Final Paper - Executive Summary
November 2000

The World Commission on Dams (WCD) has completed a global survey of the performance, impacts and decision-making aspects of 125 large dam projects - the Cross-Check Survey.

The primary aim of the Survey was to determine broad patterns and trends indicative of past performance, impacts and decision-making aspects of a selection of large dams that reflect the global population. Consequently, the findings of the Survey provided a link between the broad patterns and trends emerging at a global and regional level and the more in-depth findings provided by the other components of the WCD knowledge base. Thus, the Survey provided an entry point to "cross-check" the collective evidence of the WCD knowledge base - comprised of over 900 topic-related written submissions, four regional consultations and workshops, 17 thematic reviews, 8 case studies and three country studies. The Cross-Check Survey sought to inform the WCD knowledge base on large dams in a responsible and consequential way but without any claims for being necessarily complete.

The Cross-Check Survey does not provide assessments of individual dam projects nor does it claim to be statistically representative of the total global population of large dams. Rather, statistical significances have been largely omitted due to a relatively small sample size. Descriptive and informational objectives were chosen as a higher priority, by the WCD.

Numerous challenges were faced in applying a global survey of this nature as part of a time-limited process. Due to several factors (including logistical difficulties, inadequate records kept, controversy leading to denial of permission or lack of political expediency), firsthand information for some large dams was inaccessible. However, at the close of the Survey process, contributions were received and verified for over 80% of the original targeted sample of approximately 154 large dams.

Coverage and diversity achieved:

The Survey sample targeted a wide variety of large dams: from big controversial projects to smaller, less-contested projects, the performance and impacts of which have often gone unnoticed. These smaller projects have seldom gained recognition in existing studies despite the fact that these dams (15m to 30m in height) comprise two-thirds of the total global population of around 45 000 large dams. Within size ranges, a variety of single and multipurpose dams of different functions and age were selected and analysed from existing studies, WCD Case Studies and the 1999 ICOLD Register of Large Dams to provide diversity to the Survey sample.

In addition to this spread of dam types, the Cross-Check Survey achieved extensive geographical coverage across 52 countries in six major regions of the world.

The Survey process reflects an international, multi-stakeholder collaboration in which over 100 contributors of differing constituencies participated. The Survey was conducted via a set of questionnaires addressing, in a broad manner, questions pertaining to large dam performance, environmental and social impacts, and decision-making trends and related activities.

Data confidence:

Extensive 'safety nets' were incorporated in the Survey process to address the paucity of reliable data pertaining to certain categories, the balance of contributions from different constituencies (eg government, private, NGO) and the overall integrity of data submitted. This data refinement and enhancement process incorporated rigorous data verification, internal and external independent review phases, a WCD literature review, and iterative refinement of information by contributors. Through these processes, many sub-standard sections of individual questionnaires were enhanced to a satisfactory condition before conducting final analysis.

However, for a portion of dams in the Survey, the safety nets did not preclude dam-specific data gaps for certain categories in the questionnaires. In two cases, for example, questionnaires were excluded from the analysis reporting little or no data for the key fields required.

The occurrence of spurious data results or significant outliers arising from the analysis of the final sample of dams in the Survey was negligible, improving reliability and confidence in the emerging findings.

Indicators used in analysis of the Survey data:

Whilst drawn directly from the data fields of the Survey questionnaire, care was taken to harmonise the indicators used for deriving the findings of the Survey with internationally acceptable measures for large dam development projects. Once analysed, the challenge in discerning between permutations of results proved difficult, since those findings exhibiting marked differences across categories to those illustrating no marked differences do not necessarily imply greater significance. Where existing literature and other studies provided a platform for further investigation, emerging findings were reported primarily for those results showing a notable or marked pattern or trend in performance. This was complemented by the intuitive and collective experiences of the WCD Secretariat and Commissioners, although care was taken not to preclude certain findings ensuring that unanticipated trends or patterns were captured accordingly.

The findings gleaned from the final analysis of large dams in the Survey yielded many valuable insights and provided broad support for much of the evidence elaborated on by the WCD Case Studies, Thematic Reviews and existing literature. The emerging patterns and trends largely corroborate the supporting literature and hypotheses on large dam performance. In addition, certain findings added new and significant insights into the variability of project performance, accuracy in prediction of impacts, and marked trends in activities influencing decision-making through time, yielding a consequential set of derived indicators. Of 32 indicators derived in the Cross-Check Survey, 24 were used as primary entry points for emerging evidence in the Final Report of the World Commission on Dams.

Key findings and conclusions:

While there is great variability in the performance of the large dams contained in the Cross-Check survey the information indicates that the majority of dams in the survey under-performed with respect to the achievement of intended benefits and delivery of services. In some instances, benefits occurred for much longer periods than predicted in the studies and still continue. Adverse impacts on ecosystems occur frequently and a significant number of these adverse impacts are still unanticipated in the planning and decision-making.

For the dams in the Cross-Check survey that involve displacement of people, there was systematic underestimation of the number of families and people physically displaced and involuntarily resettled. The lack of enumeration of social records for reporting these aspects also remains a contested issue that continues to fuel controversy in the large dams debate.

The parameters for decision-making on dams have been changing over time. While there is a clear evidence of increased attention to social and environmental aspects in decision-making, technical, financial and economic activities still remain the most frequent overriding decisive factors.

The following are some of the main patterns and trends emerged for the dams in the Cross-Check survey:

I. Performance

1. There is a marked tendency towards schedule delays for large dam projects. Of 99 dams analysed the average schedule delay was 1.75 years.
2. Cost performance data shows that the average financial cost overrun of 81 projects included in the analysis of this indicator was 56%. There was a clear upward trend in cost overruns from the 1950s through to the 1970-80s. These ranged from a 9% average overrun in the 1950s, steadily increasing to 85% average overrun in the 1970s. The upward trend is arrested in the transition to the 1980s and diminishes rapidly again in the 1990s with an average overrun of 24%.
3. Irrigation components fell well short of targets in terms of irrigation command area developed, actual irrigated area achieved and to a lesser extent, the cropping intensity. It was not possible to obtain data on predicted versus actual crop values. A high degree of variability was observed between projects with actual achievement closer to targets 30 years from commissioning. Yet, data obtained on smaller dams suggested greater consistencies in performing closer to targets than larger ones.
4. In contrast to irrigation, the hydropower performance of large dams was on average closer to target. But as with irrigation dams, the variability in performance across the projects was significant. Dams that over-performed sustained the average performance in the sample. This masks the variability in performance that is weighted towards a slight shortfall in overall power delivery for the Cross-Check sample.
5. One-quarter of the 29 dams in the sample with a water supply function have to date delivered less than 50% of target. About one-quarter of dams surveyed over-performed, of which most were multipurpose projects. However, the longer planning horizon for the full development of water supply schemes (as opposed to irrigation or hydropower) is evident from the data on rate of development. Larger than necessary reservoir storage may also reflect over-estimates of water demand, or high reserve capacities for anticipated drought events.
6. The performance of flood control was difficult to measure because floods are catastrophic events and highly variable in nature. The more detailed case studies are more useful in assessing performance in this area. The varying experiences of operators of dams in the Cross-Check survey also provided anecdotal evidence of the operational conflicts that occur for about one-half of the dams for which flood control is a primary or secondary purpose.
7. With regard to single and multipurpose projects, emerging patterns suggest higher variability and lower average performance of multipurpose projects when compared to targets.

II. Impacts

1. Many ecosystem impacts (both positive and negative) resulting from the large dams surveyed remain unanticipated even in the 1990s. Furthermore, beyond the trend toward better anticipation of ecosystem impacts in recent decades, the findings conclude that two-thirds of the ecosystem impacts occurring had negative consequences.
2. For 87 projects almost 60% of the impacts identified were unanticipated prior to project construction. Furthermore two thirds of the anticipated ecosystem impacts occurring had negative consequences.
3. More than one-quarter of the dams ranked impedance of migratory fish as the most significant negative impact.
4. Mitigation was the most widely practised response to ecosystem impacts for the large dams in the Survey. Mitigation has failed or worked only sporadically in the case of most dam-induced ecosystem impacts. Of 87 projects for which ecosystem impacts were recorded, mitigation was undertaken for less than one-quarter of the anticipated ecosystem impacts (i.e. in 10% of all ecosystem impacts that occurred).
5. For 47 projects reporting the effectiveness of mitigation measures, only 20% worked very well, 40% did not mitigate the impact, and 40% were moderately effective (average). Thus, less than five percent of ecosystem impacts that occurred across the sample of dams were actually mitigated effectively.
6. The number of people displaced by the dams in the sample suggested that for 42 dams where displacements occurred, a total of 1 937 106 people were actually displaced compared to a predicted total of 1 507 476 (29% underestimated). The underestimates in people to be displaced were most prominent for small reservoir capacities decreasing with reservoir size. The shortfalls in predictions of displaced people seemed to be most marked for dams built in the upper reaches of rivers and in low-income countries.
7. Involuntary resettlement predictions of project-affected people show that for 12 dams surveyed, the total actual number of people resettled was 201 487 compared to a total of 139 922 predicted (or 44% more than predicted).
8. For the dams in the Cross-Check survey, approximately one-in-nine people displaced were not resettled by the official programme but moved on their own.
9. Of the 68 dams reporting displacement less than one-half affected indigenous people. The tendency has been towards an 'all or nothing' scenario suggesting that where indigenous people were displaced, they were the majority and more predominantly the sole ethnic group involved.
10. Those dams built in the tropics or sub-tropics showed a marked tendency for disease proliferation; one-third of the number of actual post-dam infection cases was predicted. Quantitative data received for health impacts requested in the Survey questionnaire were not available in most cases.

III. Decision-making

1. Participation and transparency in decision-making processes was neither open nor inclusive through to the 1980s. There has been a growing emphasis on transparency and participation in decision-making involving large dams, especially in the 1990s. It was found that 50% of the surveyed projects still do not plan for the public participation of affected people. Participation by affected people shows a more recent trend rising only from the 1970s and ending at around 56% of dams in the 1990s. The trend for participation of NGOs was present in 37% of the projects by the 1990s.
2. Economic cost-benefit analyses (CBAs) and financial analysis plans have been the overriding required activities for an increasing number of projects since the 1950s, but suggest signs of tailing off and decreasing since the 1980s to around 70% implementation in the 1990s. Emerging economic appraisal techniques such as risk and distribution analyses were mandated for 20% of large dam projects in the 1990s. Sensitivity analysis is more common and has become standard for donor-financed projects.
3. Over the last two decades, environmental aspects have increasingly been implemented to better inform the decision-making process for project approvals and operation. Technical, economic and financial aspects are still prioritised, having been implemented far more frequent than environmental aspects for the same dams in the 1990s.
4. Environmental impacts assessments (EIAs) are recorded for 55% of dams ? even for those commissioned in the 1990s. Baseline surveys occur at similar levels, with strategic environmental assessments (SEAs) at around one-in-four of projects in the sample.
5. Social baseline surveys and social impacts assessments (SIAs) depict a clear upward trend over time. Social baseline surveys were absent from the sample of dams in the 1950s but implementation steadily rose to reach almost one-half (46%) of the dams in the 1990s. Similarly, for social impact assessments, the frequency of implementation rises to around 40% of the dams in the 1990s. Planning programmes and monitoring of social impacts however, remain a low priority.
6. For 96 dam projects, economic growth was the overriding project development objective. Recovery of capital costs and recovery of operation and maintenance (O&M) costs were the overriding requirements specified in the planning documents. These three activities showed a rapid increase as project objectives in the planning documents, starting in the 1950s and reaching 70% of all the projects from the 1960s to 1990s.
7. Food security and poverty alleviation as project objectives show a marked increase between the 1980s and 1990s and was specified for one-half of the projects in the 1990s. Empowerment of women and indigenous/tribal communities still remain low priorities for project development objectives.
8. River basin authorities (RBAs) play an increasing role in the fulfilment of specific functions. The most notable trend emerging from these activities was that for water rights. Water rights managed by RBAs exhibits a marked upward trend applicable to 90% of the dams in 1990s.
9. Regarding decision-making activities required today, very few contributors - and in some cases water resource government departments - provided absolute clarity on which activities were mandatory decision-making requirements for new dam projects in their respective countries.

Implications and recommendations for further study:

Since the Cross-Check methodology aimed at a responsible sample of large dams that are somewhat indicative of global and regional populations, it seems reasonable to assume that the emerging patterns and trends highlighted here may be generic to many more dams.

From the collective findings on large dam performance in the Cross-Check Survey it is evident that many of the large dams surveyed require redress. This redress should focus on optimising the performance of the benefit streams or services of existing projects. Furthermore, that projects 'in the pipeline' and new future projects should conduct comprehensive reviews of the potential limitations and constraints, as early in the project planning cycle as possible, in order to at least achieve or preferably exceed development targets set.

An imminent call to action should be advocated for further independent studies to extend insights into large dam performance compared to pre-project benchmarks where possible. Furthermore, specific aspects of large dam performance need to be reviewed for more conclusive findings. Among others, these should include: flood control performance; downstream impacts; cumulative impacts, operation and maintenance (O&M) variable cost predictions, and a more detailed assessment of the effectiveness of mitigation measures undertaken. Also, the accurate assessment of performance regarding both compensation and resettlement costs requires further detailed investigation and is recommended for further studies similar in nature.

There is a need for further studies to better understand the 'softer' issues through investigation of the causal mechanisms for second and third order impacts and the corresponding influences on, and relationships to, communities at risk. This should involve a more qualitative assessment of the underlying issues and options available for environmental enhancement and avoidance of impacts.

In conclusion, as a global survey integral to the independent, participatory and multi-stakeholder approach of the World Commission on Dams, the Cross-Check Survey presents significant and consequential insights into broad patterns and trends of performance, impacts and decision-making aspects of large dams.