380
The Nobel Prizes
have no further value, the investor knows that he will have no leverage then: the
manager can pocket all the $100 with impunity. Thus the only way for the inves-
tor to be paid is to liquidate now.
In Hart and Moore (1994), the assumption that the manager can pocket the
monetary returns is replaced by the assumption that the manager can withdraw
her human capital. Suppose that a project costs $100 at date 0 and yields $120 at
date 2. The manager borrows the $100 and promises to repay this amount at date
2. At date 1 the manager could threaten to withdraw her human capital unless
the debt is reduced. If the parties have equal bargaining power, and the project
has zero value without the manager, then the debt can be renegotiated down to
$60, and an investor who foresees this will not lend money. Collateral can again
help here. If the assets have an alternative use at date 1, then the investor is at
least partially protected against strategic default.
The Hart-Moore (1994) model reminds us again of the distinction between
human and non-human assets. A project that consists mainly of human capital is
difficult to finance because an investor is subject to hold-up by the human capi-
tal. Conversely, a project that has significant non-human assets can be financed
without the fear of hold-up.
15
In some interesting recent work, Kaplan et al. (2009) have investigated the
importance of human vs. non-human assets in the context of start-ups. Their
paper, whose title suggestively begins, “Should Investors Bet on the Jockey or the
Horse?,” finds evidence that non-human assets in the form of a business plan are
an important and durable source of value. However, Bernstein et al. (2017) find
that human assets are also very important in the early stages of a start-up. Indeed,
as Rajan (2012) suggests, the balance may shift over time: part of the transforma-
tion of a start-up into a successful, mature firm may be a standardization process
that ensures that no individual’s human capital is that important.
This latest research is notable because it connects to one of the original moti-
vations of Grossman and Hart (1986) and Hart and Moore (1990). Both papers
start off with the same question: What is a firm? The answer given is that non-
human assets are an important part of any firm; they are the glue that holds the
firm together (see also Hart (1995)). The work by the above authors is helping
to clarify this.
16
4. APPLICATION TO PUBLIC VS. PRIVATE OWNERSHIP
Economists generally agree that there are some goods and services that will not
be provided at an adequate level through private markets, and that therefore
need to be financed by the government. Clear examples are national defense, the
Incomplete Contracts and Control
381
police, foreign policy, and prisons. Examples about which there might be greater
debate are health care and schools.
Government financing does not necessarily imply government provision.
The government has a choice about whether to provide the good or service itself
or to contract with a private provider. The incomplete contracting approach can
be helpful in elucidating the trade-off.
Andrei Shleifer, Robert Vishny, and I (Hart et al. (1997)) explore the public-
private choice, focusing particularly on prisons.
17
Suppose that the government,
acting on behalf of society, wants to incarcerate prisoners. The government
may have several goals: to prevent the prisoners from escaping, to treat them
humanely, and to maximize the chance that they can return to society as well-
functioning citizens. The government can own a prison and use government
employees to run it; or it can contract with a private company to run the prison.
Which is better?
The first point to note is that in an ideal world, where everything can be antic-
ipated and written into a contract, the choice does not matter since a complete
contract will be written in both cases. In a complete contracting world, owner-
ship and residual rights of control are irrelevant since all decisions are specified
in the contract. The presence of asymmetric information and moral hazard does
not change this conclusion: these factors simply lead to the addition of various
incentive-compatibility constraints in the solution for the optimal contract.
When contracts are incomplete, residual control rights become important.
The contracts that the federal or state governments write with private prison
companies are in fact quite elaborate and cover a number of aspects of prisoner
treatment including food, hygiene, health care, work, education, recreation, etc.
However, Hart et al. (1997) argue that the contracts are significantly incomplete
with respect to two important factors: the use of force by guards and the qual-
ity of personnel. As a result of this incompleteness a private contractor can use
its residual rights of control to save money by hiring cheap, unqualified guards.
These guards may not have the skill to respond to violent situations effectively.
The prison company hiring cheap unskilled guards is analogous to the coal
mine choosing to mine high-ash-content rather than low-ash-content coal. In
each case the supplier is choosing an action permitted by the contract that saves
money at the expense of quality: in one case the quality effect is borne by the
power plant, in the other case by the government or society. Of course, if the
quality reduction has greater value than the saving in costs, ex post renegotiation
of the contract should occur. Indeed, the model in Hart et al. (1997) supposes
that it does. However, there is still a distortion: the private provider will have an
excessive incentive to develop cost-saving, quality-reducing ideas.