The Theory of Water and Utility Pricing - Global Drinking Water Management and Conservation - page 88

Table 5.1 Adjusted double-log regression model (including dummy and interaction terms)
n R 2 R(bar) 2 d.f. SEE/SER Durbin-Watson statistic

36 32 0.998883 0.998779 0.782302197 2.053142

n is the number of observations
d.f. are the degrees of freedom
SEE is the Standard Error of the Estimate

The model contains no heteroscedastic errors or heteroscedasticity and provides
an exceptionally good dent to use this model

for the estimation of Marginal Costs and Shadow Ramsey Prices.

t to our dataset. We can now be con

5.6.1 Derivation of MC for Two Types of Desalination

This section uses the adjusted double-log model presented in Eq. 5.27a, b above
and estimates the MC functions for both brackish water and seawater desalination
plants. The coef cient estimators used are summarized in Table 5.2 .

First, the total cost (TC) function obtained is simpli ed as follows:

e c SEA e dð In QSEA Þ
T C ¼ e a Q b e c SEA e dð In QSEA Þ
T C ¼ e a Q b e c SEA

¼ e a e b In Q

e In T Cy

d SEA

e In Q

This yields the following Total Cost function:
T C ¼ e a Q ðbþd SEA Þ e c SEA
On a priori grounds, it makes good economic sense to separate seawater desa-
lination from brackish water desalination. Accordingly, two Marginal Cost equa-
tions are derived from the TC function above.
For a Brackish water desalinating plant:

f SEA ¼ 0 g

T C ¼ e a Q b

)

MC ¼ e a b Q b ðÞ

)

Using the coef cient estimates obtained in the adjusted log-linear regression

model, we get:

Table 5.2 Estimated
coef cients of the adjusted
log-linear model

ʱ 13.885652232

ʲ 0.740621837

ʳ 0.928025419