A Review of the Epupa Draft Feasibility Study

Steve Rivkin
Wednesday, January 7, 1998

The author reviewed all sections relating to the project’s economic viability, except for the detailed appendix chapters, which were not available to the public outside of Namibia. The author’s comments are his own, and do not reflect the opinion of the institutions with which he is associated.

The draft feasibility study examines the costs and benefits of the Epupa dam hydropower project. It reaches the conclusion that the expected benefits exceed the expected costs and that the dam is economically viable. It qualifies this conclusion by noting that certain environmental and social costs could not be quantified, therefore the ultimate decision depends upon whether the difference between the quantified benefits and costs is large enough to offset the necessarily subjective valuation of the difference between nonquantified costs and benefits.

It is important to remember that the numbers in the draft proposal are estimates, not fact. The validity of these estimates rest upon the accuracy of each of the various benefit and cost components. In some cases estimation is simple and straightforward, but as the consultants acknowledge, in other instances the there is a substantial amount of speculation. Though a 20% contingency is added to the figures and some sensitivity testing is performed, a number of assumptions seem unreasonably slanted in favor of the project. These include but are not limited to estimates of the future growth in energy demand and increases in the future price of imported power. Given the historical record, it is highly unlikely that both energy demand and the price of imported power will grow at the rate needed to make the project economically viable.

Apart from the quantifiable costs and benefits, the two other crucial determinents of project viability are the appropriate discount rate and the value of the nonquantifiable environmental and social costs. Because the project is very risky and permanent and a large fraction of total investment in Namibia, the appropriate discount rate (required rate of return) is almost certainly much higher than 10%, which makes the project economically unviable even if the nonquantifiable costs are ignored. The fact that this project would permanently eliminate the homeland for the Himba people and the precious natural resource of Epupa Falls reduces further the likelihood that this project will benefit the people of Namibia.

The Epupa hydropower project offers Namibia the least flexibility and the most risk compared to other energy options. I do not believe that it passes the narrow test of economic viability using only the quantifiable costs and benefits, much less so when all costs and benefits are considered. I strongly recommend that the project not be undertaken.

My detailed comments are divided into four sections corresponding to four components of the study’s cost-benefit analysis: 1) expected costs; 2) expected benefits; 3) the discount rate; and 4) sensitivity analysis.

The costs of the project include the direct capital and maintenance costs of the dam and power plant and the environmental and social costs that result from dam construction. The study does not adequately address the likelihood of major cost overruns for the project. Large dams are notorious for having cost- and time-overruns, even when managed to "best practice" standards such as those imposed by the World Bank. For example, Chixoy Dam in Guatemala cost 136% more than its projected cost of US$400 million (adjusted for inflation), and took 50% more time to build than planned. Cost overruns of many hundreds of percentage points are not uncommon-- China’s Gezhoube cost 270% more than planned and Australia’s Pieman River Dam cost 200% more (both adjusted for inflation), while Brazil’s Itaipu cost 480% more to build (not adjusted for inflation). The World Bank found that inflation-adjusted construction cost overruns on 70 hydropower dams it had funded since the 1960s averaged 30 percent, almost three times higher than the average ocost overruns on a similar number of Bank-financed thermal power stations.

There is no reference to disassembly costs after the projected 50-year life of the project. While these costs will be devalued substantially by being so far into the future, the omission of any such discussion is inappropriate and makes a full accounting of the project’s costs impossible. Although there are no figures or industry standards for decommissioning very large dams such as this one, there are some examples of cost estimates for smaller dams. The decommissioning costs for two dams in the state of Washington in the US, the Elwha and Glines Dams (30 meter high and 70, respectively), range from $67-80 million for both, with the cost going as high at $148-200 million if the cost of restoring the river and compensating the dam owners for loss of power and income are taken into account.

One cost adjustment that is made on the basis of extremely limited information is that the opportunity cost of unskilled labor is equal to 50% of current wages (in chapter 23 a reference is made to a published study suggesting that the opportunity cost equals 35% of wages, but rounding during calculations pushes the value to 50%). Because it is important to measure labor and capital costs in terms of lost production elsewhere in the economy, the decision to adjust the cost of unskilled labor is theoretically correct. However, if the true alternative opportunities of workers exceed 50% of current wages, or if wages are so high because workers must be compensated for having to live far from home or in difficult working conditions, the adjustment would not be valid and the costs would be understated.

The environmental and social costs are typically much more difficult to value in monetary terms. The consultant attempts to place a valuation on some of these costs, and simply describes other costs and benefits. The largest of these costs results from the creation of the reservoir and consequent inundation of land. The land is valued at $2.66 million dollars plus the compensation paid to the Himba people. This number is small relative to the total project price, but this valuation and proposed compensation might be considered unreasonably low. In Chapter 15, the consultant points out that the valuation of land and tourism is quite complicated. The consultant also estimates an annual loss from a decline in tourism of $300,000. This figure might be too low, particular as Epupa Falls becomes an increasingly rare place of natural beauty and per capita income increases throughout the world, thereby increasing the demand for ecotourism.

A social cost that is discussed but not valued is the increase in disease and social dislocation resulting from the construction project. Costs of disease mitigation are included, but there are likely to be substantial costs imposed on indigenous peoples including increased rates of HIV infection, hepatitis, water-borne diseases and other health impacts as a result of the project. The report discusses the ways in which such problems are often quantified, but chooses not to include these health costs in the cost-benefit analysis. If there is an attempt to quantify these costs, it should not be done (as the report suggests) just on the basis of income lost due to illness, rather the value of time lost outside of work should also be counted.

The Feasibility Study estimates the value of the power consumed today and predicts future growth in the demand for power. It uses these predictions to estimate the yearly consumer surplus--i.e., the benefits of the project to users of electricity. The study provides three possible growth scenarios and uses the middle one. It seems like a compromise in comparison to the other two, but upon closer inspection the middle growth scenario appears to be quite high in comparison to recent experience. Between 1988 and 1995 energy demand increased at a rate of 3.7% per year during a period in which the price of electricity fell slightly. The middle scenario used by the study predicts that energy demand will increase at a substantially higher rate during the coming years despite the fact that cost of energy is expected to INCREASE substantially during this period. This implies that trends in some or all of the determinents of energy demand will diverge sharply from recent experiences.

The report asserts that energy use is quite insensitive to changes in price, though no evidence is provided. It is likely that the low price of imported power has increased current consumption, which raises the baseline figure used to predict the future demand for power. It is also quite likely that increases in the future price of power will curb demand growth, in part by increasing the incentive to conserve energy. It is crucial to remember that if the actual increase in energy demand is even slightly below the estimated middle scenario, the project is no longer economically feasible. The very optimistic estimates of future demand increases given past experiences raise serious questions concerning project feasibility.

Estimates of the costs of alternative power sources are a crucial part of the analysis. Imports and the natural-gas power plant constitute the least-cost alternatives. The estimate of the cost of natural gas produced energy appears to be based on fairly good evidence. In contrast, estimates of the future price of imported power are based on very limited evidence and run contrary to the statements the RSA power supplier. Because a small overestimate in the price of imported power has a large impact on the desirability of the project, this is a very important issue. The current price of imported power is 1.5 cents per kWh, but the study asserts that this price is exceptionally low and will certainly increase in the future based on predictions of power demand and supply throughout the region. It further states that excess capacity will be fully digested by 2005. These predictions contradict the statements by those in South Africa, which found in the Feasibility Study (Chapter 23, page 7 of the report states that RSA’s Eskom believes that it will not need new capacity until 2010, not 2005). The timing of construction of power lines to accommodate the increased imports is also important: the further into the future they are needed, the lower the cost of the import alternative.

A clear disadvantage of Epupa in comparison to imports is a lack of flexibility. The project is only feasible economically if there is a substantial increase in the demand for electricity. If economic growth is lower than expected (either due to lower population growth or reduced per capita demand, which could come from slower economic growth, conservation or changes in industry needs), the benefits of the project will not outweigh the costs. If Namibia were to continue to import, however, it would be the foreign company that would
have to bear the burden of the decline in demand for power. By avoiding large capital investments, imports in a sense provide insurance against the uncertainty of future demand, even if Epupa were the least-cost solution under the most probably scenario.

A difficult-to-quantify benefit of the project is the increase in self-sufficiency that it will bring to Namibia. Even if estimates of future import prices overstate the actual future prices, the country would have the benefit of reduced dependence on foreign generated power. The question is whether this trend toward imports at this time is good or bad. From a purely economic position, this trend reflects appropriate decision-making on the part of the Namibian government. The price of imported power is extremely low, and there is no reason to devote resources to increase domestic production unless expected future demands justify such an investment or there are legitimate political concerns about the availability or price stability of imported power. Imported power lights homes just as well as domestically produced power, and foreign trade improves the average standard of living of both countries. Given that the trend is toward regional interdependence, especially in the energy sector, and that South Africa will likely buy Kudu gas, the argument for "energy independence" seems to be a red herring. Moreover, power generated by Epupa is also sensitive to the actions of foreign countries. Namibia shares the river with Angola, and increased water use by Angola over the next 50 years, a distinct possibility, could impose substantial additional costs and reduce the return on the investment. The experiences of Gove and other dams in Africa also reveal that hydropower stations are quite vulnerable to military attack.

The discount rate is the estimated return of the next best alternative use of a scarce resource. In this study the discount rate is assumed to be 10%, the standard rate of discount used by the World Bank for lesser developed countries. It is by no means clear that this rate applies to Namibia. If in fact there are a number of projects with expected returns greater than 10%, the appropriate discount rate for Namibia might be higher, in which case the project would not pass the test of economic feasibility.

A second important issue related to the discount rate is the notion of risk. An implicit assumption underlying the type of project analysis used here is that any single project is only a small part of the total investment portfolio, and therefore that the deviations from expected returns will offset each other. This is not true in Namibia, where the cost of the Epupa Dam project would be extremely large relative to total government investment expenditures (the $500 million dollar price tag is roughly 40% of annual government expenditures in 1995, according to World Bank figures). Consequently, the very risky and unpredictable nature of this project should be considered in the determination of the appropriate rate of return required for project viability. Even if 10% is the appropriate discount rate for smaller or less risky projects, discount rates of 13%, 15% or even higher are more appropriate for large, risky projects such as Epupa. As a consequence, the project would no longer pass the economic feasability test.

Sensitivity analysis investigates the robustness of conclusions about economic viability to changes in assumptions about future costs and benefits such as the future demand for power or future price of imported power (remember that all future events are uncertain). There are two main problems with the sensitivity analysis presented in this feasibility study. First, it does not provide results showing the sensitivity to such important assumptions as the expected growth in future demand for power. Rather it investigates changes in only a subset of the variables.

Second, the draft study reports on a number of different sensitivity analyses conducted one at a time. But since some of the problems are probably linked together, this one-at-a-time approach does not reflect some likely scenarios that could negatively affect the project’s viability. For example, if there is a one year delay in operation, there is probably also a cost over-run. However, the feasibility study looked instead at the changes to project economics if either one or the other of these related scenarios occurred. The assumption that the problems are independent minimizes the likelihood that unexpected events will reduce the return on the project to below 10%.

Finally, there is also no way to conduct an appropriate sensitivity analysis for environmental problems that are difficult to value. The study points out in chapter 20 that uncertainties in ecological response require monitoring and possible future operational constraints. It is difficult to incorporate information about the unknown, which leads to the 20% contingency cost component, which may or may not be the correct addition to costs.

This project would devote a substantial amount of resources to a single large project that is unlikely to be economically viable even ignoring the nonquantifiable costs and using the chosen discount rate. Because there is no immediate demand for the additional electricity, and because there are better options, there is no reason to undertake this uncertain, high cost, nonreversible investment at the current time. A decision to not build the dam has a number of benefits, including the possibility that improvements in other energy alternatives will make their adoption cost-efficient in addition to the previously discussed flexibility of using imported power. There are likely to be a variety of other projects in Namibia that would generate greater benefits to society (and a higher expected return), while also having a more positive impact on the distribution of income and far fewer negative social and environmental consequences. Finally, it would also give the government a chance to find a solution for its energy needs that will not destroy the homeland of the Himba people and the river that they depend on.

Prepared by Steve Rivkin
Department of Economics
Amherst College
Amherst, MA 01002
Tel: (413) 542-2106
Fax: (413) 542-2090

Current Contact Information:
University of California, Berkeley
Deptartment of Economics
549 Evans Hall, #3880
Berkeley, CA USA 94720
E-mail: rivkin@econ.Berkeley.edu

Further Reviews: