Ch. 8: Conglomerate Firms and Internal Capital Markets 447
the firm. Each band in the figure represents firms with equal market value to book value
ratios.
We can observe that for a firm to produce in two distinct industries near a 45 degree
line in the center of the graph, it has to havehigherproductivity than firms with equiv-
alent market value to book value ratios. Equivalently, if we match by productivity (or
size) single segment firms in two industries to a conglomerate firm producing in both
industries, the conglomerate firm will have alowermarket value to book value ratio
than the weighted average of the single segment firms. Thus one cannot in general con-
clude that multi-segment firms with lower market to book ratios are allocating resources
inefficiently.
We now illustrate the effect when we generalize the model allowing firms to pro-
duce across ten different industries. We illustrate this using two numerical exam-
ples that show how differences in organizational talent across industries causes firms
to choose to operate segments of different sizes and different observed productivi-
ties.
In each example we take the number of industries to be ten. We assume there are
25,000 potential firms, each of which is assigned firm-specific ability for each of the
ten industries. In terms of the previous discussion and the empirical work, high ability
is the same as high productivity. We draw the ability assignment d from a normal dis-
tribution with a mean ability of 1 and a standard deviation of 0.5. The output and input
prices and the cost parameters in all industries are held constant (in this case we set the
parameters from equation(1)as follows:p=200,r=200,α=5,β=2). In the
first example, firm ability is industry-specific. Firms’ ability to manage in one industry
is independent of their ability to manage in the other industries. Thus, the draws are
independent and identically distributed both within firms and across firms. In the sec-
ond example, there is a firm-specific effect: The draws within a firm for each of the ten
industries are correlated. We draw the common ability from a normal distribution with
a mean equal to 0 and standard deviation equal to 0.25. We add this common ability to
the random industry ability drawn earlier. Thus, part of a firms’ ability can be applied
equally to all industries. In each case we determine the industries in which it is optimal
for each firm to produce and also the amount of each firm production in each industry,
given the price of output and the prices of inputs. We keep track separately of firms
that choose to produce in one industry only, two industries only, etc., up to firms that
choose to produce in all the industries (if such firms exist). Thus, we have simulated
data on one-segment firms, two-segment firms, etc. For all firms with a given number
of segments, we rank the segments by size, and we compute the mean firm ability d for
that segment.
InFigure 4we allow the draws of firm ability in each of the 10 industries to be inde-
pendent. We call the industry in which the firm produces its “segments”. We label the
segment in which the firm produces the most its segment #1, the industry in which pro-
duces its second most, its “segment #2”, increasing this for each of the firm’s remaining
segments. The height of the graph (z-axis) gives the managerial abilityandequivalently
the size of the firm in that industry in which the firm produces. Each row of the figure