438 V. Maksimovic and G. Phillips
they have motivated several of the empirical studies we examine below. In the interest
of brevity, we do not discuss several models which deserve a separate review, including
Berkovitch, Israel and Tolkowsky (2005), Faure-Grimaud and Inderest (2005), Fluck
and Lynch (1999)andInderst and Meuller (2003).
3.1. Efficient internal capital markets
Stein (1997)analyzes how internal capital markets create value and the optimal size
and scope of such markets. In Stein’s model firms consist of either a single stand-
alone project or of several projects overseen by a headquarters. Stein assumes that each
project’s managers obtain private benefits from managing their project. These benefits
are higher for better projects. The private benefits give managers an incentive to over-
state their project’s prospects. This is known to potential investors, who therefore supply
less capital than the managers request. As a result, good projects are capital rationed if
they operate as individual firms.
Stein assumes that a conglomerate’s headquarters has the ability to monitor the
projects it oversees. It uses its information in two ways. First, it can transfer capital from
one project to another. Second, it can appropriate for itself some of the private benefits
of the project managers, albeit at the cost of diluting the incentives of the managers.
Because the headquarters can extract private benefits from several projects simultane-
ously it has the incentive to allocate capital to the better projects. The ability to transfer
funds across projects, allocating some more funds than they would be able to raise as
stand-alone firms, and others less, makes better allocation possible.
A key assumption inStein (1997)is that as the number of projects overseen by the
headquarters increases, the quality of monitoring provided by the headquarters declines.
However, as the number of projects the quarters oversees increases, the headquarters in
Stein’s model also gains in two ways. First, the value of its ability to transfer funds from
the worst to the best projects increases. Second, if the project payoffs are not perfectly
correlated the volatility of the firm’s payoffs declines and it becomes able to raise more
funds from the capital market, thereby reducing credit rationing and increasing value.
The firm reaches its optimal size when the marginal decline in value due to declining
monitoring ability is equal to the marginal increase resulting from the relaxation of
financing constraints and the funding of good projects.
The theory also has implications for the optimal scope of the firm. Stein addresses
two effects which work in opposite directions. To the extent that the returns of different
divisions of a conglomerate are uncorrelated diversification increases the value of the
headquarters’ ability to direct investment funds and raise capital externally. However,
there may be another effect at work. Because headquarters’ allocation decisions are
dependent on therankingof investment projects rather than their absolute values, and
to the extent that accurate rankings are more likely to be made if all projects are within
the same industry (because valuation errors are likely to be correlated), diversification
is costly.