The World’s Largest Open Access Agricultural & Applied Economics Digital Library

This document is discoverable and free to researchers across the

globe due to the work of AgEcon Search.

Help ensure our sustainability.

Give to AgEcon Search

AgEcon Search
http://ageconsearch.umn.edu
[email protected]

Papers downloaded from AgEcon Search may be used for non-commercial purposes and personal study only.
No other use, including posting to another Internet site, is permitted without permission from the copyright
owner (not AgEcon Search), or as allowed under the provisions of Fair Use, U.S. Copyright Act, Title 17 U.S.C.

No endorsement of AgEcon Search or its fundraising activities by the author(s) of the following work or their
employer(s) is intended or implied.
 AGRICULTURAL
ECONOMICS
ELSEVIER Agricultural Economics 28 (2003) 1-12
www.elsevier.com/locate/agecon

Welfare consequences of water supply alternatives in rural Tunisia

Slim Zekri a,l Ariel Dinarb·*
'
a Ecole Superieure d'Agriculture de Mograne, University of Tunis, 1121 Zaghouan, Tunis, Tunisia
b Rural Development Department, The World Bank, 1818 H St. Nw, Room MC 5-815, Washington DC 20433, USA
Received 14 November 2000; received in revised form 26 August 2001; accepted 14 November 2001

Abstract
Many economic studies have addressed the issue of inefficiency of public water supply in rural areas and the potential for
improved service by private companies. Many of these analyses focused on identifying either willingness to pay or comparing
average service costs and prices paid by customers. This paper performs a welfare analysis of two water supply systems-
public and self owned-in rural Tunisia. The paper calculates consumer and producer surplus and compares the performance
of the two systems from a social point of view. Results suggest that both systems are inefficient, mainly because of a production
level that is low compared with production capacity.
© 2002 Elsevier Science B. V. All rights reserved.
JEL classification: 018; Q25; R13

Keywords: Rural water supply; Social welfare; Consumer and producer surplus; Water demand and supply

1. Introduction levels of service (Briscoe and de Ferrenti, 1988).

Recent research on willingness to pay for improved
Water is an expensive commodity in rural areas of water supply services 2 in the rural areas of develop-
developing countries. Only 30, 50 and 70% of the rural ing countries indicates very clearly that households
population world wide had access to safe water supply demand a higher level of service (World Bank Water
in 1980, 1990, and 1994, respectively (Table 1). In Demand Research Team, 1993). Other findings sug-
spite of the significant improvements made to facilitate gest that they are willing to spend more to upgrade
access to safe water in rural areas, services provided their level of service, and that they are actually spend-
in many countries are still not up to standard. ing significant amounts of money to augment the low
Water supply policy in developing countries has fo- level of service provided by the public water sup-
cused to a greater extent on population coverage and ply (Atlaf, 1994; Brookshire and Whittington, 1993;
low tariffs (Atlaf, 1994). This policy, coupled with Mangin, 1991; Whittington et al., 1990).
limited government funds, lead to a situation in which Evidence on willingness to pay for improved ser-
highly subsidised water supply systems produce· low vice suggests that existing public supply systems are
associated with social inefficiency. Moreover, the lim-
• Corresponding author. Tel.: +1-202-473-0434; its on the state's ability to provide water supply at a
fax: +1-202-614-0793.
E-mail addresses: [email protected] (S. Zekri), 2 Service levels can be measured by different variables such as:
[email protected] (A. Dinar). amount of water per person per day; number of hours per day for
1 Fax: +216-2-660-563. water supply; water quality; cost structure.

0169-5150/02/$ -see front matter© 2002 Elsevier Science B.V. All rights reserved.
PH: SO 169-5150(02)00032-4
2 S. Zekri, A. Dinar! Agricultural Economics 28 (2003) 1-12

Table I schemes are undertaken as welfare activities under

Percentage of persons with access to safe drinking water various basic needs programs and, therefore, financial
1970 1975 1980 1985 1990 1994 viability cannot be applied to these schemes. Saleth
Tunisia
suggests that financial viability can be improved by
Urban 92 93 100 100 100 considering appropriate water rates structures, which
Rural 17 17 31 89 (68a) will allow capturing both the needs of the greater
Total 49 60 70 99 (87b) proportion of poor in rural communities and the need
World for financial viability. In this regard, reforms of water
Urban 77 82 82
Rural 30 50 70
supply rate design in the urban sector (Boland and
Total 44 61 74 Whittington, 2000) demonstrate the important role of
various tariff structures, and the political economy
Source: all entries are from Tables 5 and 6 in Gleick (1998),
Tables 5 and 6, unless otherwise indicated. Note: empty cells mean
consequences in developing countries that may ex-
no data. plain why various stake holders may resist certain
a INS estimates for 1996 suggest 68%. reforms.
b Based on estimates in MEAT (1997). Economic analysis of public rural water supply
services is in most cases restricted to the village level
higher service level call for alternative solutions such (e.g. Mullick, 1987; Perkins, 1994; Wahadan et al.,
as private companies, community participation and 1990; Roark et al., 1987; Mangin, 1991; Atlaf, 1994;
other decentralised arrangements. The World Bank and Whittington et al., 1990), or it attempts to esti-
Water Demand Research Team (1993) concluded that mate the potential value to consumers of an upgraded
the level of service and the level of cost recovery for service level. Parker and Skytta (2000) also refer to
rural water supply at the village level can be adjusted to this issue by distinguishing among three approaches-
community characterisation and ability to pay, so the top-down, the community-based, and the demand-
that the long-term sustainability of the system can be responsive approaches. However, their review does
achieved. not provide a clear cut among the various approaches.
Parker and Skytta (2000) reviewed 15 World Bank From the literature we have reviewed in this sec-
rural water supply and sanitation projects, with a total tion, we conclude that it is not trivial to demonstrate
investment of US$ 1 billion and 20 million benefi- whether alternative water supply arrangements, such
ciaries. Their findings are very relevant to the work as private sector or community management of the
in this paper. Parker and Skytta (2000) suggest even supply, are socially more efficient. Lack of appropri-
more detailed policy measures than The World Bank ate data and locally restricted analysis may provide
Water Demand Research Team (1993) in that they biased and partial answers.
allow various service levels per village, and request To cast light on these issues, we analyse a dataset
that projects should be adapted to the socio-economic from Tunisia (Zekri, 1999). This dataset allows us to
characteristics of each village served. While we agree compare public provision with alternative provision
that village level design may be a necessary factor of water supply services at sub-regional levels. The
in improving the likelihood of efficient performance paper focuses on various comparisons between pub-
of the system, it may not be a sufficient condition lic and local water supply agencies, including cost
for sustainability and equity. For example, data in effectiveness, price of service and welfare calcula-
Saleth (1996) indicate that water supply and sanita- tions. The next section introduces a simple economic
tion programs in India between 1974 and 1989 could framework by which welfare calculations of public
not produce sufficient revenue for financial sustain- and 'alternative' water supply arrangements can be
ability. Comparing rural with urban performance, it compared. In the third section, a short summary of
is evident that rural schemes have operational costs the rural water supply situation in Tunisia is provided.
that are two to three times higher, and run deficits This is followed by an empirical section, which pro-
that are four times larger than similar sized urban vides calculations of demand and supply of public and
schemes. The explanation given by Saleth (1996) for 'alternative' elasticities; of welfare calculations with
this phenomenon is the fact that rural water supply public and 'alternative' water supply arrangements;
 S. Zekri, A. Dinar I Agricultural Economics 28 (2003) 1-12 3

and of comparisons between cost effectiveness and ing and that the alternative supply function (SA) is
price of service of the two supply arrangements. The monotonically increasing (Fig. 1). Public supply at
paper concludes with a discussion of the social con- service level Qp (the actual service level is lower than
sequences of the two rural water supply systems in Qp because users will tum away to other alternatives,
Tunisia. leading to lower cost recovery rates and so on ... ) is
provided at price Pp. At Pp the quantity demanded
is Qp' (>Qp ), but this quantity cannot be provided
2. Analytical framework by public supply. Consumer surplus at Pp is Ppbcd',
which does not reach its potential level Ppbe, and pro-
Why and when should public rural water supply ducer surplus at this point is aPpd' Qp. Social surplus
be re-considered? We attempt to answer this question at Pp is abcQp. Consumers are willing to pay up to
using the following simple analytical framework. As- P1 to improve the service level. Therefore, alternative
sume that water supply service level (Q) is expressed supply services will be introduced.
by one variable, that the public supply function (Sp) An alternative supply, characterised by a supply
is inelastic, that demand (D) is monotonically declin- function SA will produce equilibrium values of Pz

Price

Public supply
b

'Alternative' supply

/ :
/'

/
/ ''
/
/
/
':'
d
''
/
/
'
:'
/
/
/
/
/
''
--------------1----------------------
e
,
/ d' ''
'''
/
/ D
/
/
/ '' Demand
// ''
h ''
/
/ ''
/ ''
/
/
''
/ '
/ '
a ''
a' Qp QP'
Service level

Fig. I. Social inefficiency in public rural water supply.

4 S. Zekri, A. Dinar/Agricultural Economics 28 (2003) 1-12

and Q2. Consumer surplus at P2 is P2bjand producer total Tunisian population had access to safe drinking
surplus is P2fh. Social surplus at P2 is hbf An al- water in 1994. INS (1997) estimated that 68% of the
ternative supply will therefore be socially justified if rural population had access to safe drinking water in
hbf- abcQp > 0 or if cfd > ahdQp. Furthermore, 1996 (see also Table 1 for comparison with world sta-
with two alternative supplies, SA and S' A, both go- tus). Obviously, a large share of the rural population
ing through point 'f and characterised by price elas- in Tunisia still lacks access to safe water resources.
ticities ssAand ss'A' we can see that ss'A < ssA =? The supply of drinking water in rural Tunisia is
!cfd- ahdQpl < !cfd'- a'd' Qpl· provided by two organisations: the Societe Nationale
Estimation of demand and supply elasticities make d'Exploitation et de Distribution des Eaux (SONEDE)
it possible to empirically calculate the areas under the and the Associations of Collective Interests (ACI).
demand and supply curves, and perform welfare com- The SONEDE is a public company that is responsible
parisons. Let en be the price elasticity of demand, for supplying drinking water, mainly in urban areas,
and ssp and ssA the price elasticities of public and where it supplies 90% of the water quantity to 87%
'alternative' supply, respectively. 3 Note that ssp » of the customers (and thus, exercises monopolistic
ssA. Since all quantities are known, and if we as- power in the urban sector). An ACI is an association
sume that while changing supply arrangement quan- of joint use of a water resource. The ACis are assisted
tities demanded will not change dramatically (or in by the public authorities (as part of the Agriculture
other words, that the demand function remains intact), Ministry, General Board of Rural Engineering-GBRE)
one can derive the equilibrium price and calculate ar- with a bureau of ACis at the national level as well
eas under the supply and demand curves. Suppose that as local ACis. These organisations operate when
Qp and Q2 are observed at levels of qp and q2, respec- there is no common or private safe and reliable wa-
tively, with q2 > qp. Then 1'1q = q2 -qp. By using the ter resource in a rural community to satisfy human
price information in the same way, define 1'1 P = P2 - needs.
Pp. Let the 'alternative' supply curve be qA = a+f3P, In addition to these services, many families in ru-
where a is an intercept and f3 = 1'1q I 1'1 P. Note that ral areas have their private sources such as storage
ssA = (1'1qjq)j(I'1PjP) = (!'1qji'1P)(Pjq). Since tanks to store rainfall water that is harvested from the
ssA, P and q are known, (1'1qji'1P) = ssA/(Pjq). roof or wells of the house. Harvested water is usually
This can be inserted into the 'alternative' supply equa- used only for drinking purposes. The other needs of
tion which becomes qA =a+ [ssA/(P jq)]P. A simi- the families are generally satisfied by alternative wa-
lar procedure can be used to specify the public supply ter sources of lesser quality, which are not suitable
equation and the demand equation. for drinking. The quality of the water supplied by
SONEDE and GBRE/ACI in rural areas varies ac-
cording to local conditions. Water quality is one of the
3. Water supply in rural Tunisia most important variables contributing to the success/
failure of the GBRE/ACI service.
Investments in water provision to the rural commu-
nities in Tunisia are borne mainly by the public author- 3.1. The costs of water supply in rural Tunisia
ities, loans from international agencies or by private
donations. MEAT (1997) estimated that 87% of the There is a great disparity between the cost of water
supply in rural Tunisia that is borne by both SONEDE
3 One reviewer suggested correctly that this comparison is valid
and ACI, and the revenue these providers collect from
only under the assumption that the suppliers are not natural monop- users. The difference between the per unit cost of ser-
olies. While 'alternative' supply is usually of a competitive nature, vice and the fee reflect, to a certain extent, the hid-
public supply is more likely to be natural monopoly. In the case den consumer benefits. To understand better the nature
of Tunisia, this is not the case in the rural sector since consumer and extent of the cost-price difference, in this section
associations can select the service provider (either SONEDE, ACI,
or GBRE, as is explained in the next section). In fact, the public
we employ various sources of data on water provision
authorities do decide which is the best scheme to implement after costs in rural Tunisia. Table 2 presents data on various
a study by a private consultancy. components of the cost of water supply to (urban and
 S. Zekri, A. Dinar/Agricultural Economics 28 (2003) 1-12 5

Table 2 3.2. Estimating SONEDE water costs in rural areas

Cost of water for SONEDE (all customers) and ACI (million 1996
TD)
A breakdown of SONEDE cost data into rural and
Item SONEDE ACI urban area is not available. To grasp the magnitude
Total operating and maintenance 55146 2327648 of the total cost of SONEDE water in rural areas, a
expenses (A) detailed example of one rural locality is considered.
Depreciation (B) 24043 5213320 Table 3 presents the total cost of supplying water to the
Financial fees (C) 6698
rural locality of Hichria in the district of Sidi Bouzid
Total assistance cost (D) 218210
Financial exchange fees (E) 10443 (SB), as well as the total cost for the district of SB (ru-
Total cost (A + B + C + D + E) 96331 7781075 ral and urban) in 1996. Hichria is a small rural com-
Volume of water (million m3 ) 228509 8378382 munity with 328 customers subscribed to SONEDE
Total cost per unit of water 0.422 0.929 service. The network for water distribution extends
(TD/m 3 )
over 20 km. The total cost for Hichria is 1.042 TD/m3 ,
Operating and maintenance cost 0.241 0.278
per unit of water (TD/m 3 ) whereas the total cost (rural and urban) in the district
Staff cost per unit of water 0.128 of SB is 0.464 TD!m 3 (SONEDE, 1998). The situa-
(TD/m 3 ) tion described for Hichria may be typical for other ru-
Depreciation cost per unit of 0.105 0.625 ral communities, namely remoteness that necessitates
water (TD/m 3 )
extension of the pipe system and service to a small
Financial cost per unit of 0.075
water (TD/m 3 ) number of customers. This translates into per unit cost
Source: Zekri (1999). in rural areas that are about three to four times higher
than those in urban areas.

rural without distinction) SONEDE customers and to Table 3

Water costs in the rural locality of Hichria serviced by SONEDE
ACI customers in 1996.
(1996 TD)
Scrutiny of Table 2 shows that the total cost of one
cubic meter (m3 ) of water to SONEDE customers in Hichria District SIDI
BOUZID
1996 was 0.425TD/m3 (in 1996, 1 TD = US$ 1).
Consumables 1172 10221
The operating and maintenance cost was calculated
Chemicals 538 12777
as the sum of the expenditures on consumable mate- Fuel 1746 10221
rials, chemicals, fuel, energy, staff, water purchased Power 1706 71552
and others. The numbers in Table 2 represent the cost Water purchase 0 128285
of water supplied by SONEDE to rural and urban ar- Staff cost (A) 10893 495283
Overhead costs (B) 3203 98311
eas of Tunisia. There are substantial regional differ-
Others (C) 937 59163
ences in costs (not shown, see Zekri, 1999), arising Depreciation (D) 13578 299910
from remoteness and the number of service connec- Total expenses 33773 1185723
tions in the rural community. For example, in 1996 (A+B+C+D=T)
the total unit cost ranged from 0.229 TD!m 3 in the dis- Volume of water serviced 32401 2555436
(V, million m 3 )
trict of Tunis to 0.533 TD/m3 in the district of Mahdia
Total cost (TIV, TD!m 3 ) 1.042 0.464
(SONEDE, 1997). Because of the unified prices ap- Operating and maintenance cost 0.623 0.347
plied to customers at the national level, it is clear that ((T - D)IV, TD!m 3 )
a substantial cross-subsidisation between districts or Staff cost (AIV, TD!m 3 ) 0.336 0.194
regions takes place. Besides, a tiered pricing method Overhead cost (BIV, TD!m 3 ) 0.099 0.038
Depreciation (DIV, TD!m 3 ) 0.419 0.117
is applied for SONEDE customers, with consumers
Operating and maintenance 59.8 74.7
using less than 20 m3 per quarter paying less than the cost/total cost (%)
total costs, which implies additional cross subsidies Staff cost/total cost (%) 32.3 41.8
between users (Zekri et al., 1997). Therefore, one can Overhead costs/total cost (%) 9.5 8.3
expect variation also in consumer surplus among the Depreciation/total cost (%) 40.2 25.3
various regions. Source: Zekri (1999).
6 S. Zekri, A. Dinar/ Agricultural Economics 28 (2003) 1-12

28.00

-
24.00

20.00
-
- Correlation coefficient = -0.265
E'
.g
16.00
Q
'~=- ':.

0
0
12.00
-
~

-;. -
8.00 =
-
~-
~-
-
4.00
I~ -
~-- -
W,t~s.~- .... - ... - -- - -·
0.00 - ......==-- -~--- - -
0 50000 100000 150000 200000 250000 300000
Volume Supplied (em/year)

Fig. 2. Cost vs. volume supplied by ACI.

Lahouel et al. (1993) estimate the long-term preparation costs, association management expendi-
marginal cost (LTMC) per unit of water for SONEDE tures, payment of the annual instalment/annuity and
in urban and rural areas. The estimates for the expected expenditures.
urban areas vary between 0.430TD/m3 in 1993 The average operating cost for the 637 ACis was
and 0.485 TD/m3 in 1996. However, the estimates 0.148 TD/m3 in 1996. This cost varies considerably
for the rural area ranged between 0.810TD/m3 in within regions with a maximum of 0.469 TD/m3 for
1993 and 0.893 TD/m3 in 1996. These figures show Sfax and a minimum of 0.084 TD/m3 for Kasserine.
that rural LTMC is almost twice as high as urban The reasons for these variations are similar to those
LTMC. mentioned above for SONEDE supplies in rural areas.
The average operating and maintenance cost in 1996
3.3. The AC!s' water costs for the 637 ACis was 0.278 TD/m3 . This is the cost
that the ACI water users should have normally borne.
The costs for the ACis have been determined Operating and maintenance costs also vary consid-
on the basis of available information for 637 ACis erably among districts, ranging from a maximum of
in 14 districts of Tunisia in 1996. The information 0.796TD/m3 inMahdia to a minimum of0.147TD/m3
comes from the database of the Special Assistance in Kasserine.
for Project Implementation (SAPI, 1998) and the Where data was available, a detailed analysis of
German firm IGIP. Cost information includes the water supply costs in individual ACis was also con-
following expenditures: 4 maintenance, operating and ducted. Of the 637 ACis in the sample, we used only
405 because of missing data. The information reveals
costs ranging between 0.041 and 25.88 TD/m3 . The
4 Based on Decree no. 87-261 enacted on 27 October 1987, distribution of per unit cost is plotted against the
concerning the organisation and management of the ACis. volume supplied in Fig. 2. About 72% of the ACis
 S. Zekri, A. Dinar/ Agricultural Economics 28 (2003) 1-12 7

Table 4
Comparison of total charges per consumer for SONEDE and ACI
SONEDE & ONASa SONEDE only ACI

Water consumption (m 3per year) 137 137 68

Water services
Water price (TD/m 3 ) 0.19 0.19 0.195
Fixed charges SONEDE (TD) 11.764 11.764
Wastewater treatment
Fixed charges ONAS (TD) 5.64
Price (TD/m 3 ) (block 1 = 0-20m 3 ) 0.006 0.006
Price (TD/m 3 ) (block 2 = 2~0m 3 ) 0.0078 0.078
Value added tax (%) 18 18
Total of the bill tax included (TD) 57.1 44.6 13
Total (TD/m3 ) 0.417 0.326 0.195
Average cost of water (TD/m3 ) 0.422 0.422 0.148
Price-to-cost ratio (%) 45 45 132
a ONAS stands for Office National Assainissement, and is the organisation in charge of wastewater management.

fall in the ranges of 0-40,000 m 3 per year and 0.04- and prices borne by the consumers. 5 Table 4 high-
6.0TD/m3. lights a big difference between ACI and SONEDE
To be able to compare SONEDEs and ACis per unit in regard to the cost of providing water services
rural water costs, depreciation costs as well as central and the derived price-to-cost ratio. Because the two
administrative and regional specific charges should be agencies' prices of water to consumers are quite sim-
added to the ACI cost structure. The reason is that ACI ilar (0.19TD/m3), SONEDE recovers only 45% of
customers do not pay for these fixed costs. Therefore, its costs, while the ACis recover 132% of their costs,
total cost was estimated by adding depreciation costs on average. Rates vary between 107 and 170% at the
and administrative and maintenance costs to the op- district level.
erating costs. Administrative costs are staff expenses,
travelling expenses and the budget allocated to the ACI 4. Estimating rural water demand and supply
administrative by the public authority GBRE. These elasticities
assistance costs have been estimated for the GBRE
bureau of assistance to ACis on national and local In 1996 the number of people with access to drink-
levels. Total administrative cost was divided in pro- ing water in rural Tunisia was estimated at 2.39 mil-
portion to the total number of ACis. The average to- lion. About 1.4 million people were served by the ACis
tal cost of water amounted to 0.929 TD/m3 for the (about 200,000 families), and the rest (974,000) by
637 ACis. The District of Mahdia recorded the highest SONEDE (139,000 families). About 1400 ACis oper-
total cost (2.871 TD/m 3), whereas in Kasserine total ated in 1996. The average calculated consumption was
costs reached only 0.457 TD/m3. 137m3 per year per family served by SONEDE and
68m3 per year per family served by the ACis. The av-
3.4. Price-to-cost ratio
erage price paid by ACI customers was 0.195 TD/m3 ,
and the average price paid by SONEDE customers
The price-to-cost ratio indicates the cost recovery
was 0.190TD/m3 (or 0.326TD/m3 if fixed costs are
rate of the supply agency or the level of subsidy
included). A mean minimum salary of 155 TD per
provided to consumers. To measure the price-to-cost
month prevailed in rural Tunisia in 1996.
ratios for customers in rural SONEDE and for the
members of ACis, we first compare the prices paid by 5 We use the average consumption of 68 and 137m3 per year for
each of the two types of consumers. This comparison a subscriber family of ACI and SONEDE, respectively (SONEDE,
is presented in Table 4 and is based on the actual costs 1997). A detailed discussion can be found in Appendix A.
8 S. Zekri, A. Dinar/Agricultural Economics 28 (2003) 1-12

Table 5
Demand and supply estimates for ACI and SONEDE (log-log specification)

ACI SONEDE

Demand Supply Demand Supply

Equations (1) (2) (3) (4) (5) (6)

Intercept 3.32 (7.33) 8.11 (39.94) 4.45 (12.23) 6.47 (32.55) 3.2 (a) 2.72 (51.21)
Cost -0.42 ( -5.37) -0.78 ( -23.64) -1.42 ( -4.59)
Price -1.29 ( -1 0.23) -1.30 (-6.18) -0.24 ( -0.95)
Number of families 0.513 (11.30)
connected
Length of pipe 0.055 (9.16)
system
Water quality -0.48 ( -2.08)
Minimum salary -0.08 ( -0.25)
number
Number of 405 223 369 369 40 10
observations
R-square 0.24 0.19 0.20 0.77 0.85 0.70
Adjusted R-square 0.17 0.18 0.19 0.76 0.53
D-W 1.92 2.37
Breusch-Pagan 20.43 25.81 28.53 34.12 17.12
F-test 27.43 26.25 401.75 408.84 10.35
Source Lahouel et a!. (1993) Zekri (1999)

The t-values are in parenthesis.

a Not reported.

Based on the available data for 405 ACis (Zekri, water supplied will increase as the number of families
1999), demand and supply functions for ACis were connected increases. The length of the pipe system is
estimated. Several specifications were considered, another explanatory variable in the supply function.
including various sets of explanatory variables in ad- Longer pipe systems may represent larger supply sys-
dition to the conventional variables of the price paid tems and thus be correlated with higher values of vol-
by consumers and the cost of producing the water. ume supplied.
Various functional forms were tested but only the Scrutiny of Fig. 2 suggests that although most of the
log-log form yielded reasonable and comparable re- observations are below both 6 TD/m3 and 40,000 m 3
sults. The estimated demand and supply functions are per year, it is clear that some ACis are characterised by
presented in Table 5. Due to missing values for some higher values and these observations affect the trend
of the variables, the number of observations for each of the estimated supply curve to reflect economies of
specification of the demand or supply functions dif- scale.
fers. Descriptive statistics of these variables are not A demand function for SONEDE customers in ru-
presented but available from the authors upon request. ral areas based on Lahouel et al. (1993) is used. The
Water quality is one of the explanatory variables in demand function considers the low revenue popula-
the demand function. It is measured in ranges of salin- tion and is based on quarterly data for the period
ity values (0-1.5; 1.6-2.5; >2.5 g/1) with higher salin- 1983-1992. Based on data in Zekri (1999), a supply
ity values representing lower quality. It is expected function for SONEDE was estimated using long-term
that lower quality will be negatively correlated with marginal cost data for rural areas (Table 5).
the volume consumers are willing to pay for. Our purpose in estimating supply and demand func-
The number of families connected to the ACI sup- tions for SONEDE and the ACis was to prepare tools
ply system is one of the explanatory variables of the for the calculation of welfare measures of water sup-
ACI supply function. We expect that the volume of ply in rural Tunisia. Prior to launching the welfare
 S. Zekri, A. Dinar/Agricultural Economics 28 (2003) 1-12 9

analysis, however, we can gain several interesting ob- the supply agency. Improved quality reflects a higher
servations from scrutiny of the results in Table 5. Fo- service level and should be considered by the agency.
cusing first on the price elasticity of demand, one can The ACI supply estimates also include variables in-
see that the range is -0.24 to - 1.3, for both supply dicating the number of connections and the length of
arrangements. This range of price elasticity values is the system. More connections and longer pipe systems
quite reasonable in developing countries, both in ru- affect the volume supplied positively.
ral (e.g. Briscoe and de Ferrenti, 1988; World Bank
Water Demand Research Team, 1993) and in urban 4.1. Welfare implications of SONEDE and ACI
areas (e.g. Boland and Whittington, 2000). In this re- water services
spect, the demand functions used in our analysis are
quite representative. Inserting the mean values of the variables in the es-
Comparing estimated coefficients of the ACI and timated supply and demand functions, one can obtain
SONEDE demand functions in Table 5 suggests that a comparable measure of social welfare for SONEDE
the intercepts are quite similar, but the price elasticity and ACI water supply services, using the framework
for ACI is higher than for SONEDE. This difference
reflects the fact that ACI customers have different 1.2
characteristics than those served by SONEDE. We
suspect that ACI connections reach the poorest rural
population. This is reflected in the higher absolute
elasticity values for ACI customers compared with
iE
.!=! 0.8 --D-SONEDE
those for SONEDE rural customers, and is in ac- .c
:I - S-80NEDE
.!;!
cordance with the actual volumes demanded by ACI Q 0.6
customers (68m3 per year) and SONEDE customers t::.
'!;;
0
(137m 3 per year). (J 0.4
Another issue to pay attention to is the sign of !;
the cost elasticity of supply in both the ACI and -~I!. 0.2
SONEDE functions. The negative sign of the cost
elasticity means that in both cases production is on 0
0 100 200 300 400 500
the declining part of the marginal cost function. This
Volume (cubic meter)
indicates that the systems are operating inefficiently,
reflecting the fact that both SONEDE and the ACis Fig. 3. Supply and demand curves for SONEDE water.
reach small and remote communities and, therefore,
volumes delivered are much smaller and production 1.2
costs much higher than in urban areas. When we cal-
culate social welfare under these declining marginal
cost functions, we will use the difference between the
consumer and (negative) producer surplus (instead of iE
0.8
---D·ACI
the sum of consumer and producer surpluses as in the .!=!
.c ···························--S-ACI
:I
case of a positively sloped supply function). .!;! 0.6
Q
Other interesting results in Table 5 are the negative t::.
~...
coefficient of the water quality variable in the ACI de- 0.4

mand function. The results suggest that the quality of 0

Gl 0.2
the water (salinity) is an important determinant in the ·!:!
I!.
demand. The higher the salinity the lower the quan-
0
tity demanded. We are not aware of previous studies 0 100 500 600
200 300 400
of residential water supply that include water salinity Volume (cubic meter)
as a determinant of demand. However, as we indicate
later, water quality can be seen as a service provided by Fig. 4. Supply and demand curves for ACI water.
10 S. Zekri, A. Dinar/Agricultural Economics 28 (2003) 1-12

Table 6
Welfare calculation for ACI and SONEDE water supply in rural Tunisia
ACI SONEDE
Consumer Producer Social Consumer Producer Social welfare
surplus surplus welfare surplus surplus
Based on the estimated supply and demand curves
Per family (TD) 25.2 -53.2 -28.00 17.64 (21.298 ) -13.72 3.92 (7.57 8 )
Total (million TD) 5.04 -10.64 -5.60 2.45 -1.91 0.54
Based on the observed values of mean volumes and prices
Per family (TD) 11.27 -56.98 -45.71 14.19 -7.85 6.34
Total (million TD) 2.25 -11.99 -9.74 1.97 1.09 0.88
• Value obtained with so = -0.5 for SONEDE.

suggested earlier. This can be done, under the assump- address these values in the welfare calculations con-
tion that the populations currently served by ACI and ducted below.
SONEDE are characterised by similar intercept and The social surplus is computed using both the ob-
price elasticity values. However, if we accept the ex- served mean prices and volumes of water consumed
planation in the previous section regarding the differ- as well as the values obtained from the intersections
ences between the populations served by SONEDE of the demand and supply functions in Figs. 3 and 4.
and the ACis, we can rest assure that the welfare ben- Calculation of welfare values is presented in Table 6
efit comparison can be conducted on the ground that using both the procedure suggested in the analytical
the population remains the same when the water sup- framework and linear approximation of the curves in
ply provider changes (say from SONEDE to ACI). Figs. 3 and 4. The per family surplus is then multi-
We elected to use the demand and supply functions plied by the number of families served to obtain total
for SONEDE and ACI that include only the unit price social values.
and cost variables (columns 1, 3, 5 and 6 in Table 5) The calculated consumer and producer surplus
in order to comply with the simple model presented values suggest that in the case of the ACis there is
in Section 2. These functions are plotted in Figs. 3 a small social welfare loss (5.60-9.74 million TD
and 4. per year, depending on the data source). In the case
One immediate observation is that the intersection of SONEDE there is a small social welfare gain
values (volume and price of water) in Figs. 3 and 4 (0.54-0.88 million TD, depending on the data source).
differ from those reported above. Using the intersec- In both cases the producer surplus is negative, due
tion values from Figs. 3 and 4, the annual volume to the fact that the marginal cost curve is above the
per family served by SONEDE is 49m3 and the price equilibrium price. This implies that both agencies
is 0.64 TD/m3 . In the case of the ACis, annual wa- produce in a non-optimal zone of the production
ter consumption per family is 140m3 and the price function. Because the estimated demand function
is 0.24 TD/m3 . These estimates are very interesting to for SONEDE does not have a statistically significant
contrast to the actual prices and volumes reported in price coefficient, we conducted a sensitivity analysis
Table 4. First, the average water price across ACis and for the demand elasticity. With en= -0.5, consumer
for SONEDE is 0.19TD/m3 . Second, the actual wa- surplus and social welfare increase and are closer to
ter consumption for ACI and SONEDE families is 68 the estimates based on the observed values in Table 6.
and 137m3 per year, respectively, which is reciprocal In the case of the ACis it was also possible to cal-
to the intersection values of 140 and 49m3 per year. culate a proxy welfare index that provides an alterna-
These differences arise from the nature of the basic tive estimate of social welfare. 6 Because information
data available for our analysis. One possible explana-
tion is the fact that the supply and demand equations 6 This idea was provided by one of our reviewers and we
for SONEDE were imported from other studies. We acknowledge it with much appreciation.
 S. Zekri, A. Dinar/ Agricultural Economics 28 (2003) 1-12 II

is available on the individual ACis, we calculated for inefficient zone of the production function. These in-
each ACI (with available data) the consumer surplus efficiencies may result from the fact that there are
minus the sum of supply cost and subsidy, where sub- economies of scale in water supply that are not re-
sidy was calculated as the difference (if any) between alised due to small plants or low demand (small and
the supply cost and revenue. To calculate the consumer remote communities).
surplus, we inserted the actual price and volume con- From a public policy perspective, both SONEDE
sumed by each ACI into the estimated demand func- and ACI receive substantial public subsidies. In the
tions (column 1 in Table 5). Since not all ACis in the case of ACI this is a government subsidy in the form
database have a water price figure, we conducted this of 'assistance', and in the case of SONEDE this is
analysis on a subset of ACis. This subset included 231 cross-subsidisation among regions (recall that the
ACis with a total of 30,408 families (out of 200,000). price is unified across regions). Our analysis suggests
Total consumer surplus is 1.47 million TD and total that the subsidy to both SONEDE and ACI is pro-
supply cost is 1.76 million TD. The resulting social vided because both operate in a sub-optimal manner
welfare is thus, -0.29 million TD. Although it is hard (declining marginal cost). The relatively high con-
to compare this result with those in Table 6, one under- sumer surplus measures indicate that the value of the
lying outcome is clear: the social welfare figures are water supplied to consumers is much higher than the
negative. Scrutiny of the individual ACI results sug- price they are charged for it. This implies that there
gests that the proceeds from water sales covered the may be scope for additional steps to be taken, such
supply costs in only 25 of the 231 cases in this data as improving the quality of service so as to provide
subset. justification for price increases and/or expand demand
to increase the level of supply and cover operating
and maintenance and even investment costs. From a
5. Conclusion welfare point of view, the results of our analysis sug-
gest that in heavily subsidised water supply systems,
SONEDE is a public monopolistic enterprise with a positive consumer surplus may override negative pro-
responsibility for providing water to urban communi- ducer surplus and create an overall positive social
ties in Tunisia. SONEDE's service has been extended welfare effect.
to rural areas as well. In 1990 the ACis were estab-
lished and started providing rural communities with
safe water. Our cost analysis shows that when consid- Acknowledgements
ering the total cost of water supply, the ACis are more
efficient than SO NED E. Rural customers of SONEDE The authors would like to thank two anonymous re-
paid only 18% of the total cost of water, while ACI viewers for an excellent set of comments and dedica-
members paid 21% of the total cost. In addition, the tion through the review process. The views expressed
total bill for a SONEDE customer is four times higher in this paper should not be attributed to the World
than the bill for an ACI member. This is due to differ- Bank.
ences in the volume of water consumed as well as in
fixed charges: the average ACI member uses only half
the water quantity of a SONEDE customer and pays Appendix A. Price-to-cost ratio calculations
neither fixed charges nor added value tax.
Currently, the ACI members pay just the operating The calculation of SONEDE costs takes into
costs of water supply. There is a need to include at account (given the nature of the data) the national
least the maintenance costs in order to insure the re- average total cost for rural and urban areas together.
liability and continuity of the service. SONEDE cus- This cost was 0.422 TD/m 3 in 1996. The average
tomers are cross-subsidised. The subsidies come from operating cost actually borne by ACI members or
other customers and/or other urban or rural districts. consumers was 0.142 TD/m 3 in the same year. The
Even if there are differences between the efficien- price structure for a SONEDE customer includes sev-
cies of SONEDE and ACI, both still operate in an eral components: a fixed water charge, two variable
12 S. Zekri, A. Dinar/Agricultural Economics 28 (2003) 1-12

block prices, a wastewater fixed charge and a value MEAT, 1997. Ministere de l'Environnement et de I' A.menagement
added tax (Table 4). The figures for 1996 show that a du Territoir. Profil de Ia Tunisie Application d' Actions 21:
family served by SONEDE & ONAS pays 57.1 TD Examen des Progres Accomplis Depuis Ia Conference de Rio
des Nations Unies sur l'Environnement et le Developpement.
per year. A family served by SONEDE alone pays Parker, Ronald, Skytta, Tauno, 2000. Rural water projects: lessons
44.6 TD per year and a family served by an ACI pays from OED evaluations. OED Working Paper Series, The World
only 13 TD per year on average. Thus, ACI members Bank Operations Evaluation Department, Washington DC.
pay only 23 and 30% of the amount paid by those Perkins, F., 1994. Cost effectiveness of water supply technologies
in rural Indonesia: evidence from Musa Tenggara Barat. Bull.
subscribing to SONEDE & ONAS and SONEDE,
Indonesian Econ. Stud. 30 (2), 91-117.
respectively. These figures show that it is decisive to Roark, Philip, Backer, James S., Buzzard, Shireley, Cauley, Henry
take into consideration the standard of living of rural A., 1987. Privatization Study of the Village Water Supply
families in the choice of the intervention mode for and Sanitation Project Lesotho. WASH Field Report No. 215,
the provision of drinking water when estimating their Prepared for the USAID Mission to the Government of Lesotho,
willingness to pay. WASH Activity No. 344.
Saleth, Maria R., 1996. Water Institutions in India. Commonwealth
Publishers, New Delhi.
SAPI,
References 1998. Assistance Speciale pour Ia Realisation du Projet. SAPI.
Approvisionnement en Eau Potable des Zones Rurales dans six
Atlaf, M.A., 1994. Economics of household response to inadequate Gouvernorats en Republique Tunisienne. Fonds de Coopera-
water supplies: evidence from Pakistan. Third World Plan. Rev. tion Economique du Japan, Ministry of Agriculture, Tunisia.
16 (1), 41-53. SONEDE, 1997. Essai de Calcul du Prix de Revient d'un Metre
Boland J., Whittington, D., 2000. The political economy of water Cube d'Eau en Milieu Rural. Cas de Ia Localite Hichria a
tariff design in developing countries. In: Dinar, A. (Ed.), The Sidi Bouzid, Direction Comptable et Financiere, Division Com-
Political Economy of Water Pricing Reforms. Oxford, New ptabilite Analytique et Budgetaire, Ministry of Agriculture,
York. Tunisia.
Briscoe, J., de Ferrenti, D., 1988. Water for Rural Communities: SONEDE, 1998. Etude du Secteur de L'Eau Potable en Milieu
Helping People Help Themselves. World Bank, Washington Rural Phase I: Retrospective et Prespective. Ministry of Agri-
DC. culture, Tunisia.
Brookshire, D.S., Whittington, D., 1993. Water resources issues in Wahadan, L., Dodge, C.P., Ekvall, T., Yousif, M.A., 1990. Cost
the developing countries. Water Resources Res. 29 (7), 1883- effective water and sanitation in Sudan. Water Qual. Bull. 15 (1),
1888. 11-17.
Gleick, P.H., 1998. The World's Water-The Biennial Report on Whittington, D., Apia, 0., Augustine, 0., McPhail, A., 1990.
Freshwater Resources. Island Press, Washington, DC. Strategy for cost recovery in rural water sector: a case study of
INS 1997. Rapport annuel sur les indicateurs d'infrastructure. Nsukka district, Anambra state, Nigeria. Water Resources Res.
Ministere de Developpement Economique. Tunis. 26 (9), 1899-1913.
Lahouel, M., Rejeb, M.S., Bouzaiene, M., Donas, C. Mamoghli, World Bank Water Demand Research Team, 1993. The Demand
1993. Etude Economique sur l'Eau Potable en Tunisie. Rapport for Water in Rural Areas: Determinants and Policy Implications.
premiere partie, November 1993, Ministry of Agriculture. The World Bank Research Observer, Vol. 8 (1), pp. 47-70.
Tunisia. Zekri, S. 1999. Cost of Water supply in Rural Associations in
Mangin, J.-M., 1991. Rural water supply in southern Ethiopia: Tunisia. Working Paper. Ecole Superieure d' Agriculture de
failures and alternatives. Can. J. Develop. Stud. 12 (2), 297- Mograne, University of Tunis II.
312. Zekri, S, Echi, E.L., Sghaier, M. 1997. Tunisia. In: Dinar, A.,
Mullick, M.A., 1987. Socio-Economic Aspects of Rural Water Subramanian, A. (Eds.), Water Pricing Experiences: An Inter-
Supply and Sanitation. The Book Guild, Sussex, England. national Perspective. World Bank Technical Paper No. 386.