436 V. Maksimovic and G. Phillips
capital stock. In their view a firm that operates in industries that have a greater disparity
of investment to capital stock ratios is more diversified than a firm that operates in
industries that have similar investment to capital stock ratios. Thus, for each 2-digit
SIC code industry to which COMPUSTAT assigns the firm’s segments, Lamont and
Polk calculate the median investment to capital ratio among the single-segment firms.
The measure of a conglomerate firm’s diversity in yeartis then computed asσ,the
weighted standard deviation of these median ratios for all segments.
Lamont and Polk argue that changes inσ over time can be decomposed into en-
dogenous and exogenous components. The exogenous change in diversity,σX,isthe
change in diversity betweent−1 that would have occurred if COMPUSTAT had as-
signed the firm in the current year to precisely the same 2-digit SIC codes as in the
previous year. The endogenous change in diversity,σN, is the change in diversity that
occurs because the 2-digit SIC codes assigned to the firm have changed between years
t−1 andt.
Lamont and Polk use COMPUSTAT data for 1,987 diversified firms during the period
1980–1997. They find that 80% of the variation in firms’ diversity is due to exogenous
industry shocks. In their regressions they regress the change in excess value onσX
andσNalone and with control variables such as laggedσ. They find that increases in
bothσXandσNreduce firm’s excess value. They interpret the negative coefficient
ofσXas evidence that diversification reduces firm value. This finding persists even
when plausible measurement error is taken into account.Lamont and Polk (2002)also
analyze similarly defined changes in diversity of leverage, cash flows and sales growth
one at a time. They do not find that “exogenous” changes in diversity of these variables
have a significant negative effect by themselves.
Lamont and Polk’s interpretation of their results on investment diversity are in sharp
contrast toCampa and Kedia (2002)andVillalonga (2004a, 2004b)and have not been
fully reconciled with these studies.Villalonga (2003)argues that Lamont and Polk’s
measure does not pick up “diversification” as traditionally measured in the literature—
the presence of the firm’s operations in more than one industry—but “diversity” which is
the within firm dispersion of some industry characteristics. Indeed, she reports tests that
show that measures of diversification, such as the number of two digit industries that the
firm operates in are uncorrelated with Lamont and Polk’s measure of exogenous cash
flow diversityσX. However, this observation raises the question of which measure
better captures economic differences between firms.
- Theory explaining the conglomerate discount and organizational form
The early literature on the conglomerate discount leaves several questions unanswered.
Perhaps the most fundamental of them is why there should be a conglomerate discount?
Is the existence of a discount evidence of bad investment choices or is the discount an
endogenous outcome of a process by which different types of firms optimally select