444 V. Maksimovic and G. Phillips
Arguably, if investment patterns in conglomerates can be predicted by a neoclassical
model, then the effort in explaining misallocation of resources by managers may be
better directed at examining other forms of shareholder expropriation.Maksimovic and
Phillips (2002)consider a neoclassical model where firms differ because managerial
and organizational talent or some other fixed resource varies across firms. Interestingly,
the neoclassical model for conglomerate firms was introduced after the initial models
of power within the firm. It has motivated empirical models of investment within the
conglomerate firm and also endogeneity and sample selection models.
InMaksimovic and Phillips (2002)the firm decides endogenously whether to produce
in one or in several industries. As inCoase (1937)andLucas (1978), it is assumed
that there are diseconomies of scale within firms. Firms exhibit neoclassical decreasing
returns-to-scale, so that their marginal costs increase with output. Specifically, firms use
the variable inputs of labor, and capacity units to produce output.
In each industry, firms with higher organizational ability or talent can produce more
output with the same amount of input, and thus have higher productivity, than firms with
lower ability or talent. Thus, differences in talent have greater economic significance
when output prices are high. The productivity with which any given firm operates plants
can differ across industries in which it operates. For a given output price and a given
talent level, there are decreasing returns to scale in each industry in which the firm
operates and at the level of the whole firm.
For concreteness, consider a population of firms that can operate in a maximum of
two industries, which we denote as industry 1 and industry 2, respectively. The pro-
ductivity of each firm can be modeled by a vector(d 1 ,d 2 ), where the firm’s talent in
industryiisdi. Firms that have a higher productivity in industryi,di, produce more
output for a given level of inputs if they choose to operate in industryi. All firms are
assumed to be price-takers and to produce a homogeneous output. Firms use two inputs:
industry-specific homogeneous production capacitykand laborl. Further assume that
firms can trade capacity with other firms in the same industry or build capacity at price
rper unit. For tractability, we assume that each unit of capacity produces one unit of
output. For each firm, the profit function is
d 1 p 1 k 1 +d 2 p 2 k 2 −r 1 k 1 −r 2 k 2 −αl^21 −αl^22 −β(l 1 +l 2 )^2 , (2)wherepiandriare the prices of output and capacity in industryi=1or2,αand
βare positive cost parameters, andkiis the capacity the firm maintains in industryi.
The profit function embodies the assumption of neoclassical diminishing returns within
each industry (theαli^2 terms) and the assumption that when organizational talent is a
scarce resource, costs depend on the firm’s total size (theβ(l 1 +l 2 )^2 term). A firm is
diversified ifk 1 >0 andk 2 >0 and single segment if capacity in only one of the two
industries is greater than 0.
The model can be solved at the firm level to give the firm’s optimal capacity(k 1 ,k 2 )in
each of the industries as a function of its own productivity vector(d 1 ,d 2 ), and industry-
level variables, demand(p 1 ,p 2 )and the cost of capacity(r 1 ,r 2 ). Optimal outputs by
the firms in each industry can be obtained by direct optimization. Dropping the firm