Horizontal Integration and Relational Contracting: An
Application to Local Public Services
Claudine Desrieux
Eshien Chong
St´
ephane Saussier∗
Abstract
Legal frameworks, especially in Europe, encourage private participation and
competition in the management of public services. However, many local public
authorities concentrate the various services they have in charge in the hands
of a single operator, leading to horizontal integration which a priori minimizes
the positive effects of competition. The following article tries to understand
why vertical disintegration is regularly combined with horizontal integration.
Results of our model show that under some conditions, this may lead to better
performance at lower cost for the public authority. Such a proposition is tested
and corroborated using an original database concerning the contractual choices
made by 5000 French local public authorities in 1998 and 2001.
Codes JEL: D23, H11, K12, L14
1
Introduction
In the past few decades, the European Union has been promoting private partici-
pation and competition in public services, considered as a way to increase efficiency
in the management of public services.1 More precisely, in a first communication
∗All from ADIS, Facult´e Jean Monnet, 54 Boulevard Desgranges, 92331 Sceaux Cedex, France,
E-mail of the corresponding author: claudine.desrieux@free.fr. We are grateful to Bruno Deffains,
Dominique Demougin, Eduardo Engel, Elisabetta Iossa, Steven Tadelis, Robert Gibbons for their
helpful comments on this work, as well as participants of the International Conference on “Public-
Private Partnerships, Competition and Institutions”, Paris, 7-8 December 2007, and of the American
Economic Association conference, New Orleans, 4-6 January 2008. The Conseil R´
egional d’Ile de
France is also gratefully acknowledged for financial support.
1Let us note however that both notions are different: private participation does not necessary
mean competition, as a private firm can be a monopole, and competition does not exclude the
participation of public firms, that can be in competition with private firms on some markets. In the
European union, both private participation and competition are promoted.
1

---

in 19962, the Commission explained the interplay for the citizens’ benefit between
Community measures in the areas of competition and free circulation and public ser-
vice tasks. This communication was updated in 20003 in order to increase the legal
certainty for operators as regards the application of competition and internal market
rules to their activities. In 2001, these two communications were complemented by
a Report to the Laeken European Council.4 This report responds to concerns with
regard to the economic viability of operators entrusted with public service tasks. It
highlights the guarantees offered by Article 86 (2) of the Treaty,5 Community action
and the responsibility of the Member States, in particular as regards the definition
of public service obligations (European Commission [2003]).
However, statistics about the management of public services seem rather discon-
nected with such a trend, aiming to promote competition. Many local public au-
thorities concentrate the various services they have in charge in the hands of a single
operator, which a priori minimizes the positive effects of competition. Therefore, it
seems that public authorities have been rather convinced to “vertically disintegrate”
services (at least in France), but surprisingly enough, have in parallel choose to “hor-
izontally integrate” them.
When observing management practices more precisely, most private operators are
global groups capable of providing many local public services. As a consequence,
the market for public services is rather oligopolistic, especially for “environmental
services” such as water, sanitation, waste or energy managements, as illustrated by
table 1.
2“Services of general interest in Europe”, OJ C 281, 26.9.1996, p.3
3“Services of general interest in Europe”, OJ C 17, 19.1.2001, p.4
4COM(2001) 598 final, 17.10.2001
5Article 86 (2) provides: “Undertakings entrusted with the operation of services of general eco-
nomic interest ... shall be subject to the rules contained in this Treaty, in particular to the rules on
competition, insofar as the application of such rules does not obstruct the performance, in law or
in fact, of the particular tasks assigned to them. The development of trade must not be affected to
such an extent as would be contrary to the interests of the Community”.
2

---

Table 1: Market shares in % of French urban population, year 2004.
Water
Garbage collection
Urban
management
and treatment
warming
Veolia
40%
37 %
38 %
Suez
20 %
21 %
47 %
SAUR
10 %
9 %
Independent operators
1 %
6 %
8%
In house provision
29 %
27 %
7 %
Source: Direction des affaires ´
economiques internationales, Minist`
ere de l’´
equipement
Therefore, public authorities often rely on the same operator to provide different
services they choose to contract out. This seems surprising when one thinks of the
egalitarian and transparency principles of the European Union for attribution of
markets.6
How then to explain the gap between the will to promote competition and the obser-
vation of a rather concentrated market for public services? Is competition effective
or does it reduce to a goal mentioned in formal official speeches ?
Up to now, few works have been done on this theme. As previously shown, many
works deal with public-private partnerships (PPPs) but focus on their design (Ben-
nett and Iossa [2006]), on the trade-off with public provision (Hart, Shleifer, and
Vishny [1997], Hart [2003]), or on factors causing their renegotiation (Guasch, Laf-
font, and Straub [2003], Guasch [2004]). To our knowledge, no work specifically
deals with horizontal integration in PPPs, and the reasons for the concentration of
services, with the exception of Gence-Creux [2001]. The latter documents a tendency
for local public authorities in France to rely on the same operator for providing sev-
eral different public services such as water, cable television, garbage collection etc.
He shows that a mayor who has electoral concerns may be led to favor a unique
manager even though this choice proves to be inefficient. However, no explanation
6Such concentration is not specific to France. In a guide for Nova Scotia Municipalities that might
be interested by PPPs (p.9), a warning is written about limited competition: “Where municipalities
are seeking to increase private partner participation in services that have been provided by the
public partner, there may be a limited number of firms with the experience or expertise to compete
for the contract. In such cases, a public monopoly may simply be replaced with a private monopoly
that nullifies many of the advantages of a partnership.” See http://gov.ns.ca/snsmr/muns/fin/pdf-
ppp/ppp 1.pdf
3

---

has been proposed for such a market concentration in case of benevolent government.
Our paper tries to propose such an explanation by relying on “relational contracts”,
as defined by Baker, Gibbons, and Murphy [2002] or Baker, Gibbons, and Murphy
[2004]. Contractual incompleteness is here taken for granted: Indeed, the quality
of services the public authorities want is often difficult or prohibitively costly to
specify in details ex ante, at least in a way to be enforced by courts (Hart, Shleifer,
and Vishny [1997]). As a consequence, renegotiations occur ex post. Yet, parties
may also tacitly agree on the way uncontractible parameters can be managed. As
informal dealings cannot be enforced by courts, their self-enforcement comes from
the perspective of future business between partners, and the need for a good repu-
tation. As in many other businesses (Baker, Gibbons, and Murphy [2002]), informal
relationships between public and private contractors may help to circumvent diffi-
culties in formal contracting. Case studies undertaken by the World Bank [2006]
concerning Manilla and Gabon illustrate this point.7 In the same way, the Euro-
pean Commission [2004b] has also shown that informal relationships between public
and private contractors may be helpful.8 A last example of informal agreements in
contracts between public and private partners is given by the French law. During
the selection procedures, public authorities are allowed to introduce some negotia-
tions with potential candidates in parallel to the objective selection criteria of the
competitive tendering process (Auby [1997]). This may lead to some informal talks
between parties.
7In Manilla and Gabon, the World Bank reports some informal commitments over additional
investments by the concessionaires over the contract’s lifetime. Some more general considerations
are also given about the role of informal dealings at an early stage of cooperation between public
and private partners: “talking with potential bidders at an early stage about the structure and scope
of a proposed project is a good idea. This type of informal market sounding, typically based on
an initial project briefing, a consultation paper, or a prebid road show, is often undertaken before
commencing the formal procurement process. Potential bidders generally welcome the opportunity
to participate in informal market soundings. Early recognition of bidders’ commercial concerns can
greatly enhance bidder interest and increase the overall effectiveness of the formal procurement”
(p.180).
8For instance, Case 17 of the Resource Book on PPP Case Studies accounts for the German
experience (M¨
ulheimer Entsorgungsgesellschaft mbH ), and states that “to handle the complex mul-
tidimensional objectives and to protect their interests the parties had to agree on several informal
and formalized agreements” (p.84).
4

---

In order to take into account such a relational dimension between public and pri-
vate actors, we develop a model in which a public authority decides to contract out
the management of two services, whose uncontractible investments have different
impacts on social benefit. The public authority can decide either to “horizontally”
integrate the services by delegating them to one single private operator, or she can
choose two different managers. The key question in our paper is whether such a
choice has consequences on promises about how to deal with non-contractible out-
comes. We show that in a static framework, these informal dealings prove to be
irrelevant, and whether transactions are horizontally integrated or not has no im-
pact. Private provision leads to optimal incentives for the service with low adverse
effect, but over-optimal investments for the service with high adverse effect. We thus
confirm results already obtained by Hart, Shleifer, and Vishny [1997], considering
two transactions instead of one as they did.
Nevertheless, when parties have concerns for future business, relational contracts
can encourage useful actions. We explore this and we show that a private partner
may accept to invest at a level that is socially optimal, if he is rewarded for such
a behavior, by a bonus or a promise to be chosen again in subsequent periods. His
deviation can be punished in the long run. We found that with two different services,
one with and the other without adverse effects of cost reduction on social benefits,
informal agreements are more easily sustainable when the private manager has both
contracts in charge. The bonus the public authority has to pay to achieve the so-
cial optimum is then lower, which means that the total price paid to manage both
services is lower in case of horizontal integration than in case of horizontal disinte-
gration. Our main result is thus to show that, under some conditions, horizontal
integration may force the private manager to respect the informal dealing at lower
costs. In such a perspective, horizontal integration appears as an instrument in the
service of the parties’ relationship.
We then test such a proposition on an original database combining data from the
French Environment Institute (IFEN) and the French Health Ministry (DGS), con-
cerning 5000 local public authorities and their contractual choices in force in 1998
5

---

and 2001. Our results show that the choice of the same operator in order to operate
both distribution and sanitation of water is not random and is not neutral. Our
main proposition is rejected by the data.
We believe our paper is a contribution to the relational contracting literature by
highlighting the fact that it is necessary to study several transactions, and not one
in isolation, in order to understand contractual choices and performances. Our model
provides a rational basis for this, focusing on the fact that problematic transactions
(i.e. with non contractible investments) might be adequately bundled with more
simple ones in order to ease their enforcement. Furthermore, we provide, as far as
we know, the first econometrical test of this proposition on a public private contracts
database.
The paper is organized as follows. In a first section (section 2) we present our
theoretical framework in a static context. Then, we extend our results to a dynamic
context, using a repeated game framework leading to our main propositions (section
3). Then, in a last section (section 4) we present our data and test our propositions.
Conclusions follow.
2
The theoretical model
2.1
The general framework
To study the issues at stake, we build a theoretical framework based on Hart et al.
[1997]. More specifically, we assume that a benevolent public authority (PA, to
whom we will refer to as “she”) is in charge of providing two public services to users.
We denote these services as A and B. To provide those two services, we assume
that PA has to rely on external operators through the use of contracts.9
9Contrary to Hart, Shleifer, and Vishny [1997], we will not consider the public provision case, to
focus on horizontal integration and disintegration when contracting out.
6

---

More specifically, we assume that ex ante, PA may describe and specify in a contract
some aspects of the provision of a good. However, when executing the contract, the
private operator of a service may come up with new innovative ways to adapt the
service to users’ need, or to reduce the costs of provision of these services. Such
innovations are often difficult and costly to anticipate ex ante, which leads to some
contractual incompleteness as defined by Grossman and Hart [1986], Hart and Moore
[1990] or Hart [1995]. Hence, when such innovations turn up, parties will revise
the contract ex post when it is clear to them what the relevant contingencies are.
This leads us to assume that such innovative efforts are uncontractible ex ante, but
observable ex post (and then contractible) once relevant contingencies are realized.
2.1.1
Production technologies
To fix our ideas, we will assume that, ex ante, for a given service, the cost of provision
incurred by an operator is C0
s , s ∈ {A , B}. For simplicity’s sake, this cost is assumed
to be the same for all operators, and it is known to all. In the same way, we denote
the benefits to society that come from the provision of the basic service s as B0s,
s ∈ {A , B}. Following Hart, Shleifer, and Vishny [1997], we call this good the
“basic” good, and denote its price P 0
s .
Yet, operators may undertake efforts to innovate on the service provided during the
execution phase. Two types of innovations are considered: innovations that lead
to a reduction in costs, and innovations that lead to a better quality of the pro-
vided service. Efforts devoted to cost-reducing innovations (resp. quality-enhancing
innovations) for a given service s are denoted es (resp. is), s ∈ {A , B}. Upon im-
plementing the innovations, the social benefits and costs of providing a given service
s become
Bs = B0 −
s
bs(es) + βs(is)
Cs = C0 −
s
cs(es) + is + es
7

---

where cs(es) ≥ 0 is the reduction in costs corresponding to the cost innovation for
service s, bs(es) ≥ 0 is the reduction in quality corresponding to the cost innova-
tions for service s, and βs(is) is the quality increases net of costs from the quality
innovations for service s, s ∈ {A , B}. The function bs measures how much quality
is affected because of a (noncontractible) reduction of costs for service s.
For our purpose, we assume that services A and B differ in terms of the perspec-
tives for cost-reducing innovations and quality-enhancing innovations. In particular,
we assume that for service A , there is no perspective for quality-enhancing innova-
tions, and that cost reductions do not have any impact on the quality of the service
provided. In other words, bA (eA ) = 0 and βA (iA ) = 0. This means that no ex
post renegotiation arises. Indeed, the private manager can implement cost reduc-
tion without the approval of the public authority. Moreover, there is no impact on
quality, hence PA does not need to ask for renegotiation.
On the other hand, the perspectives of innovation for service B and their impact
on costs and social benefits to the society correspond to the classical case analyzed
in Hart, Shleifer, and Vishny [1997], i.e. including adverse effects in case of cost
reduction, and potential quality innovations. This assumption is meant to capture
the fact that cost-reducing perspectives and quality-enhancing opportunities differ
across different services. Notice that we also assume that both services are not
related in any way.
We make the following standard assumptions on cs, bB and βB: bB(0) = 0, bB(eB) ≥
0, bB(eB) ≥ 0; cs(0) = 0, cs(0) = ∞, cs(es) > 0, cs(es) < 0 , cs(∞) = 0; βB(0) = 0,
βB(0) = ∞, βB(iB) > 0, βB(iB) < 0, β (∞) = 0; cB(eB) − bB(eB) ≥ 0. The
assumptions cB(eB) − bB(eB) ≥ 0 and βB(iB) > 0 say that the quality reduction
from a cost innovation for service B does not offset the quality increase.
8

---

An operator’s overall ex ante costs can therefore be written as follows:
For service A
:
C0
A − cA (eA ) + eA
For service B : C0B − cB(eB) + eB + iB
2.1.2
Contracts
Following the literature, we further assume that iB, bB, βB, es and cs, with s ∈
{A , B}, are observable to the contracting parties, but are not verifiable to outsiders
(such as a court). Therefore, these variables cannot be part of an enforceable con-
tract. Furthermore, since these variables are not contractible ex ante, PA and the
private operator(s) may renegotiate the initial contract, once the innovations are
discovered. Similar with Hart, Shleifer, and Vishny [1997], we assume that if the
parties renegotiate the contract ex post, the gains from renegotiation are divided
between them according to a Nash bargaining outcome. The timing of the one shot
static game is depicted in the following figure.
Figure 1: Timing of the game
What is crucial here is that PA may propose an additional informal contract to the
operator to share the gains from innovation that are not contracted on ex ante, thus
avoiding ex post renegotiations. An informal contract here aims to motivate the
operator to achieve first-best levels of investments eF B
s
and iF B
s
, in exchange of a
supplementary monetary transfer, denoted Ts from PA to the operator of a given
service s. Such a contract, however, may not be enforced by any third party, since
innovative efforts are non-verifiable. Consistent with the economic literature, an
9

---

informal contract is self-enforcing for each party if the payoff stream from cooperation
is higher than the payoff stream from deviation (Baker, Gibbons, and Murphy [2002],
Baker, Gibbons, and Murphy [2004]). As such the informal contract that we discuss
in this paper corresponds to a relational contract. We model such an aspect using
a repeated game framework, in which an informal contract is consider to be self-
enforcing in the shadow of the future. This issue will be further discussed later on
in this paper.
Hence, in our framework, PA is confronted with the decision to whether use a same
private operator (horizontal integration) to ensure the provision of both services,
or to delegate the provision of both services to two different operators (horizontal
disintegration). In other words, PA may choose to bundle the provision of both
services or not. We suppose that PA is benevolent, and then will take these decisions
to maximize consumers’ surplus.10
2.2
The first best
We will briefly derive the first-best case to serve as a benchmark. In this situation,
we assume contractual completeness for es and is.
As shown by Hart, Shleifer, and Vishny [1997], contracting parties will choose es
and is to maximize total net surplus from their relationship, and divide the surplus
between themselves using lump-sum transfers. As a consequence, first-best incentives
are those maximizing:
max
[−bs(es) + cs(es) + βs(is) − es − is]
es,is
10As in HSV [1997], the public authority G does not maximize the global surplus during rene-
gotiations: its utility function is given by the welfare of the rest of society, excluding the manager
M. Indeed, “ The political process aligns G’s and society’s interests (since M has negligible voting
power, his interests receive negligible weight). As will become clear, if G placed the same weight on
M’s utility as on the rest of society, the first-best could be achieved”.
10

---

Document Outline

• Introduction
• The theoretical model
  • The general framework
    • Production technologies
    • Contracts
  • The first best
  • The one-shot game
• The repeated game framework
  • Horizontal disintegration: A different operator for each service
    • Share of the uncontractible surplus
    • Total cost for the public authority
  • Horizontal integration: A same private operator
    • Share of the uncontractible surplus
    • Total cost for the public authority
  • Cost comparison and proposition
• An empirical analysis of horizontal concentration in the French water sector
  • Putting the Model to the test
  • The Data
  • Empirical methodology
  • Estimation results
  • Alternative explanation
• Conclusion

---