460 V. Maksimovic and G. Phillips
RSZ run the following regression equation separately for the segments in each cell of
the classification matrix:
m(j,t)∑k= 1IAIjkt=α+β1
qjt+γ(Diversity)jt+δ(Firm sales)jt+controls+εjt,where
∑m(j,t)
k= 1 IAIjktis the sum of the IAI across them(j, t)segments belonging to firm
jat timetin the cell, andqjtis the equally weighted averageqsoffirmjsegments
at timetand the firm’s diversity is measured as the standard deviation of the firm’s
weighted segmentqs divided by the meanq,or
Diversityjt=
√
√
√
√
m(j,t)∑k= 11
m(j, t)− 1(
wjktqjkt−∑m(j,t)
k= 1 wjktqjkt
m(j, t))/∑m(j,t)
k= 1 qjkt
m(j, t).
The control variables include the firm fixed effects and calendar year dummies.
The predictions of the RSZ model are summarized inTable 1.
Investment falls in high opportunity segments with high resources as the firm’s di-
versity increases (cell (1)). Investment increases in low opportunity segments with low
resources as diversity increases (cell (4)). Investment increases with diversity in high
opportunity resource segments (cell (2)). Investment falls with diversity in large un-
profitable segments (cell (1)).
These predictions contrast this with Efficient Internal Market models that emphasize
the positive aspects of internal capital markets: top management has the option to re-
allocate resources from divisions with low investment opportunities to divisions with
high investment opportunities. An increase in the diversity increases the value of this
option and, thus, should increase the amount of resources transferred to segments with
better investment opportunities. Thus, if firms’ internal capital markets are efficient, we
would observeγ>0 in cells (1) and (2) andγ<0 in cells (3) and (4) as increases in
diversity make transfers between segments more valuable.
Ta b l e 1
Predictions of the RSZ ModelSegments with
resources>firm avg. resourcesSegments with
resources<firm avg. resourcesSegments with
Q>firm averageQ(1)γ<0(2)γ> 0Segments with
Q<firm averageQ(3)γ<0(4)γ> 0