ArticlePDF Available

Abstract and Figures

As first noticed by Coase (1946), a standard result in utility regulation is that efficiency requires two-part tariffs with marginal prices set to marginal costs and fixed fees equal to each customer’ s share of fixed costs. Residential water customers in France face marginal prices for water that average about 8% more than marginal costs. Under price elasticity estimates that are consistent with previous results in the literature, efficiency costs represent around 8 million euros of welfare losses for 2008. Even though the impact is fairly small, current price schedules are an important pre-existing distortion which should be considered when evaluating current taxes aimed at addressing external costs. Moreover, efficiency gains from reformed tariffs could be used to fund water assistance programs focused on financially stressed households.
Content may be subject to copyright.
EPPP DP No. 2012-08
Efficiency and Equity in Two-Part Tariffs:
The Case of Residential Water Rates
Simon Porcher
December 2012
Chaire Economie des Partenariats Public-Privé
Institut d’Administration des Entreprises
Efficiency and Equity in Two-Part Tariffs: The Case of
Residential Water Rates
July, 2012
As first noticed by Coase (1946), a standard result in utility regulation is that efficiency
requires two-part tariffs with marginal prices set to marginal costs and fixed fees equal to
each customer’s share of fixed costs. Residential water customers in France face marginal
prices for water that average about 8% more than marginal costs. Rebalancing rates
from current tariffs to Coasian tariffs results in lower bills for consumers on average
but does create strong distributional consequences. Under price elasticity estimates that
are consistent with previous results in the literature, efficiency costs represent around 8
million euros of welfare losses for 2008. Even though the impact is fairly small, efficiency
gains from reformed tariffs could be used to fund water assistance programs focused on
financially stressed households.
JEL codes: D21-D82-L33-L95. Keywords: public-private partnerships, water industry, Two-Part Tariffs,
Efficient Pricing
IAE Sorbonne Business School - GREGOR
1 Introduction
In regulated markets such as energy, electricity, water and wired phone service, where
price schedules can have strong distributional consequences and economic distortions, it
is crucial that pricing appropriately encourages equity and efficiency in use. This historical
debate has given way to a rich theoretical literature examining utility pricing in relation
to the public interest. Hotelling [1938] first argues that all prices in an economy should
be set equal to marginal cost, with fixed costs paid for with government subsidies from
income, inheritance and land taxes. Coase [1946] considered that efficient pricing in regu-
lated markets implies two-part tariffs. Further theoretical developments usually consider
a Ramsey-Boiteux pricing to derive how prices should be marked up above marginal cost
(Baumol and Bradford [1970]) in order to meet the social revenue requirement. Equity is
first incorporated into the efficiency analysis by Feldstein [1972] who assumes a functional
form of the social welfare function and derives formulas for the socially optimal two-part
tariff. These Ramsey-Boiteux pricing schemes however represent second-best optima as
they suppose deviations from marginal cost pricing. The challenge in regulating markets
is that price be set such as to enforce efficiency and equity.
Water supply exemplifies this issue. Water is a large market that directly affects over
99% of French households. The French water market - including water provision and
sewage - represented a market of 5.4 billions euros in 2008. The same year, 4 billions
cubic meters of water have been billed to domestic users and industrial consumers. The
main costs for water provision can be divided in three parts. First, water provision implies
costs for extracting, treating and distributing water to the consumer. Once water enters
the network, around 10% is lost in leakages. In addition to these costs, water utilities face
the fixed and sunk costs of processing bills and taking calls. Moreover, water utilities have
to maintain networks and connections and install water meters. The scale of the costs
thus differ from one utility to another: the costs of production depend on the volumetric
charge while the scale of the fixed costs is largely invariant to the number of customers,
such as customer service or meters management, or to the size of the network, such as
In France, regulation is made through a contract between a private operator and the
municipality when the public service is outsourced and through a public council decision
when the public service is managed in-house. As a result, local monopolies are largely
unregulated: they tend to maximize profit by pricing above marginal cost, resulting in a
level of output below the socially optimal level. As in many regulated industries, in the
simplest case, the tariff is divided in two parts: a fixed fee, no matter the level of con-
sumption, and a volumetric charge depending on water consumption. A standard result
first developed by Coase [1946] is that setting marginal prices to marginal costs would
eliminate the deadweight loss associated with monopolies. The local monopoly then re-
coups its fixed costs through fixed fees equal to each customer’s share of fixed costs.
Although it is compulsory to use two-part tariffs in the French water sector, opera-
tors tend to charge fixed fees and volumetric charge that differ from the theoretical ideal.
This paper applies the standard monopoly framework to answer the following questions:
(1) How do marginal prices differ from marginal costs? (2) What are the distributional
impacts of a switch from current tariffs to Coasian tariffs? (3) Do the reformed tariffs
fit better the equity considerations? (4) What are the efficiency costs from the observed
deviations from marginal cost pricing?
This paper examines a nationally representative dataset of 4,500 French municipalities
for 2008. The dataset contains demographic and economic information about households
at the municipal level, but also a large set of information on water demand and supply,
such as consumption, spendings, rates and some water utilities characteristics. We find
that marginal prices differ from marginal costs. Even if the range of the deviation is
limited - a 8% deviation is observed for the volumetric charge - these markups impose a
deadweight loss by leading customers to consume too little water and to support fees that
do not represent capital costs. Rebalancing rates to match the Coasian tariffs imply large
increase in welfare for consumers, especially those living in cities with lower incomes.
This is due to the fact that the correlation between water consumption and income is
significantly positive but weak. Consequently, reformed price tariffs benefit more to large
consumers more than low incomes. As a matter of fact, after the transition to Coasian
tariffs, cities in the first fourth quintiles regarding the per-unit income would experience
decreases in bills that are almost similar, between 21.45 and 20.07 euros per year. We
thus consider alternative water assistance programs focusing directly on cities with lower
per-unit incomes. We particularly find that a free fixed fee policy could be implemented
for poor cities, without loss of profits for firms, at the cost of 1.90 euros per non recipient.
We then compare the costs of these assistance policies to the current efficiency costs.
Under conservative levels of price elasticities, a transition to marginal cost pricing implies
efficiency gains of 8 million in 2008, a level that is low compared to the global profits of
water industries in France1. However, these efficiency gains are sufficient to fund assis-
tance programs such as decreased fixed fees for poor households.
The paper finally highlights several explanations for the current price distortion, such
as firms’ profit maximization (small versus large consumers?), resource scarcity (markup
versus Pigouvian taxes?) and management structure (public versus private?). We then
briefly discuss the validity of the results, precisely regarding consumers’ responses to
marginal prices and the link with related markets, such as sanitation.
The paper contributes to the literature on public utility regulation in several ways.
First, it shows that contrary to other regulated industries, water supply in France has
low-margins. However, deviations from marginal cost can have strong welfare and dis-
tributional impacts. Second, several assistance policies are empirically tested and shows
that at low-cost for water suppliers, it is possible to fund some assistance programs. These
assistance programs have stronger distributional consequences than tariff reforms. The
results of the paper are similar to those of Garcia and Reynaud [2003] who estimated the
benefits of efficient water pricing in France using a sample of 50 water utilities for four
years. Even if the authors found that marginal prices were on average lower than marginal
costs while fixed fee were marked up above each customer’ share of fixed costs, they find
a low-price elasticity as in this paper, resulting in rather small welfare gains of efficient
pricing. However, they conclude on the positive impact of rebalancing rates under some
1These are national estimations and profits include industrial and residential consumption. At the scale of
our dataset, the deadweight loss from current tariffs for residential customers is 5.36 million euros for 2008 and
the global profits of water industries for residential customers are 3 billion euros.
social objectives. In this paper, we complement this approach by simulating the impact
of some social policies.
The paper proceeds as follows. Section 2 presents relevant background information
about the organization and the regulation of the French water market. Section 3 describes
the two datasets, their validity and performs a test of marginal cost pricing. Section 4
examines the distributional consequences of a transition to Coasian tariffs when demand
elasticity is null. Section 5 performs an estimation of price elasticities, computes the
efficiency effects of marginal cost pricing and examines the reasons for current markups.
Section 6 discusses the results. A brief conclusion follows.
2 The French Water Market
2.1 Organization and Regulation
In France, as in most European countries, municipalities must provide local public services
that have public good characteristics. Water provision and sewage are two of these public
services and can be managed by two different operators2. However, if the responsibility
for public services’ provision is public, its management can be either public or private.
Although some municipalities manage production through direct public management and
undertake all operations and investments needed for the provision of the service, the
dominating contractual form is delegated management3. In this case, a private operator,
independent of the local government, is hired to manage the service and operate facilities,
through one of the four different private-public arrangements. The most common is the
lease contract in which the operator manages the service, invests in the network and gets
a financial compensation through consumer receipts. Under a concession contract, the
external operator also undertakes construction risks, as it must finance a large part of
investments over the duration of the contract. These contractual agreements differ from
the previous ones in that operators share risk in exchange for greater decision rights and
claims on revenues. Other contracts can be chosen by the local authority such as the
gerance in which it pays an external operator a fixed fee, or an intermediary management
contract, i.e. a gerance contract but with a small part of the operator’s revenues depend-
ing on its performance. Such contracts provide few incentives to reduce costs and transfer
no risks and decision rights to a private operator. Although there are a large variety of
contracts, the participation of the private sector is characterized by a concentration on
three major companies. These companies share with their subsidies more than 90% of the
private market and other private companies operate mainly in small cities.
Contrary to other industrialized countries, there is no price-cap or rate-of-return reg-
ulation for water utilities in France as there is no national regulator. Such regulation
2Water provision refers to the production and the distribution of water and sewage implies wastewater
collection and treatment. We focus in this paper on water provision.
3An official report by Dexia, a French financial intermediary, states that 63% of French medium-sized cities
contract out the services of drinking water treatment and distribution and 58% also contract out their sewerage
services. It is however difficult to have a precise estimation of how many municipalities and communities have
contracted out both services with the same operator. In our database, more than 60% of the municipalities are
managed by private operators. According to the Cour des Comptes [2011], the highest financial court in France,
71% of the population is covered by a private operator for water provision and 56% for water sewage.
has been replaced by a contract, in the case of a private operator, or a decision of the
municipality board, in the case of public operation. In the case of delegated management,
rules have been defined to ensure that standards are respected during the operation to
limit the opportunistic behavior of operators and guarantee competition between firms.
First, since the Sapin Law (1993) a national legislative framework governs the form of
the private sector participation and the conduct of the bidding process. Second, a strong
regulation on contract duration and delegatee obligations was implemented in 1995 with
the Barnier Law. As a matter of fact, water quality in France has increased and is now
relevant for more than 99% of the tests and a lot of investments have been implemented
to deter leaks. However, because regulation is made through contracts between the two
parties, depending on the respective power of negotiators, with some contracts signed
a century ago, there are doubts about the possibility of the parties to regularly adapt
tariffs to the needs of the utilities. Even if they did, water tariffs may not be efficient nor
equitable from the economic point of view.
2.2 Tariffs
Applying an efficient tariff for water is difficult to achieve. To be efficient, the design of
the tariff must satisfy several conditions. The main objective of the pricing scheme is to
generate revenues covering costs. However, the pricing rate should also allow different
costs between users with heterogeneous financial means as much as it has to provide in-
centives for efficient use of the resource. As these criteria may be contradictory, finding a
rate structure balancing efficiency and equity is not an easy task.
Previous studies on efficient pricing focused on which price schedule yields the highest
level of utility, using the framework of the second-best pricing, the so-called “Ramsey-
Boiteux” pricing. When searching for utility maximization under linear prices solved
by Ramsey [1947], Boiteux [1956] shows that the welfare-maximizing price markup is
proportional to the inverse of the elasticity of demand. “Ramsey-Boiteux” pricing ensures
the welfare maximization under a budget constraint. In this framework, a monopolist
facing inverse demand function pi(xi)for good i, a social planner constrained to using
linear prices can maximize social surplus by setting prices
pi∂C (X)
1 + λ(1)
where λis a non-negative constant. Such a framework is for example used by Garcia
and Reynaud [2003] to reform French water tariffs but also by Diakité et al. [2009] to
implement social pricing in Côte d’Ivoire. However, this optimal solution implies that
the utility knows demand-elasticities for each consumer and that regulators or parties to
the contract constrain themselves to linear prices. In practice, network industries such
as water but also electricity or gas have long implemented two-part tariffs. Water tariffs
in France have two compulsory components since 1994. On the one hand, each customer
must pay a fixed charge corresponding to provisions for capital stock renewal and debt
service. On the other hand, a marginal tariff corresponds to operating expenses of the
volumetric charge. For a baseline annual household water consumption of 120 cubic me-
ters, the fixed-part of the tariff represents 25% of the total price. Moreover, there are
additional fees going to the Basin Agency and a value-added tax for the State.
A standard result in regulation is that efficiency requires marginal prices to equal
marginal costs. In the water industries, the obligation to have a two-part tariff facilitates
pricing at marginal cost because the volumetric charge can be set equal to marginal cost
and the fixed monthly fee set to cover fixed costs. Pricing at marginal cost may have many
drawbacks. Indeed, it is inappropriate when managers have no budget constraints as they
would have no incentive to reduce costs. Moreover, marginal cost pricing implies that the
utility runs a deficit if there are increasing returns to scale. This deficit might lead to dis-
tortionary taxes if there are no lump-sum transfers. As first suggested by Coase [1946], an
alternative solution to marginal cost pricing is to use two-part tariffs with a marginal price
corresponding to the marginal cost and the fixed fee set to cover the total fixed costs. In
water industries with declining average costs and constant marginal costs4, this would im-
ply setting the fixed monthly fee equal to each customer’s share of the utilities’ fixed costs.
Efficient pricing may however not be achieved in water industries for two reasons. On
the one hand, water utilities face volatile revenues. For example, water consumption is
often higher during summers than winters while some touristic areas face high consump-
tion levels during national vacations. Over the years, billed volume of water tend also
to decrease, probably due to changing consumer behavior towards sustainable water use
and to less consuming intermediary goods. This revenue volatility is a source of concern
for water utilities. On the other hand, operators and city councils set tariffs such as the
expected revenues from water sales covering the forecasted expenses, which is close to
an average-cost pricing. In practice, water tariffs thus differ from the theoretical ideal of
marginal cost pricing.
There are at least two reasons why marginal cost pricing has not been implemented.
The first one comes from the diminished profits that would occur for the water industries
if fixed fees remain the same. The second one lies in the distributional implications of
such a reform. Such a decrease in marginal prices would especially benefit large consumers
rather than small consumers. To the extent that income and water consumption are re-
lated5, this would mean that higher incomes would face larger decrease in their bills than
lower incomes would.
One might argue that water tariffs already include distributional considerations be-
cause rates can include non-linear pricing schemes. These pricing schemes aim at taking
into account resource sustainability and distributional considerations. In our dataset,
1,260 municipalities have non-linear tariff schemes. Even if we have little details about
the tiers - we know the kink points at which consumers switch from one tier to another
- we observe only 152 municipalities with a two-tier tariff limitation below 300 cubic me-
ters, which is higher than the average consumption of the top 10% residential consumers.
Most of the multi-tier tariff schemes thus benefit huge consumers such as industries, public
administrations and agricultural holdings.
4Because of the fixed tariff, average costs are declining with consumption. Marginal costs are supposed here
to be constant as scale effects used in alternative regressions are very weak. Discussions with professionals let
us know that marginal cost depends first of all on the age of the plant more than on the volumetric charge.
5This assumption is tested below. The result is a significant positive but weak correlation between income
and water consumption.
2.3 Water-poor in France
In France, 13.5% of French households have an income lower than 60% of the median
income. For the lowest 10% incomes, the share of constraint households’ expenditures
has risen from 24% to 48% between 1979 and 2005 (Mareuge and Ruiz [2008]). Water
affordability and access has been a hot topic in France as the French Parliament has
been voting the right for an existing governmental agency to pay a part of the bill of
households with financial difficulties, e.g. experiencing overindebtedness or unsanitary
housing. While access to water is a recognized right in international conventions, public
and private operators jointly created in 2000 a special fund to subsidize poor households
which could not pay for their water bills. There are however very few statistics about
water poverty in France. According to Smets [2004], there are 3 million French people
experiencing difficulties to pay their water, electricity, gas or phone bills. The same year,
over 700,000 households have asked to reschedule their water bills.
Defining water poverty is difficult as the threshold depends on local conditions. This is
especially true for the French case where prices and incomes differ from one municipality
to another. According to Smets [2004], the affordability index for households with an
income below 40% of the median income varies from 2.5 to 3.5% in developed countries.
A threshold of 3% was also proposed by the OECD and by the United Nations specifically
for France (Reynaud [2007])6. Using this definition, Reynaud [2008] finds that 4.31% of
French households are water-poor in 2006. As we only consider the first part of the bill
representing exactly 50% of the whole tariff with value-added taxes, we consider water-
poor as households paying more than 1.5% of their income in their water bill. Using
this definition, there are 479,974 out of 16.5 million households in our dataset potentially
experience water poverty. On average in our database, French households pay water pro-
vision bills lower than 0.7% of their income, a figure that is consistent with the UNRISD
report by Reynaud [2007]7.
This definition of poverty is however limited. First, “water-poor” may not be house-
hold facing financial stress. A simple example can illustrate the limits of the definition.
Households with swimming-pools can consume large amount of water resulting in con-
sistent water bills. Second, from one consumer behavior perspective, water consumption
may only be the result of utility-maximizing behaviors. For these reasons, we will use a
broader definition of poverty and needs-based on the national poverty threshold.
3 Data and Research Design
We developed a unique dataset by combining data from the French Environment Institute
(IFEN-SOeS), the French Health Ministry (DGS) and the French National Institute for
Economics and Statistics (INSEE) on 5,215 representative municipalities in 2008. Because
of missing data, our results are extracted from a 4,500 observations dataset. We match this
6Several studies such as Fitch and Price [2002] for the UK and Reynaud [2008] for France conclude that
water poverty means that the share of income spent by households for water services is equal or higher than 3%
for the three lowest deciles. They however consider a bill including water provision and sewage. Hence, being
water-poor can result from one decision for the highest deciles.
7According to a report by Reynaud [2007] for the UNRISD, the average percentage of income spent on paying
water charges is 1.20% in 2001 for French households.
large dataset with a sub-sample of 650 observations on net results in the water industries
for 2009. The unit of observation is a municipality.
3.1 IFEN-SOeS database
The IFEN-SOeS, collected by the French Environment Institute and the Environment
Minister, is a nationally-representative municipal survey of the public service of water.
This sample is representative of the total French population and the local public author-
ities where they are living: all sizes of local authorities are proportionally represented
and municipalities with more than 5000 inhabitants are all included. The IFEN-SOeS
database provides detailed information about public water services and municipalities’
characteristics. There has been four data collection in the last ten years. The data col-
lection proceeds as follows. Municipalities fulfill the database, then data is checked by
the Environment Minister. The IFEN-SOeS is the only national representative dataset
on public water services.
The database includes a lot of information at the municipal level about water con-
sumption by domestic customers8and municipalities’ characteristics that can influence
water consumption. We know for example whether the city is located in a touristic area
or not or in which region the city is located. The latest variables are important con-
trols when one tries to explain water consumption: on the one hand, touristic areas face
larger levels of consumption during some periods of the year; on the other hand, water
consumption is higher in some regions such as the south of France. Moreover, we can
create dummies to take into account the density of water consumption on the network.
Using regulatory indicators provided by the French Observatory of Water and Aquatic
Environments (ONEMA in French), we consider a city to be rural if the ratio of billed
water and the length of mains is smaller than 10 and to be urban if this ratio is larger than
30. Cities with a ratio between 10 and 30 are considered semi-urban. These dummies
provide helpful controls to normalize consumption levels from one municipality to another.
Table (2) reports covariate means and standard deviation by consumption-unit house-
hold income quintile. The first quintile for example includes cities in which the median
income is between 0 and 159%. Annual per consumption-unit median income increases
from an average of 14,275 euros in the first quintile to an average of 23,755 euros in
the fifth quintile. Panel (A) in Table (2) shows some cities economic and demographic
characteristics such as its touristic and urban status. Mean annual consumption and ex-
penditure are relatively stable from one quintile to another in Panel (B). Mean annual
consumption goes from 136.145 cubic meters per year in the first quintile to 139.541 in the
fifth quintile for a relatively close expenditure. Marginal prices are similar in the quintiles
1 to 4 but very different in the fifth quintile where they are 7 to 10 cents more expensive.
This difference in marginal prices is fulfilled by lower fixed fees in the fifth quintile. Cities
with higher incomes face fixed fees equal on average to 38.611 euros while the first and
second quintiles respectively pay 48.93 and 49.456 euros for their fixed fees.
8Domestic customers include households but also small firms and agricultural firms. In some cases, big firms
are also included in domestic customers. We however do not take into account exports and a part of billed
water sold to non-domestic customers, usually big firms with a particular tariff rate.
Panel (C) describes water utilities characteristics that are useful to understand the dif-
ferences in prices or costs of water production and distribution. On the one hand, ground
water is usually associated with higher treatment complexity because it is more polluted
than underground water. On the other hand, underground water is more costly to extract.
Its impact on costs is thus not clear. Treatment complexity has a direct impact on costs
and thus on the price of water. As Table (2) shows, higher quintiles are associated with
higher complexity and lower underground water that explains the differences in marginal
An important feature of the IFEN-SOeS dataset is that, in addition to characteristics
about the contract such as ownership structure, it provides high-quality information about
water bill structure. Even if we have little information about differentiated rates, we have
a lot of information about the bill composition of a baseline bill for a household, defined
by the National French Statistics Institute as a consumption of 120 cubic meters a year
per household. At the baseline consumption level, we know the amount of the fixed-part
and we can compute the marginal price per unit. As there are different rate schemes, one
might consider that observed marginal prices do not fit non-linear pricing schemes. Table
(1) shows the result of our test for consumption split in different tiers of the marginal
tariff rate. For all the municipalities with multi-tier marginal tariffs, we reject the null
hypothesis H0of an average consumption higher than the second-tier break even point
with a p-value less than 0.001. Overall, the test provides strong evidence of average
consumption levels lower than the second-tier of the marginal price. The hypothesis of a
single unit price experienced by households is thus validated.
3.2 INSEE database
The INSEE database gives us information about household characteristics at the munici-
pal level that is presented in Panel (A) of Table (2). We have the number of households,
the population structure of the municipality and median income per households. We will
briefly discuss the representativeness of this dataset.
We use median declared fiscal incomes as a proxy for a typical household standard
of living. Incomes include labor and capital incomes before tax and deductions and do
not include cash and non-cash benefits from public assistance. We however assume that
income is a good proxy for the standard of living. Using weighted incomes, we find a me-
dian income of 17,923 for a single person, while the standard of living - including benefits
and subtracting taxes - is 17,170 according to INSEE. However, our measure of incomes
has two drawbacks. First, it is upward biased for low-income as the average income in the
lower quintile is higher than it is for the standard of living (14,275 versus 10,530 euros).
Second, it is downward biased for higher incomes as the average income in the top quintile
is lower than the one of the standard of living (23,755 versus 35,580 euros). Our measure
of household incomes is thus more concentrated than the distribution of the standard of
In order to gauge the financial stress on poor households, we must measure the impact
of water tariffs on a household adjusted for its composition. To do so, we took into account
household composition at the municipal level to compute an income per consumption unit.
INSEE defines consumption units in the following way: household members aged less than
15 years old count for 0.3 unit, the first household member aged more than 15 counts for
a single unit and other members aged more than 15 count for 0.5 unit. We can thus build
an adjusted household income which takes into account that there are differences in the
standard of living across households depending on the number of household members.
Panel (A) in Table (2) shows that demographic structures are quite similar except for the
proportion of adults above 60 that is higher in lower quintiles.
INSEE defines the poverty threshold as an income of 9804 euros per year for a single
unit of consumption for 2008. As we consider median municipal incomes before taxes and
without subsidies or benefits at the municipal level - we cannot take into account isolated
single parents with children - where poverty is usually higher. Using municipal-level units,
we have to consider reforms regarding “poor cities” rather than poor households.
There are no formal definitions of what a “poor city” is. Studies made by INSEE
usually define poor cities as cities with high-level of unemployment, a large share of
households living on public benefits and annual incomes per households below 12,000 eu-
ros. For simplicity, we consider as “poor” cities with a median income per unit below
the minimum wage for a full-time employed person, that is 12,450 euros a year9. In our
dataset, “poor cities” are thus cities with at least 50% of their households not earning the
full-time minimum wage per unit. This definition is restrictive for several reasons. One
that can be particularly strengthened is that it does not take into account inequalities
within cities, as could approximate consumer-level studies. In this case, a high price of
water can have no negative impact on the average consumption of the city and at the same
time be very distorsive for poor consumers. However, using a municipal-level analysis is
useful for at least two reasons. First, as there is no national regulator, prices could tend
to be higher in rural areas with incomes generally lower than in large cities such as Paris
or Lyon. Second, cities represent an interesting laboratory to simulate the impact of the
rebalanced tariffs. In the latter case, one could extend the municipal-level results to the
district-level within a given city. Overall, city-level data provides a large heterogeneity in
prices and consumption.
3.3 OSEA database
To better understand water rate schedules in France, data on revenues, costs, the number
of customers and billed volumes has been collected for 139 big water utilities for 2009.
The data collection has proceeded as follows. We launched a data collection on the top
720 cities in France, representing 320 water utilities. We got data for 297 and, because of
missing data, obtained a complete sample of 139 water utilities. As these water utilities
all include at least one city with 15,000 inhabitants, they usually share their network with
small cities around. We finally have a dataset covering revenues and costs for 650 cities of
the IFEN-SOeS dataset. For 139 water utilities, the dataset contains information about
the global revenues and costs so one can compute a net revenue equal to revenues minus
costs. It is impossible to have detailed information about costs and investments in order
to extract water production and distribution costs on the one hand and capital cost on
the other hand. The dataset is completed using numerous variables that we can find also
in the IFEN-SOeS database such as the number of customers, billed volumes of water and
9In 2008, the minimum wage in France is 1,037.53 euros per month corresponding to 12,450.36 a year
water production specific indicators such as water sources and treatments.
OSEA dataset is useful to have information about the cost structure of the 4,500
IFEN-SOeS cities. However, we have to make several assumptions. First, we assume
that marginal costs and revenues are moving in the same way between 2008 and 2009
as our data was mostly available only for 2009. Second, data is often aggregated at the
contract level. A contract usually implies water production and distribution for several
cities, i.e. a territory. So one might assume that customer density and consumption habits
are the same from one city to another within the same territory, which is not always the
case. When it is possible to split cities one from another, we do so. Thirdly, we have
sometimes data aggregating different contracts from the same operator within the same
territory. This case is particular because marginal costs are the same within the territory
but marginal prices differ from one contract to another while we are only able to extract
one marginal price for the whole territory. Finally, we have to assume that results issued
from the OSEA database have an external validity and are thus expandable for the other
French municipalities. The next subsection discusses the potential selection-bias that can
occur from this study.
3.4 Sample-Selection Bias
Due to data collection, our merged sample is truncated. One question that arises is
whether results from this sub-sample can be generalized to the whole representative sam-
ple. To check the sub-sample external validity, we apply a simple Heckman [1979] selection
model. In the first stage, we use a Probit model of the probability of observing the data re-
garding a function of regressors independent from observed marginal costs. The selection
equation is:
where Viis a latent variable equal to one if the city is included in the sample, βthe
vector of coefficients for the selection equation, Zithe vector of covariates for city iand
ηithe random disturbance for a given city i. The vector of covariates includes dummies
for the urban, semi-urban or rural status and a dummy equal to 1 if water is privately
The second-stage of the model regresses net revenues per customer on billed water per
customer to test for marginal cost pricing. A similar model is used in Borenstein and
Davis [2011] and Davis and Muehlegger [2010] for example. The following equation gives
us the average margin per billed unit and per customer:
NRCi=α0+α1qi+α2Xi+ Φii(3)
where net revenue per customer from water sales, NRCi, is regressed on the annual
consumption per customer of a given utility, qi.Xiis a vector of variables shifting costs
- treatment types and water origins - crossed with the consumption per customer qiand
Φiis the inverse Mills ratio derived from the selection equation. The coefficient αis
the average mark-up per unit i.e. the difference between marginal prices and operating
costs. We exploit differences in water sources and water treatments to generate different
mark-ups10. The constant α0is the average extra-amount paid in fixed fees, i.e. the
difference between fixed fees and capital expenditures. The inverse Mills ratio Φimakes
this mark-up on fixed price vary from one city to another.
Table (3) shows the Heckman-selection regression results. Results can be interpreted
in the following way. From the selection equation, we observe that our sub-sample tends
to over-represent semi-urban, urban and privately managed cities. The highly-significant
coefficient of the inverse Mills ratio means that there was a selection bias from our sub-
sample. We can however control for this bias by correcting our predicted results from the
second-stage equation. Results from the test of marginal cost shows that marginal prices
tend to differ from marginal costs. Indeed, for each volumetric unit sold, a consumer pays
on average 0.1239 euros more than the marginal cost of water provision. Considering
cross-variables, bad water quality seems to be positively marked-up on per-unit prices
while more complex treatments lead to lower per-unit mark-ups. Regarding fixed prices,
interpreting the sign of the mark-up is less straightforward: while the constant suggest a
negative loss for water producers, the inverse Mills ratio has a significant positive coeffi-
cient. Using the model and the coefficients from the regression, we build counterfactual
bills using a second database with 4,500 observations at the municipal level. The results
are detailed in the next section.
4 Switching to Marginal Cost Pricing
4.1 Graphical Analysis
In this subsection, we use computed city-level natural water consumption and expenditure
to describe the rate schedules faced by French residential customers. Figure (1) plots a
fitted least squares regression line of average annual consumption and expenditure (the
solid line). There is large variation across households in annual consumption but the fig-
ure shows a strong correlation between consumption and expenditure. There is, however,
a large degree of heterogeneity in expenditure across the country. In many cases, different
households consuming the exact same amount of water in the same basin pay consider-
ably different amounts. Costs may vary across utilities based on the mix of residential,
commercial and industrial customers, scale economies, age of the meter and transporta-
tion costs when water is imported. Once again, data are computed from overall municipal
consumption and not from customers’ bills. Several limitations result: we cannot consider
whether seasonal differences in consumption have an impact on the average annual bill
for example; we can neither compare bills from different households of the same city. We
can only conclude on differences on the typical bill of a consumer in a given city.
For simplicity, we assume that consumption elasticity is null and that revenue is neu-
tral to consumption. A simple reason why null consumption elasticity can be a reasonable
assumption is that consumers can have limited attention to complex and less salient price
incentives. This situation arises when consumers do not know their marginal price of
water (Carter and Milon [2005]). While several studies assume that income and water
10In other regressions, we also included dummies for touristic areas, operators or whether municipalities are
interconnected, but the results remained stable. In order to keep an intelligible form of the cost function, we
decided to apply a simple model focusing on production costs.
consumption are strongly related (Diakité et al. [2009] for example), this assumption can
be relaxed here by the fact that income and water consumption are weakly correlated.
Figure (3) plots an fitted least square of the two variables. Each observation is a city. The
figure illustrates a positive correlation but little of the variation in water consumption is
explained by income variation. The OLS regression reveals a 0.0006 R2. Part of this lack
of correlation comes from differences of consumption in geographic divisions. However,
even in the same regions, income explains a small fraction of the variation in water con-
sumption. This weak correlation illustrates the difficulties to have strong distributional
impacts with tariff reforms. Any tariff reform must take into account household compo-
sition and structure to target water assistance programs and have stronger equity effects,
something that we consider using per-unit income.
In Figure (1), the dashed-line plots the bill faced by households under marginal cost
pricing. As the fitted least square line is flatter under marginal cost pricing, customers con-
suming the same amount of water than in the current rate scheme would face significantly
lower bills. Overall, less than 3% of customers would face higher prices under marginal
cost pricing. Households with low levels of annual consumption could face increasing bills
due to higher fixed-fees, while household with high levels of annual consumption would
tend to pay less. In the following subsection we examine distributional consequences in
detail, comparing the characteristics of households with different levels of incomes, house-
hold composition and consumption.
Factors that can create differences in rate schedules are urban density and organi-
zational choice to provide water. Figure (2) shows different bills reflecting alternative
consumption in rural (solid black line), semi-urban (dash line) and urban (dash-dot line)
areas when the utility is publicly and privately managed. This graph does not take into
account controls for selection effects that could explain differences in rates between pub-
lic and private management. However, one can see that under private management, the
slope of the line is higher than under public management, meaning that prices increase
faster under private management. Another noteworthy element is that under private
management, urban areas face higher marginal prices than semi-urban areas.
4.2 Rebalancing rates in water tariffs
Table (4) describes the rate schemes for different types of water utilities. We present
marginal and fixed prices for different organizational choices and different consumption
density. The unit of observation is a municipality. Results are unweighted by the number
of households. So when considering marginal price in public and private management
for example, we consider average price between municipalities, not between households.
Household-level results would be different as there are heterogeneity in the number of
inhabitants between and within the different categories. For example, if all the inhab-
itants of Paris support an increase in prices, this has a more important impact at the
national level than it could have in a small city. However, as the nature of our data is
municipally-leveled, we present change in tariffs at the city-level.
The first column shows current water tariffs while the second column gives the rebal-
anced rate schedules when the Coasian tariff is implemented. In many cases, different
households consuming the exact same amount of water in the same region pay consid-
erably different amounts. This heterogeneity in water prices is at first sight surprising.
In most cases, water production is quiet cheap and does not change a lot across regions
or basins. However, differences arise from the cost of local distribution and other fixed
costs that are recovered in the utility’s volumetric charge or fixed costs. The difference in
per-unit price between public and private management is a little bit more than 18 cents,
representing a 16.8% deviation from mean price. The gap between private and public
management is even wider when one considers the fixed-part of the tariff. There is indeed
a 12.63 euros difference per customer, representing a 27% deviation from the mean fixed-
In column (2), marginal tariffs are rebalanced such as the water industry tend to-
wards to a null profit, in the idea of Coase [1946]. In column (2), the reformed rates
are derived from the marginal cost model corresponding to equation (1) and Table (3).
All prices logically decrease on average but some heterogeneity is found between organi-
zational choices and different consumption densities. While marginal prices decrease in
rural areas, they tend to increase in urban areas and to remain stable in semi-urban areas.
On average, marginal price is set 0.154 euros higher than marginal cost under public
management while unit price is 0.119 higher than marginal cost under private manage-
ment. Differences between organizational choices are higher under marginal cost pricing:
on average, unit price under private management will be 0.218 euros more expensive while
it is 0.183 under current rates. Public managed utilities thus tend to have higher per-unit
margins than privately managed utilities. The gap between public and private manage-
ment is even wider if one considers the fixed-part of water rates. While in column (1), the
gap is 12.63 euros, it is 17.06 euros in column (2). One might consider that this wider gap
between public and private management is counterintuitive. In the public debate, public
management is often viewed as being cheaper because it has lower margins than private
management. We argue here that per-unit prices under public management could be even
cheaper while private managers tend to keep low per-unit margins to remain competitive11.
Another factor that creates differences in rate schedules within divisions is population
density. Consumers in urban areas face higher unit prices than consumers in rural areas.
The gap is however balanced by the differences in fixed costs. Urban customers pay on
average 34.89 euros per year for their subscription while rural customers pay 57.81 euros
per year. This is surely because a part of fixed costs in urban areas is recovered by
the volumetric charges while in rural areas where consumption density is lower, utilities
secure their revenues through high fixed tariffs. Note that rural areas represent 40%
of our observations but only 1,670,649 households versus 9,391,694 households living in
urban areas and 5,590,629 living in semi-urban areas. Even if cities experience on average
decreasing fixed fees, households experience overall increasing fixed fees when they switch
from current tariffs to Coasian rates. Column (2) shows that current water tariffs are far
from being well-designed and could be rebalanced in order to slightly increase fixed-price
and lower marginal prices. This would also fit firms’ willingness to ensure sustainable
11Accounting rules in public budget are clear. All margins are automatically used to fund next year operating
expenses or can be used as provisions for future investments. However, these provisions i) are against lower
prices for consumers, ii) do not represent the cost of water supply and are distorsive and iii) do not imply larger
investments under public management.
profits using access fees and to maintain the optimal level of investments12.
4.3 Counterfactual Bills
Table (5) describes the distributional impact of a change to marginal cost pricing as-
suming zero demand elasticity. Panel (A) reports results by household income quintile.
Households in the first quintile would pay on average 22.32 euros less under marginal cost
pricing and only 1.1% of the households of this quintile would experience a bill increase.
Households in the fifth quintile would experience smaller decreases in bills and 4.67% of
this class would experience increase in prices.
Results in panel (B) by adjusted income quintile are somewhat similar to the previ-
ous results. When one considers household composition, households in the first quintile
face larger decreases in bills than households from the fifth quintile. The former would
annually pay 21.45 euros less while the latter would pay on average 15.90 euros less. The
pattern of the change comes from the fact that lower adjusted income quintiles can be
those with higher consumption if the lower income is due to numerous members in the
household. For example, a family of two adults and three children would have a lower
adjusted income than in panel (A) while their consumption would remain the same.
Panel (C) examines consumption quintiles. As Figure (1) shows, the transition from
current tariffs to marginal cost pricing is assumed to advantage households consuming
the biggest amount of water. The first quintile in panel (C) has a probability of 4.33%
of experiencing increase in bills because of increasing fixed-prices. Panel (D) focuses only
on water-poor and poor cities. Applying marginal cost pricing leads to lower prices for
water-poor and households below the poverty line. Municipalities with water-poor ex-
perience a 54.26 euros decrease in their bills and municipalities with incomes below the
poverty line experience a 22.19 euros decrease in their bills. The gap between the two
groups of households comes from the fact that water poverty is correlated with consump-
tion and incomes while the poverty line depends only on income considerations. A few
municipalities with water-poor citizens or median incomes below the annual minimum
wage experience increased bills under Coasian tariffs.
Even if Table (5) is instructive to understand the impact of reformed tariffs, there are
two drawbacks to the correct interpretation of the table. On the one hand, one might
argue that household income may not be a good indicator of the financial stress that
households face. Cutler and Katz [1992] state for example that permanent income is a
more accurate measure of the distribution of resources than current income. Poterba
[1989] argues that households can base their spending on their expected lifetime income,
meaning that consumption would provide a more accurate measure of households’ re-
sources. On the other hand, our residential approach to water consumption does not take
into account households’ appliances, that can be a proxy for expected lifetime income.
There is unfortunately no available data on durable goods owned by households at the
municipal level. However, this could be an interesting point to explore using a household-
level dataset. Ideally, we could also have information on consumers’ housing such as the
number of bathrooms they have, whether they rent or own their housing and whether
12One of the theoretical features of public-private contracts is that, in a principal-agent model, the agent in
charge of providing the service will underinvest if it has no incentives to do other.
they live in multiple-unit buildings or not.
4.4 Including Water Assistance Programs
Table (5) gives clear-cut results in favor of efficient pricing for consumers. However, its
redistributive impact can be considered insufficient and can be criticized in terms of out-
comes for operators who would experience substantial profit losses. In this section, we
consider that the regulatory profile would ensure marginal cost pricing for the volumetric
charge. We then assume two situations corresponding to Part I and Part II in Table (6).
In Part I, a Coasian tariff is implemented and firms have to bear null profits in favor of
consumers. In Part II, we assume that firms charge per-unit consumption at the marginal
cost but increase fixed fees in order to maintain the same level of profits than under cur-
rent tariffs. We run four reforms that could be discussed at the national level. In panels
(A) and (B) of Part I, we consider two reforms. The first one provides free fixed fees for
households in poor cities. The second one consists in a refund of increased fixed fees that
can result from Coasian tariff schedules, no matter if the city is considered as being poor
or not. The result of the later reform can be expressed in the following way. Cities with
increased fixed fees under rebalanced tariffs will be funded in order to face the current
fixed fees. We then compare their distributional impacts regarding current price schedules.
In panel (C) and (D) of Part II, we consider marginal cost pricing with increased
fixed-fees such as water industries keep the profits constant and we apply a free-fixed fees
policies for poor cities and for cities with median incomes below 159% of the poverty line.
Table (6) reports the results of these simulations on five categories: cities with median
income below 159% of the poverty line, water-poor cities, poor cities, the annual cost per
non-recipient and the overall cost in millions.
Panel (A) in Table (6) shows the impact of free-fixed fees on poor cities before rebal-
ancing tariffs. Because households below the poverty line represent 576,399 households
out of 16.7 million in our dataset, it is relatively costless to fund a free-fixed fee policy by
non-recipient households. The impact on tariffs in poor cities is a decrease of 29.14% of
the water bill, representing 50.51 euros. On average, cities with a median income below
159% of the poverty line experience a decrease of their water bill by 9.372 euros per year
but 79% of this category has to participate in the funding of poor cities. The annual cost
per non recipient is 1.44 euros per year for an overall cost of 23 million euros.
Panel (B) is the case in which tariff reform is guaranteed with no increase in fixed-fees
in any city regarding the current tariffs, no matter whether the city is considered poor
or not. In this case, households living in a municipality within the first quintile face an
average decrease of 0.33 euro in their annual bills. Poor cities experience a decrease of 1.25
euros on average of the water bill and no poor cities would experience increased tariffs,
meaning that poor cities are all cities facing increasing fixed rates when we switch from
current to Coasian tariffs. The annual cost of this program is 1.20 euros per non-recipient
household and the overall cost is 19.4 million euros, both are below what is observed in
panel (A). Costs of reforms in panels (A) and (B) are comparable but they do not target
the same cities. Programs described in panel (B) will especially advantage urbanized ar-
eas that are more represented within the 5th quintile (a quarter of the cities) than within
the 1st quintile (around 13% of the cities) as it is shown in Table (2).
Results in Part I of Table (6) provide a better understanding of the costs of tariff
reforms. While households would on average largely benefit from Coasian tariffs, small
consumers could be disadvantaged regarding large consumers. Panels (A) and (B) give
solutions to mitigate the distributional impacts of reforms. Note that these reforms could
be implemented under current tariffs.
In part II of Table (6), we assume marginal cost pricing and rebalanced fixed fees such
as firms do not support profit losses under the 0-demand elasticity assumption. In this
case we assume marginal cost pricing for the volumetric charge and higher fixed fees to
maintain constant profits for the firm. One of the arguments against marginal cost pric-
ing when firms maintain their profits is that it results in larger fixed fees that can affect
particularly poor households. We offer here two alternative reforms that can mitigate the
distributional impacts of a transition to marginal cost pricing with a significant increase
in fixed fees. This solutions can associate efficiency in pricing at marginal cost and equity
by decreasing bills in poor cities.
Panel (C) shows the result of a free fixed fee policy in poor cities funded by non-poor
cities’ households. Because of increased fixed fees for all the households, cities within the
first quintile and poor cities would experience larger decreases in their bills. The cost per
non-recipient would be 0.46 euro higher. For the same level of consumption, each house-
hold in non-poor cities would have to pay 1.90 euros more than under current pricing.
Overall costs are 30.60 million euros, 7 million more than under Coasian tariffs.
Panel (D) shows the impact on water bills of a free fixed fees policy for cities with
a median income in the first quintile. In this case, the scope of the policy is wider as
the number of households targeted largely outpasses the number of households living in
poor cities (3,319,712 vs. 576,399 households). As one can expect, the policy has a larger
impact on the mean annual bill of cities within the first per-unit income quintile with an
average decrease of 88.15 euros per year. The annual cost per non-recipient is 11.86 euros,
representing 6.64% of the typical bill of a non-recipient, a 166 million euros overall annual
cost. Matching efficiency with equity is thus possbile if the implementation of marginal
cost pricing for the volumetric charge is combined with transfers between cities.
Under rate reforms such as those presented in panel (C) and (D), poor cities would
experience larger decreases in their annual water bills at a low cost for a non-recipient. In
more ambitious reforms such as the one presented in panel (D), cities with median incomes
in the first quintile would have average bills decreased by more than 69 euros, a result
that is more than three times higher than under marginal cost and capital cost pricing
without water assistance programs. These results suggest that it may be possible for wa-
ter assistance programs to take into account distributional considerations without losses
of revenues for water utilities, a solution that is more credible than perfect Coasian tariffs.
Nevertheless, designing consistent water assistance programs is difficult. First, threshold-
effects are important. Households in cities with incomes just above the defined poverty
threshold would face increased tariffs to fund households below the poverty line. Second,
it implies that water utilities fix their rates considering household incomes instead of their
costs. Even if they were subsidized by other customers, this would imply limitations in
their capacity to negotiate contracts that reflect their needs.
It is also worth emphasizing that these mean impacts obscure substantial heterogeneity
across households. Because households differ substantially in their level of water consump-
tion, the lump sum payment can be far too much for small consumers and not incentive
enough to sustain water resources for others. Moreover, utilities differ in their needs to
invest in capital. Suppressing fixed-fees for a whole set of utilities, even if they get na-
tional subsidies, could be alarming as the level of investments would depend on other
subsidies rather than their capacity to raise fixed prices. Finally, these reforms would face
political challenges, as municipalities are keen on administering their contracts, even if
the proposed reform would probably better match the needs of poor households than the
current tariffs. For all these reasons, efforts should go in the direction of efficient pricing,
potentially closer to marginal cost pricing.
5 Welfare Effects of Changing Retail Prices
In order to evaluate the total deadweight loss from the observed departures from marginal
cost and capital cost pricing, we first estimate the price-elasticity of demand for each
per-unit of consumption income quintile. We then calculate the welfare changes and the
deadweight loss associated with the current pricing schedules compared to efficient pricing.
5.1 Consumer elasticities
The counterfactual bills we have considered thus far show how household expenditure on
water would change under marginal cost pricing if demand elasticity were zero, which im-
plies huge efficiency consequences of the change. With non-zero elasticity, it is interesting
to see whether households would consume more water, thus leading to a proper dead-
weight loss. Table (7) reports demand elasticities for the five household adjusted-income
quintiles. In order to compute elasticities, we regressed the log of annual consumption
per household on the logs of marginal price, income and demand shifters such as regional
fixed-effects, urban density, touristic area, household size and the share of population aged
between 15-64 years old.
Demand is significantly negatively correlated with marginal prices. The elasticity point
estimates for the first quintile is -0.281 while it is -0.223 for the last quintile. The second
and the third quintile face higher elasticities than the first one with respectively -0.287
and -0.304. These results are consistent with previous studies on the French water market
(Nauges and Thomas [2003]; Garcia and Reynaud [2003]) but also in developing countries
(Nauges and Whittington [2009]) and other markets such as gas or electricity in the USA
(Borenstein and Davis [2011]; Ito [2010]). This estimation includes income elasticity by
using crossed variables between per-unit of consumption quintiles and marginal prices.
Even if one could consider linear effects of income elasticity, here we take into account
different price-elasticity intensities following revenue distribution.
Municipalities’ demographic and geographical characteristics have strong effects on
water consumption. Regional fixed effects are significant to explain differences in level of
consumption. Touristic areas are associated with higher level of consumptions because
of a high level of seasonal consumption. Urban and semi-urban areas tend to have less
per-household water consumption than rural areas. Average household size and structure
matters. The larger the number of family members, the larger the consumption. A large
share of 15 to 64 year old inhabitants is also associated with lower levels of consumption,
perhaps because cities with a lot of working inhabitants are often urbanized and thus
correlated with less-consuming capital goods.
From our demand-elasticity results, we can conclude that changing retailing prices
would improve economic efficiency as consumers would change their behavior in response
to the price changes. The efficiency impact can however be limited because consumer
behavior is difficult to predict, and decreased marginal prices do not automatically lead
to increased consumption. This is especially true in cities experiencing increased fixed
fees such as urbanized areas. Computing welfare changes implies taking into account
the linear welfare impact of marginal-cost pricing, the increased consumption that results
from lower prices and the change in fixed fees.
5.2 Welfare Effects including Marginal Quantity Changes
Counterfactual bills presented so far showed welfare changes under the assumption of zero
demand elasticity. In this subsection, we use elasticities from Table (7) to compute the
deadweight loss of restrained water consumption due to inefficient pricing. We assume
here that the tariff change does not lead any consumers to enter or exit the market. Table
(8) reports deadweight loss generated by using existing pricing tariffs relative to marginal
cost prices. We separately report mean welfare changes for each adjusted-income quintiles
at the municipal level and for the whole set of households taking, weighting the municipal
observations by the number of households. To compute welfare changes, we consider a
constant elasticity for each income quintile. Price-elasticity is thus the same to a certain
threshold of per-unit of consumption income. On average, in the sample, lowering the
volumetric charge implies a 12% decrease. With a -0.22 to -0.3 price elasticity, this yields
an increase in consumption of 5.5 cubic meters for the average consumer of a city com-
pared to the initial level of 136.8 units. We consider the change in consumer welfare as the
area to the left of the demand curve that computes the area of the difference between the
original price and the marginal cost and the new level of consumption, and we substract
the difference between annual fixed fees. The deadweight loss corresponds to the triangle
ABC in figure (4).
Overall, the current marginal price schedule creates 5,357,913 euros in deadweight loss,
relative to efficient pricing. The dataset represents a market of more than 16.7 million
households and a gross market of 3.05 billion euros13 so the deadweight loss represents
approximatively 3% of the considered market. As a thought exercise, one can compute the
deadweight loss for the whole water market in 2008 as the full dataset is representative of
French municipalities. As there are 26.615 million households, the deadweight loss for the
water market in 2008 could be set to 8 million euros for household consumption. Even in
the case of counterbalanced fixed-part tariffs in order to maintain water industries’ prof-
13For simplicity, we excluded taxes that are proportional to the volumetric consumption of water, such as
value-added taxes but also a whole set of fees related to water production and distribution. When it is possible
to dissociate domestic from industrial consumption, we do so. We also exclude sanitation and sewage from our
analysis as we do not have information about the cost structure of these services.
its, the deadweight loss would remain the same, as it is the result of differences between
marginal prices and costs. These results help clarify the overall debate about tariffs in
In Table 6, we find that a free-fixed fee policy in poor cities has an annual cost of
23 million euros under Coasian tariffs (see panel (A) in Part I of 6), while the efficiency
cost of non marginal-cost pricing is 5.36 million euros. For the price elasticity of demand
found above, the deadweight loss from transferring these funds is lower than 25%, mean-
ing than the distortionary impact of a 20% take-up of fixed fees in poor cities for example
could be offset under Coasian tariffs. Under marginal cost pricing with current profits,
a full take-up of fixed fees in poor cities would cost 30.60 million euros (see panel (C) in
Part II of 6). This is far more than the efficiency cost of current tariffs. In this context,
water assistance programs could fund a minor part of fixed fees, e.g. a subsidy of 5 to
10 euros per household that could barely offset the negative impact of increased fixed fees.
The effect of marginal cost pricing on water conservation is also another feature of
the deadweight loss analysis that must be discussed. Under marginal cost pricing and the
assumption that customers respond to their marginal price, a typical household would
consume 5.5 cubic meters more per year on average than under current tariffs, a result
that goes against the argument for sustainable water use. In an extensive way, one could
imagine that consumers paying cheaper bills would invest in less-consuming durable goods
and thus promote water conservation.
These estimates provide a valuable preliminary assessment of the welfare consequences
of the observed departures from marginal cost pricing. However, it is necessary to un-
derline that the calculation of the deadweight loss is sensitive to the estimation of the
elasticity demand. This has two limitations. First, demand elasticity might differ when
one considers marginal price and average price (Borenstein [2010], Ito [2010]), or differ-
ent estimates of long-term elasticity (Nauges and Thomas [2003]). We will discuss these
limitations in the following section. Second, consumer elasticities assume that individuals
respond to a pricing scheme in a way that the standard economic model predicts. Heck-
man [1983] shows for example that in nonlinear price schedules, the absence of bunching
around the kink points could imply that individuals respond to other perceptions of price
rather than the actual marginal price they are paying. Cognitive difficulties to understand
rate schemes or simply missing information about their marginal price of water could also
limit the possibility of evolving consumption when marginal price decreases.
5.3 Possible Explanations for Maintaining Efficiency Costs
Departures from efficiency pricing may have three explanations (a similar discussion is
made by Davis and Muehlegger [2010] for the US gas industry). The first one lies in
firms’ profit maximization. In the last years, water operators in France have been justify-
ing the increasing marginal prices of water by the diminishing demand from consumers.
Increasing marginal price was thus a means to maintain stable profits. Moreover, some
commentators argue that fixed fees are too large regarding capital costs because firms
want to maximize their profits using fixed fees. The marginal cost of a new customer is
indeed null and does not vary with the utilities’ characteristics. In practice, small cus-
tomers are sensitive to fixed fees while large customers are sensitive to unit fees when
they make their demand decision. In Table (4), the transition from current schedules to
Coasian pricing shows that utilities currently advantage small customers in urban areas
and large consumers in rural areas. We undoubtedly lack information as we do not have
the details about the stock of capital and the forthcoming investments. However, water
companies have probably different pricing strategies depending on cost structures and
water utilities’ characteristics that can explain different styles in departures from Coasian
Environmental considerations provide a second alternative explanation for setting high
per-unit margins. In this view, departures from marginal costs could be justified by the
need to address environmental externalities (such as water pollution) and sustainable wa-
ter use14. In the standard view of externalities, the gap between marginal prices and
costs is comparable to a Pigouvian tax that would reflect marginal damages. In this
case, current tariffs15 reflect the socially optimal level of exchange on the market be-
cause marginal prices equal the sum of private marginal costs and the costs of marginal
damages. However, while this assumption is reasonable in competitive markets, they are
less reasonable for regulated markets such as water in France. As noticed by Davis and
Muehlegger [2010], in regulated markets, the standard Pigouvian solution is only verified
and thus not distortionary if prices are set equal to marginal cost. An alternative view is
that tariffs reflect the need for sustainable water use, including a discount rate in current
tariffs. However, recent renegotiations in France tend to prove that tariffs probably reflect
more the market structure than the real need for sustainable use16.
Moreover, in France, negative externalities and resource protection are considered in
the tariff structure of water. Two fees, one to protect resources and one to struggle against
pollution, have been implemented. These fees are per-unit taxes that finance Basin Agen-
cies’ in order to subsidize projects which struggle against pollution and ensure resource
protection. The per-unit rates of these fees are fixed by the Agencies and depends on
the geological characteristics of the Basin. These characteristics are the origin of water
and the condition of the sources for the resource protection fee and pollution intensity
for the pollution fee. On average, the pollution fee is a 0.21 euro tax per unit while the
resource protection rate is a 0.52 euro tax per unit. These fees are largely higher than the
margins from current tariff, that are around 0.15 euro. Moreover, per-unit margins are
higher in rural than urban areas while pollution and resource protection fees are higher
in urban areas than in rural areas. Margins are thus not justified by the search for more
sustainable use, neither by the scope of struggling against negative externalities.
14One might argue that the difference between marginal prices and marginal costs could reflect different level
of leaks between utilities. As Garcia and Thomas [2001] noticed, when demand increases, utilities face two
choices. On the one hand, they can repair leaks, which is costly as it is largely labor-intensive. On the other
hand, they can produce more water, which is less costly as it is electricity-intensive. Utilities with low leak-ratio
may have to deal with higher costs. This explanation can explain why utilities have different marginal prices,
as some include water scarcity in their pricing strategies, but not why marginal prices and marginal costs differ.
15To the best of our knowledge, there are few studies evaluating the price of scarce resources. Moncur and
Pollock [1988] consider for example the change of marginal cost that would occur at the complete use of the
current water source. In their study, they consider that water demand would be satisfied through a desalination
technology or a trans-basins diversion, leading to a marginal cost twice higher than the current one.
16Recently, the price of Antibes, a city in the south of France where water stress is important, has been
divided by 1.5.
These fees should be the main instrument to ensure environmental considerations and
regulatory rules should incite firms to fix water rates regarding costs rather than sustain-
able use. These fees could however be reformed in order to be set by progressive tiers
matching the marginal private impact of consumption on resource safety, assuming that
consuming more water has a more negative impact on resource sustainability. However,
the distributional impact would be uncertain as the correlation between consumption and
income is positive but very flat. For this reason, agencies could consider regional price
elasticities and incomes to define the levels of the fees.
A third explanation to current efficiency costs is private operators’ participation in
the market. Private operators’ participation has been growing since the 1980s and is
often pointed as being responsible for high marginal prices. On the contrary, public pro-
vision is often regarded as an alternative approach for lowering per-unit prices. However,
in our OSEA sample, public provision is associated with higher net results than private
management, thus leading to higher distorsions. There are several reasons for this situa-
tion. According to the highest French financial court (Cour des Comptes [2011]), public
providers tend to underestimate the depreciation rate of capital in order to get higher net
results and to refund their water debt; on contrary, private providers tend to overestimate
capital depreciation to decrease their results and the amount that they have to pay in
taxes. Moreover, in municipalities with less than 3,000 inhabitants, public managers can
use the profits of their water services to finance other prerogatives of the municipality.
Finally, public and private management face different tax rates, particularly on labor.
Private firms have to pay extra-taxes to fund their retirement schemes; in public manage-
ment, these fees are paid through taxation at the national level. In the latter case, this
means that current lower public management fees are associated with tax distortions in
other parts of the economy. In this case, a general rule following Hotelling [1938] could
be to directly fund fixed-costs using public subsidies to break the differences in taxation
between public and private management.
6 Discussion and Further Extensions
6.1 To Which Price Do Consumers Respond?
The previous analysis maintains the assumption that households have perfect information
and respond to marginal cost pricing, an assumption that is common to several papers,
e.g. Saez [2004] on income taxation, Reiss and White [2005] on electricity pricing and
Olmstead et al. [2007] on water pricing. These may not be reasonable assumptions. Al-
though water bills are reasonably clear about the distinction between the fixed part tariff
and the volumetric charge, many customers have not thought much about the distinction.
As a matter of fact, a large number of surveys show that a majority of people do not
know the marginal price of their nonlinear tax, electricity and water rates. For example,
Carter and Milon [2005] find that only 6% of households know their marginal price of
water. Rebalanced prices could then have no clear effects.
Customers who are not aware of the existing two-part tariffs, or that do not under-
stand the two-part tariff, might respond to the total bill, rather than the volumetric
charge17. Such an assumption would consistently change the previous results as price-
elasticities critically depends on whether consumers respond to marginal or average price.
Recent empirical evidence on the electricity distribution in the United States shows that
customers respond to average price rather than marginal, expected marginal or average
price (Ito [2010], Borenstein [2010]). Ito [2010] finds evidence that Californian households
respond to average price rather than marginal prices concerning electricity. Although
these results are interesting, they do not fit overall water market regulation in France as
Californian households face four and five-tier increasing block tariffs. As we have shown
in our test of non-linear pricing schemes in section 2, the structure of rates in the French
water industries is simpler and allows households to distinguish average and marginal
volumetric prices. However, this leaves interesting studies to do in France in geographic
areas where there are two or three-part marginal tariffs.
Borenstein [2010] uses electricity consumption household-level data from Californian
utilities and suggests that individuals may use expected marginal price rather than their
average price in the presence of uncertainty. Such utility-maximization models can be
implemented with annual or monthly series. Our whole dataset contains data for four
separate years -1998, 2001, 2004 and 2008 - which makes results less consistent. Indeed,
consumers may not choose their level of consumption for a given year using marginal or
average prices from their consumption level four years ago.
As a thought exercise, it would be interesting to consider how the welfare implications
would change under the alternative hypothesis that households respond to average prices.
Under a transition to marginal cost pricing, households with high consumption levels ex-
perience decreases in both average and marginal price, implying welfare gains regardless
of how well the customer understands the tariff. In contrast, households with low con-
sumption levels could experience decreasing marginal price with increasing average price,
potentially moving consumption in the wrong direction. The total change in welfare could
be positive or negative.
As an extension, we computed elasticities under average prices. Results are shown
in Table 9. Price-elasticities when consumers respond to average price varies between
-0.606 for the first quintile to -0.581 for the fifth quintile. This leads to a deadweight
loss of 9,105,368 euros for 16.7 million households, a higher value than when consumers
respond to marginal prices. The reason is that price-elasticities are higher under average
price responses. As a result, distributional consequences could be more equitable for poor
households under this assumption as the deadweight loss would be higher. Efficiency gains
could fund for example 50% of the fixed fees for poor households through water assistance
programs. However, the interval of price elasticities suggests increases in consumption
that would weaken the achievement of water conservation.
However, one should bear in mind that elasticites computed with average price raise
several endogeneity and identification problems as Borenstein [2010] and Ito [2010] no-
ticed. Indeed, as average price depends directly on the level of consumption, the OLS
average-price elasticity estimates are probably biased. An instrumented regression should
be used, including as instruments consumption shifters that could explain different sta-
17de Bartolome [1995] finds for example that many individuals in laboratory experiments use their average
tax rate as if it is their marginal tax rate when making economic decisions based on tax tables
ble consumption levels. Further extensions, using panel data, would provide consistent
demand-elasticity estimates when consumers respond to average price.
Overall, if customers respond to average price rather than marginal price, then the
welfare gains from rebalancing water tariffs could be slightly different. This raises other
questions such as the design of water bills or transparency about marginal and average
prices and about fixed fees and volumetric charges. Because of this lack of information,
consumers have probably under-maximizing behaviors. In particular, suggested reforms
should be clearly explained to consumers, in order to incite them to change their behaviors
in the expected way.
6.2 Distortions in Connected Markets
A complete empirical investigation of the distortions on connected markets is far beyond
the scope of the paper. However, one might consider that sanitation tariffs are also im-
portant to consider. Sanitation costs and prices have been growing in recent years for at
least two reasons. First, regulation on pollution has been hardened by the need to im-
prove water quality. Second, private participation within this particular sector has been
growing because of the large amounts of investments to undertake. Negative net results
in sanitation could thus explain the need for margins in water distribution.
Further studies could investigate the global efficiency costs of the water and sanitation
markets. As the markets are related, a part of the distortion in one market could be
the results from the other market. An interesting question lies particularly in the scope
economies that could benefit operators that bundle both public services. Desrieux et al.
[2012] for example find strong evidence of scope economies between water and sanitation
markets in France leading to reduced bills under bundled services. The study of net re-
sults from these two connected markets would be interesting as a part of the investments
are shared between the two sectors.
Another connected market is the quality and protection of forest lands. Abildtrup
et al. [2011] for example shows using a French sample of cities in France that the propor-
tion of forest land at the local level has a significant negative impact on water production
costs. Forest preservation is costly but can lead to the preservation of water resources.
Further studies could examine this point, by comparing the marginal cost of protecting
forest lands and the marginal impact of this protection on marginal water production costs.
Further studies could focus on the impact of distortions between connected markets.
There could be especially some tax distortions between directly and privately managed
water utilities that could explain differences in prices and margins at the local level.
7 Conclusion
In this paper, we used nationally-representative city-level data to characterize the transi-
tion to marginal cost pricing in French water industries. The results confirm that price
reform would have positive distributional consequences, but tends to be similar from one
quintile to another. Needs-based reforms, such as free fixed fees in poor cities, could
likely increase the distributional consequences in favor of households at the bottom of the
income distribution.
We have three main results. First, we find that departures from marginal cost pricing
are not very important - an 8% gap between marginal prices and costs - regarding other
regulated industries. However, margins result in a transfer from consumers to producers
that results in a 201 million euros gain for operators at the expense of consumers. Second,
we compute estimates of the price elasticity of demand that are consistent with previous
literature and we estimate the efficiency costs of current rate structure to be around 8
million euros for the French water market for 2008. In short, the current tariffs induce
a level of consumption that is too small for a range of households because of inefficient
prices. Third, efficient pricing does not level out the existing differences between con-
sumers. Water assistance programs can be implemented to erase the negative impact of
marginal cost pricing, especially when fixed fees increase to maintain firms’ profits. These
programs can be funded by customers themselves through cross-transfers. However, such
transfers result in distorsions that should not exceed the efficiency gains of marginal cost
pricing. Transfers could thus only cover a part of fixed fees for households living in poor
The broader conclusion is that policy makers, firms and municipalities should bear
in mind the trade-off between equity and efficiency when implementing rate structures.
Stronger regulation in France could lead to the broader use of redistributive tariffs or to
the constitution of funds to directly finance households experiencing difficulties to pay
their bills. Because of the strong implications of the subject, more analyses, using real
world data, are needed to study the impact and the magnitude of rebalanced tariffs and
assistance programs.
Abildtrup, J., Garcia, S., and Stenger, A. (2011). The effecs of forest land use on the cost
of drinking water supply: A spatial econometric analysis. Working Paper.
Ballard, C. and Fullerton, D. (1992). Distortionary Taxes and the Provision of Public
Goods. Journal of Economic Perspectives, 6(3):117–131.
Baumol, W. J. and Bradford, D. F. (1970). Optimal Departures from Marginal Cost
Pricing. American Economic Review, 21(6):265–283.
Boiteux, M. (1956). Sur la gestion des monopoles publics astreints à l’équilibre budgétaire.
Econometrica, 24(1):22–40.
Borenstein, S. (2010). The Redistributional Impact of Non-Linear Electricity Pricing.
NBER Working Paper.
Borenstein, S. and Davis, L. W. (2011). The Equity and Efficiency of Two-Part Tariffs
in U.S. Natural Gas Markets. Journal of Law and Economics.
Carter, D. and Milon, J. (2005). Price knowledge in household demand for utility services.
Land Economics, 81(2):265–283.
Coase, R. H. (1946). The Marginal Cost Controversy. Economica, 13:169–182.
Cour des Comptes (2011). Les services publics d’eau et d’assainissement: des evolutions
encourageantes. Report, Cour des Comptes.
Cutler, D. M. and Katz, L. F. (1992). Rising Inequality? Chances in the Distribution of
Income and Consumption in the 1980’s. The American Economic Review, 82(2):546–
Davis, L. W. and Muehlegger, E. (2010). Do Americans Consume Too Little Natural Gas?
An Empirical Test of Marginal Cost Pricing. Rand Journal of Economics, 41(4):791–
de Bartolome, C. A. (1995). Which Tax Rate Do People Use: Average or Marginal?
Journal of Public Economics, 56(1):79–96.
Desrieux, C., Chong, E., and Saussier, S. (2012). Putting All One’s Eggs in One Bas-
ket: Relational Contracts and the Management of Local Public Services. Journal of
Economic Behavior and Organization, Forthcoming.
Diakité, D., Semenov, A., and Thomas, A. (2009). A Proposal for Social Pricing of Water
Supply in Cote d’Ivoire. Journal of Development Economics, 88:258–268.
Feldstein, M. (1972). Equity and Efficiency in Public Sector Pricing: the Optimal Two-
Part Tariff. The Quarterly Journal of Economics, 86(2):175–187.
Fitch, M. and Price, H. (2002). Water poverty in england and wales. Report, Center for
Utility Consumer Law and Chartered Institute for Environmental Health.
Garcia, S. and Reynaud, A. (2003). Estimating the Benefits of Efficient Water Pricing in
France. Journal of Resource and Energy Economics, 26:1–25.
Garcia, S. and Thomas, A. (2001). The Structure of Municipal Water Supply Costs:
Application to a Panel of French Local Communities. Journal of Productivity Analysis,
Heckman, J. (1983). Comment on Stochastic Problems in the Simulation of Labor Supply.
In Feldstein, M., editor, Behavorial Simulations in Tax Policy Analysis, Handbook of
Econometrics, pages 70–81. University of Chicago Press.
Heckman, J. J. (1979). Sample Selection Bias as a Specification Error. Econometrica,
Hotelling, H. (1938). The General Welfare in Relation to Problems of Taxation of Railway
and Utility Rates. Econometrica, 6(3):242–269.
Ito, K. (2010). Do Consumers Respond to Marginal or Average Price? Evidence from
Nonlinear Electricity Pricing. EI @ Haas Working Papers.
Mareuge, C. and Ruiz, N. (2008). Dispersion des revenus et hétérogénéité des structures
de consommation, vers la notion de pouvoir d’achat effectif catégoriel. Appendix to the
Report to the Prime Minister 73, Conseil d’Analyse Economique.
Moncur, J. E. and Pollock, R. L. (1988). Scarcity rents for water: A valuation and pricing
model. Land Economics, 64(1):62–72.
Nauges, C. and Thomas, A. (2003). Long-Run Study of Residential Water Consumption.
Environmental and Resource Economics, 26:25–43.
Nauges, C. and Whittington, D. (2009). Estimation of Water Demand in Developing
Countries: An Overview. World Banck Research Observer, 25(2):263–294.
Olmstead, S., Hanemann, M. W., and Stavins, R. (2007). Water Demand Under Al-
ternative Price Structures. Journal of Environmental Economics and Management,
Poterba, J. M. (1989). Lifetime Incidence and the Distributional Burden of Excise Taxes.
The American Economic Review, 79(2):325–330.
Ramsey, F. (1947). A Contribution to the Theory of Taxation. Economic Journal, 37:47–
Reiss, P. C. and White, M. (2005). Household Electricity Demand Revisited. Review of
Economic Studies, 72(3):853–883.
Reynaud, A. (2007). Social policies and private sector participation in water supply - the
case of france. Report, United Nations Research Institute for Social Development.
Reynaud, A. (2008). Assessing the impact of public regulation and private participation.
In Prasad, N., editor, Privatization of Water: Social Policy, Regulation and PSP in
Water Supply, Handbook. Palgrave MacMillan.
Saez, E. (2004). Using Elasticities to Derive Optimal Income Tax Rates. Review of
Economic Studies, 68(1):205–229.
Smets, H. (2004). La solidarité de l’eau potable: aspects économique. Report, Académie
de l’Eau.
Figure 1: Rebalancing Water Rates
Figure 2: Rebalancing Water Rates, by Urban Density and Organizational Type
Figure 3: Water Consumption and Income
Figure 4: Water Consumption and Income
Note: The line MC is the constant marginal cost and the line MP is
the constant marginal price. P(q) is the inverse Marshallian demand
function. The deadweight loss is the ABC region in the graph.
Table 1: A test of non-linear pricing schemes in the French public water services
H0Degrees of freedom Pr6=H0Confidence Interval
Consumption Second-tier threshold 1270 100% 0.001
Note: H0is the hypothesis that the average consumption is higher than the threshold of the second-
tier of the tariff. We reject the null hypothesis with a confidence interval of 0.001.
Table 2: Descriptive Statistics by Per-Unit Adjusted Income Quintiles
Percent of Poverty-Line <159% 160-173% 173-187% 187-211% >211%
A. Cities Economic and Demographic Characteristics
Mean Annual Median Income 14274.95 (1152.104) 16252.64 (401.4486) 17639.02 (407.7273) 19401.02 (672.9867) 23754.71 3338.452
Number of Household Members 2.401 (0.256) 2.347 (0.219) 2.38 (0.255) 2.44 (0.242) 2.536 (0.224)
Proportion of Children under 15 0.181 (0.035) 0.179 (0.034) 0.187 (0.034) 0.189 (0.033) 0.194 (0.027)
Proportion of Adults above 60 0.262 (0.734) 0.256 (0.065) 0.233 (0.0633) 0.220 (0.064) 0.200 (0.051)
Touristic Area 0.098 (0.297) 0.153 (0.361) 0.162 (0.369) 0.160 (0.367) 0.096 (0.294)
Urban Area 0.133 (0.34) 0.092 (0.29) 0.117 (0.321) 0.139 (0.346) 0.249 (0.433)
B. Water Consumption and Expenditure
Mean Annual Consumption 136.145 (57.52) 132.738 (49.972) 135.97 (98.144) 139.527 (70.748) 139.541 (57.964)
Mean Annual Expenditure 192.651 (80.63) 187.527 (70.231) 191.193 (96.62) 188.668 (81.585) 196.66 (80.444)
Expenditure as a Fraction of Income 0.009 (0.004) 0.007 (0.003) 0.007 (0.004) 0.006 (0.002) 0.005 (0.002)
Marginal Price 1.074 (0.32) 1.065 (0.324) 1.087 (0.351) 1.058 (0.324) 1.154 (0.337)
Fixed-Price 48.93 (27.175) 49.456 (27.198) 48.79 (27.176) 44.119 (24.758) 38.611 (23.905)
C. Water Utilities Characteristics
Proportion of Privately Managed 0.658 (0.475) 0.612 (0.488) 0.63 (0.483) 0.634 (0.482) 0.69 (0.463)
Underground Water 0.739 (0.439) 0.688 (0.464) 0.652 (0.477) 0.653 (0.476) 0.574 (0.494)
Ground Water 0.10 (0.30) 0.144 (0.352) 0.136 (0.343) 0.152 (0.359) 0.15 (0.357)
Treatment Complexitya2.919 (1.299) 2.935 (2.990) 2.990 (1.285) 3.006 (1.238) 3.184 (1.281)
Net Revenues per Customer (Subsample) b23.959 (27.079) 29.618 (33.055) 27.058 (27.797) 26.416 (26.291) 25.736 (29.694)
Note: These municipal-level data come from IFEN-SOeS, INSEE and OSEA datasets. The sample includes 4,500 municipalities, 900 per quintile,
except for the subsample on marginal cost, that includes data for 650 cities. These municipalities represent 16.7 million households. The first per-unit
of consumption quintile of municipalities represent 3,319,712 households; the second 3,105,233; the third 3,315,489; the fourth 2,941,573 and the top
quintile 3,970,965. Means and standard deviations (in parentheses) are calculated using 2008 euros.
aFor ease in reading, there are six possible treatments numbered from 1 to 6, treatments 1 to 3 are easy, 4 and 5 are complex and 6 is intermediary
complex. Water treatment performed by the operator before the water is distributed are important cost-shifters. Indeed, water treatment does not
only approximate the complexity of service provision but also the level of specific investments needed to operate the service. A telltale story is that
underground water is generally more stable over time which has two advantages. First, it reduces uncertainty about the evolution of costs. Second,
treatment costs are usually lower when water is pumped from the underground. Under mixed sources of water, costs might be higher than under ground
or underground sources as the utility might need a treatment factory for each type of water. Treatments are sixfold and coded between 1 and 6 in the
IFEN-SOeS dataset. In the simplest case, there is no treatment. In this case, the treatment variable takes value 1. When raw water needs disinfection,
treatment takes value 2. The value is equal to 3 if raw water needs a heavy disinfection treatment and equals 4 if water needs a heavy disinfection
treatment plus extra-controls. The variable takes 5 and 6 when mixed treatments are needed, the most difficult treatment being 5.
bFulfilled for the subsample including costs and revenues.
Table 3: A test of marginal cost pricing in the French public water services
Variables NRA
q×Ground Water 0.0247
q×Mixed-Water 0.0673***
q×Treat2 -0.0874***
q×Treat3 -0.152***
q×Treat4 -0.183***
q×Treat5 -0.121***
q×Treat6 0.0756
Constant -13.29*
N 650
Marginal effect of q0.1240***
Results from the Selection Equation
Variables V
Semi-Urban 0.759***
Urban 1.654***
Private Management 0.597***
Constant -2.278***
N 5,215
Pseudo R20.1991
Robust Standard Errors in Parentheses
*** p<0.01, ** p<0.05, * p<0.1
Table 4: Rate Schemes Implemented in Different Types of Water Utilities
Current Rate Scheme Reformed Rate Scheme
(1) (2)
Marginal Price Fixed-part Marginal Cost Capital Cost
Public Management 0.968*** (0.00725) 38.04*** (0.543) 0.814*** (0.00771) 33.31*** (0.532)
Private Management 1.151*** (0.00632) 50.67*** (0.529) 1.032*** (0.00647) 50.37*** (0.490)
Rural 1.097*** (0.00734) 57.81*** (0.604) 0.950*** (0.00772) 52.14*** (0.60)
Semi-Urban 1.054*** (0.00793) 37.04*** (0.899) 0.920*** (0.00822) 37.33*** (0.517)
Urban 1.151*** (0.0123) 34.89*** (0.516) 1.065*** (0.0133) 38.94*** (0.922)
Note: This table reports how customers expenditure on water would change under Coasian tariffs. Bootstrap
standard errors based on 1000 replications in parentheses with *** p<0.01, ** p<0.05, * p<0.1.
Table 5: Impact on Bills of Rebalanced Tariffs
Mean Change in Euros Mean Change in Percent % Experiencing Bill Increase
A. By Income Quintile
1st Quintile -22.32*** (0.709) -12.43*** (0.336) 1.332*** (0.377)
2nd Quintile -21.36*** (0.541) -11.97*** (0.289) 1.332*** (0.389)
3rd Quintile -20.89*** (0.697) -11.58*** (0.320) 2.558*** (0.534)
4th Quintile -19.33*** (0.465) -11.28*** (0.284) 3.444*** (0.586)
5th Quintile -16.95*** (0.482) -9.563*** (0.271) 4.672*** (0.706)
B. By Adjusted Income Quintile
1st Quintile -21.45*** (0.598) -11.95*** (0.319) 1.776*** (0.440)
2nd Quintile -21.85*** (0.545) -12.56*** (0.313) 1.333*** (0.388)
3rd Quintile -21.57*** (0.747) -11.93*** (0.305) 1.556*** (0.423)
4th Quintile -20.07*** (0.514) -11.45*** (0.275) 2.667*** (0.528)
5th Quintile -15.90*** (0.498) -8.933*** (0.265) 6.007*** (0.820)
C. By Consumption Quintile
1st Quintile -12.99*** (0.305) -9.401*** (0.260) 4.329*** (0.686)
2nd Quintile -16.39*** (0.315) -10.73*** (0.242) 2.000*** (0.466)
3rd Quintile -18.29*** (0.396) -11.50*** (0.287) 1.444*** (0.416)
4th Quintile -20.88*** (0.478) -11.94*** (0.315) 2.667*** (0.528)
5th Quintile -32.32*** (0.953) -13.25*** (0.370) 2.892*** (0.536)
D. By Poverty Status
Water-Poor -54.26*** (5.272) -10.96*** (0.880) 0
Poor Cities -22.19*** (1.434) -11.68*** (0.765) 2.030*** (0.988)
Note: This table reports how customers expenditure on water would change under Coasian tariffs. Bootstrap
standard errors based on 1000 replications are shown in parentheses with *** p<0.01, ** p<0.05, * p<0.1.
In panel (D), water-poor are defined as cities in which the average water bill represents more than 1.5% of
the household median income. Poor Cities are defined as cities in which the annual median per-unit income
is lower than 12,450 euros a year. Poor and water-poor cities represent respectively 576,399 and 126,466
Table 6: The Impact of Four Reforms on Lower Incomes
Mean Change in Euros Mean Change in Percent % Experiencing Bill Increase
I. Assistance Programs Under Coasian Tariffs
A. Free Fixed-fees for Poor Cities
20% lower per-unit incomes -9.372*** (0.824) -5.299*** (0.455) 79.02*** (1.368)
Water-poor -14.79*** (3.394) -4.221*** (0.969) 79.05*** (3.990)
Poor Cities -50.51*** (1.813) -29.14*** (0.908) 0 -
Annual cost per non-recipient 1.442*** (0.000) 0.999*** (0.007) 100 -
Overall Cost (in millions euros) 23.2*** (0.000) - - - -
B. No Increase in Fixed Fees
20% lower per-unit incomes -0.329*** (0.072) -0.168*** (0.053) 79.02*** (1.368)
Water-poor -1.252*** (0.349) -0.245*** (0.091) 79.05*** (3.990)
Poor Cities -0.245*** (0.151) -0.082*** (0.102) 0 -
Annual cost per non-recipient 1.204*** (0.000) 0.870*** (0.008) 100 -
Overall Cost (in millions euros) 19.4*** (0.000) - - - -
II. Assistance Programs Under Marginal Cost Pricing and Current Profits
C. Free Fixed Fees for Poor Cities
20% lower per-unit incomes -21.12*** (1.876) -8.318*** (0.672) 79.02*** (1.368)
Water-poor -35.56*** (7.837) -7.653*** (1.658) 79.05*** (3.990)
Poor Cities -107.73*** (4.954) -43.63*** (1.288) 0 -
Annual cost per non-recipient 1.901*** (0.000) 1.065*** (0.006) 100 -
Overall Cost (in millions euros) 30.60*** (0.000) - - - -
D. Free Fixed Fees for Cities with Median Income <159% of the Poverty Line
20% lower per-unit incomes -88.15*** (1.855) -40.98*** (0.565) 0 -
Water-poor -81.73*** (8.874) -19.23*** (2.011) 43.81*** (4.657)
Poor Cities -102.99*** (5.242) -41.69*** (1.448) 0 -
Annual cost per non-recipient 11.86*** (0.000) 6.642*** (0.040) 100 -
Overall Cost (in millions euros) 166*** (0.000) - - - -
Note: This table reports how customers expenditure on water would change under Coasian tariffs in part (I) and under
marginal cost pricing with increased fixed fees such as firms’ profit is unchanged in part (II). Bootstrap standard errors
based on 1000 replications are shown in parentheses with *** p<0.01, ** p<0.05, * p<0.1. Costs per non-recipients
are computed using weights for the number of households. The 20% cities with the lower incomes represent 3,319,712
households; Poor and Water-poor cities represent respectively 576,399 and 126,466 households.
Table 7: Price-Elasticity of Demand
Variables Ln(Consumption)
Ln(MP) ×1st Quintile -0.281***
Ln(MP) ×2nd Quintile -0.304***
Ln(MP) ×3rd Quintile -0.287***
Ln(MP) ×4th Quintile -0.269***
Ln(MP) ×5th Quintile -0.223***
Semi-Urban -0.163***
Urban -0.120***
Household Size 0.217***
Touristic Area 0.138***
Share of Population 15-64 YO -0.805***
Region FE Yes
Constant 5.120***
Observations 4,500
R-squared 0.197
Note: Robust Standard Errors in Parentheses.*** p<0.01, ** p<0.05, * p<0.1.
Demand Elasticity is computed for current marginal prices.
Table 8: Welfare Change and Deadweight Loss Estimates for 2008
Mean Annual Welfare Change in euros
1st Quintile 22.31*** (0.561)
2nd Quintile 22.98*** (0.566)
3rd Quintile 22.63*** (0.777)
4th Quintile 20.92*** (0.526)
5th Quintile 17.06*** (0.473)
Water-Poor 52.94*** (5.545)
Poor Cities 24.53*** (1.473)
Consumers’ Welfare Change (in millions) 201 (0.000)
Deadweight Loss (in millions) 5.358 (0.000)
Note: This table reports how customers welfare change for per-unit
of consumption income quintiles. Bootstrap standard errors based
on 1000 replications are shown in parentheses with *** p<0.01,
** p<0.05, * p<0.1. Consumers’ welfare change under Coasian
tariffs includes the deadweight loss and subtracts increased fixed
fees. The deadweight loss is the net efficiency gains from marginal
cost pricing.
Table 9: Price-Elasticity of Demand when Consumers Respond to Average Price
Variables Ln(Consumption)
Ln(AP) ×1st Quintile -0.606***
Ln(AP) ×2nd Quintile -0.630***
Ln(AP) ×3rd Quintile -0.624***
Ln(AP) ×4th Quintile -0.608***
Ln(AP) ×5th Quintile -0.581***
Semi-Urban -0.109***
Urban -0.111***
Household Size 0.206***
Touristic Area 0.121***
Share of Population 15-64 YO -0.816***
Region FE Yes
Constant 5.435***
Observations 4,500
R-squared 0.274
Note: Robust Standard Errors in Parentheses.*** p<0.01, ** p<0.05, * p<0.1.
Price-Elasticity is computed for current average prices.
... Variations in the formulation and the regulatory process involved in the implementation of WRSs motivate the analyses of correlations between changes in water pricing and issues of public policy. It is challenging for water utilities to adopt a rate structure that meets all relevant objectives: economic efficiency, financial self-sufficiency, water conservation, and equity and affordability, (Mullin, 2008;Porcher, 2014). Prioritization of a particular objective may raise concerns on other issues of public interest. ...
... Similarly, Diakité et al. (2009) designed and calibrated an optimal tariff for Côte d'Ivoire, whereby an initial "social" block would be offered at a lower price than cost, to find that the gains accruing to consumers from increasing the number of price blocks and moving to that optimal tariff would outweigh the producer surplus lost. More recently, Porcher (2014) found that, although fixed fees and volumetric charges differ from their theoretical ideal, departures from marginal-cost pricing in the volumetric rates are not substantial in France's water sector, where the use two-part tariffs is compulsory, when compared to other regulated industries. However, they have strong distributional effects and current tariffs result in a sizable transfer of welfare from consumers to producers. ...
... 14 This can have consequences for the actual effects of a conservation policy. For example, Porcher (2014), estimates the deviation between marginal prices and costs in France and then separately considers the welfare effects of a water tariff reform assuming a response to average price versus a response to marginal price, obtaining a different result in each case. ...
We analyze the relationship between local municipal characteristics and their choice of Water Rate Structure (WRS), applying a Multinomial Logit Model (MLM) to the most recently available dataset of Canadian municipal data. The results reveal that, to some extent, local characteristics explain the municipalities' choice of WRS consistent with expectations. However, beyond the effect of determinants that describe the utilities’ operational environment, there remains a degree of variability in tariff setting that likely stems from the balancing of efficiency objectives with considerations of fairness and political acceptability.
... For example, a number of studies derived Ramsey prices or Feldstein prices (Monteiro and Roseta-Palma, 2011;Diakit� e et al., 2009;Garcia-Valinas, 2005;and Renzetti, 1992). Others attempted to derive tariffs that implement marginal cost pricing (e.g., Porcher, 2014;Hall, 2009;Grafton and Kompas, 2007;Briand, 2006;Garcia and Reynaud, 2004;Krause et al., 2003;and Cueva and Lauria, 2001). In addition to using economic efficiency to motivate the design of alternative tariffs, nearly half of the studies identified in the systematic review calculated the customer welfare effects of moving from the current tariff to a new tariff. ...
... Half of the studies identified in the systematic literature review addressed the issue of cost recovery, and the majority of these studies addressed cost recovery by imposing it as a constraint in the simulation or optimization model (e.g. Wichelns, 2013;Monteiro and Roseta-Palma, 2011;Loehman, 2008;Hall, 2009;Diakit� e et al., 2009;and Garcia-Valinas, 2005;Porcher, 2014;Reynaud, 2016;Wolak, 2016;Nauges and Whittington, 2017). However, Rosenberg (2010) and Cueva and Lauria (2001) explicitly compared the level of cost recovery Almost half of the studies included in our review addressed notions of social equity 5 or examined distributional issues in some way. ...
... These studies provide insight into consumer behavior in particular contexts, add to the body of literature on consumer response to price changes, and provide concrete examples of how prices can be set to improve economic efficiency. Overall, several studies found that although it is possible to derive welfare-improving tariffs, the welfare gains from doing so were relatively modest (Groom et al., 2008;Garcia and Reynaud, 2004;Diakit� e et al., 2009;Porcher, 2014;and Nauges and Whittington, 2017). ...
Across the globe, many low- and middle-income countries are investing in their first generation of piped water and sanitation infrastructure. At the same time, the water and sanitation infrastructure in many industrialized countries is reaching, or has reached, the end of its useful life. Governments will need to mobilize substantial resources to finance this global water and sanitation infrastructure transition and user charges (tariffs) will play an integral role in supporting these efforts. This paper presents the results of a systematic review of the empirical literature on the design and evaluation of tariffs for municipal water and sanitation services, highlighting ways in which insights from the literature might inform policy and identifying areas for future research. Overall, we find that the empirical literature on pricing municipal water and sanitation services is diverse. Studies identified through our systematic review are published in a wide range of journals and vary considerably with respect to their primary aims, methods, number of tariffs analyzed, and the metrics used to evaluate different tariffs. The majority of studies examine two or fewer metrics of tariff performance, limiting the extent to which the literature characterizes the tradeoffs policy makers face when setting tariffs for municipal water and sanitation services. Finally, the majority of studies in the literature focus on water pricing in industrialized countries, highlighting an opportunity for research on water pricing in low- and middle-income countries.
... D'abord, la tarification dans les services publics est généralement réalisée par un tarif binôme où les coûts fixes sont payés avec la part fixe, et les coûts variables avec la part variable (ce qui est le cas dans l'expérience de Dunkerque). Cette hypothèse a été testée par Porcher [2014] qui montre que, dans le secteur de l'eau en France, le coût marginal diffère du prix marginal alors que le prix de la part fixe est quasiment égal au coût fixe par abonné. Dès lors, la tarification progressive ne s'appliquant qu'à la partie variable du tarif, nous n'intégrons pas la problématique des coûts fixes dans ce modèle. ...
... -Au niveau agrégé, les consommateurs ne réagissent pas aux variations du prix moyen (non significatif) mais réagissent au prix marginal, avec une élasticité au prix marginal de -0,2. Cette élasticité se rapproche d'autres calculs d'élasticités obtenus dans l'eau potable qui la situaient entre -0,08 et -0,22 (Nauges et Reynaud [2001] ; Porcher [2014]). D'autres auteurs avaient relevé des élasticités comprises entre -0,2 et -0,66 pour l'eau potable en France (Nauges et Thomas [2003] ; Garcia et Reynaud [2004] ; Porcher [2014]) et dans les pays en développement (Nauges et Whittington [2010]). ...
... Cette élasticité se rapproche d'autres calculs d'élasticités obtenus dans l'eau potable qui la situaient entre -0,08 et -0,22 (Nauges et Reynaud [2001] ; Porcher [2014]). D'autres auteurs avaient relevé des élasticités comprises entre -0,2 et -0,66 pour l'eau potable en France (Nauges et Thomas [2003] ; Garcia et Reynaud [2004] ; Porcher [2014]) et dans les pays en développement (Nauges et Whittington [2010]). ...
L’approche des tarifs de l’eau potable en France a changé depuis 2010, puisque les collectivités peuvent recourir à des tarifs progressifs croissants par blocs. Ces tarifs, bien qu’initialement conçus comme des solutions de second rang aux pertes sèches du monopole, se révèlent complexes à mettre en œuvre pour satisfaire d’autres objectifs. La réaction sous-optimale des consommateurs au signal-prix et les problématiques de redistribution questionnent l’efficacité du mécanisme. Cette contribution analyse théoriquement les propriétés d’un tarif progressif, puis évalue empiriquement la réaction des consommateurs au signal-prix à partir d’une expérience naturelle menée à Dunkerque. Les résultats indiquent une bonne réaction au prix marginal des consommateurs situés dans les tranches extrêmes, tout en questionnant l’équité d’un mécanisme fortement distorsif.
... However, many issues are raised in terms of equity. For example, in industrialized countries, consumption is more related to household size than to its income level (for a debate about France, see [7] while in developing countries, water demand tends to be more closely related to the household financial constraints [26]. Demand-driven solutions are increasingly viewed as a necessary complement, if not a substitute, for supply-oriented policy measures. ...
Full-text available
This paper proposes an original assessment of sustainable water demand management policy in Tunisia over the last five decades. Using quarterly data from 1988 to 2015, describing six heterogeneous regions, we employ the novel and robust Quantile-on-Quantile (QQ) approach. We show that in Tunisia, the incentive water tariff used to ensure the sustainability of water management was unfair. Low-income water consumers are negatively affected by the nonlinear tariff more than high- income consumers. In addition, the QQ results show spatial variability and regional disparity characterizing the relationship between water price and consumption for different quantile levels. The paper argues for a decentralized water pricing system to achieve the goal of social equity and an alternative water management policy.
... First, it extends the revelation of cost and/or effort by a regulation contract (See to the theoretical and empirical analyses about water leakages considered as a key indicator of the service quality in the water sector. These analyses are provided by ; , , Elnaboulsi (2009), , Porcher ( , 2015 and . Regulators and operators are constrained by the volume of water distribution, in which the level of water leakages affects water prices, and hence the performance of managing water services. ...
In this thesis, we are looking for solutions enabling a local authority to improve transparency and governance in the delegated management of its services, e.g., water and sanitation services. Based on the theoretical and experimental literature on the social value of information, we analyze the role that public information can play in defining delegation agreements. We point out that a local authority can use the available information to induce operators to reveal privately-held information. Furthermore, we show that public information can help local authorities to coordinate operators' strategic decisions to improve infrastructure quality. Then, we analyze the decisive role of information precision in the definition of contractual schemes. By comparing the precision of public information with that of private information at the negotiation stage, we determine the threshold of public information enabling a local authority to choose the most appropriate management contract. Furthermore, as competition between potential operators wishing to win the delegation contract generates misleading beliefs about the quality of the infrastructure, we show that any increase in information precision limits these beliefs and reduces the informational rents a local authority may concede to an operator. Finally, in the selection phase, we combine information disclosure with the well-known Yardstick Competition regulatory process. We show that, for a given level of precision of privately-held information, disclosure enhances the efficiency of this regulatory scheme.
... Deadweight loss refers to the loss of economic efficiency when equilibrium for a good is not achieved. It leads to loss of welfare mainly due to tariffs and pricing (Coughlin, 2010;Dixon & Rimmer, 2010;Irwin, 2010;Porcher, 2014). In the case of tariffs, deadweight loss is the burden created due to the loss of benefits to the stakeholders. ...
Full-text available
With the onset of the US-China trade war in July 2018, the trade patterns between China, the US, and India have undergone a tremendous change. The number of products in which China had a competitive advantage in terms of exports to the US has declined in the last 9 months. A number of developing countries may be benefitted from the ongoing tariff war between the US and China, like Vietnam, Brazil, India, and Korea. In the present study, an attempt has been made to analyse the impact of the US-China trade war on exports of India to the US. The sector which has been selected is the chemical sector comprising of organic and inorganic chemicals as chemicals are one of the top-exported products from India to the US. To analyse the impact, the difference-in-differences technique of regression has been applied. The results indicate that after July 2018, i.e., the commencement of the US-China trade war, the impact on firms exporting chemicals from India to the US has been significant and firms in India may be a potential source for chemicals for the US provided the right policy measures are exercised in India. The results indicate that the trade war between the US and China has had a positive impact on the chemical exports from India to the US. The chemical exports from India to the US have increased post-July 2018, though not at a steep rate. This indicates that India has the potential to export chemicals to the US
... Past utility regulatory developments and research have been dominated by issues of economic efficiency and affordability. In this regard, much has been written about the tradeoffs associated with rate-of-return (ROR) and price-cap-regulation (PCR) (Aubert & Reynaud 2005); benefits of private versus public utility ownership (Abbott & Cohen 2009); the role of markets (Crase et al. 2000); optimal tariff design (Porcher 2014); and role of incentives-based regulation to promote competition-by-comparison (De Witte & Marques 2010). ...
The growing interest in customer engagement (CE) has triggered a new wave of reforms, particularly in utility regulation. Within the water sector, there has been a shift from a focus on cost‐reflective pricing toward customer‐centric pricing processes designed to identify customer preferences and expectations. The Victorian water sector in Australia offers a unique opportunity to explore the outcomes of these CE trends given the recent first‐time application of a novel Performance, Risk, Engagement, Management, and Outcomes (PREMO) framework. Based on in‐depth interviews with senior industry representatives from a diverse sample of Victorian water utilities, this paper critically analyses the scope, design, and incentive mechanisms that underpin the new regulatory process. Findings indicate that CE has potentially beneficial aspects to both the regulator and the regulated utility, including more transparency in capital projects and a better understanding of customer preferences in service delivery.
Rapidly‐growing concern among scholars and policy makers over residential drinking water affordability in the United States highlights the need to identify and assess the efficacy of potential solutions to address this problem. Accordingly, in this advanced review of the literature, we examine the state of scholarly evidence over the last 30 years on the prevalence and effectiveness of strategies to address household drinking water affordability in the United States. We classify interventions into four categories: rate structure designs, water efficiency programs, recurring bill assistance, and crisis relief. Our findings are twofold. First, the conceptual literature on affordability interventions is fairly robust, but demonstrates both tradeoffs and complementarities across the four approaches. Second, despite employing a PRISMA approach, we identify few empirical studies that demonstrate the effectiveness of affordability interventions in practice, especially the targeted approaches of recurring bill assistance and crisis relief. The literature on affordability interventions thus appears to lag considerably behind scholarship identifying and defining the problem of affordability. Accordingly, we suggest key questions throughout our review that need to be answered, thereby providing an agenda for future research on drinking water affordability solutions. This article is categorized under: • Human Water Abstract Intervention pathways improving household water affordability.
In this paper, I show that not only do distortionary electricity tariffs create welfare loss, they create a platform for growing welfare losses with expected technological change. I estimate the welfare loss attributable to existing British electricity tariffs, finding that they are equivalent to between six and 18% of domestic consumption value. Losses are greater than unpriced distributional and environmental countereffects and therefore common arguments against reform are invalid. Expected technological change will increase this welfare loss. Deployment of distributed energy resources (e.g. solar) benefits adopters at the expense of non-adopters as tariffs are recalibrated to recover fixed costs. Reform on Coasian principles avoids these welfare losses and redistributional effects. The structure of electricity tariffs will therefore determine whether technological change is beneficial to consumers. In providing these estimates, I provide both analytical and numerical insight. I combine household-level micro-data with information on utility cost and tariff structure. I propose a simulation methodology to elicit the welfare effects of technological change.
Financially self-sustainable management of urban water infrastructure assets in IBTs structure is a significant challenge faced by policy makers. In this paper, a novel causal loop diagram is developed to model dynamics of feedback loops and interactions between physical, financial, and consumers’ sectors in order to assess financially sustainable management strategies of water distribution networks in IBTs structure. Developing a system dynamics (SD) model for policy analysis of urban water management according to IBTs structure, is the main contribution of this research. In the presented model, results of different financial and rehabilitation strategies are compared by means of financial, environmental, social, and service performance indicators. The proposed model is implemented for Isfahan province in Iran, demonstrating that prices set to finance full cost of the service through higher blocks of consumption decrease annual fee hike rate of lower blocks of consumption. However, penalizing higher blocks has no considerable impact on the amount of total water volume purchased. Households outflow to the lower blocks slows down the growth rate of utility’s revenue that consequently results in deterioration of network pipes and leakage. Larger price differentials between consumption blocks causes reduction in the utility’s revenue and underlines the need for debt-based financing.
Full-text available
French municipalities often contract out the provision of local public services to private companies, and regularly choose the same private operator for a range of different services. We develop a model of relational contracts that shows how this strategy may lead to better performance at lower cost for public authorities. We test the implication of our model using an original database of the contractual choices made by 5000 French local public authorities in the years 2001, 2004 and 2008.
Describes pricing practices commonly used by urban water utilities and defines the efficient water price in terms of long-run marginal extraction costs per unit plus a rent element reflecting the value of water in situ, ie, in its original aquifer or surface source. Section II develops a model, following Hanson (1980), to measure this rent. Section IV presents estimates for the case of the Honolulu Board of Water Supply. The Board, which serves a population of about 800 000 on the island of Oahu, anticipates the need for desalting brackish groundwater within 20 years. While still a rough approximation, estimates of scarcity value suggest that at present raw water should be valued on the order of twice the current unit quantity charge. Thus, adding the value of raw water to existing costs suggests that the current charge should be roughly tripled. Section V examines alternative extraction cost functions to generalize the possible implications of this model, while section VI offers some final comments. -from Authors
Cet article constitue la mise au point et le développement d'une communication antérieure sur le même sujet [4]. Il concerne le problème posé par l'incompatibilité de la règle de vente au coût marginal préconisée par les Welfare Economics, et les conditions d'équilibre budgétaire imposées par les Pouvoirs Publics aux entreprises nationalisées. La méthode consiste en une ``maximation de Pareto'' appliquée à un modèle général dont les liaisons de structure comportent à la fois des liaisons entre les quantités et des liaisons entre les valeurs. Les résultats obtenus, rassemblés dans la section 8, sont commentés et comparés aux solutions antérieurement proposées à ce problème. Un résumé de cet article est constitué par les sections 1, 3, et 8, qui peuvent être lues sans prendre connaissance des calculs. Le modèle développé dans la section 2 a pour objet de faciliter l'interprétation donnée dans la section 3, et d'expliquer plus clairement l'intervention des prix et revenus dans les liaisons du système; mais la maximation exposée dans la section 5 est effectuée sur le modèle transformé qui, à la fin de la section 4, rassemble sous une forme condensée les liaisons du système. La section 9, enfin, à l'occasion de la critique faite de la politique ``traditionnelle'' des monopoles publics, commente certains points particuliers des résultats exposés dans la section 8.
I. Introduction, 175. — II. The optimal price, 176. — III. The welfare loss due to incorrect pricing, 182. — IV. An example, 183. — V. Concluding remarks, 187.