Wayne C. White

September 26, 1997

Review of Louis Berger Economic Impact Study

AN UNFORMATTED COPY OF DR WHITE’S REVISED REVIEW IS PRINTED BELOW. IF YOU
REQUIRE A FORMATTED VERSION, OR A COPY OF DR WHITE’S RESPONSE TO NTEC
COMMENTS, PLEASE EMAIL AVIVA@IRN.ORG AND SPECIFY WHETHER YOU NEED IT IN MAC
OR IBM FORMAT.

26 September 1997

Mr. Somphone Phanousith
Director of Cabinet
Science Technology and Environment Organization
PO Box 2279
Vientiane
Lao PDR

Fax. 0015 85621213472

Dear Mr Somphone

Review of Louis Berger Economic Impact Study

Following please find a revised version of Dr Wayne White’s review of the
Louis Berger Economic Impact Study for Nam Theun 2. This version includes
consideration of the revised edition of the Louis Berger report, dated 28
July 1997, and comments from Mr Scott Thomas, Team Leader of the Economic
Impact Study.

Dr White’s review of the June report asserted that the “pessimistic” or
“nightmare” scenario for the project would be significantly worse than
presented if the Study team had addressed all the legitimate project risks.
The July report appears to reinforce this. IRN’s understanding of Dr
White’s revised review is that Louis Berger’s assessments of rate of return
and net present value to Government of Lao were maintained by making some
radical and arbitrary cuts in valuation of both lost forestry earnings and
the Land Opportunity Cost.

According to Dr White, “The July version of the study has lower project
benefits in terms of project revenues, yet presents a higher net present
value and rate of return, especially for the Lao government. It is able to
do so because it greatly reduced its tabulation of land opportunity cost
including forestry, which increases the doubt as to the appropriateness of
the calculation.”

In the July version of the study, the land opportunity cost in the nominal
dollar calculation of cost/benefit was reduced from $311.4 million to
$134.5 million, despite the change in interest rate which should have
resulted in an increase in land opportunity cost.

Also attached is a response from Dr White to NTEC’s comments on his review.
Dr White refutes a number of NTEC’s assertions, including the claim that
the Turnkey Contractor would assume all the risks associated with
construction cost overruns and construction delays. According to Dr White,
“turnkey international contracts do vary in actual risk exposure, and they
are not immune to problems caused by cost overruns or construction delays.”

I hope that you find this information useful. Please contact myself or
Patrick McCully if you have any queries on Dr. White’s or IRN’s work.

Yours sincerely

Aviva Imhof
Mekong Program Coordinator
International Rivers Network

cc James D. Wolfensohn, President, World Bank
Jannik Lindbaek, Executive Vice President, IFC
Jean-Michel Severino, Vice-President East Asia and Pacific, World Bank
Jayashankar Shivakumar, Lao PDR Country Director, World Bank
David McDowell, Director-General, IUCN
World Bank Executive Directors
Scott Thomas, LBI, Inc.

__________________________

Review of Economic Impact Study: Nam Theun 2 Hydroelectric Project

Wayne C. White, Ph.D.
Foresight Associates

Note: The original version of this review was issued 1 August 1997.
Subsequently Louis Berger International issued a revised edition of the
report dated 28 July 1997, and Mr. Scott Thomas, Team Leader of the
Economic Impact Study, issued a letter in response to our review. This
revised version of the review includes consideration of the revisions to
the Study, and Mr Thomas’ letter.

All original content has been retained, superseded items are set
off in brackets, additions and changes are given in italicized type.

I. Overview

The scope of this report, in accordance with its terms of
reference, is to critically review the Economic Impact Study of Nam Theun 2
Dam Project prepared by Louis Berger International, Inc. dated June 12,
1997, revised and reissued dated July 28, 1997.

This review finds the Berger study, while in many ways adequate in
its application of conventional analytical methods, has shortcomings which
affect the legitimacy and utility of its conclusions. In contrast, an
economic analysis that accurately reflects the full range of issues
surrounding hydropower development as a means of poverty alleviation and
sustainable development, would have best served the Lao people.

A significant weakness of the Berger study is in fulfilling its
charge to, as part of the project evaluation, construct “realistic
scenarios for…pessimistic conditions.” The Berger study presents what it
alternately terms a “pessimistic” or “nightmare” scenario, yet this
economic calculation, for example, excludes any consideration of
sedimentation in the reservoir, limits its worst case construction cost
overrun to 20 % [15%], and does not consider single or multi-year
construction delays. Familiarity with international hydropower projects
demonstrates that these factors are not only pessimistically possible, they
are typical. For this reason, the statement contained in the study:
“Sensitivity analysis of the model indicates that the net present value of
the project remains positive under all but the most pessimistic of
scenarios” may be technically correct, but is fundamentally misleading.

The study is relatively even more selective in calculating project
benefit to the Lao government than it is in calculating overall project
return. As the report states (Annex 1, para. 48) “the flow of benefits to
the GOL is distributed more heavily to the later periods in project life.”
The later periods of project life are precisely when sedimentation effects
will be felt in diminished production and/or need for capital reinvestment
to maintain operation. The later periods of project life also carry a
greater risk of Thailand turning to cheaper alternate sources of power.
[By using a lower discount rate for GOL analysis (7%) than project analysis
(9%), and given the high distribution to the government in later periods,
the Berger calculation further favors an optimistic net present value for
the government. Further, the fact that the study uses a base Thai
inflation rate of 3%, and a worst case of 4%, when the present inflation
rate in Thailand is 6%, has a strong impact in bringing the supposed
earnings in later periods to a present value.] (The subsequent version of
the study does rationalize discount rates at 7%, and utilize a base
inflation rate for Thailand of 6%. More on this below.)

We also find the study to have assumptions which systematically
raise the “optimistic” and “pessimistic” case projections. The base
electrical tariff rate used is, in our opinion, high. [The Thai inflation
rate assumption also presents a systematic bias.]

II. Review of assumptions

Scoping the IRR calculation. We agree with the statement (page 3, para. 7)
that cost overruns, demand shortfalls and hydrological variations are among
the most obvious sources of risk. We cannot agree that they can be
downplayed, simply because private sector, project finance may be employed.
The presumption that “market signals” demonstrate viability, when
appearing in a economic study that should help the Lao government evaluate
the developer proposals, is tautological at best. The existence of BOT
finance does not change, in the case of hydrology, the performance of the
weather.

The presumption in the “market signal” discussion by Berger seems
to be that if shortfalls do occur, the project developers, NTEC, will
absorb any losses and the government will be not only fully protected, but
continue to receive full dividend, royalty and tax payments–this is not at
all a safe assumption, not before the contracts are negotiated and signed,
and quite probably not after they have been either.

Even if losses were limited to NTEC, this affects the GOL as an
equity participant.

Responsible characterization of these risks, however, would provide
invaluable background information to the Lao negotiating team as they enter
into the next round of negotiations–assuming the project proceeds to that
point.

Discount rate. [The study uses a 9% discount rate for project analysis.
It uses a 7% rate for the GOL case, a lower rate which increases the
calculated present value of the projected GOL earnings far in the future.
The study uses (Appendix 1, page 1) a discount rate of 10% for calculating
the present value of lost future forestry earnings from the Nakai Plateau,
a rate higher than the 7% or 9% decreases the calculated present value of
lost sustainable forestry earnings.]

The subsequent version of the study does rationalize the discount
rate at 7%. The revised calculations, however, do not show the present
value of lost forestry earnings increasing as would be expected, but
rather decreasing! This was accomplished by lowering the unit value of
wood in the calculation. (See Annex 2, para. 40) The June version shows
the range of forestry present value from a low of $22million, to a high of
$45mil. For a period of 50 years, the present worth factor at 10% is
9.915, and at 7% is 13.801, so we would correspondingly expect the forestry
present value to be about 39% higher at the 7% discount rate, therefore $31
mil. low, $63 mil. high. Instead the July version of the study cites
values of $20 mil. low, $38 mil. high.

Similarly, the Land Opportunity Cost used in the nominal dollar
calculation of Cost/Benefit (Annex 1, pages 28, 29) fell from $311.4
million to $134.5 mil. between the two studies, despite the change in
interest rate which trended the opposite direction!

Inflation rate. [The study uses a base rate for Thai inflation of 3%, with
2% as “rosy” and 4% as “nightmare;” the current rate of inflation in
Thailand is 6%. As the study says (Annex 1, p.16, para. 62) “lower
inflation works to the advantage of NT2.” The study does not adequately
represent the worst case for inflation.] The study now uses a base case
Thai inflation rate of 6%.

The study’s sensitivity test for inflation shows the project net
present value to the Lao government at the end of 30 years falling by $51
million with inflation rate only 1percentage point higher than the 3% base
case, and by $82 million when carried out to 2050. It is beyond the scope
of this review to calculate a true worst case including the higher
inflation scenario, but an inflation assumption of 5% would in itself lower
the net present value to the GOL by a further tens of millions of dollars.
The earnings to the Lao government are shown to drop in the later report,
which combines the effects of higher inflation, lower tariffs, and other
factors in the calculation.

Construction cost estimates. We maintain that it was a positive and
appropriate development for the government to employ “an outside general
engineering consultant” (page 4, para. 10) for tasks including generating
“independent cost estimates.” It is appropriate for Berger to utilize
Lahmeyer cost estimates as a construction cost basis.

We do feel that Berger responsibly should have subjected the
Lahmeyer-generated base construction costs to more realistic scenarios for
cost overruns and construction delays.

By application of the [9%] 7% real discount rate to the Lahmeyer
projected construction cost schedule, we derive a construction cost of
[US$995 million] US$931 million in the first year of project operation. As
a statement of construction costs excluding construction delays and other
cost overruns, it is roughly consistent with previous estimates.

[[
Construction cost (US$millions, 1996 dollars)
Expenditure     Carryover debt @ 9% real discount rate  Annual Sum
1998    $17.40          $17.40
1999    $117.55 $18.97  $136.52
2000    $231.63 $148.80 $380.43
2001    $231.63 $414.67 $646.30
2002    $120.55 $704.47 $825.02
2003    $14.00  $899.27 $913.27
2004            $995.46 $995.46
]]
Construction cost (US$millions, 1996 dollars)
Expenditure     Carryover debt @ 7% real discount rate  Annual Sum
1998    $17.40          $17.40
1999    $117.55 $18.62  $136.17
2000    $231.63 $145.70 $377.33
2001    $231.63 $403.74 $635.37
2002    $120.55 $679.85 $800.40
2003    $14.00  $856.43 $870.43
2004            $931.36 $931.36

Berger utilizes a base case with no cost overrun, and a worst
economic case of a [15%] 20% cost overrun, 30% in the financial risk
calculation. According to the 1996 World Bank Technical Paper No. 325,
Estimating Construction Costs and Schedules: Experience with Power
Generation Projects in Developing Countries, hydropower projects supported
by the World Bank have an “average cost overrun of 27 percent.” Based on
the experience elaborated in that document, it would have been appropriate
for Berger to have used no-cost-overrun only for the best case scenario, a
27% cost overrun for the base case, and a 50 to 100% cost overrun for the
“nightmare” or “worst case.”

As factors leading to possible construction delays and cost
overruns, we note geological uncertainty, which will affect tunnelling
operations, and possibly require structural shoring or ‘plugging’ of the
hills which will form the basin rim as the reservoir begins to fill. The
Berger study has noted (page 41, para. 8) “In regard to cost over-runs, a
preliminary review of the contract indicates that NTEC would be exposed to
additional costs related to ground conditions outside of a ‘Defined Work
Area,’ for Force Majeure events, and for lack of adequate water during
acceptance testing.” We do note the relatively good transportation access
to the site, yet we feel that logistics will present a very real challenge.

The study (Annex 1, page 18,19) presents a comparison of the
projected cost of electricity from NT2 in comparison with many other Asian
hydropower projects. This information indicates that the calculated cost
for NT2 is markedly lower than the average value of projects in Laos, and
in Southeast Asia. What is not confirmed yet is whether this comparison
demonstrates that, a) NT2 is a low cost hydropower project, or b) the NT2
cost is being appreciably underestimated.

Ecosystem preservation. The study makes its assumption clear (page 4,
para. 12) to accept the argument that project development will favor
protection of a conservation area, while failure to develop the project
will mean the would-be conservation area goes unprotected while the
reservoir site gets degraded anyway. We must note that this stance is seen
as arbitrary and unsupported by some interested parties. Arguments against
this stance would point out that it ignores the question of reversibility,
that greater logging of the Nakai Plateau to a certain point does not rule
out future ecosystem recovery while a reservoir would. Also, it could be
argued, what is the source of confidence that the watershed of the
reservoir won’t be clearcut after construction, resulting in both
environmental chaos and massive sedimentation of the reservoir?
Expenditures for preservation do not ensure results. Would the regional
development company, really have an interest in the medium term of
protecting the watershed rather than cashing in on the timber?

Without the reservoir, we would find it economically rational for
the government to pursue, and donors to support, programs of sustainable
forestry.

The calculation of the lost future timber harvests (Appendix 1,
page 2) ignores the rising real value of timber in international markets,
especially the Tokyo Exchange, in addition to the concerns mentioned above
under ‘Discount rate.’

Environmental and social cost mitigation. We would like to stress that
mitigation relies not only on the amount developers are willing to spend
(page 5, para. 13) but the efficacy of their efforts. Annex 2 of the
report provides an excellent enumeration of possible environmental and
social impacts, and makes observations that would facilitate mitigation
planning. Without careful assignment of authority and meaningful
conditionality in the contractual agreements, however, the unmitigated
costs will reach 100% despite any mitigation spending.

We support the study for the assumption (page 5, para. 14) that 50%
of the environmental and social costs will not be mitigated as a factor in
this economic analysis, as a more realistic and less conservative
assumption than might have been used, without in any way endorsing the
reality which this reflects. We also note the corollary to the assumption
that 50% of the environmental and social costs will not be mitigated: that
watershed degradation will lead to some heightened degree of sedimentation,
and that some citizens will experience a dimunition in living standard, in
violation of World Bank policy.

Reinvestment. We note and find appropriate the discussion (page 5, para.
16) of the ability of the Lao government to re-invest project earnings;
the study is correct to term NT2 an ‘enclave’ project. Such reinvestment
is a necessary component if development of the NT2 project is to contribute
to poverty alleviation in Laos. We believe that for net capital within the
Lao economy to have experienced a net increase at the end of service life
is a key criteria in assessing NT2 as a vehicle of sustainable development.

We note, however, that within the project analysis earnings to the
GOL are taken as a proxy for benefit to the Lao economy, functionally
assuming perfect reinvestment in that context. Although the study purports
to analyze “institutional capacity constraints” and present a “strategy for
their mitigation,” such discussion is not equivalent with implementing
successfully a program that achieves perfect reinvestment capability.

Hydrological risk. [The calculation uses (page 9, para. 6) a monte carlo
simulation to represent hydrological constraints to power generation. Our
apprehension is that the monte carlo simulation expresses hydrological
variation, but does not adequately address the possibility of hydrological
shortfall.] The monte carlo simulation has been removed from the primary
benefit/cost calculations, which now carry constant annual generation
estimates. The possibility of any hydrological variation is now treated
only within the sensitivity analysis. The project hydrological models are
built upon rainfall measurements taken in too few locations over too short
a span of time to provide a high level of confidence. By its nature,
modelling from inadequate data “inherently results in a lower variability
in the estimated flows than occurs during the observed sequence.”
Limitations included “an extremely poor network of rainfall stations.”

Again, generation rates of the so-called ‘base case’ ‘most likely
scenario’ are now set at idealized constant levels in the calculation.

Tariffs. The Berger study cites (Annex 1, para. 53) the tariff given in
the now expired power purchase agreement, which was 4.55 cents/kWh
escalated at 3% during construction and 35% of the rate of inflation during
operation. The study then goes on to note that Thailand’s avoided cost of
power is lower than this amount and falling, therefore “the new tariff can
be expected to be lower than the previous agreement.” The study then goes
on (para. 54) to use a base rate of “5.75 cents/kWh in 2004” increasing “at
35% of inflation during operation as in the previous agreement.” Please
note that an amount of 4.55 cents in 1996, raised by 3% a year inflation
for eight years, is equivalent to 5.76 cents in 2004.

To summarize the previous paragraph, Berger uses a base rate tariff
in its calculations which it states (in para. 53) is artificially high. In
explaining its use of this number it cites “conversations with NTEC and the
World Bank.”

The study now uses a tariff of 5.70 cents which it holds constant
in nominal terms with no inflation factor.

In 1995, the average tariff at which the Lao utility exported
electricty was 3.2 cents per kWh. Raised for inflation by the referenced
amount, this is equivalent to 3.3 cents/kWh in 1996, only 72% of the
referenced 4.55 cents/kWh. It is equivalent to 4.05 cents in 2003, or 71%
of the 5.70 cent tariff.

Berger also recognizes a scenario in which combined cycle gas
plants in Thailand create a competitive tariff price of 4.0 cents/kWh in
2003, or thirty percent cheaper than the assumed base case. The study
(Annex 1, para. 70) notes “It is very possible that both the capital costs
of combustion turbines and the production cost of natural gas will stay
flat or decline in real terms in the near future.” Yet the study
marginalizes the significance of low gas-fueled tariffs by citing natural
gas scarcity. We do not feel that the natural gas scarcity argument is
conclusive, and that the study should have given much more credence to a
scenario of continued downward tariff price pressure.

Sedimentation. We agree that (Annex 1, para. 9) sedimentation “may pose a
risk to future NT2 generation output and therefore the revenues and
benefits associated with NT2.” We do not feel that it was appropriate to
perform this study and reach conclusions without factoring in any
sedimentation scenarios.

The report states that, “no comprehensive data nor studies were
available to the project team at the time of the report.” SMEC made an
assessment of sediment loads in 1991; the SMEC data has also been the
foundation for an assessment of sedimentation in Nam Theun 2-Study of
Alternatives by Lahmeyer in February of this year, and A Review of Nam
Theun 2 Project Economics by White in July 1996.

We feel that data exists from a variety of reservoirs in Asia under
similar geological and hydrological conditions that could have been used to
generate at least approximate assumptions for sedimentation rates and the
resulting economic impact. Further, regional experience indicates a true
worst case scenario would consider extreme watershed degradation.

If the study had generated a sedimentation scenario for a watershed
protection worst case, and carried it through to a Lao government project
return calculation, it would have provided a valuable service in
demonstrating to the Lao government its self interest in watershed
protection if the project is built.

III. Review of conclusions

Returns. The Berger study finds (page 9, para. 4) that under the “most
likely” scenarios, the economic internal rate of return for the project
from a project (‘global’) perspective, and GOL perspective, is not less
than [14%] 16%, and the net present value of an initial $100 million GOL
investment is more than [$300 million] $345 million. It goes on to say
“Sensitivity analysis of the model indicates that the net present value of
the project remains positive under all but the most pessimistic of
scenarios.”

The objective of this study (page 1, para. 1) “is to provide the
Government of the Lao People’s Democratic Republic (GOL) with a
comprehensive and well balanced analysis of the economic viability of the
proposed Nam Theun 2 hydroelectric project.” From the indicated returns it
would be logical for the GOL to perceive that the project is not only
viable but attractive, with negligible risk.

A well balanced analysis should point out the real possible
downside of project development, as disappointing and unwelcome as that
news may be. Clearly, the implications of tariff revenues being 20 to 30
percent lower than assumed in the study, and a cost overrun of even 40
percent or less, even if not combined with hydrological shortfall and
diminished project life, present a very real possibility of the project
having a negative NPV, subjecting the GOL to loss of it total investment
while incurring environmental, social, and opportunity costs.

The July version of the study has lower project benefits in terms
of project revenues, yet presents a higher net present value and rate of
return, especially for the Lao government. It is able to do so because it
greatly reduced its tabulation of land opportunity cost including forestry,
which increases the doubt as to the appropriateness of the calculation.

As for the reasoning that the project developers will shield the
GOL from critical cost overruns or earnings shortfalls, it is a given that
the developers are not interested in losing tens or hundreds of millions of
dollars. They will, under such conditions, find a way of passing along
losses, or they will find a pretext for abandoning the project.

Diverging outcomes. The GOL should note (page 14, para. 19) that under the
“nightmare” scenario, which for reasons given above we do not at all
consider to represent the worst case, in the first 30 years the GOL loses
[$95 million] $69 million, while the project over all retains a positive
net present value. The study cites the current draft of the Concession
Agreement (page 41, para. 9) to point out that not only is the GOL being
asked to accept diminished royalties in the event of hydrological
shortfall, but also if the developers simply have not achieved their
‘hurdle’ rate of return. The discussion further points out (para. 10)
ways in which the GOL is being exposed to greater risk than the developer,
as the resource levy holiday “significantly exposes the GOL to the risk of
a renegotiated tariff structure in the later operating phase of the
project,” and because the developer is seeking “2-to-1 reimbursement of
development costs” by the commercial operation date.

Alternatives. The Berger study does not, we believe, sufficiently present
the no-build option in its conclusions. We do not believe that the
proposed Nam Theun 2 project, even were it to be of net benefit, is the
only vehicle by which the Lao PDR can pursue its national development
objectives.

A GOL development strategy of sustainable forestry of the Nakai
Plateau has many advantages over the NT2 scheme. It does not require a
US$100 million investment. It could begin today. It does not carry a risk
that the GOL could lose $95 million or more. It has a net present value
from timber sales alone of between $45 and $275 million. It would not
carry a cost of $95 million for combined environmental and social costs and
mitigation measures, although it would require some fraction of that for
management and protection measures. It could provide employment for Lao
citizens, especially if combined with wood products processing and
ecotourism development. It does not carry the same risks for increased
prostitution and accompanying public health concerns, nor waterborne
diseases such as Schistosamiasis. It is compatible with Lao macroeconomic
development needs including enhanced rural development, farm to market
roads, agricultural extension, and education.

Wayne C.
White, Ph.D

August 1, 1997
revised September 5, 1997