generate cheaper and cleaner energy and trigger agricultural development (4.3)
Dams are built . . . to protect river valleys against the ravages of floods. (2.1)

OED characterizes criticisms of large dams as being made purely on the grounds of their impacts on the environment and local people – the claims of critics that many dams have performed poorly in economic and technical terms are nowhere recognized:

Big dams have been important in the development of many countries. They have contributed to improved and expanded power generation, irrigation, and domestic and industrial water supplies. Such dams, however, have [caused] significant adverse impacts on the environment and local communities. As a result big dam projects have become controversial . . . (para 1.1)

The review also starts with the now–discredited assumption that unless a river is dammed or diverted for industrial or commercial purposes it serves no function.

Although nearly half of available freshwater resources worldwide still flows unused to the sea . . . (1.5)

In ecological terms, however, all river water is "used". There are no global statistics available on non–industrial/commercial human use of the resources of floodplains and rivers which would substantiate the claim that nearly half of river flow is "unused" (assuming that here "available freshwater resources" refers to river flows rather than to groundwater in aquifers – which do not flow to the sea or do so only very slowly).

The review implies without substantiation than large dams have increased the human carrying capacity of the Earth and similarly implies that dams have ended droughts, floods and famines.

"Without the exploitation of rivers, the world would be a much different place, and would be able to support fewer viable human settlements. Life for many people in the great river basins of the world would be a cycle of drought, floods and famines." (2.4)


The OED review admits in several places the inadequacy of the data which the review team gathered:

the shortage of accurate, up–to–date information about costs, outputs, and social and environmental impacts, made it difficult to arrive at precise evaluations for many of the projects. (Executive Summary, 7)
there have been only a few baseline surveys of socioeconomic and environmental conditions, and monitoring and evaluation studies during and after project implementation. (3.1)
based on current guidelines and practices, most information collected does not meet satisfactory standards of accuracy or reliability. More work will be needed to make a full and reliable assessment of social and environmental impacts and to draw the operational implications. (3.2)
A limitation of this study relates to the difficulty of quantifying and valuing the impacts of large dams on the basis of a desk review. The shortage of accurate, up–to–date information about even the most basic dimensions on costs and outputs for some of the dams, and social and environmental impacts for most of the dams, makes it difficult to arrive at a definitive evaluation. (5.2)

Despite these warnings, the review goes on to evaluate the precise economic costs and benefits of each of the 50 projects reviewed and the financial costs of mitigating the social and environmental impacts; it assesses whether the social and environmental impacts were "major positive", "positive", "major negative", "negative" or "not significant"; and then goes on to use this highly dubious data to conclude that

13 of the projects in the review can be regarded as acceptable, 24 as potentially acceptable, and 13 as unacceptable . . The finding that 37 of the large dams in this review (74 percent) are acceptable or potentially acceptable, suggests that, overall, most large dams were justified . . . (Executive Summary, 8).

The introduction to the Précis uses this analysis to state that

In most of the cases reviewed benefits have far outweighed costs, including the costs of adequate resettlement programs, environmental safeguards and other mitigatory measures. These results suggest that the Bank should continue supporting the development of large dams . . .

Nowhere does the Précis (the only product of the review available outside the Bank) mention the limited and poor quality data on which this important conclusion is based.

The conclusion that

the Bank’s guidelines for resettlement, issued in 1980, have induced major improvements in the management of resettlement aspects of large dam projects (Executive Summary, 12)

also flies in the face of an admission that the data on which the conclusion is based is so flawed as to be largely meaningless:

Of the 36 projects approved before 1980, resettlement was judged to have been satisfactory in only 39 percent and unsatisfactory (or unknown and possibly unsatisfactory) in 61 percent. In 14 projects approved after 1980, resettlement appears to have been satisfactory in 79 percent, unsatisfactory in 14 percent and no resettlement was required in 7 percent. The significance of this finding cannot be determined, however, as resettlement was not systematically discussed and monitored before 1980, and the judgements in the project completion reports appear to be largely based on the views of the implementing agency. (emphasis added) (5.7)

The introduction to the Précis states that "the policies the Bank has adopted on resettlement and environmental aspects of projects have improved performance ‘on the ground’". No mention is made in the Précis of the flaws in the resettlement data. The conclusion that environmental policies have brought improvements "on the ground" is not substantiated in the review which in fact states that

The effectiveness of the Bank’s [non–resettlement dam–related] guidelines could not be determined, as the large dams in this review were mostly planned and implemented before the current guidelines and policies were developed. (5.8)

The optimistic conclusions in the Précis introduction form the apex of a process of incremental censorship in the documentation related to this review. Negative comments on project performance in the ‘Profiles of Large Dams’ background document to the OED report are cut or weakened in the body of the main report, those which survive are emasculated further in the executive summary of the report, weakened again in the Précis, and then once more in the Précis introduction.


Although full of praise for the benefits which large dams provide to humanity, OED paradoxically seeks to downplay the role which the World Bank has played in promoting large dams in developing countries. OED states that

". . . the Bank is directly or indirectly associated with about 2 percent of all dams worldwide and 3 percent of those in developing countries." (1.3)

These statistics are skewed, however, by the huge number of large dams in China (around 19,000). Excluding China (where the Bank had funded only eight dams up to 1994), the World Bank has actually been "directly or indirectly associated" with around ten per cent of large dams in developing countries.

The statistics quoted by the OED also play down the role of the Bank in promoting the very largest projects in developing countries. The importance of the World Bank in major dam schemes is illustrated by the fact that it has directly funded four out of the five highest dams in developing countries outside China, three out of the five largest reservoirs in these countries, and three of the five largest hydroplants.


The evaluation of projects as "unacceptable", "acceptable" and "potentially acceptable" is largely based on a cost–benefit analysis of each project based on the cost of their construction and operation and maintenance compared with the value of their power generation and irrigation benefits. The review correctly notes that

By and large, large dams are justified and succeed or fail on the basis of their efficient delivery of major outputs." (5.11)

However, the report fails to assess whether any of the 50 projects reviewed have efficiently delivered their major outputs and nowhere evaluates the actual outputs from the dams, appearing to rely mainly on projected outputs although this is not made clear in the report. It thus appears that the review is largely assessing whether the initial decision to fund each dam was correct based on the information available in project documents at the time of loan approval. It does not however assess whether this information turned out to be correct. The review thus fails to address one of the key criticisms of the large–dam building process – that projections of project benefits are routinely exaggerated by consultants and other project proponents in order to secure project approval.

Power Generation

OED’s calculations of the value of power generation benefits for the 50 dams reviewed are so seriously flawed as to be largely useless. For many projects power benefits appear to have been highly exaggerated. OED assumes a value for electricity generated of between 5¢/kWh and 8¢/kWh depending on the type of power system they serve. These values are higher than current bulk power prices which currently average around 4¢/kWh in industrialized countries. Hydropower sales between Laos and Thailand are currently being negotiated for between 4.2¢/kWh and 4.5¢/kWh. OED uses a value for electricity from Nam Ngum Dam in Laos of 8¢/kWh – yet the bulk of power generated at Nam Ngum is sold to Thailand at an average price of just 3.5¢/kWh.1

OED assumes a value of power generated by Akosombo Dam, Ghana, of 5¢/kWh. Yet over half of Akosombo’s power is bought by the Volta Aluminium Company which originally paid a rate of only 0.26¢/kWh. This rate has since been increased but even by 1990 was only 1.8¢/kWh.2

OED lists power generation data for each dam reviewed without explaining the source of the data or whether the data is for long–term average annual power production, production for any given year, or projected production. As power production from dams is directly proportional to inherently difficult to predict factors with high interannual and long–term variabilities – rainfall and streamflow – it is vital when assessing dam performance to discover actual average annual production. Finding data for long–term power production from individual dams is difficult (although the OED should presumably have had access to this data from dam agencies).

International Rivers has been able to obtain from mostly industry sources seven years or more of generation data for three of the 50 dams (see Table 1). For each of these dams (Akosombo, Chiew Larn and Nam Ngum) the figure from the non–OED source indicates that long–term average power generation is greatly lower than OED indicates (respectively 28, 42, and 32 percent lower).3 International Rivers was able to find non–OED multi–year data for only one other project – a three year average for Saguling. In this case the datum matches the OED figure.

International Rivers has found non–OED data giving power generation for a single stated year for a further six dams in the review (Bayano, Cirata, Fortuna, Guri, Kariba, Kulekhani). While a one–year figure gives no allowance for production variations due to wet or dry years it is significant that for five of these six dams the non–OED source gives a generation figure less than OED (from 5 to 50 percent less). For the other dam, Cirata, the non–OED source gives a figure one percent higher. For Bayano, the non–OED source gives a figure five percent lower than the OED figure. This non–OED figure (from a dam industry survey) may however be unrealistically high: according to the ‘Profiles of Large Dams’ background document "extreme degradation of the [Bayano] watershed" meant that the dam was generating at only 17–20 percent of capacity in 1988 (no generation figures are given for other years). This reduced capacity at Bayano is not mentioned in the OED review itself.

For 7 dams, International Rivers has obtained power generation data from industry sources which like the OED data give no indication of whether they are multi–year averages or single year figures. For four of these projects the non–OED data give power outputs between 17 and 55 percent less than OED. For the remaining three the non–OED data are no more than two percent higher than OED.

OED indicates that power production at Chixoy has averaged 1,300 GWh annually since completion in 1985 – yet the World Bank’s Project Completion Report for Chixoy gives a figure for total generation from all power plants in Guatemala between 1986 and 1988 of only 1,188 GWh/yr. Nearly a third of Guatemala’ total generating capacity is at plants other than Chixoy. The ‘Profiles of Large Dams’ document notes that the figure of 1,300 GWh/yr is "well below" the predicted generation for Chixoy of 1,540 GWh/yr – but this fact is not mentioned in the main OED report.

Put together, the above evidence strongly indicates that the power generation figures in the review are not based on actual project performance and mostly exaggerate, sometimes by a significant margin, actual power generation.

It is extremely surprising that OED makes no mention of the droughts which in recent years have caused hydropower production to plummet in many countries, with especially severe economic impacts in countries which depend on hydro for a large part of their total supply. These countries include Ghana, Guatemala, Honduras, Kenya, Sudan, Zambia, and Zimbabwe – all of which have hydropower projects covered in the OED review.4 The omission of the impacts of overoptimistic hydrological assumptions during project planning on hydropower is especially difficult to understand as this has been recognized in other OED studies and Bank publications and lending decisions. For example, the January 1996 OED Précis ‘Bank Lending for Electric Power in Africa’ mentions that the 1978–1985 Sahel drought "depleted reservoirs and eroded utilities’ financial reserves". World Bank News of January 20, 1994, reported that the Bank had approved a loan for a coal–fired plant in Zimbabwe to make up for an energy shortage caused when low rainfall "drained the lakes and rivers that Zimbabwe’s mighty hydropower plants relied on to fuel the country’s energy needs. As waters receded, Zimbabwe’s hydropower output fell behind demand and the country’s power supply plummeted by nearly one–fifth." (Zimbabwe has in fact only one large hydro plant – Kariba South – which is included in the OED review).

One of the more extraordinary statements in the review is that Akosombo Dam provides Ghana with "a reliable supply of low cost electricity" (4.25). It is in fact well known that power supply from Akosombo (and Kpong downstream) has been anything but reliable. When the dams were crippled by a serious drought in 1982–1983, Ghanaian electricity supplies had to be rationed for the next three years. Because of this experience, Akosombo was redesigned with a minimum operating level several metres lower than originally planned. In mid–1994, however, after two more dry years, the level of Volta Reservoir fell below the new minimum operating level and Ghanaian electricity consumers suffered black–outs lasting 20 hours a week.5 In a February 1995 report on a loan for a thermal plant in Ghana, World Bank News stated that among Ghana’s "current problems" was "an over–dependence on hydroelectric power".6

The ‘Profiles of Large Dams’ background document does in fact state that electricity supplies from Akosombo have been restricted due to drought but this information is not given in the main body of the OED report. Similarly the background document states that in the early 1990s severe drought "greatly reduced" energy generation at Kariba – but again this is not mentioned in the OED report.

Yet another indication of the unreliable nature of the data on power generation given by OED is that the ‘Profiles of Large Dams’ states that the four dams reviewed in the Chao Phya/Meklong basins in Thailand (Bhumibol, Khao Laem, Sirikit and Srinagarind) together "produce 3,500 GWh in an average year". However, the power generation figures given for the four individual dams in Table 3 in the main report – the figures used to calculate the economic value of power generation for each dam – add up to 4,689 GWh, 25 percent more than the dams generate in an "average" year according to the background document.

The review fails to mention the well–known problem of low reservoir levels in Thailand both reducing hydropower generation and causing conflicts between using water for power or for irrigation. This is not just a problem due to low rainfall in recent years: in their authoritative 1973 history of the World Bank, Edward Mason and Robert Asher of the Brookings Institution state that the "electricity–generating capabilities" of the river on which the Bhumibol Dam was built, "were substantially overestimated".7 According to a recent book on the Chao Phya river, hydrologists predicted in the early 1960s that an annual average of 1.8 billion cubic metres of water would flow into the Bhumibol and Sirikit reservoirs. After the dams’ completion, however, inflows averaged only 1.2 billion cubic metres per year.8

Further clouding the reliability of the generation figures is that they include "incremental generation of downstream dams" although no explanation is given in the review as to which statistics include generation from other dams and how this "incremental" generation was estimated. The inclusion of "incremental" generation is in any case questionable on methodological grounds, chiefly because 1) no economic, social or environmental costs are included for these downstream dams and 2) the costs of dams upstream providing "incremental" generation for the OED dams are not included in OED’s calculations.

The power generation from Akosombo Dam in Ghana, for example, is boosted by OED from 5000 GWh/yr to 6100 GWh/yr by including all of the power produced by the downstream Kpong Dam – although Kpong’s construction cost of some $445 million ($1995) is not factored into the evaluation of Akosombo’s cost.9 OED’s calculation of benefits from Akosombo also assumes that Kpong’s additional power was available from the time of Akosombo’ completion in 1972 – while in fact Kpong was not completed until 1982. Furthermore, OED’s calculation of 6100 GWh/yr of power from Akosombo and Kpong contrasts with figures from a 1993 OED review of resettlement at Kpong which gives generation from the two plants in 1985 as 2,987 GWh, and in 1987 as 4,676 GWh.

Similarly power generation benefits at Guri Dam in Venezuela are increased from 50,000 to 66,000 GWh/yr by including all the power generated at Macagua I (360 MW) and Macagua II (2540 MW) dams downstream – but not the costs of the downstream dams. Construction costs of Macagua II alone were more than $1.7 billion ($1996). Furthermore, although the economic benefits of power generated at Macagua II are included as if they had been generated from Guri’s completion in 1985, Macagua II did not in fact begin commercial operation until 1996 and is not scheduled to be fully completed until July 1998.10

Irrigated Agriculture

As with their power production figures, the OED’s calculations of irrigation benefits are not based on actual project performance and appear to exaggerate significantly their actual value. Statistics on the extent of irrigated areas and of yields of irrigated crops are well known to be unreliable, especially for the major canal irrigation systems serviced by large dams. OED states that "little up–to–date information was available on the specific yields and crops being produced in the command areas" of the 17 large dams reviewed with irrigation components. OED should also have stated that there is little reliable information available on the areas of land which have received irrigation water for most if not all of the reviewed projects; and also little data on important statistical factors such as how much of this land was previously cropped, the size of previous yields, which crops were previously grown, and what inputs were needed. OED does not appear to allow for inter–annual variations in water available for irrigation, and in particular for the reduced area which can be irrigated when reservoir storage is depleted by low rainfall.

OED’s valuation of the benefits of Gazivode Dam in Yugoslavia assumes that the dam has been irrigating 30,000 hectares since dam completion in 1984. The profile of Gazivode in the background document, however, states that sales of water for irrigation did not begin until 1985 (the last year for which OED have data on the project) and were "far below the volumes estimated at appraisal". According to the background document, the maximum cropped area served by the project was planned to be 28,200 hectares and it was estimated that this would not be reached until 1994.

One illustration of how unreliable the statistics on areas irrigated may be is the difference between the figures given by OED for areas irrigated by dams in Thailand and those for the drought year of 1990/91 from the Thai NGO Coordinating Committee on Rural Development (NGO–CORD) (Table 2).

Table 2: Comparison of figures given for area irrigated by dams reviewed in Thailand (NGO–CORD figures are for 1990/91).

Area Irrigated (1,000 ha) Dam OED NGO–CORD Bhumibol/Sirikit 1,100 480 Khao Laem/Srinagarind 360 130 Sirikit 600 54

Source for NGO–CORD statistics: ‘Water shortage reaches crisis point: Second–rice crop banned as more dams are to build (sic)’ Thai Development Newsletter, No. 23, 1993.

Despite the paucity of reliable data, OED have calculated "a preliminary estimate" of the value of irrigation benefits for each project "based on generic information on the dominant types of crops." (2.15) OED does not give details for how it arrives at its "preliminary estimates" of irrigation benefits. No explanation is given of what are the assumptions of yields per hectare for different crops (other than for wheat in Pakistan), or what is the assumed value per tonne of the crops, or of how the assumptions were arrived at. It is clear, however, that OED have assumed that the per hectare value of any given irrigated crop is the same worldwide. However irrigated crop yields vary widely between project and region. Irrigated foodgrain yields in the most productive states in India are at least 250 percent higher than in the least productive.12 Yields also vary widely between different types of irrigation: according to canal irrigation expert Robert Chambers, land in India receiving groundwater irrigation produces roughly twice as much additional foodgrains per hectare as land receiving water from dam and canal systems. It is likely that OED is treating yields from dam and canal systems as if they were average yields although they are routinely below average.

The only project–specific discussion of how OED have calculated their irrigation benefits is for the Mangla and Tarbela Dams in Pakistan. Here they calculate an "annual irrigation benefit" of $260 million based on an assumed yield of wheat of 4 tonnes per hectare.13 Other sources, however, give average per hectare wheat yields in Pakistan in 1990–91 as 1.9 tonnes per hectare. Furthermore, while a likely highly inflated figure for the economic benefits of irrigation in Pakistan is included in the cost–benefit analyses for Mangla and Tarbela, no estimate is included of the massive costs of building irrigation canals, drains and related infrastructure over an area of 800,000 hectares (five per cent of the country’ irrigated land) – and of operating and maintaining this infrastructure. As an illustration of the scale of the costs involved in providing and managing this infrastructure, the World Bank alone lent some $2.1 billion (inflation unadjusted) for 17 separate irrigation infrastructure and management projects in Pakistan between 1976 and 1991.14

The values for irrigation benefits which are used in the OED’s calculations of total project benefits are therefore based on a series of unreliable assumptions about cropped areas, yields and dominant crops grown and exclude many of the costs of supplying irrigation. The values for irrigation benefits given by OED should thus be regarded as untenable.

As with power generation, OED fails to compare projected with actual irrigation benefits. OED makes no mention, for example, that Kainji Dam, Nigeria, was to irrigate 30,320 hectares, yet at the start of the 1990s only one per cent of this area was under irrigation.15

OED appears seriously misinformed on the impact of waterlogging and salinization on canal irrigation, stating that:

"waterlogging and salinity do not appear to have been a factor associated with most dam projects, and where they have it is likely that these conditions were present before the dam was built" (3.3h)

This assertion is nonsensical. Drylands now suffering from waterlogging and salinity due to dam and canal irrigation projects were not waterlogged or salinized when they were used for rain–fed agriculture or grazing. There are also probably few cases where land previously irrigated by groundwater or small–scale or traditional surface water schemes and now irrigated by large dams and canals was previously beset by salinization and waterlogging. The associated problems of salinization and waterlogging are a massive problem for irrigated agriculture and while they are clearly not associated with dams which do not have irrigation components they invariably accompany to at least some degree large dam and canal irrigation projects.

According to a 1990 study quoted in a 1993 World Bank technical paper on the subject, around two to three million hectares of irrigated land become so badly affected by salinity each year that they have to be abandoned. This compares with an estimated annual rate at which new irrigation schemes are now built of two million hectares. Researchers at the Australian National University calculate that the salinization of irrigated soils worldwide is costing some $11.4 billion a year in lost income due to reduced yields. The associated salinization of water resources, CRES believes, costs a couple of billion dollars more in damages to industrial water users and water distribution systems.16

Water Supply, Flood Control, and Navigation

OED includes calculations of the economic value of water supply benefits for three dams. Again these do not appear to be based on actual project performance. For Chungju Dam in South Korea, water sales for each year since dam completion in 1984 are assumed to be the same as sales in 1995, the only year for which OED were given data by project authorities. Yet no evidence is presented of the volume or proceeds of water sales in other years. For Gazivode Dam, Yugoslavia, the background document notes that sales of water to industry on project completion in 1985 were only a quarter of those forecast. In 1985 water sales were forecast to reach the level previously expected for 1985 only in 1994. OED have no information on whether this level of sales was reached yet appear to include the economic benefit of the originally forecast sales for each year from 1985.

OED state that in all three cases where water supply was a major objective of a project "a large dam represented the least cost solution for the supply of water when compared to other dam sites" (2.20). This statement however appears to be based purely on project appraisal documents. OED did not actually review any of these least costs estimates to verify their accuracy in the light of project performance and economic changes since project construction.

OED gives the economic value of flood control benefits for the Chungju and Gazivode Dams. The estimate used for the Chungju Dam is the figure estimated at project appraisal and is not based on any data showing by how much the dam actually has reduced flood damage. The Gazivode figure is based on annuitizing an estimate of damage avoided in a 1979 flood but takes no account of the success or failure of the project in mitigating flood damage since then.

The effectiveness of dams in controlling floods and the methodologies for calculating the economic benefit of avoided flood damage are both highly controversial. OED gives no mention of this debate, nor does it give any figures to show whether predicted flood control benefits of any of the 50 dams reviewed were actually realised. Similarly OED does not review whether any of the dams under review worsened downstream flood damage. This has happened at numerous dams as a result of poor reservoir management and increased development in vulnerable areas downstream due to the false sense of security offered by a large dam (while a dam may stop "normal" annual floods it may not be able to hold back an exceptionally large flood event).

The background notes for the Cerron Grande Dam in El Salvador quote the Staff Appraisal Report for the project as saying that the dam would reduce flooding downstream. In practice, farmers downstream of the dams on the Rio Lempa, of which Cerron Grande is the largest, have suffered repeated and serious crop losses due to badly timed reservoir releases. In 1995 these losses were estimated at over $2 million.17

Although OED mentions navigation as a "side objective" for large dam projects it does not attempt to quantify these benefits. OED also makes no attempt to review whether predicted navigation benefits were actually realised. Although the background document mentions that navigational locks "constructed at great cost" at Kainji Dam "are barely used" this is not included in the main report.

Mischaracterization of Lifetime of Dams and Rising O&M Costs

OED’s estimates of operation and maintenance costs given in their calculations of project economics are for no project based on actual O&M costs. Instead of reviewing actual costs, OED "postulates" that O&M costs are 1 percent of "overnight" project costs (2.7). No justification is given for this figure, but presumably it was chosen as dam appraisals commonly estimate O&M costs at this level. This thus appears to be another instance where OED erroneously assumes that pre–construction predictions are post–construction facts.

At Nam Ngum Dam in Laos, major repairs were needed beginning in 1980, 8 years after the plant was commissioned. Between 1980 and 1990 generating equipment repair costs alone at Nam Ngum equalled approximately $17.3m, yet OED allows for annual O&M costs at Nam Ngum of only $1.3 million.18

Because OED use a constant figure for the annual O&M cost of each project, they make no allowance for rising O&M costs over time or for the need for major project overhauls due to the deterioration of equipment and civil structures. Perhaps the most comprehensive analysis of O&M costs over time is an Ontario Hydro review of data from 430 hydro dams in the US and Canada. This study found that on average hydro dam operating costs rose dramatically after around 25–35 years of operation due to the increasing need for repairs. The authors of this Ontario Hydro study state that "It is worth noting that many maintenance experts and manufacturers agree that most hydro plants require major work sometime between 25 and 35 years old".19

One result of failing to include data showing increasing O&M costs over time is that this allows OED to erroneously conclude that "in most cases, there is no practical limit to [dam projects’] useful life" (2.8) OED thus avoids the issue of the economic costs of dam decommissioning and fails to assess what will happen when revenues from operation of a dam are exceeded by the costs of dam O&M, and in particular the problem of where money will come from to monitor dam safety and undertake emergency repairs or dam removal. The dam industry is now finally beginning to take seriously the issue of dam decommissioning (the American Society of Civil Engineers is currently writing a technical paper on the subject) yet OED fails to even mention this extremely important issue.

Mischaracterization of Distribution of Benefits

The failure to understand the issue of dam decommissioning leads OED to conclude in its section on the distribution of benefits from dams that

future generations are clearly favoured by current investments in large dams in terms of the stock of physical capital and the associated flow of tangible benefits . . . Future generations . . . benefit from a reliable supply of electricity and water while incurring operation and maintenance costs that are only a fraction of the investment. (2.32)

This conclusion is not based on any actual empirical evidence of the long–term reliability of the supplies of electricity and water from large dams, or of O&M costs, and takes no account of instances where future generations will suffer disbenefits from having to pay for upkeep or decommissioning of uneconomic or unsafe dams. It also makes no allowance for the problems which arise when communities (or in some cases entire regions or countries) which have become dependent on dams for water supply, irrigation or flood control have to cope with reduced reservoir capacity due to sedimentation.

OED claims that dams assist the Bank’s strategy of alleviating poverty through "improving electricity, water supply, flood control and navigation [and irrigation] at minimum cost" which leads to "the promotion of economic growth" and "the creation of employment opportunities". This generalisation is not backed by any analysis showing that in general large dams supported by the Bank have proved a least–cost method of effectively providing these services.

OED also claims that dams contribute to poverty alleviation by the "provision of social safety nets" which dam projects supposedly address through "the mitigation of resettlement impacts" and "through the policies advocated by the Bank on power and water pricing" (2.25). Yet reviews of Bank–financed resettlement operations have shown repeatedly that the majority of resettled people are left worse off after being moved – "the mitigation of resettlement impacts" therefore in reality more often leads to poverty creation than alleviation. OED itself partly contradicts the statement that power and water pricing policies linked to dams offer a "social safety net" by correctly remarking that "the distribution cost and low load density makes grid electricity a very expensive source of energy for rural areas which few consumers can pay for, and few developing countries can afford to subsidize on a massive scale" (2.29).

OED states that "the irrigation schemes served by the projects in this review have been designed to benefit mostly small farmers" (2.26). No evidence is given, however, to verify this statement or to show whether over the long term small farmers actually have benefited more than large farmers. OED also gives no evidence to show how large dam and canal irrigation schemes compare with small–scale, vernacular and groundwater irrigation systems in mostly benefiting small farmers.20

The background profile for the Dez Dam, Iran, notes that although the project was designed to benefit mostly small farmers: "The smallholder development project originally envisaged by the Bank was transformed into a large–scale mechanized farming project". The result was "social upheaval, depopulation, decreased incomes and employment and a heavy drain on Government budget (sic), without any compensating gain in production or a more rapid pace of development." This account of the failure of the Dez scheme to benefit small farmers is not mentioned in the main OED review.

Mischaracterization of Environmental Impacts of Alternatives

In its section on the environmental impacts of alternatives to large dams, OED does not assess what the impacts of alternatives to any of the specific projects reviewed might have been, nor does it review the accuracy of statements on the impacts of alternatives in project appraisal reports. It is noteworthy that a recent OED review of the effectiveness of environmental assessments highlighted that "most Category A [the environmental category which includes dams] project EAs fail to give serious consideration to alternative designs and technologies . . . and those that do often explore weak, superficial or easily dismissed options."21

OED mischaracterizes the impact of "water management alternatives" by stating that these are "similar in nature to those of large dams". The impact of non–structural flood management measures, for example, are in no way similar to large dam impacts (non–structural flood management includes measures such as watershed restoration and protection, zoning policies which dissuade floodplain development, and improved flood warning systems). OED nowhere discusses the efficacy of non–structural flood management (or flood management using structures only to protect the most vulnerable urban areas) compared with flood control from large dams.

OED also fails to mention demand side management and improved distribution infrastructure as a strategy of reducing growth in urban water consumption; and the massive potential water savings which can be made by improving irrigation efficiency (Robert Chambers calculates that better management and modest improvements in irrigation infrastructure in Pakistan could supply an amount of water equivalent to that supplied by three Tarbela Dams).22

OED claims that if the

204 billion kilowatt–hours/year of electricity generated by projects in this review had to be generated by fossil fuels, it would lead to emissions of about 220 million tons per year of carbon dioxide, equivalent to about 4 percent of the countries’ total anthropogenic emissions. (2.37)

This statement seriously exaggerates the global warming benefits of dams for several reasons: 1) as is explained above, the power generation figures used by OED for the dams under review are much higher than actual generation; 2) the figure for avoided carbon dioxide assumes that all alternative plants would be coal–fired, making no allowance for less carbon–intensive gas plants, nor for the possibilities of avoiding the need for extra generation through demand–side management; 3) no allowance is given for the emissions of greenhouse gases from reservoirs, in some cases comparable to those from equivalent sized gas or even coal–fired power plants (although OED discusses this problem later in para 3.34); and 4) no allowance is given for increased greenhouse gas emissions due to land use changes catalyzed by dams such as deforestation and conversion of dryland to irrigated agriculture.


The introductory paragraphs to the section in OED’s review of social and environmental impacts note the poor quantity and quality of the evidence available from project documents for the 50 dams under review:

Bank reports describe social and environmental impacts, beneficial and adverse, only for the more recent projects . . . there have been only a few baseline surveys of socioeconomic and environmental conditions, and monitoring and evaluation studies during and after project implementation . . . . based on current guidelines and practices, most information collected does not meet satisfactory standards of accuracy or reliability (3.1–3.2)

Despite this admitted lack of basic data, OED goes on to make largely positive evaluations of the impacts of dams and the possibility of mitigating these impacts at a relatively low cost. The conclusions of OED on the impacts of the dams under review cannot be regarded as credible.


The "main findings" of the review on resettlement are that:

resettlement impacts have been the major adverse impacts. While such impacts have been associated with nearly half the projects in this review indications are that the implementation of the Bank’s guideline (sic) in 1980 has been effective in improving this aspect of project performance" (3.3a)

The claim that adverse resettlement impacts are "associated with nearly half the projects" is based on the finding that according to "available information in the Bank project reports, the outcome of resettlement has been satisfactory in 25 of the large dams, and unsatisfactory or unknown and possibly unsatisfactory is 24 others" (3.5). Yet as OED later notes, "resettlement components were not systematically discussed and monitored" before 1980 and the "judgements in the Project Completion Reports [the sources used by OED] are largely based on the views of the implementing agency, rather than on any attempt to elicit the views of the affected people" so that "the significance of these findings [on satisfactory/unsatisfactory impacts] cannot be determined" (3.6). OED goes on to note that "a ‘satisfactory’ resettlement rating in this paper is not necessarily synonymous with compliance with the subsequent Bank guidelines" (3.6).

Despite it being extremely unclear what the actual significance of "satisfactory" resettlement is, the "finding" that less than half the dams had "unsatisfactory or unknown and possibly unsatisfactory" resettlement impacts and the implication that more than half therefore had "satisfactory" resettlement impacts is repeated in both the Executive Summary and the Précis of the review.

The claim that the Bank’s 1980 resettlement policy has been effective in improving resettlement performance (the Executive Summary states that the policy has "induced major improvements" in resettlement management) is also repeated in both the Executive Summary and Précis. This conclusion is based more on wishful thinking than actual analysis of data. More reliable is a 1993 OED overview of Bank resettlement performance which concluded that "Bank guidelines have not led to improved monitoring efforts which would permit an assessment on resettlement outcome, particularly in respect of the crucial criterion of incomes".

The Bank’s 1994 review of projects involving resettlement appraised between 1986 and 1993 (i.e. long after the adoption of the Bank’s resettlement policy) could find only one Bank–funded dam out of 39 reviewed where "incomes for all households rose after resettlement" (the "fundamental goal" of Bank resettlement policy is "to restore the living standards and earning capacities of displaced people"). Yet even for this example, Khao Laem Dam in Thailand, the 1993 OED report on which this optimistic conclusion is based notes that four out of five resettlers surveyed considered themselves economically worse off after resettlement. The survey which supposedly found that "incomes for all households rose", furthermore, was based on a small sample of families at resettlement sites, at which only 60 percent of those resettled were living.


The "main findings" of the review on the health impacts of dams are that:

health impacts are a risk . . . but these problems have in most cases been controlled at a moderate cost (3.3b).

Again this conclusion has been made in the absence of almost any supporting evidence. No information on disease impacts was reviewed for seven of the 12 projects which OED states are located in areas where serious waterborne tropical diseases were endemic. Furthermore, the geographic distribution of the dams reviewed indicates that many more than 12 are likely to have had an impact on the incidence of tropical diseases. OED does not assess whether or not any of the reviewed projects introduced or increased the incidence of these diseases in non–endemic areas. OED also does not take into account the risks of disease in the remote areas where dams tend to be built due to a sudden influx of outside construction workers which can often increase the local incidence of diseases such as tuberculosis, measles, diarrhoeal infections, leishmaniasis, syphilis and AIDS.

Of the five projects for which OED reviewed health impacts they state that "mitigation programs are reported to have reduced the incidence of disease below pre–existing levels in three cases". These three cases include Gazivode in Yugoslavia for which the only source of information on health impacts is the project authorities (and which is not in an area where waterborne diseases would be expected to be a severe problem – the background profile for Gazivode does not mention for which diseases the incidence decreased). Another of the three cases is Bayano in Panama where again the only source of information is the project agency which according to the background profile for the project states that "health conditions improved as a result of medical extension." The background profile for Bayano also notes that "laboratory studies on pre– and post–lake conditions indicate that there was an explosion in the population of disease vectors of yellow fever and encephalitis".

The third health impact "success" is Roseires Dam, although OED also includes Roseires as one of two projects where disease incidence "appears to have increased at least temporarily". It is difficult to evaluate the health impacts of Roseires as the dam is part of Sudan’s huge Gezira irrigation scheme where there had been a massive increase in the incidence of schistosomiasis since the 1940s. Roseires Dam itself was completed in 1966 although its associated irrigation schemes (the main source of schistosomiasis transmission) were not finished until the mid–1980s. In the late 1960s and early 1970s the incidence of urinary schistosomiasis at Gezira was around 20 percent in adults and 50 percent in children. Since the mid–1970s, however, public health measures have lowered the incidence to under 15 percent. Rates of the more serious intestinal version of schistosomiasis stood at 73 per cent by the mid–1970s. After this a programme of improved sanitation, public health education, the spraying of molluscicides and chemical treatment of infected people, the rate of schistosomiasis at the section of the Gezira scheme served by Roseires Dam had fallen to 15 per cent by 1988 according to WHO.23

OED states that a 1995 completion report for another World Bank–funded project at Gezira claims that a schistosomiasis control programme "brought the incidence of the disease down from 54 percent to 7 percent at a cost of $3.5 million over eight years". It is not clear how these statistics compare with the figures quoted earlier from WHO. What is clear however is that since Roseires Dam was built huge numbers of the two million people on the Gezira scheme have suffered the debilitating effects of schistosomiasis. OED’s conclusion that "there may have been an initial problem with schistosomiasis" which has now been "brought under control at moderate cost" appears gravely to understate the impact of schistosomiasis at Gezira.

OED’s second instance of a project where disease incidence "appears to have increased at least temporarily" is Akosombo in Ghana, where urinary schistosomiasis rates among villagers rocketed from less than 10 per cent in riverside villages in 1966, to around 90 per cent in children living near Volta Reservoir in 1969. A programme on Lake Volta, supported by WHO and UNDP, involving the spraying of molluscicides and treatment of infected people, halved the incidence of schistosomiasis in the project area between 1975 and the end of the program in 1981, but it still left 37 per cent of the people in the program area infected, and it had only covered a 60 kilometre stretch of Lake Volta’s 5000 kilometre–long shoreline.24 OED says that "there are reports that, more recently, this problem has been mitigated by drug therapy and molluscicides" but gives no further details.

OED adds a "notional mitigation cost" of $10 million to the initial investment costs of "all ten projects where there appears to have been an issue about such impacts" (3.17). OED gives no further explanation as to why the figure of $10 million was chosen or why it chose to add it to only 10 projects.

Natural Habitats and Fisheries

OED states that there are "eleven projects where adverse impacts have occurred or might have occurred" (3.23).25 These conclusions on natural habitats do not appear to be based on any informed understanding of the ecological impacts of dams in general or on the overall impacts of any of the specific projects. The spurious nature of this conclusion is illustrated by one of the dams which OED states had a "possible negative impact" – this is Akosombo Dam which created the most extensive reservoir in the world, flooding nearly 8,500 square kilometres of land, 4 per cent of the area of Ghana. According to OED, Sobradinho Dam, which created Brazil’s largest reservoir by flooding 4,200 square kilometres of land, had no impact on natural habitats. Other dams which according to OED did not impact natural habitats include dams which created large reservoirs in tropical forest areas such as Srinigarind, Bhumibol and Chiew Larn.

Dams are recognized as the main reason why one–fifth of freshwater fish species are now either endangered or extinct. OED, however, dismisses the impacts of dams on wild fisheries in a couple of sentences noting that out of the dams under review only Bayano, Fortuna and Itezhitezhi "are reported to have had an adverse impact on local fisheries, either in the reservoir area or downstream" (3.24). Given that almost all dams will inevitably alter riverine ecosystems and thus fish habitats, and given the available evidence of the impacts of dams on fisheries, it should be assumed until proved otherwise that all the dams in this review caused negative impacts on natural fisheries. At least some evidence available in Bank reports on the negative impacts of the dams under review on wild fisheries has been ignored: at least two Bank documents have reported that the regulation of the Volta River by Akosombo Dam caused the "loss of the thriving clam industry which had formerly employed about 2,000 women".26

The only detailed information OED reviewed on the impacts of any of these dams on wild fisheries was for Bayano, "where only 13 of 61 native fish species, and only two of 12 crustacean species appear to have survived" (3.21). OED comments that in this case "the adverse impact can be regarded as major" and that the project is therefore judged to be "unacceptable". It is likely however that Bayano only appears to be the worst project in terms of fisheries impacts because no detailed fisheries studies were done for any of the other projects.27

OED itself notes that baseline surveys and follow–up studies are essential in assessing ecological impacts yet such surveys were not required by the World Bank before 1989 and so "natural habitat issues were only addressed case by case, sometimes with reasonable success, but usually in a manner that would not meet current standards" (3.23). All of the dams under review were appraised before 1989.

"Given the extreme paucity of even qualitative information", OED states "it was not possible to arrive at an adequate estimate of mitigating the habitat losses" (3.23). Nonetheless, OED goes on to create an arbitrary figure of $10 million as "a notional cost" for mitigating natural impacts which it then adds to the cost only of the eleven dams "where adverse impacts have occurred or might have occurred". Despite having little data on natural impacts and no comprehensive data on the success of mitigation measures in compensating for the impacts of dams on riverine and riparian ecosystems, the Executive Summary of the review states that the "costs of adequately mitigating for [impacts on natural habitats] rarely affects the economic viability of the project".

OED claims that the affects on dams on wild fisheries "can often be mitigated by the development of new fisheries in the reservoir" (3.3d). This claim is misleading for several reasons:

  • far fewer fish species are suited for reservoir than riverine habitats and therefore fish diversity is almost always reduced by reservoirs;
  • the deliberate introduction of non–native species into reservoirs for fisheries reasons has been a leading reason for the decline in wild fish diversity worldwide;
  • the record of reservoir fisheries is in general extremely poor – while initial catches are high, these tend to rapidly decline;
  • local people frequently have less access to reservoir than riverine fish as reservoir fishing requires different skills and equipment, and more capital, than river fishing.

OED gives no data for any of the dams under review comparing pre–reservoir and post–reservoir yields in the riverine, floodplain, estuarine and off–shore areas affected by dam–induced hydrological and ecological changes. OED also fails to compare projected with actual fish yields for any reservoirs. For at least some of these dams, actual yields have been far lower than projected: at Chiew Larn, Thailand, the projected value of fish production was 23 million baht/year. Actual production in the nine years after reservoir filling in 1987 (which should be by far reservoir’ most productive years) averaged only 5 million baht.28 Projected yield at Kainji, Nigeria, was around 10,000 tonnes/year. The yield exceeded this for the first three years after impoundment, but since 1977 has reportedly stabilized at around 4,500 tonnes.29

A World Bank Staff Appraisal Report on a 1991 aquaculture loan to India noted that yields from reservoirs in India are "very low" for a number of reasons including "lack of finance for stocking and for fishing operations . . . Proper management is usually non–existent . . . stocking is usually inadequate, with fingerlings too small and therefore resulting in high mortality rates."

OED attaches an arbitrary – and in effect meaningless – figure of $10 million for the cost of fisheries mitigation "in the four projects where the impact on fisheries was reported to be negative" (3.27). (According to para 3.24 only three projects "are reported to have had an adverse impact on local fisheries"; Table 6 lists only two projects with "negative" fisheries impacts).

Downstream Impacts

As with the other impacts under review, OED reviewed very little data on downstream impacts and displays little knowledge of the subject. OED’s conclusion that "downstream impacts were found to be major in one project and significant in some others" (3.3g) can not be regarded as credible. Again OED adopts an arbitrary figure of $10 million to "mitigate" for downstream impacts without any evaluation of whether this figure is realistic for any of the projects.

The best illustration of OED’s lack of understanding of this issue is their treatment of the downstream impacts of Kainji Dam. In one of the few places where OED cites a non–Bank or non–dam agency source, they state that Kainji

is reported to have adversely affected hundreds of thousands of people living downriver by reducing yam and fisheries production. Also, former wetlands that had been seasonally inundated no longer provided essential grazing for livestock at the end of the dry season or water for flood recession cultivation of rice and other crops.

OED, however, then states that these findings "are difficult to understand, as the annual flow of the Niger at the dam site is only about a third of the total flow further downstream, after the confluence with the Benue, and it should be possible to operate the reservoir so as to minimize harm" (3.40). Yet a simple glance at a map shows that the confluence with the Benue is several hundred kilometres downstream of Kainji leaving a huge area of riverine and riparian ecosystems impacted by flow regulation at Kainji and unaffected by any water flowing into the Niger from the Benue. Also while it may be possible to operate the reservoir "to minimize harm downstream" (although evidence for this is not given), Kainji Reservoir is not operated to minimize downstream impacts. OED also give no indication of what the impacts of a different operation regime at Kainji would be on power production at the dam. In Table 6 OED chooses to characterize the downstream impacts of Kainji as "negative" – but not "major negative".


The main conclusions of this report can be summarized as:

"the finding that 37 (74 percent) of the dams in the review are acceptable or, with adequate mitigation, potentially acceptable, suggests that, overall, most large dams have the potential of performing as the starting point for the development of large river basins, the catalyst for integration of national power systems, the key element in a flood control program, or simply as the least cost solution to the provision of power, water and other benefits without causing irreversible or major adverse affects in the environment or society." (4.12)
"Environmental impacts are important . . . but the costs of adequately mitigating for these impacts should only rarely affect the economic viability of the project." (5.11)
"Adverse impacts that occurred could have been appropriately mitigated given adequate commitment, knowledge and resources." (5.12)

As has been explained above the data and analyses presented by OED can in no way be accepted as leading to these conclusions. These conclusions can thus not be regarded as credible and appear to be motivated by a desire by the OED and World Bank to justify their past dam lending record, fend off criticisms by dam opponents, to legitimate lending for more large dams in future, and to legitimate the Bank’s desire to facilitate private sector dam building by offering guarantees and other incentives to private companies and by offering to take on the costs of the social and environmental "externalities" of dam building such as resettlement and environmental mitigation measures. In fact the report will only serve to heighten public concern over the World Bank’s dam building record and over the Bank’s ability to carry out competent and unbiased assessments of its lending operations.


1. Matsumoto, S. (1996) ‘Key Issues on Economic Aspects of Nam Ngum Dam in Lao PDR’, JVC Laos, January.

2. Gitlitz, J. (1993) ‘The Relationship Between Primary Aluminium Production and the Damming of World Rivers’, International Rivers Working Paper 2, August.

3. According to the 1995 International Water Power and Dam Construction Handbook, production from Akosombo and Kpong in 1991 was six percent less than the OED figure for the dams.

4. For example, in 1990 Guatemala suffered a series of blackouts because of a lack of water at Chixoy Dam. Two years later, another power shortage because of Chixoy’s low reservoir resulted in electricity being rationed for more than a month, costing the country $2 million a day in lost industrial production (Yearly, R. (1992) ‘The Lights Go Out in Guatemala’, Report on Guatemala, Oakland, California; Barry, T. (1992) Inside Guatemala. IHERC, Albuquerque, 115). The 300 MW turbines of El Cajón Dam represent 70 per cent of the installed capacity of Honduras. Low rainfall through the early 1990s has meant that the level of El Cajón Reservoir has steadily fallen. By mid–1994, the dam was able to generate at only half capacity at best and Honduran electricity consumers were suffering black–outs of up to 14 hours a day (Gollin, J.D. (1994) ‘Trees Down, Lights Out in Honduras’, Christian Science Monitor, 15 November; ‘Aún no se llena El Cajón’, Tiempo, Tegucigalpa, 30 September, 1994).

5. Annan, C.K. (1989) ‘Was Ghana’ Akosombo Dam the Best Option?’, World Water, September; Graham, Y. (1995) ‘Volta power grows dim’, African Agenda, Vol. 1, No. 1.

6. Ghana to Power Up With IDA Funding’, World Bank News, 23 February, 1995.

7. Mason, E.S., and Asher, R.E. (1973) The World Bank Since Bretton Woods. The Brookings Institution, Washington, DC, 687.

8. Van Beek, S. (1995) The Chao Phya: River in Transition, OUP, p. 153.

9. Calculation of Kpong cost calculated from data in OED (1993) ‘Early Experience with Involuntary Resettlement: Impact Evaluation on Ghana – Kpong Hydroelectric Project’.

10. Perazzo, P.H. and Antonini, M.A. (1995) ‘Macagua nears completion in Venezuela’, Int. J. of Hydropower and Dams, November.

11. Canal irrigation researcher Robert Chambers writes in a text book on the subject that: ‘In an earlier outline for this book I set aside a chapter for sorting out irrigation definitions and statistics. Heroic patience and several lifetimes were evidently needed . . . Had I not backed off, this book would never have been written. I shall use statistics throughout, but the reader is asked to take them with judiciously liberal doses of salt’ (Chambers, R. (1988) Managing Canal Irrigation: Practical Analysis from South Asia. CUP)

12. World Bank (1991) ‘India: Irrigation Sector Review. Volume II – Supplementary Analysis’, 6.

13. Ghassemi, F., et al. (1995) Salinisation of Land and Water Resources: Human causes, extent, management and case studies. CAB International, Wallingford, UK, 371.

14. See Safadi, R. and Plusquellec, H. (1991) Research on Irrigation and Drainage Technologies: Fifteen Years of World Bank Experience. World Bank Discussion Paper 128, p. 33; Umali, D.L. (1993) Irrigation–Induced Salinity: A Growing Problem for Development and the Environment. World Bank Technical Paper 215, August, p.75.

15. Ayeni, J.S.O., Roder, W. and Ayanda, J.O. (1994) ‘The Kainji Lake Experience in Africa’, in Cook, C.C. (ed.) Involuntary Resettlement in Africa: Selected Papers from a Conference on Environment and Settlement Issues in Africa. World Bank Technical Paper 112.

16. Umali, D.L. (1993) Irrigation–Induced Salinity: A Growing Problem for Development and the Environment. World Bank Technical Paper 215, August, 3. HR Wallingford (1990) ‘ODU Studies: Soil Salinization, Land Reclamation and Drainage’, Project Sheet, Wallingford, UK, 1. Ghassemi, F., et al. (1995) Salinisation of Land and Water Resources: Human causes, extent, management and case studies. CAB International, Wallingford, UK, 12–13, 48.

17. Gladman, A. (1997) ‘Dam Operation Changed to Protect Farmers’, World Rivers Review, February 1997.

18. Matsumoto, S. (1996) ‘Key Issues on Economic Aspects of Nam Ngum Dam in Lao PDR’, JVC Laos, January.

19. T. Wong, ‘Determining O&M Costs Over the Life of a Hydro Station’, in ‘Hydro in the ‘90s’, Hydro Review Worldwide, Kansas City, 1994.

20. For an analysis of a large dam and canal irrigation scheme intended to benefit small farmers which ended up impoverishing small farmers and enriching large farmers see Goldman, M. (1994) ‘"There’s A Snake On Our Chests": State and Development Crisis in India’s Desert’. Ph.D. Thesis, University of California, Santa Cruz, December.

21. OED (1996) ‘Effectiveness of Environmental Assessments and National Environmental Action Plans: A Process Study’, June 28.

22. Chambers (1988) op. cit.

23. Hunter, J.M. et al. (1993) Parasitic Diseases in Water Resources Development: The Need for Intersectoral Negotiation. WHO, Geneva, pp. 45–46; Amin, M.A. (1977) ‘Problems and Effects of Schistosomiasis in Irrigation Schemes in the Sudan’, in Worthington, E.B. (ed.) Arid Land Irrigation in Developing Countries: Environmental Problems and Effects. Pergamon, Oxford.

24. Hunter et al. (1993) op. cit., 35, 73.

25. Paragraph 3.18 states that "overall, there are eleven projects in the review where there appears to have been a significant negative impact on natural habitat, and two additional ones where such an impact might have occurred". This is presumably in error – the data in Table 6 and paragraph 3.23 indicate that OED believes there are nine dams "where there appears to have been a significant negative impact on natural habitat" and two with questionable impacts.

26. See e.g. OED (1993) ‘Early Experience with Involuntary Resettlement: Impact Evaluation on Ghana – Kpong Hydroelectric Project’.

27. OED’s ignorance of the impacts of dams on fisheries is illustrated by its claim that in North America and Europe fish ladders and hatcheries have handled the "problem of migratory fish" "with some success". In reality, the record of mitigating the impacts of dams on migratory fisheries in these areas has been extremely poor. Dams are the main reason why the annual run of adult salmon and steelhead trout in the Columbia River basin in the US Pacific Northwest has declined from an average of between 10 and 16 million fish in the 19th century to only some 1.5 million fish today. The National Marine Fisheries Service have estimated the cost of salmon fishery losses due to dams in the Colombia Basin for the period between 1960 and 1980 alone at $6.5 billion. Out of the original 400 or so salmon and steelhead stocks of the US Pacific coast as a whole only 214 remain, out of which 169 are at high or moderate risk of extinction. The US Atlantic salmon population has declined from half a million in the early 18th century to a few thousand, mainly hatchery–reared, fish in the 1990s. Dams have also eliminated Atlantic salmon from France’s Dordogne, Meuse, Moselle, Garonne and Seine rivers. Fish hatcheries, which the World Bank regularly recommends as a "mitigation" measure in dam projects are extremely controversial among fisheries scientists in the US. Some $350 million is spent annually on salmon hatcheries in the US Pacific Northwest, yet not only has the number of adult salmon in the Colombia River plummeted, but hatchery fish are degrading the genetic diversity of the remaining wild salmon and helping push them toward extinction. A 1995 report by the US National Research Council warned that current hatchery policies in the Pacific Northwest were "based on deep ignorance". "It isn’t enough to focus only on the abundance of salmon", the NRC concluded. "The long–term survival of salmon depends crucially on a diverse and rich store of genetic variation." Some fish biologists in the Northwest now believe that all the hatcheries should be shut down.(See e.g. Bates, S.F. et al. (1993) Searching out the Headwaters. Island Press, Washington, D.C., 98; ‘Status Report: Columbia River Fish Runs and Fisheries 1938–94’, Oregon and Washington Departments of Fish And Wildlife, August, 1995; Nehlsen, W. et al. (1991) ‘Pacific salmon at the crossroads: stocks at risk from California, Oregon, Idaho and Washington’, Fisheries, Vol 16, No. 2; Carle, D.N. (1994) ‘Restore the Endangered Wild Atlantic Salmon’, RESTORE: The North Woods, January; Gillis, A.M. (1995) ‘What’s at stake in the Pacific Northwest salmon debate?’, BioScience, Vol. 45, No. 3, March, 127.)

28. S. Janchitfah (1996) ‘No Real Benefits From Dams’, Sunday Post, Bangkok, 10 March.

29. Ayeni, J.S.O., Roder, W. and Ayanda, J.O. (1994) ‘The Kainji Lake Experience in Africa’, in Cook, C.C. (ed.) Involuntary Resettlement in Africa: Selected Papers from a Conference on Environment and Settlement Issues in Africa. World Bank Technical Paper 112.