CMPO Working Paper Series No. 03/061
CMPO is funded by the Leverhulme Trust.
Incomplete Contracts and Public Ownership:
Remarks, and an Application to Public-Private
Partnerships
Oliver Hart
Harvard University
July 2002
Abstract
The question of what should determine the boundaries between public and private firms in an advanced
capitalist economy is a highly topical one. In this paper I discuss some recent theoretical thinking on
this issue. I divide the paper into two parts. First, I make some general remarks about the relationship
between the theoretical literature of privatization and incomplete contracting theories of the firm.
Second, I use some of the ideas from this literature to develop a very preliminary model of public-
private partnerships.
JEL Classification: D23, H1, H4
Acknowledgements
This paper was given at a CMPO session of the Royal Economic Society Conference 2002 (University
of Warwick) on ‘The Private Delivery of Public Services’. It is forthcoming in the Economic Journal
(2003). I am grateful to the National Science Foundation through the National Bureau of Economic
Research for financial support, and to Donald Franklin, Paul Grout, Rohan Pitchford, and Andrei
Shleifer for helpful comments.
Address for Correspondent
Economics Department
Littauer Center
Harvard University
1875 Cambridge Street
Cambridge
MA 02138
1
1. Parallels between theories of the firm and of privatization
Let me begin by discussing the very close parallel between the theory of the firm and the theory
of privatization.
1
In the vertical integration literature one considers two firms, A and B. A might be a
car manufacturer and B might supply car-body parts. Suppose that there is some reason for A and B
to have a long-term relationship (e.g., A or B must make a
relationship-specific investment). Then there are two principal ways in which this relationship
can be conducted. A and B can have an arms-length contract, but remain as independent firms; or A
and B can merge and carry out the transaction within a single firm. The analogous question in the
privatization literature is the following. Suppose A represents the government and B represents a firm
supplying the government or society with some service. B could be an electricity company (supplying
consumers) or a prison (incarcerating criminals). Then again, there are two principal ways in which this
relationship can be conducted. A and B can have a contract, with B remaining as a private firm, or the
government can buy (nationalize) B.
There are, of course, some important differences between the two situations. First, if B is an
electricity company, it will likely have direct dealings with consumers, independent of its relationship with
the government. In this case the contract the government has with a private electricity company can be
thought of as an attempt to regulate the company’s dealings with consumers. There is no obvious
analogy in the case of vertical integration. Second, decisions to privatize or nationalize are often highly
political, presumably because of the government’s unique position in society, whereas vertical integration
decisions are usually strictly economic. Third, the government is often thought of as a very different
agent from a private firm: it is concerned with social welfare rather than just profit. Here, however, the
2
distinction is less sharp than it might seem at first sight since there are a number of firms (particularly
nonprofits or cooperatives) that have broader concerns than just profits.
In spite of these differences, the issues of vertical integration and privatization have much more
in common than not. Both are concerned with whether it is better to regulate a relationship via an arms-
length contract or via a transfer of ownership. Given this, one might have expected the literatures to
have developed along similar lines. However, this is not so. Whereas much of the recent literature on
the theory of the firm takes an “incomplete” contracting perspective, in which inefficiencies arise because
it is hard to foresee and contract about the uncertain future, much of the privatization literature has taken
a “complete” contracting perspective, in which imperfections arise solely because of moral hazard or
asymmetric information.
My own view is that this is unfortunate. One of the insights of the recent literature on the firm is
that, if the only imperfections are those arising from moral hazard or asymmetric information,
organizational form--including ownership and firm boundaries--does not matter: an owner has no special
power or rights since everything is specified in an initial contract (at least among the things that can ever
be specified). In contrast, ownership does matter when contracts are incomplete: the owner of an asset
or firm can then make all decisions concerning the asset or firm that are not included in an initial contract
(the owner has “residual control rights”).
Applying this insight to the privatization context yields the conclusion that in a complete
contracting world the government does not need to own a firm to control its behavior: any goals--
economic or otherwise--can be achieved via a detailed initial contract. However, if contracts are
incomplete, as they are in practice, there is a case for the government to own an electricity company or
3
prison since ownership gives the government special powers in the form of residual control rights.
Even if this position is accepted, it does not follow that one can take an “off the shelf” model
from the theory of the firm literature and apply it to privatization. In the standard “property rights”
model found in that literature, ownership serves to elicit appropriate ex ante investments, particularly
those in human capital.
2
If firm A acquires firm B, then A, having more residual control rights, has
greater bargaining power when uncontracted-for contingencies arise; A earns a greater return on her
investment and therefore invests more. Conversely, B’s incentive to invest falls since B’s bargaining
power is lower. The optimal allocation of ownership trades off these two effects.
Applying this logic to the privatization context, one concludes that, if the government buys an
electricity company or prison, the benefit is that some government bureaucrat who is in charge of the
prison will invest more (have more ideas, be more entrepreneurial); but the cost is that the manager of
the prison--who used to be an owner but is now an employee--will invest less. The latter effect--that a
government employee will be less entrepreneurial than an owner-manager--seems very plausible, but the
idea that government ownership leads to more entrepreneurship by bureaucrats seems less so.
For this reason the literature has explored other trade-offs.
3
Consider, for example, the model
in Hart et al (1997) (HSV). HSV compare two cases. (1) The government can own a facility, a prison,
say, and employ a manager to run it; or (2) the government can contract with a company owned by the
prison manager to run the prison for a period of time. HSV ignore investments on the government side,
but suppose that the prison manager can make two kinds of investment. He can invest in efficiency-
enhancing ideas that raise the quality of prison services, e.g., develop new rehabilitation programs; he
can also spend time figuring out how to cut costs and quality, while staying within the letter of the
4
contract. A government employee has little incentive to engage in either activity since it is easy for the
government (as owner) to “hold up” the employee without rewarding him appropriately. In contrast, a
private prison owner-manager is less subject to hold up. The good news about this is that private
ownership encourages the first, innovative type of investment. The bad news is that private ownership
also encourages the second, quality-shading kind of investment. The choice between public and private
ownership depends on which of these effects is more important.
In summary, the HSV model differs from the standard property rights model of the firm in two
ways. First, only one party (the prison manager) invests, but he makes two kinds of investments (as in
the multi-tasking model of Holmstrom-Milgrom (1991)). Second, the contract between the government
and the prison provider plays a crucial role--it defines the extent to which quality shading can occur. In
contrast, in the standard property rights model, long-term contracts are assumed to be sufficiently
incomplete to be useless.
2. Public-private partnerships (PPPs)
In this section, I use an HSV-type model to understand the costs and benefits of public-private
partnerships.
4
To repeat what I said in the introduction, this model is extremely preliminary. For
simplicity, I will now ignore the choice between public and private ownership and assume that all
provision is private. I will take a key property of a PPP to be that facility construction and service
provision are bundled, i.e., in the case of a prison the government contracts with a private party--
henceforth known as the “builder”--to build and run the prison (the builder may then subcontract with
someone else to run the prison).
5
In contrast, under “conventional” provision, the government contracts
