Cash Holdings and Corporate Diversification: Ran Duchin

THE JOURNAL OF FINANCE • VOL. LXV, NO.
3 • JUNE 2010
Cash Holdings and Corporate Diversification
RAN DUCHIN∗
ABSTRACT
This paper studies the relation between corporate liquidity and diversification. The
key finding is that multidivision firms hold significantly less cash than stand-alone
firms because they are diversified in their investment opportunities. Lower cross-
divisional correlations in investment opportunity and higher correlations between
investment opportunity and cash flow correspond to lower cash holdings, even after
controlling for cash flow volatility. The effects are strongest in financially constrained
firms and in well-governed firms, and correspond to efficient fund transfers from low-
to high-productivity divisions. Taken together, these results bring forth an efficient
link between diversification and corporate liquidity.
CASH HOLDINGS OF U.S. companies are enormous and growing over time. As of
fiscal year 2006, nonfinancial and nonutility firms in the Compustat universe
reported aggregate cash holdings of over 1.7 trillion dollars, representing 9.2%
of the total market value of these firms’ equity.1 The growth in cash holdings
is equally impressive. According to Bates, Kahle, and Stulz (2009, BKS hence-
forth), listed U.S. industrial firms’ average ratio of cash to assets increased from
10.5% in 1980 to 23.2% in 2006. This massive increase in cash has captured
the attention of both academics and the media. For example, a recent article in
The New York Times states that, “Publicly traded American firms hold so much
cash that, as a group, they could pay off all their debt and still have money left
over.”2
While the dramatic increase in cash holdings is receiving growing attention,
another noteworthy pattern is not widely recognized: The average cash hold-
ings of stand-alone firms are almost double the cash holdings of diversified
firms. From 1990 to 2006, diversified firms held on average 11.9% of their as-
sets in cash, whereas stand-alone firms held more than 20.9% of their assets
∗ Ran Duchin is at Ross School of Business, University of Michigan. I am grateful for very
helpful comments from Harry DeAngelo, John Graham, Campbell Harvey, Gareth James, John
Matsusaka, Micah Officer, Oguzhan Ozbas, Breno Schmidt, Berk Sensoy, René Stulz, the associate
editor, two anonymous referees, and seminar participants at Arizona State University, Hebrew
University, University of Arizona, University of Florida, University of Illinois, University of Michi-
gan, University of North Carolina, University of Oregon, University of Pittsburgh, University of
Rochester, University of Washington, Vanderbilt, Virginia Tech, and Yale.
1 Cash holdings are defined as cash and short-term investments. Note that aggregate cash
holdings including financial firms and utilities were 6.8 trillion dollars in 2006, representing
27.6% of their total market value of equity, or 10.2% of their book assets.
2 “Behind Those Stock Piles of Corporate Cash,” by Mark Hulbert, The New York Times, October
22, 2006.
955
956 The Journal of FinanceR
Figure 1. Average annual cash holdings (%) for specialized and diversified firms. This
figure plots the average annual cash-to-assets ratios of specialized and diversified firms in the
sample, which consists of nonfinancial and nonutility firm-years from 1990 to 2006 with non-
missing data on cash holdings and on the industry codes of each business segment, and with total
market capitalization of $10M or more. Cash to assets is cash and short-term investments scaled
by total assets. Specialized firms are firms that reported a single business segment on Compustat,
whereas diversified firms are firms with two business segments or more.
in cash. Figure 1 breaks the universe of Compustat firms into stand-alone ver-
sus diversified firms, and compares their average cash holdings from 1990 to
2006. The figure shows not only the dramatic increase in cash holdings that
others have noted but also the pronounced difference in cash policies of diver-
sified and specialized firms, a pattern that has not received much attention. As
more in-depth results below will show, these differences cannot be explained
by economies of scale (e.g., Beltz and Frank (1996), Mulligan (1997)), growth
opportunities or cash flow volatility (e.g., Opler et al. (1999)).
This paper suggests that diversified firms are well positioned to smooth
investment opportunities and cash flows because both the opportunities and
the outcomes of their divisions are not perfectly correlated. As a consequence,
diversified firms will hold less precautionary cash, and increasingly so as
cross-divisional correlations decrease. The results suggest that when stud-
ied separately, lower correlations in investment opportunity or cash flow, as
well as higher correlations between investment opportunities and cash flows
(i.e., smaller “financing gaps,” as studied in Acharya, Almeida, and Campello
(2007)), are associated with lower cash holdings, whereas when studied jointly,
the correlations in investment opportunity play a more dominant role that sub-
sumes the effect of cash flow correlations. Importantly, diversification in cash
Cash Holdings and Corporate Diversification 957
flow still affects cash through the correlation between investment opportunities
and cash flows, that is, the financing gap.
These results are consistent with the predominant approach to understand-
ing corporate demand for cash, namely, the precautionary saving theory in-
troduced by Keynes (1936). According to the theory, firms hold cash to protect
themselves against adverse cash flow shocks that might force them to forgo
valuable investment opportunities due to costly external financing. While pre-
vious studies have focused primarily on the relation between cash flow and cash
holdings (e.g., Opler et al. (1999), Almeida, Campello, and Weisbach (2004)),
this paper introduces the notion of investment opportunity risk and studies how
it affects cash holdings (jointly with cash flow risk), and how it corresponds to
corporate diversification and internal capital markets. A central finding is that
the joint uncertainty of investment opportunity and cash flow is important
mainly when firms are financially constrained. This finding is consistent with
Modigliani and Miller’s (1958) insight that cash only affects firm value when
markets are not frictionless, as well as with the results in Almeida et al. (2004)
that cash holdings and cash flows are correlated only in financially constrained
firms.
Importantly, diversification is measured directly through the cross-divisional
correlations in investment opportunity (measured by Tobin’s Q) and cash flow
(measured by earnings less interest and taxes), as well as the financing gap
(measured by the correlation between Tobin’s Q and earnings less interest and
taxes).3 When investment opportunities or cash flows across divisions are less
correlated, firms enjoy greater diversification (or coinsurance). The findings
suggest that diversified firms hold less cash exactly when their cross-divisional
correlations in investment opportunity are lower, regardless of their number
of divisions. They also hold less cash when the correlation between their cash
flows and investment opportunities is higher. Interestingly, the correlation in
cash flow has little effect on cash holdings once the correlation in investment
opportunity and the financing gap are taken into account. As I discuss later in
the paper, these results suggest that diversification in cash flow affects cash
holdings mainly through the financing gap.
The magnitudes of the documented effects are both statistically and econom-
ically significant. For example, an increase of one standard deviation in the
cross-divisional correlation in investment opportunity leads to an increase of
4.4% (9.1%) in the cash holdings of the average (median) firm. These results
continue to hold after controlling for other determinants of cash holdings such
as cash flow and investment levels, size, leverage, agency costs, and unobserv-
able firm characteristics. (see, e.g., Dittmar and Mahrt-Smith (2007), Harford,
Mansi, and Maxwell (2008), and Opler et al. (1999) for studies of the determi-
nants cash holdings).
To gain further insight into the relation between liquidity and diversifica-
tion, I study the interaction between cash holdings and internal capital mar-
kets (ICMs). As advocated by Stein (1997), Matsusaka and Nanda (2002), and
3 See the Appendix for detailed variable definitions.

others, diversification has the potential benefit of insulating firms from the
rationing and costs of external capital markets through the workings of ICMs.
Using an indirect measure of cross-divisional transfers proposed by Rajan,
Servaes, and Zingales (2000), I find that cross-divisional diversification in in-
vestment opportunity is positively related to transfers across divisions, and
that firms hold less cash when transfers are abundant. Furthermore, hold-
ing less cash is correlated with efficient cross-divisional transfers to divisions
with better investment opportunities. While Rajan, Servaes, and Zingales find
that diversity in investment opportunities leads to inefficient fund allocations
across divisions, my results imply that holding less cash due to diversification
in investment opportunity is associated with efficient flows of funds to the more
productive divisions.
The results show further that the relation between cash holdings and diver-
sification is stronger in well-governed firms. Combined with the evidence on
ICMs, these findings suggest that the reduction in cash holdings due to diver-
sification is efficient. This suggests that saving on the costs of holding cash
might be another important benefit of diversification, to be weighed against
the various costs of diversification examined in previous studies (e.g., Rajan
et al. (2000), Scharfstein and Stein (2000), Matsusaka and Nanda (2002), Ozbas
(2005)).4 However, one important caveat is that these results do not offer con-
clusive evidence as to whether or not diversification overall adds or destroys
firm value.5
To directly test how diversification affects corporate liquidity, I adopt a lon-
gitudinal approach similar to Graham, Lemmon, and Wolf (2002) and Hyland
and Diltz (2002) and study how acquisitions affect cash holdings. The results
show that acquisitions that are more diversifying in investment opportunity
(i.e., a lower correlation between the investment opportunities of the acquirer
and the target) are followed by greater reductions in cash holdings. Interest-
ingly, firms do not adjust their cash holdings instantaneously; cash holdings
are gradually reduced over the 3 years that follow the acquisition.
Finally, the results have implications for the dramatic upward-sloping trend
in cash holdings documented by BKS. I show that the average cross-divisional
correlations in investment opportunity or cash flow have increased over time,
which suggests that the average firm has adjusted its cash holdings upward
in response to the increase in the joint risk of investment opportunity and
cash flow. These results are in line with the 1990s trend of an increasing
percentage of same-industry mergers documented by Andrade, Mitchell, and
Stafford (2001). Thus, firms have a stronger precautionary demand for cash
because they are less diversified in their investment opportunities and cash
flows.
4 The costs of holding cash include a liquidity premium, double taxation, and the agency costs of
free cash flow.

5 See Lang and Stulz (1994) and Berger and Ofek (1995) for evidence supporting a “diversification
discount.” See Campa and Kedia (2002), Chevalier (2004), and Villalonga (2004a, 2004b) for recent
evidence questioning the existence of a discount.
This paper contributes to a growing body of research on cash holdings. The lit-
erature has highlighted various benefits and costs associated with cash, such as
the financing of corporate investments (e.g., Kim, Mauer, and Sherman (1998),
Acharya et al. (2007)), and Denis and Sibilkov (2009), taxes (e.g., Foley et al.
(2007)), and agency (e.g., Jensen (1986), Dittmar, Mahrt-Smith, and Servaes
(2003), Pinkowitz, Stulz, and Williamson (2006), Dittmar and Mahrt-Smith
(2007), Harford et al. (2008)). Furthermore, Graham and Harvey (2001) report
that managers consider financial flexibility to be an important feature of their
financing policy. I focus on the role of cash in reducing the firm’s exposure to
investment risk, and suggest that diversified firms do not hold as much cash
because they are less exposed to this risk.6
The paper proceeds as follows. Section I outlines the theoretical arguments,
while Section II describes the data and methodology. Section III presents
the main results, and Section IV considers some implications and extensions.
Section V gives concluding remarks.
I. Theory
To formulate the main hypotheses investigated in the paper, consider a com-
pany that produces stochastic cash flows from existing assets and has invest-
ment opportunities that arrive randomly. The firm operates in imperfect capital
markets, and therefore cannot raise sufficient funds to finance all of its invest-
ments today and in the future. To transfer funds to future states of the world,
the firm can decide to save a portion of today’s cash flow and carry it to the
future as cash holdings. Because of the financing frictions the company faces,
carrying cash is costly since it forces the firm to forgo valuable investment op-
portunities today. The benefit of carrying cash, however, is the ability to finance
future expected investment opportunities. The optimal amount of cash equates
the marginal profitability of current investments with the marginal expected
profitability of future investments. Hence, this optimal cash level is a function
of the joint distribution of investment opportunities and cash flows over time.
This simple framework nests the essence of the precautionary demand for cash.
One characterization of the optimal amount of cash is that it increases in the
volatility of cash flows from existing assets. This result is formalized by Kim
et al. (1998) and tested by Opler et al. (1999), who find that, consistent with
the theory, firms operating in more volatile industries hold significantly more
cash as a fraction of their assets.
However, as noted above, a full account of the precautionary demand for
cash implies that optimal cash levels are determined by the joint distribution
of investment opportunities and cash flows. In other words, it is not only the
uncertainty surrounding cash flows that affects cash holdings; rather, the un-
certainty surrounding investment opportunities and the simultaneity of cash
6 When precautionary demand for cash decreases due to diversification, it is optimal to hold less
cash because of the costs associated with holding excess cash. These include a liquidity premium,
double taxation, and the agency costs of free cash flow. These costs are collectively discussed in
Opler et al. (1999). More recent literature largely focuses on the agency costs of cash (see, e.g.,
Harford (1999), Harford et al. (2008)).
flows and investment opportunities also should affect corporate cash holdings.
Firms with more volatile investment opportunities should optimally hold more
cash, as should firms whose cash flows and investment opportunities tend to
arrive in different time periods.7
To understand how diversification might affect corporate cash holdings, let
us consider a simplified example of a multidivision firm with two business
segments. Each segment operates in an industry characterized by a stochastic
stream of cash flows and investment opportunities that arrive randomly. Mul-
tidivisionalism can affect the firm’s overall level of cash holdings through the
effect of diversification on the joint distribution of investment opportunities
and cash flows across divisions.
When the cross-divisional correlation in investment opportunity is low, firms
are able to optimally hold less cash because they are less likely to encounter
multiple investment opportunities in both divisions simultaneously. A lower
correlation in investment opportunity across divisions decreases the marginal
value of cash holdings, and therefore reduces the precautionary demand for
cash. Similarly, when the cross-divisional correlation in cash flow is low, firms
are able to optimally hold less cash because they are less likely to experience
simultaneous adverse cash flow shocks in both divisions. Such adverse shocks
would increase the marginal value of holding cash and as a result increase the
precautionary demand for cash. A third dimension is the intradivisional corre-
lation between investment opportunities and cash flows. When the correlation
between cash flow and investment opportunity is high, the optimal amount of
cash decreases because investments can be financed using internally generated
cash flows without the need to resort to costly cash holdings.
This analysis suggests that diversified companies are well positioned to hold
less cash due to a coinsurance effect across divisional investment opportunities
and cash flows. In particular, it demonstrates that the key determinants of
cash holdings are the cross-divisional correlations in investment opportunity
and cash flow, as well as the degree of simultaneity (or correlation) between
divisional cash flows and investment opportunities.
This analysis also highlights a number of additional empirical implications
of the theory. First, if diversified firms hold less cash due to a lower precau-
tionary demand for cash, then this behavior should be particularly strong for
financially constrained firms.8 Recall that a key assumption of the theory is that
firms cannot finance all their investments using external funds. If they could,
then consistent with Modigliani and Miller (1958), cash would not add value.
Second, the benefit of diversification and coinsurance in facilitating lower cash
holdings assumes the existence of, and the alignment of incentives with, effi-
cient internal capital markets. Therefore, the lower cash balances in diversified
firms should be accompanied by good governance and efficient cross-divisional
transfers of funds to the more productive divisions of the company.
7 Acharya et al. (2007) investigate the theoretical implications of the financing gap (correlation)
between cash flows and investment opportunities.

8 Consistent with this implication, Han and Qiu (2007) find that cash holdings are only signifi-
cantly related to cash flow volatility in financially constrained firms.

The empirical investigation that follows tests these implications directly. I

construct measures of the cross-divisional correlation in investment opportu-
nity and cash flow as well as the intradivisional correlation between investment
opportunities and cash flows, and show that their impact on cash holdings is
consistent with the theory. Furthermore, I show that the effects are stronger in
constrained firms. I also compute different measures of corporate governance
and cross-divisional transfers, and find results consistent with the optimality
of reduced cash holdings due to diversification in investment opportunity. The
next section describes the data and the empirical methods in detail.
II. Data and Empirical Methods

A. Construction of Correlation Measures
To study the relation between risk, diversification, and cash holdings, I con-
struct direct measures of volatility and cross-divisional correlation in invest-
ment opportunity and cash flow. The building blocks of the analysis are annual
averages of investment opportunity and cash flow across all stand-alone firms
in each three-digit North American Industry Classification System (NAICS)
code industry, applied as indirect measures of divisional investment opportu-
nity and cash flow, where investment opportunity is measured by Tobin’s Q and
cash flow is measured by earnings less interest and taxes. (See the Appendix
for detailed variable definitions.) While the use of average industry stand-alone
companies to proxy for the investment opportunities of conglomerate divisions
has been criticized by previous studies (e.g., Campa and Kedia (2002), Villa-
longa (2004a)), I follow this methodology mainly due to nonavailability of direct
measures of investment opportunities at the division level. Another concern is
that segment reporting itself might be inaccurate, as suggested by Denis, De-
nis, and Sarin (1997) and Hyland and Diltz (2002). I address this issue directly
by studying corporate acquisitions, which are less vulnerable to poor segment
reporting, and by verifying that the results continue to hold after the introduc-
tion of SFAS 131, which requires greater segment disclosure (see, e.g, Berger
and Hann (2003)).
In previous studies, diversification is measured by the number of different
business segments or industries the firm reported. Opler et al. (1999) docu-
ment an inverse relation between cash holdings and the number of business
segments. While this measure is likely to be correlated with cross-divisional
correlations, which tend to decrease as the number of segments increases, it
suffers from a number of problems. First, it is unclear which dimension of
diversification it captures. For example, multidivision firms might view their
noncore segments (i.e., those segments that operate in secondary industries)
as quasi-liquid entities, which can be liquidated in the event of an increased
demand for liquidity and therefore serve as a substitute for cash.
Second, the number of segments is a crude proxy for diversification. For in-
stance, all segments might operate in the same industry, in which case there is
virtually no diversification. Or it might be the case that they operate in differ-
ent industries, but these industries are closely related to each other. Finally, it
might also be the case that firms with more segments hold less cash because
they spent it acquiring these segments. It is therefore important to measure
cross-divisional correlations directly, and to include the number of business
segments as a control.
Next, I describe how volatilities and correlations in investment opportunity
and cash flow are measured. For each year t, define a time window of [t − k, t −
1] spanning the k years preceding year t. Let INV it,k denote the investment
opportunity stream of the average stand-alone firm in industry i over [t − k, t −
1] and let CFit,k denote the corresponding cash flow stream. The volatility of
the investment opportunity and cash flow of industry i is then defined as the
standard deviation of the investment opportunity stream and cash flow stream,
respectively

σ (INV)it,k = σ INV it,k , (1)

σ (CF)it,k = σ INV it,k . (2)
Throughout the paper, volatilities are estimated over a 10-year window, with
a minimum requirement of 5 years of nonmissing data within the 10-year
window.
Next, consider a firm with N business segments. Relying on industry-level
measures of investment opportunities and cash flows, the volatilities of the
firm’s investment opportunity and cash flow in year t are defined as9

N N

σ (INV)t,k =
j
wi w j ρ(INV)i, j σ (IINV)it,kσ (INV)t,k, (3)
i=1 j=1

N N

σ (CF)t,k =
j
wi w j ρ(CF)i, j σ (ICF)it,kσ (CF)t,k, (4)
i=1 j=1
where ρ(I NV )i, j and ρ(C F)i, j are the correlations between the investment op-
portunity streams and the cash flow streams of industries i and j, respectively,
and wi is the weighting of segment i in the firm, given by the ratio between the
segment’s sales and the total sales of the firm. Note that, compared to previ-
ously employed measures of industry-level volatility in the empirical literature
(e.g., Opler et al. (1999)), these measures have the advantage of taking into
account all the industries in which the firm operates.
To measure the cross-divisional correlations in investment opportunity and
cash flow, I calculate a measure of “no-diversification” average volatility by
9 This is identical to the standard formulation of a portfolio’s standard deviation.

assuming a pair-wise correlation of one between all segments

N N

σ (INV)t,k =
j
wi w j σ (INV)it,kσ (INV)t,k, (5)
i=1 j=1

N N

σ (CF)t,k =
j
wi w j σ (CF)it,kσ (CF)t,k. (6)
i=1 j=1
Throughout the paper I call these measures Industry Q volatility and In-
dustry cash flow volatility, respectively, and include them in my regression
estimations.
We can now define the firm’s cross-divisional correlation in investment op-
portunity and in cash flow as follows:
corr(INV)t,k = σ (INV)t,k − σ (INV)t,k, (7)
corr(CF)t,k = σ (CF)t,k − σ (CF)t,k. (8)
Note that (7) and (8) are always less than or equal to zero, and measure
the difference between volatility with correlation and volatility without corre-
lation. Thus, they capture the overall effect of correlations on the volatility of
investment opportunity and cash flow. As the differences in (7) and (8) become
more negative, the correlation effect is stronger and vice versa. Thus, higher
values (i.e., less negative values) of (7) and (8) imply a higher correlation and
a smaller level of diversification in investment opportunity and cash flow.
So far, the analysis considers the volatility and correlation in investment
opportunity and cash flow separately.10 However, the division-level correlation
between investment opportunity and cash flow is also important. For exam-
ple, if investment opportunity and cash flow are perfectly positively correlated,
firms should hold less cash even if volatilities are high because investments
can be financed using internally generated cash flows. Conversely, if they are
perfectly negatively correlated, firms should hold more cash even if volatili-
ties are low because cash flow will not cover available profitable investments.
Therefore, we should also control for the intra-industry correlations between
investment opportunity and cash flow, that is, the “financing gap” studied in
Acharya et al. (2007). This correlation is calculated as the sales-weighted av-
erage intraindustry correlation across all business segments:

N
ρ(INV, CF)t,k = wi ρ(INV, CF)it,k. (9)
i=1
The next subsection describes the sample construction and gives summary
statistics.
10 The overall correlation between correlation in investment opportunities and correlation in
cash flows in the sample is approximately 0.35.
B. Sample
The sample includes all firms available from Compustat’s North America In-
dustrial Annual file and Compustat’s Segments file. All data are CPI-adjusted
into 1990 dollars. Compustat’s Industrial Annual file is used to retrieve data on
firms’ cash holdings and short-term securities, book assets, sales, operational
cash flows, market-to-book ratios, leverage, capital expenditures, dividend pay-
ments, stock repurchases, and net working capital. I use Compustat’s Segments
file to retrieve data on a firm’s business segments, including the number of busi-
ness segments within each firm and the industry of each segment (represented
by three-digit NAICS codes). Since the Segments file might contain repeated
data years if the reported segments appear on multiple source documents, I
only consider the latest source year of each segment-year observation.
I exclude financial firms and utilities, but do not exclude industrial firms with
financial segments because excluding these would eliminate from the sample
many large conglomerates that maintain a finance division. Following Almeida
et al. (2004), I also eliminate firm-years for which data on cash holdings are
missing, those for which cash holdings exceed the value of total assets, those
for which market capitalization is less than $10 million (in 1990 dollars), and
those displaying asset or sales growth exceeding 100%. Following Berger and
Ofek (1995), I eliminate all firm-year observations for which I do not have each
segment’s industry (NAICS code), and I require that the sum of segment sales
be within 1% of the total sales of the firm to ensure the integrity of segment
data.
Finally, I eliminate all firm-year observations for which one or more corre-
lation measures are missing. That is, if correlation in investment opportunity,
correlation in cash flow, or correlation between investment opportunity and
cash flow is missing, the firm-year observation is excluded from the sample. I
include this filter because my empirical investigation aims specifically at study-
ing the effect of the joint distribution of investment opportunity and cash flow.
Because the measures of volatility and correlation in investment opportunity
and cash flow are based on past moving windows of 10 years (with a required
minimum of five valid observations), the sample period begins in 1990 due to
data availability on Compustat’s segments file. Thus, the sample covers the
17-year period from 1990 to 2006 and consists of 50,905 firm-year observations
on 9,357 firms.
Table I describes the various variables employed in this study. The table
reveals a wide variation in cash holdings, with a mean of 18.8% and a standard
deviation of 21.9%. The median firm has cash equal to 9.5% of its assets,
which suggests that the cross-sectional distribution of cash holdings is right-
skewed. The independent variables in the sample also reveal wide variation.
For example, Tobin’s Q, which is used as a proxy for investment opportunity,
has a mean of 1.7 and a standard deviation of 0.9.
Table I also reveals significant variation in cross-divisional correlations: The
standard deviation in the Q correlation is 3.7 times its absolute mean, and the
standard deviation in cash flow correlation is 3.8 times its mean. Note that for
Table I
Summary Statistics
This table reports summary statistics for the sample, which consists of nonfinancial and nonutility
firm-years from 1990 to 2006 with nonmissing data on cash holdings and on the industry codes
of each business segment, and with total market capitalization of $10M or more (in 1990 dollars).
See the Appendix for variable definition.
Standard Number of
Variable Mean Median Deviation Observations
Cash/assets 0.188 0.095 0.219 50,905

Q correlation −0.003 0.000 0.011 50,905
Industry Q volatility 0.214 0.190 0.128 50,905
Cash flow correlation −0.004 0.000 0.015 50,905
Industry cash flow volatility 0.262 0.178 0.252 50,905
Firm cash flow volatility 0.078 0.048 0.080 34,613
Q-cash flow correlation 0.173 0.249 0.440 50,905
Tobin’s Q 1.744 1.457 0.943 50,905
CAPEX/assets 0.067 0.046 0.071 50,905
Cash flow/assets 0.034 0.081 0.195 50,905
Book leverage 0.221 0.172 0.240 50,844
Payout/assets 0.026 0.001 0.079 50,905
NWC/assets 0.093 0.082 0.211 49,548
Number of segments 1.426 1.000 0.922 50,905
Firm size 5.042 4.833 1.719 50,905
both correlations, the median is zero because the sample is largely dominated
by stand-alone firms. To deal with potential inference problems that arise due
to the overwhelming majority of specialized firms in the sample, subsequent
analysis considers subsamples that exclude some or all single-segment firms.
Specifically, throughout the paper and the Internet Appendix,11 the results
are reported for three different samples: (1) all firms, (2) only diversified (mul-
tidivisional) firms, and (3) a “balanced” sample composed of all diversified
firms and a randomly chosen subsample of 20% of the specialized firms. The
main reason to consider the three samples is that the Compustat universe is
largely dominated by single-segment firms, which have zero diversification by
construction. However, a single-segment firm may still have some degree of
diversification and operate in multiple industries but not report this because
of the size of the other division or because it is “related enough” to the primary
segment. It is therefore important to conduct the tests when such firms are
excluded. At the same time, excluding all specialized firms might underesti-
mate the relation between diversification and cash, because the difference in
diversification is greatest between single-segment and two-segment firms.12
11 The Internet Appendix is available at http://www.afajof.org/Supplements.asp.

12 To see this, note that the average correlation in investment opportunity is zero for stand-alone
firms, −0.009 for two-segment firms, −0.014 for three-segment firms, and −0.021 for firms with
four segments or more. Furthermore, an analogous result in the diversification discount literature,
reported by Lang and Stulz (1994) and subsequent studies, implies that the discount is most
Thus, I also consider a “balanced” sample that includes a proportional number

of nondiversified firms but is not dominated by such firms.
III. Main Results

A. Differences in Means
The empirical analysis begins in Table II by looking at difference-in-means
estimates of firm-level cash holdings associated with: (i) cross-divisional corre-
lation in investment opportunity (Tobin’s Q), (ii) cross-divisional correlation in
cash flow, and (iii) intradivisional correlation between investment opportunity
and cash flow. The results are reported for each of the three samples, and in
all panels the differences-in-means and t-statistics are calculated annually and
averaged across years (see, e.g., Fama and MacBeth (1973), Petersen (2009)).
The results in Panels A, C, and E show that firms with higher cross-divisional
correlations in investment opportunity or cash flow, and lower correlations be-
tween investment opportunity and cash flow, hold significantly more cash.
These results are robust across the different samples. As expected, the largest
differences are recorded in the full sample, whereas the smallest differences
are recorded in the sample of only diversified firms. While the full sample might
overestimate the impact of cross-divisional correlations, the multidivision sam-
ple might underestimate their impact due to the nonlinearity of diversification
in the number of segments discussed above. Thus, if we take the difference in
the “balanced” sample to be the most reliable, then average cash holdings in
firms with high cross-divisional correlations in investment opportunity (cash
flow) are 6.1% (6.3%) higher than in firms with low correlations (see Panel C).
The results in Panels B, D, and F show that within similar-size bins, firms
with lower cross-divisional correlations in investment opportunity hold signif-
icantly less cash. The results also confirm that there are economies of scale
in cash holdings (see, e.g., Beltz and Frank (1996), Mulligan (1997)): Indeed,
larger firms hold less cash relative to their total assets. Nevertheless, the ef-
fect of diversification in investment opportunity is still highly economically and
statistically significant within each size bin across all three samples.
Interestingly, the differences in average cash holdings between high- and
low-correlation firms are smaller for larger firms. In panel D, for example, the
difference between high- and low-correlation firms is 6.9% for small firms and
2.7% for large firms. This is an intriguing result that I investigate in Table
IV. As shown in Table IV, below, the effect of diversification on cash hold-
ings is mostly concentrated in financially constrained firms, consistent with
Modigliani and Miller’s (1958) result that the precautionary demand for cash
disappears in frictionless markets. As larger firms tend to be less constrained,
it is not surprising that diversification has a smaller impact on cash holdings
in those firms. Finally, note that the results continue to hold with difference-
in-medians. These results are reported in the Internet Appendix.
significant between one- and two-segment firms, and less so between two-segment firms and firms
with more than two segments.
Table II
Average Annual Cash Holdings
This table presents difference-in-means estimates of firm-level annual cash holdings. The estima-
tion involves a two-step procedure: (1) estimation of annual differences-in-means for each year from
1990 to 2006, and (2) time-series averaging of annual differences-in-means and t-statistics. The
sample consists of nonfinancial and nonutility firm-years with nonmissing data on cash holdings
and the industry codes of each business segment, and with total market capitalization of $10M or
more. Panels A and B use the full sample. Panels C and D use a “balanced” sample, which consists
of all diversified firms, and a randomly chosen subsample of 20% of the stand-alone firms in the
sample. Panels E and F use only diversified firms that reported two business segments or more.
See the Appendix for variable definitions.
Panel A: All Firms—One-Way Sorting on Correlation

High
Variable Low High Minus Low t-Statistic
Q correlation 0.117 0.194 0.078 6.508

Cash flow correlation 0.115 0.195 0.080 6.717
Q–Cash flow correlation 0.191 0.146 −0.045 4.310
Panel B: All Firms—Two-Way Sorting on Correlation and Size
Low Q High Q High
Size Correlation Correlation Minus Low t-Statistic
Small 0.180 0.279 0.099 3.094
Medium 0.138 0.184 0.047 2.145
Large 0.082 0.110 0.029 2.291
Panel C: “Balanced” Sample—One-Way Sorting on Correlation
High
Q correlation 0.117 0.177 0.061 5.021
Panel D: “Balanced” Sample—Two-Way Sorting on Correlation and Size
Low Q High Q High
Small 0.176 0.245 0.069 2.515
Medium 0.123 0.162 0.039 1.757
Large 0.081 0.108 0.027 1.807
Panel E: Diversified Firms—One-Way Sorting on Correlation
High
Q correlation 0.117 0.145 0.028 2.104
Panel F: Diversified Firms—Two-Way Sorting on Correlation and Size
Low Q High Q High
Small 0.169 0.188 0.019 2.091
Medium 0.111 0.132 0.021 1.882
Large 0.080 0.106 0.026 1.877
B. Regression Evidence
The univariate results of the previous subsection suggest that the corre-
lations in investment opportunity and cash flow, as well as the correlation
between investment opportunity and cash flow, have sizable effects on cash
holdings. In this subsection I estimate panel regressions that control for vari-
ous other variables known to affect corporate cash holdings, and I also include
all these measures simultaneously. Hence, these regressions estimate the over-
all effect of the joint risk in investment opportunity and cash flow on cash
holdings, controlling for other determinants of cash holdings.
Table III reports estimates from panel regressions explaining firm-level cash
holdings for each of the three samples (All firms, Balanced sample, Diversified
firms). All specifications are OLS regressions with year fixed effects and robust
standard errors clustered by firm. The regressions do not control for leverage,
payout, and capital expenditures because the three are choice variables that
are jointly determined with cash holdings. However, the Internet Appendix
reports the results for an extended specification that includes leverage, payout,
and capital expenditure as additional controls. In the Internet Appendix I also
adopt an instrumental variables approach to deal with the endogeneity concern.
Panel A considers each diversification measure separately and shows that
when studied in isolation, the correlation in investment opportunity (Tobin’s Q),
the correlation in cash flow (earnings less interest and taxes), and the correla-
tion between cash flow and investment opportunity are all significantly related
to cash holdings after controlling for various other cross-sectional determinants
of cash holdings.
The magnitudes of the effects are nontrivial: Based on the balanced sam-
ple (columns 4 to 6), a one-standard deviation increase in the correlation in
investment opportunity (cash flow) corresponds to an increase of 5.4% (4.2%)
in average cash holdings. A one-standard deviation decrease in the correlation
between investment opportunity and cash flow corresponds to an increase of
7.0% in average cash holdings. These effects are all statistically significant at
the 1% level.
Panel B of Table III studies all diversification measures together, and es-
timates two regression specifications, with and without firm fixed effects, for
each of the three samples. The main takeaway from Panel B is that diversifica-
tion in investment opportunity is an important determinant of cash holdings,
which is both statistically and economically significant: Based on column 3, for
example, a one-standard deviation increase in the correlation in investment
opportunity corresponds to an increase of 4.4% (9.1%) in the cash holdings of
the average (median) firm. These results hold even after controlling for the cor-
relation and volatility in cash flow, and for the correlation between investment
opportunity and cash flow.13
13 For robustness, I also estimate the regressions controlling for firm-level cash flow volatility.
One drawback of including firm-level volatility is that it decreases the sample size substantially
because in many cases companies do not have enough available observations over the past 10
Importantly, the results indicate that diversification mainly affects cash hold-
ings through investment opportunities. The correlation in cash flow is never
statistically significant in Table III once we account for the correlation in in-
vestment opportunity and the correlation between investment opportunity and
cash flow. However, industry cash flow volatility remains an important deter-
minant of cash holdings. Furthermore, diversification in cash flow still affects
cash holdings through the correlation between cash flow and investment op-
portunity, that is, the financing gap. The magnitude of the effect is nontrivial:
Based on column 3, a one-standard deviation decrease in the correlation be-
tween cash flow and investment opportunity corresponds to an increase of 6.8%
in average cash holdings.
One interpretation is that in the context of diversification, cash flow affects
cash holdings mainly through its availability to fund investments (i.e., the cor-
relation between investment opportunity and cash flow, or the financing gap,
discussed in Acharya et al. (2007)). Nevertheless, the weaker results obtained
for the correlation in cash flow and the correlation between investment oppor-
tunity and cash flow may simply be due to the fact that industry cash flow can
be a poor proxy for the availability of internal funds at the divisional level.
In contrast, it is likely that industry Q is a better proxy for division-level in-
vestment opportunities. For example, same-industry firms may have different
past performances but similar future prospects. It may thus not be surprising
that diversification seems to operate mostly through the correlation in invest-
ment opportunity. The above measurement error issue, as well the possibility
that correlations change little over time, may also explain why the effects of
the correlation in investment opportunity and the financing gap decrease after
including firm fixed effects.
Finally, note that the correlation measures completely absorb the effect
of the number of business segments on cash holdings documented by Opler
et al. (1999). Thus, the relation between multidivisionalism and cash is fully
explained by cross-divisional coinsurance and the financing gap. Furthermore,
these findings are inconsistent with alternative hypotheses for the relation
between diversification and cash. If diversified firms were holding less cash
because they had spent it acquiring their divisions (see, for example, Harford
(1999), who finds that cash holdings drive inefficient acquisitions), then the
number of segments and not the correlation measures should be a significant
determinant of cash holdings. Similarly, if diversified firms were holding less
cash because of the “quasi-liquid” nature of noncore segments (as suggested by
years (with a minimum of five observations). This can be seen from Table I, which shows that
the number of firm-level cash flow volatility observations is only 34,613 (compared to 50,905
available observations for the industry-based measures). These results are reported in the Internet
Appendix, and they support the robustness of the findings here. Also, similar statistical and
economic significance obtain with Fama and MacBeth (1973)–based regressions that estimate
separate cross-sectional regressions in each sample year and average the coefficients and standard
errors across years. The Internet Appendix reports these results.
Table III
The Cross-section of Corporate Cash-Holdings
970
This table presents estimates from panel regressions explaining firm-level cash holdings for fiscal years 1990 to 2006. The sample consists of
nonfinancial and nonutility firm-years, with nonmissing data on cash holdings and the industry codes of each business segment, and with total
market capitalization of $10M or more. See the Appendix for variable definitions. All regressions include year fixed effects. Standard errors (in
brackets) are heteroskedasticity consistent and clustered at the firm level.
All Firms “Balanced” Only Diversified

(1) (2) (3) (4) (5) (6) (7) (8) (9)
Panel A: Separate Diversification Measures

Q correlation 0.570∗∗∗ 0.566∗∗∗ 0.534∗∗∗
[0.113] [0.110] [0.108]
Cash flow correlation 0.311∗∗∗ 0.307∗∗∗ 0.280∗∗∗
[0.079] [0.076] [0.075]
Q-CF correlation −0.042∗∗∗ −0.027∗∗∗ −0.023∗∗∗
[0.004] [0.004] [0.005]
Industry Q volatility 0.202∗∗∗ 0.199∗∗∗ 0.192∗∗∗ 0.231∗∗∗ 0.221∗∗∗ 0.216∗∗∗ 0.254∗∗∗ 0.237∗∗∗ 0.243∗∗∗
[0.023] [0.023] [0.022] [0.024] [0.024] [0.024] [0.033] [0.032] [0.033]
Industry cash flow 0.070∗∗∗ 0.071∗∗∗ 0.056∗∗∗ 0.046∗∗∗ 0.050∗∗∗ 0.039∗∗∗ 0.022 0.029∗ 0.026∗
volatility [0.011] [0.011] [0.011] [0.012] [0.012] [0.012] [0.015] [0.015] [0.015]
Cash flow/assets −0.160∗∗∗ −0.160∗∗∗ −0.153∗∗∗ −0.137∗∗∗ −0.137∗∗∗ −0.134∗∗∗ −0.070∗∗∗ −0.070∗∗∗ −0.071∗∗∗
[0.010] [0.010] [0.009] [0.013] [0.013] [0.013] [0.019] [0.019] [0.019]
The Journal of FinanceR
Tobin’s Q 0.069∗∗∗ 0.068∗∗∗ 0.066∗∗∗ 0.065∗∗∗ 0.065∗∗∗ 0.064∗∗∗ 0.050∗∗∗ 0.050∗∗∗ 0.051∗∗∗
[0.002] [0.002] [0.002] [0.002] [0.002] [0.002] [0.004] [0.004] [0.004]
NWC/assets −0.122∗∗∗ −0.121∗∗∗ −0.110∗∗∗ −0.108∗∗∗ −0.107∗∗∗ −0.100∗∗∗ −0.084∗∗∗ −0.083∗∗∗ −0.082∗∗∗
[0.011] [0.011] [0.010] [0.016] [0.016] [0.016] [0.021] [0.021] [0.021]
Number of segments −0.007∗∗∗ −0.008∗∗∗ −0.011∗∗∗ −0.008∗∗∗ −0.009∗∗∗ −0.011∗∗∗ −0.001 −0.002 −0.004∗
[0.002] [0.002] [0.001] [0.002] [0.002] [0.002] [0.002] [0.002] [0.002]
Firm size −0.024∗∗∗ −0.024∗∗∗ −0.024∗∗∗ −0.019∗∗∗ −0.019∗∗∗ −0.019∗∗∗ −0.015∗∗∗ −0.015∗∗∗ −0.015∗∗∗
[0.001] [0.001] [0.001] [0.001] [0.001] [0.001] [0.002] [0.002] [0.002]
Year F.E. Yes Yes Yes Yes Yes Yes Yes Yes Yes
R2 0.33 0.33 0.335 0.325 0.325 0.326 0.276 0.275 0.273
N Obs 49,548 49,548 49,548 19,076 19,076 19,076 11,504 11,504 11,504
(continued)
Table III—Continued
All Firms “Balanced” Only Diversified

Baseline Firm Fixed Effects Baseline Firm Fixed Effects Baseline Firm Fixed Effects
(1) (2) (3) (4) (5) (6)
Panel B: Joint Diversification Measures

Q correlation 0.459∗∗∗ 0.278∗∗ 0.473∗∗∗ 0.281∗∗ 0.456∗∗∗ 0.225∗
[0.124] [0.119] [0.120] [0.130] [0.118] [0.118]
Industry Q volatility 0.192∗∗∗ 0.038∗∗ 0.221∗∗∗ −0.013 0.251∗∗∗ −0.033
[0.022] [0.017] [0.026] [0.031] [0.033] [0.033]
Cash flow correlation 0.103 0.096 0.110 0.050 0.107 0.063
[0.089] [0.073] [0.085] [0.076] [0.082] [0.075]
Industry cash flow volatility 0.056∗∗∗ −0.009 0.034∗∗∗ 0.012 0.025 0.038∗
[0.011] [0.010] [0.012] [0.014] [0.015] [0.021]
Q-CF correlation −0.042∗∗∗ −0.007∗∗ −0.025∗∗∗ −0.006 0.003 −0.003
[0.004] [0.003] [0.005] [0.005] [0.005] [0.005]
Cash flow/assets −0.153∗∗∗ 0.081∗∗∗ −0.138∗∗∗ 0.090∗∗∗ −0.070∗∗∗ 0.089∗∗∗
[0.009] [0.009] [0.013] [0.020] [0.019] [0.020]
Tobin’s Q 0.066∗∗∗ 0.023∗∗∗ 0.062∗∗∗ 0.024∗∗∗ 0.050∗∗∗ 0.019∗∗∗
[0.002] [0.002] [0.002] [0.003] [0.004] [0.004]
NWC/assets −0.111∗∗∗ −0.156∗∗∗ −0.098∗∗∗ −0.144∗∗∗ −0.085∗∗∗ −0.139∗∗∗
[0.010] [0.019] [0.016] [0.043] [0.022] [0.033]
Number of segments −0.001 −0.001 −0.002 −0.001 −0.001 −0.002
[0.002] [0.002] [0.002] [0.002] [0.002] [0.003]
Firm size −0.024∗∗∗ −0.005∗ −0.019∗∗∗ −0.006 −0.015∗∗∗ −0.004
[0.001] [0.003] [0.001] [0.005] [0.002] [0.006]

Year F.E. Yes Yes Yes Yes Yes Yes
Firm F.E. No Yes No Yes No Yes
R2 0.336 0.833 0.320 0.876 0.277 0.850
N Obs 49,548 49,548 19,013 19,013 11,504 11,504
Significance at the 1%, 5%, and 10% levels is represented by ∗∗∗ , ∗∗ , and ∗ , respectively.
971
Opler et al. (1999)), we would expect the number of segments rather than the
correlations to explain cash holdings.14
C. Financial Constraints
As noted above, Modigliani and Miller (1958) point out that in frictionless
capital markets cash has zero net present value. This suggests that the pre-
cautionary saving motive of cash only matters to firms with costly external
financing. Almeida et al. (2004), for instance, show that cash flows and cash
are only correlated when firms are financially constrained, which suggests that
they face high costs of external financing.
The same rationale applies to the relation between cash holdings and diver-
sification in investment opportunity and cash flow. Unconstrained firms that
can tap external capital markets without deadweight costs have no reason
to adjust their cash holdings to the level of diversification. If cash flows are
low and/or investment opportunities are abundant and cannot be financed in-
ternally in full, unconstrained firms can obtain external funds at little or no
cost. However, diversification should be a concern for financially constrained
firms.
To examine the impact of financial constraints, I adopt a similar approach as
that of Almeida et al. (2004), and divide the sample into financially constrained
firms and financially unconstrained firms. However, because diversified firms
are on average larger firms, they might be less financially constrained. This in-
teraction between diversification and the degree of financial constraints might
contaminate the analysis. To deal with this concern, I employ two measures.
First, I repeat the tests across the three different samples (All firms, Balanced
sample, Diversified firms). Second, I estimate the breakpoints that separate
constrained from unconstrained firms based on specialized firms only, and ap-
ply them to each of the three samples. This should neutralize the effect of di-
versified firms on the classification of firms into constrained and unconstrained
firms.
14 The Internet Appendix addresses a number of additional robustness issues. First, I verify
that the correlation-based diversification measures continue to have a significant effect on cash
holdings after a new, improved segment reporting standard (SFAS 131) was issued in 1997 (e.g.,
Berger and Hann (2003)). Next, I make use of the GIM index (Gompers et al. (2003)) to show that
the results continue to hold after controlling for the possibility that diversified firms hold less cash
because they are more prone to agency problems (e.g., Shleifer and Vishny (1989), Grossman and
Hart (1982), Morck, Shleifer, and Vishny (1990), Harford (1999)). Third, I address the endogeneity
of financial and investment decisions by reestimating the regressions with an instrumental vari-
ables approach using lagged variables as instruments, following Almeida, Campello, and Weisbach
(2004) and Fazzari and Petersen (1993). The results indicate that the results here are not sensi-
tive to this change in specification. Finally, I verify the robustness of the diversification measures
when estimated over a 15-year window instead of a 10-year window. I find that the correlation
in investment opportunity and the financing gap remain significant at the 10% level or better.
Their magnitudes, however, decrease substantially, possibly because the estimation is based on
investment opportunities and cash flows that are further away in time.
For robustness, I use five different measures of financial constraints: (i) the fi-
nancial constrains index of Whited and Wu (2006),15 (ii) firm size (Gilchrist and
Himmelberg (1995)), (iii) payout ratio (Fazzari, Hubbard, and Petersen (1988)),
(iv) bond ratings (e.g., Whited (1992), Kashyap, Lamont, and Stein (1994)), (v)
commercial paper ratings (Calomiris, Himmelberg, and Wachtel (1995)).16 In
all cases, the annual median value of each measure, across specialized firms
only, is used as the cutoff point between unconstrained and constrained firms.
Each panel in Table IV corresponds to a different measure of financial con-
straints and estimates the previous cash regressions (see Table III) separately
for financially unconstrained and constrained firms. For brevity, I only report
the coefficients on the correlation-based measures of diversification. Table IV
reports the results for the balanced sample; the Internet Appendix reports the
results for the full sample and the diversified sample.17
The results in Table IV suggest that, consistent with the theory, cash holdings
are mostly sensitive to diversification when firms are financially constrained.
Strikingly, across all five measures of financial constraints, the coefficients on
the diversification measures are always more pronounced in constrained firms.
The economic magnitudes of these effects are nontrivial. For example, based
on the Whited and Wu (2006) index, a one-standard deviation increase in the
correlation in investment opportunity corresponds to an increase of 2.5% in
average cash holdings of unconstrained firms, and 4.6% in cash holdings of
constrained firms. For the correlation in cash flow and the correlation between
investment opportunities and cash flows, the effects are statistically insignifi-
cant in unconstrained firms. For constrained firms, an increase of one standard
deviation in the correlation in cash flow corresponds to an increase of 2.0% in
average cash holdings, and a decrease of one standard deviation in the correla-
tion between investment opportunity and cash flow corresponds to an increase
of 11.8% in cash.
Importantly, the effect of the correlation between investment opportunity
and cash flow (i.e., the financing gap) on cash holdings is always statistically
significant at the 1% level for financially constrained firms. It is reassuring to
find that consistent with the theory, the financing gap significantly affects cash
holdings in the constrained sample even though it may not be robust for the
entire sample (see Table III). The overall picture portrayed by Table IV suggests
that the joint uncertainty of investment opportunity and cash flow significantly
affects cash holdings, mainly in financially constrained firms. More specifically,
greater diversification in investment opportunity and a smaller financing gap
15 The
Whited and Wu (2006) index is defined as:
WW index = −0.091 × CashFlow − 0.062 × DividendDummy + 0.021 × LongTermDebt
− 0.044 × Size + 0.102 × IndustrySalesGrowth − 0.035 × SalesGrowth.
16 Recently, Campello, Graham, and Harvey (2009) use a more direct measure of financial con-
straints based on a survey of CFOs in the United States, Europe, and Asia. They find a more severe
effect of the 2008 credit crisis on constrained firms.
17 Table IV (A2) shows that even based on the conservative classification method into constrained
and unconstrained firms, quite a few diversified firms are classified as financially constrained.
Based on the Whited and Wu (2006) index, for example, 3,499 firm-year observations are classified
as financially constrained, and 7,673 firm-year observations are classified as unconstrained.
Table IV
Financially Constrained vs. Unconstrained Firms (Balanced Sample)
This table presents estimates from panel regressions explaining firm-level cash holdings for fiscal
years 1990 to 2006. The baseline sample consists of nonfinancial and nonutility firm-years with
nonmissing data on cash holdings and the industry codes of each business segment, and with total
market capitalization of $10M or more. The regressions are estimated separately for financially
constrained and financially unconstrained firms in a balanced sample consisting of all multiseg-
ment firms and 20% of the stand-alone firms in the baseline sample. Financial constraints are
measured based on: (i) the Whited and Wu (2006) financial constraints index, (ii) firm size, (iii)
dividend payouts and stock repurchases, (iv) commercial paper ratings, and (v) bond ratings. The
table reports regression coefficients estimated from a full specification regression similar to the re-
gression specification in Table III. Variable definitions are given in the Appendix. Standard errors
(in brackets) are heteroskedasticity consistent and clustered at the firm level.
Unconstrained Constrained Difference
Panel A: Financial Constraints Measured by the Whited and Wu (2006) Index

Q correlation 0.235∗∗ 0.358∗∗ 0.123∗∗
[0.136] [0.125] [0.054]
CF correlation 0.092 0.211 0.119
[0.089] [0.172] [0.118]
Q-CF correlation 0.008∗ −0.057∗∗∗ −0.065∗∗∗
[0.005] [0.008] [0.007]
N Obs 11,232 6,925
Panel B: Financial Constraints Measured by Firm Size
Q Correlation 0.391∗∗∗ 0.436∗ 0.045∗
[0.136] [0.237] [0.026]
CF correlation −0.013 0.421∗∗ 0.434∗∗
[0.086] [0.177] [0.211]
Q-CF correlation −0.002 −0.072∗∗∗ −0.070∗∗∗
[0.005] [0.009] [0.008]
N Obs 13,118 5,958
Panel C: Financial Constraints Measured by Shareholder Payouts
Q correlation 0.291∗∗ 0.503∗∗ 0.212∗∗∗
[0.145] [0.201] [0.041]
CF correlation 0.158∗ 0.188 0.030
[0.092] [0.152] [0.127]
Q-CF correlation 0.008∗ −0.049∗∗∗ −0.057∗∗∗
[0.005] [0.007] [0.008]
N Obs 9,014 8,630
Panel D: Financial Constraints Measured by Commercial Paper Ratings
Q correlation 0.171 0.525∗∗∗ 0.354∗∗∗
[0.133] [0.161] [0.111]
CF correlation −0.031 0.227∗ 0.258∗
[0.100] [0.116] [0.135]
Q-CF correlation 0.011∗∗ −0.040∗∗∗ −0.051∗∗∗
[0.005] [0.006] [0.007]
N Obs 14,263 4,813
(continued)
Table IV—Continued
Unconstrained Constrained Difference
Panel E: Financial Constraints Measured by Bond Ratings

Q correlation 0.314 0.463∗∗∗ 0.149∗∗∗
[0.239] [0.124] [0.041]
CF correlation −0.240 0.146∗ 0.386∗
[0.164] [0.089] [0.199]
Q-CF correlation 0.018∗∗∗ −0.029∗∗∗ −0.047∗∗∗
[0.007] [0.005] [0.009]
N Obs 13,884 5,192
between investment opportunity and cash flow push firms to hold less cash.
Diversification in cash flow, however, does not affect cash holdings directly after
controlling for diversification in investment opportunity and the financing gap.
Finally, similar results also obtain by employing a Fama–MacBeth-based
regression approach. These results are reported in the Internet Appendix. Fur-
thermore, a similar qualitative pattern, with smaller magnitudes, holds when
the regressions are estimated with firm fixed effects. These results are also
reported in the Internet Appendix.
IV. Implications and Extensions

A. Internal Capital Markets
The findings above suggest that diversified firms can hold less cash because
diversification reduces the ex-ante probability of financing shortages that might
lead to underinvestment. One way in which diversified firms can finance in-
vestments without resorting to their cash reserves is cash flow transfers from
divisions without investments to divisions with investments. Thus, diversifi-
cation in investment opportunity might facilitate transfers across divisions,
which, in turn, reduce the demand for precautionary cash.
The literature on internal capital markets shows that one potential cost
to being diversified is mismanagement of internal cash flows, which are not
always allocated to high-growth divisions (see, for example, Shin and Stulz
(1998), Rajan et al. (2000)). While the overall efficiency of internal capital
markets is beyond the scope of this paper, this subsection examines whether
active internal capital markets and cross-divisional transfers reduce the firm’s
demand for cash. The evidence presented below suggests that holding less
cash is associated with efficient cross-divisional transfers to high-productivity
divisions. Furthermore, the reduction in cash holdings due to diversification
in investment opportunity and cash flow is shown to be partially driven by
efficient internal capital market allocations.
To study the role of internal capital markets, I examine the relation between
cross-divisional correlations, transfers, and cash. One potential problem, how-
ever, is that cross-divisional transfers cannot be observed directly. To overcome
this difficulty, I follow Rajan et al. (2000) and measure transfers by the dif-
ference between the investment a segment makes when it is part of a diver-
sified firm and the investment it would have made had it been on its own.
Note that this measure is based on the assumption that transfers of funds
across divisions correspond to changes in divisional investments. The invest-
ment a segment would have made on its own is approximated by the weighted
average of the ratio of capital expenditures to assets of single-segment firms
in the same industry (defined by three-digit NAICS codes). As noted by Rajan
et al. (2000), it is possible that diversified firms have more funds, perhaps be-
cause their cost of capital is lower. One therefore needs to further subtract the
average industry-adjusted capital expenditure-to-assets ratio averaged across
the segments of the firm. Thus, cross-divisional transfers are measured as

CAPEXj CAPEXjss N
CAPEXj CAPEXjss
− − Wj − , (10)
Assetsj Assetsss
j Assetsj Assetsss
j
j=1
where j = 1 . . . N denotes segment j, ss refers to single-segment firms and wj is

segment j’s share of total firm assets.
The first four columns of Table V estimate panel regressions explaining firm-
level cash holdings of diversified firms, augmenting the regression models in
Table III with various measures of cross-divisional transfers and efficiency.
Column 1 of Table V examines the relation between cash holdings and overall
ICM activity level, as measured by the sum of the absolute values of divisional
transfers (see equation (10)) across all divisions. My hypothesis suggests that
higher levels of ICM activity should reduce cash holdings. The results in column
1 suggest that cash holdings are indeed negatively related to the overall amount
of cross-divisional transfers. This implies that more active internal capital
markets, with greater amounts of fund transfers across divisions, facilitate
lower cash holdings.
A natural question that arises is whether these transfers are efficient trans-
fers from low-productivity to high-productivity divisions, or inefficient trans-
fers from high-productivity divisions to low-productivity divisions. Rajan et al.
(2000) find that diversity in divisional investment opportunity is associated
with inefficient transfers. The question addressed here is more nuanced: Is
the reduction in cash due to cross-divisional transfers driven by efficient or
inefficient transfers?
To answer this question, I further distinguish between efficient and in-
efficient transfers by classifying divisions into low-productivity and high-
productivity divisions based on whether their industry Q is lower or higher than
the firm weighted-average Q. If it is higher, the division is classified as a high-
productivity division. If it is lower, it is classified as a low-productivity division.
For each company, I sum the transfers made to high- and low-productivity firms
and examine their relation to cash holdings. The second column in Table V re-
ports these results. It shows that inefficient transfers to low-productivity firms
are not significantly related to cash. However, efficient transfers are signifi-
cantly negatively related to cash holdings. This suggests that efficient transfers
Table V
Internal Capital Markets
This table presents evidence from panel regressions on the relation between cash holdings and cross-divisional transfers. The sample consists of
multidivision, nonfinancial, and nonutility firm-years with nonmissing data on cash holdings and the industry codes of each business segment, and
with total market capitalization of $10M or more. Columns 1 through 4 estimate regressions explaining firm-level cash holdings, whereas columns 5
through 8 estimate regressions explaining firm-level reduction in cash due to diversification, that is, the reduction in cash holdings explained by the
company’s diversification in investment opportunity, cash flow, and the correlation between investment opportunity and cash flow. In columns 1 and
5, total transfers is the sum of the absolute value of fund transfers across divisions, where fund transfers are measured as in Rajan et al. (2000):
N

CAPEXj CAPEXjss CAPEXj CAPEXjss
− − Wj − ,
Assetsj Assetsss
j Assetsj Assetsss
j
j=1
where j = 1 . . . N denotes segment j, ss refers to single-segment firms, and wj is segment j’s share of total firm assets.
In columns 2 and 6, the efficiency of the cross-divisional transfers is measured by summing all transfers to high-productivity and low-productivity
divisions, where a division is classified as high (low) productivity if its average industry Tobin’s Q is higher (lower) than the firm-weighted Tobin’s Q.
In columns 3 and 7, the efficiency of the cross-divisional transfers is measured by the value added, defined by Rajan et al. (2000) as
⎛ ⎞
N
N
CAPEXj CAPEXjss CAPEXj CAPEXjss
Assetsj (Qj − Q̄) ⎝ − − Wj − ⎠
Assetsj Assetsss
j Assetsj Assetsss
j
j=1 j=1
,
Total Assets
where j = 1 . . . N denotes segment j, ss refers to single-segment firms, and wj is segment j’s share of total firm assets.
In columns 4 and 8, the efficiency of the transfer is measured by the absolute value added by cross-divisional transfers, defined by Rajan et al. (2000)
as
N

CAPEXj CAPEXjss
Assetsj (Qj − 1) −
Assetsj Assetsss
j
j=1
,
Total Assets
where j = 1 . . . N denotes segment j and ss refers to single-segment firms.

All other variables are defined in the Appendix. All regressions include year fixed effects. Standard errors (in brackets) are heteroskedasticity
consistent and clustered at the firm level.
(continued)
977
978
Table V—Continued
Cash/Assets Reduction in Cash Due to Diversification

(1) (2) (3) (4) (5) (6) (7) (8)
Total transfers −0.173∗∗∗ −0.011∗∗∗

[0.029] [0.002]
Transfers to low- −0.002 −0.008∗
productivity divisions [0.054] [0.004]
Transfers to high- −0.088∗∗ 0.015∗∗∗
productivity divisions [0.041] [0.004]
Value added −1.474∗∗∗ 0.069∗∗
[0.416] [0.030]
Absolute value added −0.599∗∗∗ 0.018∗∗
[0.119] [0.008]
Q correlation 0.413∗∗∗ 0.457∗∗∗ 0.454∗∗∗ 0.438∗∗∗
[0.116] [0.117] [0.118] [0.117]
Avg. industry Q volatility 0.224∗∗∗ 0.231∗∗∗ 0.231∗∗∗ 0.230∗∗∗ 0.003 0.003 0.004 0.004
[0.033] [0.033] [0.033] [0.033] [0.002] [0.002] [0.002] [0.002]
[0.079] [0.079] [0.079] [0.079]

Avg. industry cash 0.030∗∗ 0.032∗∗ 0.032∗∗ 0.030∗∗ 0.001 0.001 0.001 0.001
flow volatility [0.015] [0.015] [0.015] [0.015] [0.001] [0.001] [0.001] [0.001]
Q–cash flow correlation 0.003 0.003 0.004 0.004
[0.005] [0.005] [0.005] [0.005]
Control variables Yes Yes Yes Yes Yes Yes Yes Yes
R2 0.298 0.293 0.293 0.295 0.049 0.047 0.044 0.044
N Obs 11,498 11,498 11,498 11,498 11,498 11,498 11,498 11,498
Significance at the 1%, 5%, and 10% levels is represented by ∗∗∗ , ∗∗ , and ∗, respectively.
facilitate lower cash holdings, while inefficient transfers are not significantly
related to cash holdings. Thus, efficient transfers allow firms to economize on
their cash holdings.
The magnitude of these effects is economically significant. Based on column
1, a one-standard deviation increase in the amount of total cross-divisional
transfers is associated with a reduction of 9.5% in average cash holdings. Based
on column 2, a one-standard deviation increase in the amount of efficient cross-
divisional transfers, from high-productivity to low-productivity divisions, is
associated with a reduction of 6.2% in average cash holdings. These effects are
statistically significant at the 5% level or better.
Columns 3 and 4 use two alternative measures of internal capital mar-
kets’ efficiency. Column 3 examines the relation between cash holdings and
the value added by cross-divisional transfers, defined by Rajan et al. (2000)
as
⎛ ⎞
N
CAPEXj CAPEXj ss
N
CAPEXj CAPEXjss
Assetsj (Q j − Q̄) ⎝ − − Wj − ⎠
Assetsj Assetsss
j Assetsj Assetsss
j
j=1 j=1
.
Total Assets
(11)
In column 4, the efficiency of the transfer is measured by the absolute value
added by cross-divisional transfers, defined by Rajan et al. (2000) as

N
CAPEXj CAPEXjss
Assetsj (Q j − 1) −
Assetsj Assetsss
j
j=1
. (12)
Total Assets
In both columns, cash appears to be reduced by value-enhancing allocations.

This suggests that efficient transfers across divisions facilitate cash reductions.
Note that these results are not inconsistent with the overall inefficiency of
diversity in Q shown in Rajan et al. (2000). In their case, the inefficiency
resulted from other dimensions of the corporation, unrelated to its cash policy.
Reductions in cash, however, appear to be associated with efficient transfers
from low-Q divisions to high-Q divisions.
In columns 5 to 8, I specifically examine whether the reduction in cash due
to diversification is associated with the efficient workings of internal capital
markets. This line of investigation tries to underpin the effect of diversification
in investment opportunity on cash holdings. It examines whether companies
that choose to hold less cash due to diversification ex ante, make ex post in-
vestments by utilizing their internal capital market. The dependent variable in
columns 5 to 8 is the reduction in cash holdings implied by the company’s diver-
sification. This measure is calculated as the difference between the predicted
cash holdings from the previous regression models (see Table III), estimated
without the diversification measures, and the predicted cash holdings from the
previous regression model in Table III that include diversification. A bigger
difference between the two implies a larger reduction in cash holdings due to
diversification.
The results presented in columns 5 to 8 indicate that the reduction in cash
due to diversification is associated with more active (column 5) and efficient
(columns 6 to 8) internal capital markets. The reduction in cash is larger when
the company makes more cross-divisional transfers, and when these transfers
are efficient transfers from low-Q divisions to high-Q divisions.
Thus, diversification in investment opportunity across divisions, which leads
to reduced cash holdings, does imply that companies utilize their internal cap-
ital markets through efficient transfers to make investments.
B. Efficiency and Governance

The evidence presented so far suggests that diversification allows firms to
hold less cash. Thus, given the costs associated with holding cash, the ability
to hold less cash due to diversification should positively affect firm value.
To examine the efficiency of the cash-diversification relation, I test whether
the relation between cash and diversification is stronger in well-governed firms,
where there is reason to expect efficient considerations of cash management
to play a stronger role. Table VI reports the coefficients on the correlation-
based measures of diversification from the previous cash regressions, estimated
separately for poorly governed and well-governed firms in the balanced sample.
Four different measures of governance are employed, and each panel in
Table VI corresponds to a different measure of governance. The governance
measures include: (i) the GIM index, (ii) the total percentage of shares held
by institutional investors, (iii) the number of blockholders that hold 5% or
more of the company’s shares, and (iv) the percentage of shares held by the
largest institutional shareholder. These measures are collectively examined in
Gompers, Ishii, and Metrick (2003), Cremers and Nair (2005), and Bebchuk,
Cohen, and Ferrell (2009).
Let us start with Panel A, where governance is measured by the GIM index.
The results suggest that the correlations in investment opportunity and cash
flow have a stronger impact on cash holdings in well-governed firms. Further-
more, the differences between poorly governed firms and well-governed firms
are highly significant at the 5% level or better. The magnitudes of the effects
are significant as well: A one-standard deviation increase in the correlation in
investment opportunity (cash flow) corresponds to an increase of 2.7% (1.3%)
in average cash holdings of poorly governed firms, and 5.6% (4.2%) in well-
governed firms.
Panels B, C, and D employ the three alternative measures of governance.
The results are generally consistent across all measures of governance: The
cross-divisional correlations in investment opportunity and cash flow have a
significantly stronger effect on the cash holdings of well-governed firms. Fur-
thermore, with the exception of the GIM index, the effect of the financing gap is
also significantly stronger for well-governed firms. For example, based on Panel
B, the effect of the financing gap is insignificant in poorly governed firms. In
Table VI
Governance (Balanced Sample)
This table presents estimates from panel regressions explaining firm-level cash holdings for fiscal
years 1990 to 2006. The baseline sample consists of nonfinancial and nonutility firm-years with
nonmissing data on cash holdings and the industry codes of each business segment, and with
total market capitalization of $10M or more. The table reports coefficients from the extended
specification in Table III, estimated in a “balanced” sample of all diversified firms and 20% of
the single-segment firms chosen randomly. The regressions are estimated separately for poorly
governed and well-governed firms. Governance is measured based on: (i) the Gompers et al. (2003)
(GIM) governance index, (ii) the percentage of shares held by institutional investors, (iii) the
number of institutional investors that hold 5% or more, and (iv) the maximal percentage of shares
held by a single institutional investor. All regressions include year fixed effects. Standard errors
(in brackets) are heteroskedasticity consistent and clustered at the firm level.
Poor Governance Good Governance Difference
Panel A: Governance Measured by the GIM Index

Q correlation 0.494∗∗ 0.814∗∗∗ 0.320∗∗∗
[0.213] [0.214] [0.084]
CF correlation 0.180 0.336∗∗ 0.156∗∗
[0.159] [0.154] [0.076]
Q-CF correlation 0.002 0.004 0.002
[0.007] [0.010] [0.011]
N Obs 2,907 2,984
Panel B: Governance Measured by Total Shares (in %) Held by Institutional Investors
Q correlation 0.410∗ 0.650∗∗∗ 0.240∗∗∗
[0.223] [0.164] [0.071]
CF correlation 0.139 0.283∗∗ 0.144∗
[0.165] [0.110] [0.0078]
Q-CF correlation −0.003 −0.018∗∗ −0.015
[0.006] [0.009] [0.006]
N Obs 5,022 6,292
Panel C: Governance Measured by Number of Block Holders (5% or More)
Q correlation 0.563∗∗∗ 0.556∗∗∗ −0.007
[0.202] [0.156] [0.136]
CF correlation 0.196 0.231∗∗ 0.035
[0.133] [0.114] [0.153]
Q-CF correlation −0.01 −0.011∗ −0.001
[0.008] [0.006] [0.009]
N Obs 5,364 5,950
Panel D: Governance Measured by Maximal Block Holder
Q correlation 0.474∗∗ 0.627∗∗∗ 0.153∗∗
[0.199] [0.164] [0.074]
CF correlation 0.068 0.361∗∗∗ 0.293∗∗∗
[0.130] [0.130] [0.088]
Q-CF correlation −0.007 −0.014∗∗ −0.007
[0.008] [0.007] [0.007]
N Obs 5,272 6,017
well-governed firms, a decrease of one standard deviation in the correlation

between investment opportunities and cash flows corresponds to an increase
of 8.6% in average cash holdings.
These results suggest that diversification has a bigger effect on cash hold-
ings in well-governed firms. Because managers of well-governed firms are more
likely to act efficiently, these results suggest that it is efficient for firms to
hold less cash when they are more diversified. The Internet Appendix repeats
the above analysis in the two remaining samples of all firms and diversified
firms. In the Internet Appendix, I also augment the specification with firm
fixed effects and repeat the analysis estimating Fama–MacBeth-based regres-
sions in the balanced sample. The results are consistent with the results in
Table VI.
C. Acquisitions
Recall that the results in Table III suggest that the magnitude of the ef-
fects and their statistical significance are smaller when firm fixed effects are
introduced into the regression estimation. To test directly whether changes
in diversification affect cash holdings, I next examine how acquisitions affect
cash holdings. Such an approach has the additional advantage of mitigating
concerns about inaccurate or strategic reporting of segment data (e.g., Villa-
longa (2004b)).
To directly test whether diversification is driving firms to hold less cash, I
examine whether firms reduce their cash holdings by engaging in diversifica-
tion. Table VII studies how corporate acquisitions affect postacquisition cash
holdings, taking into account preacquisition cash holding, acquisition charac-
teristics (value, method of payment), and other determinants of cash choices.
This approach allows me to distinguish between acquisitions per se and their
degree of diversification, and to accommodate slow adjustment rates of cash to
the degree of diversification. A similar longitudinal approach is taken by other
papers such as Graham et al. (2002), and Hyland and Diltz (2002), who study
the effect of diversification on firm value.
I obtain data on mergers and acquisitions from the Securities Data Corpo-
ration (SDC) and include all completed acquisitions of public and private com-
panies, as well as subsidiaries. Given that acquisition activity might generate
short-term (transitory) fluctuations in cash holdings pre- and postacquisition,
I consider changes in cash holdings over the 1- to 3-year period following the
acquisition. Table VII reports estimates from panel regressions explaining the
postacquisition cash holdings of acquiring firms. The table considers three sub-
samples from 1990 to 2006: all acquiring firms, diversified acquiring firms, and
stand-alone acquiring firms. Three regression specifications are considered for
each subsample. The specifications differ in the postacquisition period being
considered: They consider cash holdings 1 to 3 years after the acquisition.
For each specification, I exclude companies that made additional acquisitions
during the relevant postacquisition period because this would change the com-
pany’s postacquisition degree of diversification. All specifications control for
Table VII
Acquisitions
This table reports estimates from panel regressions explaining acquiring firms’ postacquisition cash holdings over the 1 to 3 years following the
acquisition. The sample consists of all firms that completed an acquisition between 1990 and 2006 based on the SDC database. Postacquisition cash is
averaged over the 1 to 3 years following the acquisition, and preacquisition cash is estimated over the 1 to 3 years prior to the acquisition. Acquisition
Q correlation is the correlation in Tobin’s Q between the acquirer and the target companies over the 10 years before the acquisition. Acquisition cash
flow correlation is defined analogously with respect to cash flows. The cash acquisition dummy is set to one if the acquisition’s method of payment is
cash, and to zero otherwise. Log(deal value) is the logarithm of the total value of the acquisition. Other control variables are defined in the Appendix
and are estimated over the subsequent 1 to 3 years as well. The first three columns correspond to all acquiring firms, whereas columns 4 through
6 correspond to multisegment acquiring firms only. The last three columns correspond to single-segment acquiring firms only. Standard errors (in
brackets) are heteroskedasticity consistent.
All Acquirers Only Diversified Acquirers Only Stand-Alone Acquirers

1 Year 2 Years 3 Years 1 Year 2 Years 3 Years 1 Year 2 Years 3 Years
(1) (2) (3) (4) (5) (6) (7) (8) (9)
Preacquisition cash 0.797∗∗∗ 0.690∗∗∗ 0.636∗∗∗ 0.756∗∗∗ 0.654∗∗∗ 0.653∗∗∗ 0.810∗∗∗ 0.702∗∗∗ 0.634∗∗∗
[0.008] [0.011] [0.014] [0.013] [0.019] [0.024] [0.010] [0.014] [0.018]
Acquisition Q correlation 0.074∗∗ 0.197∗∗∗ 0.185∗∗∗ 0.085∗ 0.097∗∗ 0.137∗∗∗ 0.070 0.311∗∗∗ 0.305∗∗∗
[0.035] [0.050] [0.062] [0.046] [0.045] [0.043] [0.051] [0.073] [0.092]
Acquisition cash flow correlation 0.051∗ −0.015 0.011 0.025 0.105∗ 0.158∗∗ 0.065 −0.116∗∗ −0.116
[0.028] [0.041] [0.051] [0.039] [0.056] [0.073] [0.040] [0.058] [0.071]
Cash acquisition dummy 0.004 0.005 0.007 0.002 0.007 0.010 0.006∗ 0.003 0.004
[0.002] [0.003] [0.004] [0.003] [0.005] [0.006] [0.003] [0.005] [0.006]
Log (deal value) 0.003∗∗∗ 0.007∗∗∗ 0.009∗∗∗ 0.002 0.003 0.009∗∗∗ 0.005∗∗∗ 0.010∗∗∗ 0.010∗∗∗
[0.001] [0.001] [0.002] [0.001] [0.002] [0.003] [0.001] [0.002] [0.002]
Cash flow/assets −0.057∗∗∗ −0.105∗∗∗ −0.142∗∗∗ −0.064∗∗∗ −0.121∗∗∗ −0.175∗∗∗ −0.054∗∗∗ −0.098∗∗∗ −0.132∗∗∗
[0.007] [0.010] [0.013] [0.011] [0.018] [0.023] [0.009] [0.013] [0.016]
Tobin’s Q 0.003∗∗ 0.005∗∗∗ 0.005∗∗ 0.005∗∗ 0.007∗∗ 0.008∗∗ 0.002 0.004 0.005
[0.001] [0.002] [0.002] [0.002] [0.003] [0.004] [0.002] [0.002] [0.003]
Book leverage −0.064∗∗∗ −0.091∗∗∗ −0.121∗∗∗ −0.068∗∗∗ −0.084∗∗∗ −0.139∗∗∗ −0.062∗∗∗ −0.094∗∗∗ −0.109∗∗∗
[0.006] [0.008] [0.010] [0.009] [0.013] [0.016] [0.008] [0.011] [0.013]
Payout/assets −0.019 −0.027 −0.038 −0.021 −0.045∗ −0.067∗∗ −0.012 0.035 0.071
[0.016] [0.023] [0.028] [0.017] [0.024] [0.028] [0.035] [0.054] [0.075]
(continued)
983
984
Table VII—Continued
All Acquirers Only Diversified Acquirers Only Stand-Alone Acquirers

1 Year 2 Years 3 Years 1 Year 2 Years 3 Years 1 Year 2 Years 3 Years
(1) (2) (3) (4) (5) (6) (7) (8) (9)
CAPEX/assets −0.066∗∗∗ −0.083∗∗∗ −0.070∗∗∗ −0.063∗∗ −0.106∗∗∗ 0.000 −0.071∗∗∗ −0.071∗∗∗ −0.100∗∗∗
[0.015] [0.020] [0.025] [0.025] [0.035] [0.045] [0.018] [0.025] [0.031]
NWC/assets −0.026∗∗∗ −0.028∗∗∗ −0.024∗ −0.006 −0.023 −0.011 −0.034∗∗∗ −0.030∗∗ −0.028∗
[0.007] [0.010] [0.013] [0.011] [0.017] [0.023] [0.009] [0.013] [0.017]
Number of segments 0.001 0.001 0.001 0.002 −0.002 −0.002 −0.002 −0.005 −0.002
[0.002] [0.002] [0.002] [0.001] [0.002] [0.002] [0.002] [0.003] [0.004]
Firm size −0.001 −0.003∗ −0.005∗∗ 0.000 0.000 −0.003 −0.002 −0.005∗∗ −0.005∗
[0.001] [0.002] [0.002] [0.002] [0.002] [0.003] [0.002] [0.002] [0.003]
R2 0.747 0.621 0.567 0.676 0.537 0.520 0.766 0.646 0.587

N Obs 7,302 5,457 4,118 2,895 2,156 1,628 4,407 3,301 2,490
both whether the acquisition was paid with cash and the acquisition value, as
well as the usual set of control variables, recorded 1, 2, and 3 years after the
acquisition.
The acquisition’s degree of diversification is measured by the correlations
in investment opportunity and cash flow between the acquiring and target
firms, measured by the correlation between their industry average investment
opportunities (Tobin’s Q) and cash flows over the 10-year period prior to the
acquisition.
The main result in Table VII is that a higher correlation in investment
opportunity between the acquiring firm and the target firm leads to higher
cash balances following the acquisition. Three years after the acquisition, the
results are statistically significant at the 1% level and the magnitudes of the
effects are nontrivial: An increase of one standard deviation in the correlation
in investment opportunity between the acquirer and the target corresponds
to an increase of 6.4% in cash holdings in the sample of all firms (column 3),
4.3% in the sample of diversified firms (column 6), and 8.0% in the sample of
single-segment firms (column 9).
There are two other notable patterns worth mentioning. First, note that
the effects are bigger in specialized acquiring firms than they are in diver-
sified acquiring firms. These results are consistent with the results in Table
III, where the magnitude of the effects is significantly bigger when specialized
firms are included in the analysis. These results are due to the nonlinearity
in the changes in diversification: The decrease in correlations due to the diver-
sification of a previously specialized firm is on average much larger than the
decrease in correlation due to the diversification of an already-diversified firm.
The second notable result is that the effects are generally weaker and less
significant when we consider the immediate year subsequent to the acquisition.
The effects become significantly stronger after 2 years and more so after 3 years,
consistent with a slow, noninstantaneous adjustment of cash to changes in the
degree of diversification.
D. Why Do Diversified Firms Hold More Cash Than They Used To?
The evidence presented thus far supports the notion that higher degrees of
diversification correspond to lower ratios of cash to total assets. However, this
does not explain why diversified firms tend to hold more cash than they used to.
As Figure 1 shows, the average cash-to-assets ratio has increased dramatically
from 1990 to 2006 in diversified firms (with two business segments or more).
BKS show that increasing cash flow volatility is one of the main driving forces
behind the increase in cash balances. Indeed, this is consistent with a number
of recent studies (e.g., Irvine and Pontiff (2009)) that document a market-wide
increase in idiosyncratic risk and cash flow volatility. Thus, the precaution-
ary demand of firms for cash reserves is greater due to increasing levels of
risk.
Consistent with the findings in BKS, Panel A of Figure 2 shows that the
average cash flow volatility has also increased during the same period. Panel B
Panel A: Average Cash Flow Volatility Panel B: Average Investment

Opportunity Volatility
8
.12
7
.1
%
6
.08
%
.06
5
.04
4
1990 1995 2000 2006 1990 1995 2000 2006

Fiscal Year Fiscal Year
Panel C: Average Correlation Panel D: Average Correlation

in Cash Flow in Investment Opportunity
-1.1 -1 -.9 -.8 -.7 -.6
-3
-3.5
%
%
-4
-4.5
-5
1990 1995 2000 2006 1990 1995 2000 2006

Fiscal Year Fiscal Year
Figure 2. Average volatility and correlation in investment opportunity and cash flow.
This figure plots the average annual volatility and correlation in investment opportunity and cash
flow for all diversified firms in the sample, which consists of nonfinancial and nonutility firm-years
from 1990 to 2006, with nonmissing data on cash holdings and on the industry codes of each
business segment, and with total market capitalization of $10M or more. Diversified firms are
firms that reported two business segments or more on Compustat. See the Appendix for variable
definitions.
of Figure 2 shows further that the increase in average cash holdings has been
accompanied by an increase in investment opportunity volatility. Given the
previous finding of a positive relation between cash holdings and investment
opportunity volatility, this suggests that the time trend in cash holdings is
related to both cash flow and investment opportunity risks.
Interestingly, Panels C and D show that the cross-divisional correlations
in cash flow and investment opportunity have also increased from 1990 to
2006. Thus, the increase in cash holdings has been accompanied not only by an
increase in volatilities, but also by an increase in the correlations in investment
opportunity and cash flow. It is important to note that these are two different
effects. The volatility measures employed here are sales-weighted industry-
level volatilities that do not incorporate cross-divisional diversification. Thus,
they are very similar to the cash flow volatility measure used by BKS, which
is also an industry-level measure of volatility. However, the results here show
that the degree of diversification has decreased from 1990 to 2006, which might
further explain why diversified firms hold more cash.
Table VIII
The Time Series of Aggregate Cash Holdings in Diversified Firms
This table presents evidence from time-series regressions explaining annual average corporate cash
holdings. The sample consists of multidivision, nonfinancial, and nonutility firms with nonmissing
data on cash holdings and the industry codes of each business segment, and with total market
capitalization of $10M or more. Independent variables include annual averages of correlations
and volatilities in investment opportunity and cash flow, as well as the annual average correlation
between investment opportunity and cash flow. Additional control variables include annual average
cash flow/assets, Tobin’s Q, the number of business segments, and firm size (see the Appendix for
variable definitions).
(1) (2) (3) (4) (5)
Q correlation 9.506∗∗
[3.428]
Cash flow correlation 8.134∗∗∗
[1.332]
Avg. industry Q volatility 0.653∗∗∗
[0.125]
Avg. industry cash flow volatility 0.313∗∗∗
[0.024]
Q–Cash flow correlation 0.133∗∗
[0.057]
Cash flow/assets −1.006∗∗∗ 0.056 −0.412 −0.208∗ −0.915∗∗
[0.283] [0.255] [0.235] [0.111] [0.308]
Tobin’s Q 0.133∗∗ 0.076∗∗ 0.068 0.057∗∗∗ 0.139∗∗
[0.050] [0.034] [0.040] [0.017] [0.055]
Number of segments 0.399∗∗ −0.019 0.214∗ 0.07 0.391∗∗
[0.158] [0.119] [0.116] [0.056] [0.169]
Firm size 0.031 0.090∗ −0.003 −0.016 −0.09
[0.066] [0.043] [0.045] [0.020] [0.075]
R2 0.904 0.963 0.953 0.990 0.891
N Obs 17 17 17 17 17
The time-series relation between cash and investment opportunity/cash flow

risk is further examined in Table VIII. This table estimates time-series regres-
sions explaining annual average corporate cash holdings from 1990 to 2006.
The results suggest that the above increases in correlation and volatility in
investment opportunity and cash flow are significantly positively related to the
upward time trend in cash holdings.
These effects are economically significant. For example, according to column
1, the increase in average Q correlation from 1990 to 2006 (from −1.8% to
−0.9%) corresponds to an increase of 8.2% in average cash holdings. Further-
more, these effects are statistically significant at the 5% level or better.
Overall, these findings suggest that multidivisional firms are less diversified
in their investment opportunities and cash flows than in the past, and therefore
do not enjoy the benefits of investment and cash flow coinsurance as much
as before. These results are in line with the 1990s trend of an increasing
percentage of same-industry mergers documented by Andrade et al. (2001).

Thus, these firms are more exposed to investment and cash flow risks, and
have a stronger precautionary motive for holding cash. It is important to note,
however, that the increase in cross-divisional correlations is not the only reason
cash holdings have increased over time. If it were the only reason, we would
not have witnessed a similar increase in the cash holdings of specialized firms,
as can be seen in Figure 1.
V. Conclusions
The interaction between corporate liquidity and corporate diversification is
interesting theoretically as well as practically. From a theoretical point of view,
diversified firms enjoy the benefit of coinsurance, which reduces their exposure
to risk and allows them to hold reduced amounts of cash in comparison to their
stand-alone counterparts. From a practical point of view, diversified U.S. firms
hold a large fraction of total corporate cash. In 2006, for example, diversified
firms held approximately 72% of aggregate corporate cash. Thus, it is important
to understand the unique determinants and implications of diversified firms’
cash holdings.
This paper shows that multidivision firms hold approximately half as much
cash as specialized firms do, and that this difference can be attributed to di-
versification in investment opportunity and cash flow. The results emphasize
the impact of the joint uncertainty in investment opportunity and cash flow
on cash holdings. Specifically, diversification mainly affects cash through the
cross-divisional correlation in investment opportunity and the financing gap,
that is, the correlation between a firm’s investment opportunities and cash
flows. More diversified firms, with lower cross-divisional correlations in in-
vestment opportunity and smaller financing gaps, are well positioned to hold
less cash. Previous literature has solely focused on cash flow volatility, and
this paper augments the picture by considering the overall joint uncertainty in
investment and cash flow.
Exploiting the imperfect correlations between divisions is also in line with the
coinsurance effect, introduced by Lewellen (1971). In his work, the imperfect
correlations between divisions’ cash flows increase the debt capacity of firms by
reducing the probability of default. This paper suggests that diversified firms
also hold more net debt (defined as debt minus cash) as a fraction of assets.
Thus, diversification also affects firms’ leverage through its impact on net debt
(see also Ahn, Denis, and Denis (2006)).
Another key finding is that diversification is mainly correlated with lower
cash holdings in financially constrained firms. Thus, the precautionary theory
of corporate cash holdings mainly applies to financially constrained firms, as
suggested by the early work of Modigliani and Miller (1958). Cash has no
benefit if firms are not financially constrained and can tap external capital
markets without incurring deadweight costs.
Underlying the hypothesis that diversification allows firms to hold less cash
is the assumption that the reduction in cash holdings is optimal because firms
save on the costs of holding cash. Consistent with this assumption, the findings
in this paper also suggest that diversification primarily reduces cash holdings
in well-governed firms, where managers are more likely to behave optimally.
Furthermore, the reduction in cash holdings is correlated with efficient flows of
funds to high-productivity divisions. These findings are altogether consistent
with the notion that diversification allows firms to optimally hold less cash.
Holding less cash should positively affect the value of diversified firms because
they save on the costs of holding cash.
Finally, it is worth mentioning that holding cash is not the only way in
which firms can manage liquidity. In particular, bank lines of credit are also a
possibility (e.g., Sufi (2009)). Therefore, the interaction between diversification
and the demand for bank lines of credit might be an interesting topic for future
research.
Appendix: Variable Definitions

Note: All names in parentheses refer to the annual Compustat item name.
Accounting Variables
Cash/assets = cash and short-term investments (che)/book as-
sets (at)
Cash flow/assets = (income before extraordinary items (ib) + de-
preciation and amortization (dp))/book assets
(at)
Tobin’s Q = market value of assets (book assets (at) + mar-
ket value of common equity (csho ∗ prcc) − com-
mon equity (ceq) − deferred taxes (txdb))/ (0.9 ∗
book value of assets (at) + 0.1 ∗ market value
of assets)
Book leverage = (debt in current liabilities (dlc) + long-term
debt (dltt))/book assets (at)
Payout/assets = (total dividends (dvt) + purchase of common
and preferred stock (prstkc))/book assets (at)
Capital expenditure
(CAPEX)/assets = capital expenditure (capx)/book assets (at)
Net working capital
excl. cash (NWC)/assets = (current assets (act) − current liabilities (lct) −
cash (che))/book assets (at)
Number of segments = number of business segments reported by the
firm
Firm size = natural logarithm of book assets (at)
Diversification and Volatility Variables

Industry Q volatility = segments’ sales-weighted Q volatility assuming
a correlation of one between all segments. Seg-
ments’ Q volatility is measured as the volatility
of the average stand-alone firm’s Q in its indus-
try over the past 10 years, where an industry
is defined by its three-digit NAICS code
Q correlation = difference between industry Q volatility and
the Q volatility obtained after accounting for
the cross-divisional Q correlations
Cash flow correlation and
Industry cash flow volatility = are defined analogously with respect to cash
flows
Q-cash flow correlation = sales-weighted segments’ correlation between
Tobin’s Q and cash flow. A segment’s correlation
between Tobin’s Q and cash flow is measured as
the correlation of the average stand-alone firm
in the segment’s industry over the past 10 years
Firm cash flow volatility = volatility of firm-level cash flow over the past
10 years
REFERENCES
Acharya, Viral, Heitor Almeida, and Murillo Campello, 2007, Is cash negative debt? A hedging
perspective on corporate financial policies, Journal of Financial Intermediation 16, 515–554.
Ahn, Seoungpil, David Denis, and Diane Denis, 2006, Leverage and investment in diversified firms,
Journal of Financial Economics 79, 317–337.
Almeida, Heitor, Murillo Campello, and Michael Weisbach, 2004, The cash flow sensitivity of cash,
Journal of Finance 59, 1777–1804.
Andrade, Gregor, Mark Mitchell, and Erik Stafford, 2001, New evidence and perspectives on
mergers, Journal of Economic Perspectives 15, 103–120.
Bates, Thomas, Kathleen Kahle, and René Stulz, 2009, Why do U.S. firms hold so much more cash
than they used to? Journal of Finance 64, 1985–2021.
Bebchuk, Lucian, Alma Cohen, and Allen Ferrell, 2009, What matters in corporate governance,
Review of Financial Studies 22, 783–827.
Beltz, Jess, and Murray Frank, 1996, Risk and corporate holdings of highly liquid assets, Working
paper, University of British Columbia, Vancouver.
Berger, Phillip, and Rebecca Hann, 2003, The impact of SFAS No. 131 on information and moni-
toring, Journal of Accounting Research 41, 163–223.
Berger, Phillip, and Eli Ofek, 1995, Diversification’s effect on firm value, Journal of Financial
Economics 37, 39–65.
Calomiris, Charles, Charles Himmelberg, and Paul Wachtel, 1995, Commercial paper and corpo-
rate finance: A microeconomic perspective, Carnegie Rochester Conference Series on Public
Policy 45, 203–250.
Campa, Jose Manuel, and Simi Kedia, 2002, Explaining the diversification discount, Journal of
Finance 57, 1731–1762.
Campello, Murillo, John Graham, and Campbell Harvey, 2009, The real effects of financial con-
straints: Evidence from a financial crisis, Journal of Financial Economics, forthcoming.
Chevalier, Judith, 2004, What do we know about cross-subsidization? Evidence from merging
firms, Advances in Economic Analysis & Policy: Vol. 4: Issue 1, Article 3.
Cremers, Martijn, and Vinay Nair, 2005, Governance mechanisms and equity prices, Journal of
Finance 60, 2859–2894.
Denis, David, Diane Denis, and Atulya Sarin, 1997, Agency problems, equity ownership and cor-
porate diversification, Journal of Finance 52, 135–160.
Denis, David, and Valeriy Sibilkov, 2010, Financial constraints, investment, and the value of cash
holdings, Review of Financial Studies 23, 247–269.
Dittmar, Amy, and Jan Mahrt-Smith, 2007, Corporate governance and the value of cash holdings,
Journal of Financial Economics 83, 599–634.
Dittmar, Amy, Jan Mahrt-Smith, and Henri Servaes, 2003, International corporate governance
and corporate cash holdings, Journal of Financial and Quantitative Analysis 38, 111–133.
Fama, Eugene, and James MacBeth, 1973, Risk, return, and equilibrium: Empirical tests, Journal
of Political Economy 81, 607–636.
Fazzari, Steven, Robert Hubbard, and Bruce Petersen, 1988, Financing constraints and corporate
investment, Brookings Papers on Economic Activity 19, 141–195.
Fazzari, Steven, and Bruce Petersen, 1993, Working capital and fixed investment: New evidence
on financing constraints, RAND Journal of Economics 24, 328–342.
Foley, Fritz, Jay Hartzell, Sheridan Titman, and Garry Twite, 2007, Why do firms hold so much
cash? A tax-based explanation, Journal of Financial Economics 86, 579–607.
Gilchrist, Simon, and Charles Himmelberg, 1995, Evidence on the role of cash flow for investment,
Journal of Monetary Economics 36, 541–572.
Gompers, Paul, Joy Ishii, and Andrew Metrick, 2003, Corporate governance and equity prices,
Quarterly Journal of Economics 118, 107–155.
Graham, John, and Campbell Harvey, 2001, The theory and practice of corporate finance: Evidence
from the field, Journal of Financial Economics 60, 187–243.
Graham, John, Michael Lemmon, and Jack Wolf, 2002, Does corporate diversification destroy
value? Journal of Finance 57, 695–720.
Grossman, Sanford, and Oliver Hart, 1982, Corporate financial structure and managerial incen-
tives, in John J. McCall, ed.: The Economics of Information and Uncertainty (University of
Chicago Press).
Han, Seungjin, and Jiaping Qiu, 2007, Corporate precautionary cash holdings, Journal of Corporate
Finance 13, 43–57.
Harford, Jarrad, 1999, Corporate cash reserves and acquisitions, Journal of Finance 54, 1969–
1997.
Harford, Jarrad, Sattar Mansi, and William Maxwell, 2008, Corporate governance and firm cash
holdings in the U.S., Journal of Financial Economics 87, 535–555.
Hyland, David, and David Diltz, 2002, Why firms diversify: An empirical examination, Financial
Management 31, 51–81.
Irvine, Paul, and Jeff Pontiff, 2009, Idiosyncratic return volatility, cash flows, and product market
competition, Review of Financial Studies 22, 1149–1177.
Jensen, Michael, 1986, Agency costs of free cash flow, corporate finance, and takeovers, American
Economic Review 76, 323–329.
Kashyap, Anil, Owen Lamont, and Jeremy Stein, 1994, Credit conditions and the cyclical behavior
of inventories, Quarterly Journal of Economics 109, 565–592.
Keynes, John Maynard, 1936, The General Theory of Employment, Interest and Money (Harcourt
Brace, London).
Kim, Chang-Soo, David Mauer, and Ann Sherman, 1998, The determinants of corporate liquidity:
Theory and evidence, Journal of Financial and Quantitative Analysis 33, 305–334.
Lang, Larry, and René Stulz, 1994, Tobin’s Q, corporate diversification, and firm performance,
Journal of Political Economy 102, 1248–1280.
Lewellen, Wilbur, 1971, A pure financial rationale for conglomerate merger, Journal of Finance
26, 527–537.
Matsusaka, John, and Vikram Nanda, 2002, Internal capital markets and corporate refocusing,
Journal of Financial Intermediation 11, 176–211.
Modigliani, Franco, and Merton Miller, 1958, The cost of capital, corporation finance and the theory
of investment, American Economic Review 48, 261–297.
Morck, Randall, Andrei Shleifer, and Robert Vishny, 1990, Do managerial objectives drive bad
acquisitions? Journal of Finance 45, 31–48.
Mulligan, Casey, 1997, Scale economies, the value of time, and the demand for money: Longitudinal
evidence from firms, Journal of Political Economy 105, 1061–1079.
Opler, Tim, Lee Pinkowitz, René Stulz, and Rohan Williamson, 1999, The determinants and im-
plications of corporate cash holdings, Journal of Financial Economics 52, 3–46.
Ozbas, Oguzhan, 2005, Integration, organizational processes, and allocation of resources, Journal
of Financial Economics 75, 201–242.
Petersen, Mitchell, 2009, Estimating standard errors in finance panel data sets: Comparing ap-
proaches, Review of Financial Studies 22, 435–480.
Pinkowitz, Lee, René Stulz, and Rohan Williamson, 2006, Does the contribution of corporate cash
holdings and dividends to firm value depend on governance? A cross-country analysis, Journal
of Finance 61, 2725–2751.
Rajan, Raghuram, Henri Servaes, and Luigi Zingales, 2000, The cost of diversity: The diversifica-
tion discount and inefficient investment, Journal of Finance 55, 35–80.
Scharfstein, David, and Jeremy Stein, 2000, The dark side of internal capital markets: Divisional
rent-seeking and inefficient investment, Journal of Finance 55, 2537–2564.
Shin, Hyun-Han, and René Stulz, 1998, Are internal capital markets efficient? Quarterly Journal
of Economics 113, 531–552.
Shleifer, Andrei, and Robert Vishny, 1989, Management entrenchment: The case of manager-
specific investments, Journal of Financial Economics 25, 123–139.
Stein, Jeremy, 1997, Internal capital markets and the competition for corporate resources, Journal
of Finance 52, 111–133.
Sufi, Amir, 2009, Bank lines of credit in corporate finance: An empirical analysis, Review of Finan-
cial Studies 22, 1057–1088.
Villalonga, Belen, 2004a, Does diversification cause the “diversification discount”? Financial Man-
agement 33, 5–27.
Villalonga, Belen, 2004b, Diversification discount or premium? New evidence from the business
information tracking series, Journal of Finance 59, 479–506.
Whited, Toni, 1992, Debt, liquidity constraints and corporate investment: Evidence from panel
data, Journal of Finance 47, 425–460.
Whited, Toni, and Guojun Wu, 2006, Financial constraints risk, Review of Financial Studies 19,
531–559.

Cash Holdings and Corporate Diversification: Ran Duchin

Uploaded by

Copyright:

Available Formats

Cash Holdings and Corporate Diversification: Ran Duchin

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Cash Holdings and Corporate Diversification: Ran Duchin

Uploaded by

Copyright:

Available Formats

THE JOURNAL OF FINANCE • VOL. LXV, NO.

Cash Holdings and Corporate Diversification

3 See the Appendix for detailed variable definitions.

free cash flow.

between cash flows and investment opportunities.

cantly related to cash flow volatility in financially constrained firms.

The empirical investigation that follows tests these implications directly. I

II. Data and Empirical Methods

9 This is identical to the standard formulation of a portfolio’s standard deviation.

assuming a pair-wise correlation of one between all segments

corr(CF)t,k = σ (CF)t,k − σ (CF)t,k. (8)

Cash/assets 0.188 0.095 0.219 50,905

11 The Internet Appendix is available at http://www.afajof.org/Supplements.asp.

Thus, I also consider a “balanced” sample that includes a proportional number

III. Main Results

Panel A: All Firms—One-Way Sorting on Correlation

Q correlation 0.117 0.194 0.078 6.508

All Firms “Balanced” Only Diversified

Panel A: Separate Diversification Measures

All Firms “Balanced” Only Diversified

Panel B: Joint Diversification Measures

[0.001] [0.003] [0.001] [0.005] [0.002] [0.006]

Unconstrained Constrained Difference

Panel A: Financial Constraints Measured by the Whited and Wu (2006) Index

Unconstrained Constrained Difference

Panel E: Financial Constraints Measured by Bond Ratings

IV. Implications and Extensions

where j = 1 . . . N denotes segment j, ss refers to single-segment firms and wj is

where j = 1 . . . N denotes segment j and ss refers to single-segment firms.

Cash/Assets Reduction in Cash Due to Diversification

Total transfers −0.173∗∗∗ −0.011∗∗∗

[0.079] [0.079] [0.079] [0.079]

In both columns, cash appears to be reduced by value-enhancing allocations.

B. Efficiency and Governance

Poor Governance Good Governance Difference

Panel A: Governance Measured by the GIM Index

well-governed firms, a decrease of one standard deviation in the correlation

All Acquirers Only Diversified Acquirers Only Stand-Alone Acquirers

All Acquirers Only Diversified Acquirers Only Stand-Alone Acquirers

R2 0.747 0.621 0.567 0.676 0.537 0.520 0.766 0.646 0.587

Panel A: Average Cash Flow Volatility Panel B: Average Investment

1990 1995 2000 2006 1990 1995 2000 2006

Panel C: Average Correlation Panel D: Average Correlation

1990 1995 2000 2006 1990 1995 2000 2006

(1) (2) (3) (4) (5)

The time-series relation between cash and investment opportunity/cash flow

percentage of same-industry mergers documented by Andrade et al. (2001).

Appendix: Variable Definitions

Diversification and Volatility Variables

You might also like