Date post: | 05-Apr-2018 |
Category: |
Documents |
Upload: | thuthuy3384 |
View: | 225 times |
Download: | 0 times |
of 35
7/31/2019 Determinants and Consequences
1/35
Corporate working capital management:
Determinants and Consequences*
Rabih Moussawi
Wharton Research Data ServicesUniversity of Pennsylvania
Philadelphia, PA 19104Email: [email protected]
Mark LaPlanteSchool of Management, SM31University of Texas at Dallas
Richardson, Texas 75080Email: [email protected]
Robert KieschnickSchool of Management, SM31University of Texas at Dallas
Richardson, Texas 75080Email: [email protected]
Nina BaranchukSchool of Management
University of Texas at DallasRichardson, TX 75080
Current Draft: November 4, 2006
* The authors wish to thank Alexander Butler, Jarrod Harford, and Robert MacDonaldfor helpful comments on an earlier version of this paper.
7/31/2019 Determinants and Consequences
2/35
Corporate working capital management:
Determinants and Consequences
ABSTRACT
Recent business surveys suggest that firms over-invest in working capital. We examinethis suggestion, and find evidence consistent with it. Given this evidence, we then focuson what factors influence corporate working capital management. We find that industrypractices, firm size, future firm sales growth, the proportion of outsider directors on aboard, executive compensation (current portion), and CEO share ownership significantlyinfluence the efficiency of a companys working capital management. Overall, ourevidence suggests that managers respond positively to incentives and monitoring inmanaging their firms working capital.
7/31/2019 Determinants and Consequences
3/35
1
1. Introduction
Several recent business studies suggest that corporations, on average, over-invest
in working capital. For example, REL Consultancy Group has for years conducted an
annual survey of corporate working capital management practices for CFO Magazine,
which CFO Magazine then reports. TheREL 2005 Working Capital Survey concludes
that U.S. corporations had roughly $460 billion unnecessarily tied up in working capital.
Similarly, ITworld.com recently posted the results of a study arguing that poor working
capital management practices cost IT companies billions of dollars annually
(ITworld.com, 2002).
Do US corporations over-invest in working capital? If so, to what extent is this
due to agency problems? We address both of these questions in this study. To see how
important the efficiency of a corporations working capital management can be, we use
an example given in Shin and Soenen (1998). Shin and Soenen (1998) point out that
Wal-Mart and Kmart had similar capital structures in 1994, but because Kmart had a cash
conversion cycle of roughly 61 days while Wal-Mart had a cash conversion cycle of 40
days, that Kmart likely faced an additional $198.3 million per year in financing
expenses.1 Such evidence demonstrates that Kmarts poor management of its working
capital contributed to its going bankrupt. As their 2005 U.S. survey report points out,
there is a high positive correlation between the efficiency of a corporations working
capital policies and its return on invested capital.
To present our examination of these issues, we organize this paper as follows. In
Section 2, we briefly review the literature on working capital management. Our sample
and data sources are described in Section 3. Section 4 provides an analysis of the effect
of working capital management on firm value, and Section 5 provides an analysis of what
factors may influence working capital management. Finally, Section 6 concludes the
paper with a summary of its principal findings.
Using data on a panel of U.S. corporations from 1990 through 2004, we find
evidence of a significantly negative relationship between firm value and investment in
working capital that is consistent with over-investment in working capital. According to
our estimated equations, an additional $1 million investment in working capital is
1 Shin and Soenen (1998), page 37.
7/31/2019 Determinants and Consequences
4/35
2
associated with a roughly 119 to 129 thousand dollar reduction in firm value. To put this
in perspective, a firm that under-utilizes debt by $1 million, can increase firm value by
roughly $140 thousand at current rates by increasing its interest tax shield. Consequently,
it is clear that working capital management decisions have corporation valuation effects
of the same magnitude of corporate capital structure decisions and so probably warrant
just as much attention.
Turning to what influences a firms management of working capital, we find that
a firms working capital policy is influenced by its industrys working capital policies, its
size, its expected sales growth, the proportion of outside directors on its board, the
current compensation of its CEO, and its CEOs share ownership. Consequently,
managerial incentives and the monitoring of management are significant influences on a
firms working capital management performance.
2. Review of Prior Literature
Shin and Soenen (1998) point out that a corporations working capital is the result
of the time lag between the expenditure for the purchase of raw materials and the
collection from the sale of finished goods. As such, it involves many different aspects of
corporate operational management: management of receivables, management of
inventories, use of trade credit, etc. Consequently, there are streams of research on
individual aspects of working capital management (cash and marketable securities, e.g.
Mauer, Sherman and Kim (1998), trade credit, e.g. Rajan and Peterson (1997), etc.).2
However, Schiff and Lieber (1974), Sartoris and Hill (1983), and Kim and Chung (1990)
all emphasize the need to consider the joint effects of these individual policies,
particularly with respect to inventory and credit decisions. For this reason, we only
discuss the prior literature that focuses on overall working capital management.
With respect to the effect of working capital management on firm value, we find
no direct evidence. While Schiff and Lieber (1974), Sartoris and Hill (1983), and Kim
and Chung (1990) model the effects of working capital management practices on firm
2 Some authors include cash and marketable securities in their characterization of working capital and otherexclude these items. We will follow the later tradition as it is consistent with Schiff and Lieber (1974),Sartoris and Hill (1983), Kim and Chung (1990) and other papers that focus on the joint effects of workingcapital management practices and it is consistent with the business survey described earlier.
7/31/2019 Determinants and Consequences
5/35
3
value, they do not provide evidence on whether firms actually do maximize their value by
their working capital management choices. The study that comes nearest to addressing
this issue is the study by Shin and Soenen (1998), which examines the relation between
different accounting profitability measures and net trade cycles, a summary efficiency
measure of a firms working capital management. Shin and Soenens study implies,
without providing direct evidence, that firms that manage their working capital more
efficiently (i.e., shorter net trade cycle) experience higher operating cash flow and are
potentially more valuable.3 However, this last implication does not necessary follow
because firms that have longer net trade cycles are also investing in short-term assets
which may pay off in subsequent periods. Further, their evidence does not speak to
whether the market sees firms as over-investing in net working capital. So the question
as to whether firms over-invest in net working capital on average iis unanswered by prior
research.
As for the determinants of working capital practices, we find even less prior
evidence on which to draw. Nunn (1981) uses the PIMS database to examine why some
product lines have low working capital requirements, while other product lines have high
working capital requirements. In addition, Nunn is interested in permanent rather than
temporary working capital investment as he uses data averaged over four years. Using
factor analysis, he identifies factors associated with the production, sales, competitive
position, and industry. Reinforcing the role of industry practices on firm practices,
Hawawini, Viallet, and Vora (1986) examine the influence of a firms industry on its
working capital management. Using data on 1,181 U.S. firms over the period 1960 to
1979, they conclude that there is a substantial industry effect on firm working capital
management practices that is stable over time. From these studies, we conclude that sales
growth and industry practices are important factors influencing a firms investment in
working capital.
What the above review illustrates is that while there are models to describe how
working capital management practices influence firm value, there is practically no
3 Shin and Soenens measures of firm profitability are ratios of its operating income plus depreciation overeither total assets or total sales. One interesting aspect of Shin and Soenen s evidence is their conclusionthat their net trade cycle variable is measuring something different than what is measured by the currentratio, which typically includes cash and marketable securities.
7/31/2019 Determinants and Consequences
6/35
4
evidence that firms manage their working capital so as to maximize their value. Further,
there is little evidence on what factors influence a firms management of working capital,
particularly whether agency cost issues are concerns.
3. Sample and Sample Data
To address these questions, we examine samples of U.S. public corporations from
1990 through 2004. We begin by identifying all U.S. corporations with Compustat and
CRSP data over this time period. Next, we exclude all firms in financial service
industries (SIC 6000 to SIC 6800) as working capital has a very different meaning in
these industries. Table 1 column 2 presents the number of firms in this sample by year.
This sample is what we examine when we study the effect of investment in net working
capital on firm value.
To study what influences working capital management performance, we add data
from a number of different data sources, which reduces our sample in different analyses.
First, we use the Investor Responsibility Research Center (IRRC) Governance database to
obtain data on certain corporate governance features over the 1990 through 2004 time
period. 4 The availability of these data underlies our choice of time period to study.
Specifically, IRRC collects data on governance provisions in effect for at least 3,155
major US corporations consisting of the S&P 1500 firms and other companies selected
primarily on the basis of market capitalization and high institutional ownership levels
over the years 1990 to 2004.5
As the original IRRC data is biennial and sometimes
triennial, we use the filling method adopted by Gompers Ishii and Metrick (2003) and
lately followed by Bebchuk Cohen and Ferrell (2004) in building our sample for all the
4 IRRCs governance database is an electronic translation of a part of IRRCs published Corporate
Takeover Defenses volumes (September 1990, July 1993, July 1995, February 1998, November 1999,February 2002, and January 2004)5 As IRRC does not provide a manual of its dataset on WRDS, we refer to studies on IRRC data for moreinformation on sampling and individual provision specifications. Some of these sources are Gompers, Ishiiand Metrick (2003), Gillan, Hartzell and Starks (2003), and others. Note here that Gompers Ishii andMetrick (2003) assume that though IRRC governance data is a noisy measure of a firms governanceprovisions, there is no reason to suspect any systematic bias (p. 113). Gompers Ishii and Metrick (2003)report that over the sample period from 1990 to 2002, IRRC tracked more than 93 percent of the totalcapitalization of the combined New York Stock Exchange (NYSE), American Stock Exchange (AMEX),and NASDAQ markets (p. 111).
7/31/2019 Determinants and Consequences
7/35
5
years over the sample period. The maximum number of firms included in this restricted
sample by year is identified in column 3 of Table 1.6
We say the maximum number because the number of firms with IRRC data is
larger than the number of firms with S&P Execucomp data and the number of firms with
IRRCs Directors data.7 We use the IRRCs Directors database to collect information on
the board of directors of sample firms. We use S&P Execucomp database to collect
information on CEO compensation and share ownership. We will discuss the use of
these data further in our analysis of what factors influence a firms working capital
performance.
4. Do firms over-invest in working capital?
4.1 Capital Cash Flow Analysis
While earlier examples focused on the costs associated with investment in
working capital, they do not address the potential benefits. Clearly a company has to
have stock on hand in order to make some sales. Further, competition between firms may
require them to provide customers with interim financing in the form of trade credit
which becomes a receivable to the supplier. Thus, the net effect of investment in
working capital is not as straightforward as earlier examples suggest.8
To discern if firms over-invest in working capital, we use the capital cash flow
valuation model used in Kaplan and Ruback (1995) as our guide. Assume, as they did,
the following simple DCF valuation model:
(1) ( )( )
( ) ( )
( )1 1
( )( )0 (0) (0)
1 1
L s
F t tt t
OCF t INV t INV t CCF t PV CASH CASH
r r
= =
= + = ++ +
where PVF(0) is the current value of the firm; CASH(0) is the current value of its cash
assets; CCF(t) equals its Capital Cash Flows; OCF(t) equals REV(t) EXP(t) + OTH(t),
REV(t) equals net revenues; EXP(t) equals Cost of Goods Sold + SGA + Taxes; OTH(t)
6 It is worth noting that our inferences in the valuation section of our paper are unchanged if we restrict theanalysis to just this sub-sample. In fact, the magnitude of the reduction in firm value from over-investmentin working capital is even larger.7 Execucomp data starts in 1992 and is constrained to the S&P 500 companies until 1994 when Execucompcoverage extends to all S&P 1500 constituents. IRRCs Directors data, on the other hand, tracks directorsof S&P 500, S&P Midcap, and S&P Smallcap companies, starting in 1996.8 See Schiff and Lieber (1974), Sartoris and Hill (1983), or Kim and Chung (1990) for further discussion ofsome of these trade-offs.
7/31/2019 Determinants and Consequences
8/35
6
= Depreciation + Amortization + After tax cash from asset sales; INVL(t)= investment in
long-term assets; and INVS(t)= investment in net working capital, where net working
capital equals accounts receivable plus inventory less accounts payable and accrued
expenses. Like Kaplan and Ruback, we add the firms current cash balances and use a
net working capital definition that excludes investment in cash balances. This approach
is particularly valuable in our context as it allows us to separate out cash management
issues, which have been the focus of a separate literature (e.g., Faulkender and Wang
(2005) and Pinkowitz and Williamson (2005)), from working capital management issues.
Based upon this DCF valuation model, we develop two regression models to
ascertain the relationship between firm investment in net working capital and its market
value. Thus we are able to address the question of whether or not the market sees firms
as over-investing in net working capital.
Regression model 1:
Assume that the firms capital cash flows grow at a constant rate, so that we can
rewrite (1) as:
(2) ( )( ) ( )
( )
( 1) 1 1( )
L s
F
OCF t INV t INV t PV t CASH t
r g
+ + + = +
which can be re-written as:
( ) ( ) ( )1 1 1
( ) ( 1) 1 1F L S
PV t CASH t OCF t INV t INV t r g r g r g
= + + + +
.
This relationship can then be re-expressed as the following regression model:
(3) ( ) ( ) ( )* * *0 1 2 3 4( | ) ( ) ( 1) 1F L SE PV t X CASH t OCF t INV t INV t = + + + + + +
where:
( ) ( )* * *1 1 1
( ) ( ) ; ( ) ; ( ) .L L S SOCF t OCF t INV t INV t and INV t INV t
r g r g r g
= = =
To interpret the marginal effect of INV(t), we would need to recognize that:
(4)( )
( ) ( ) ( )4
( ) 1 1 ( ) 11F
s s s
PV t OCF t OCF t
INV t INV t r g r g INV t r g
= = =
7/31/2019 Determinants and Consequences
9/35
7
So, investment in working capital is at the optimal level if4 is insignificantly different
from zero. Since ( )1
r g
>0, then 4 > 0 implies that there is under-investment in net
working capital and 4 < 0 implies that there is over-investment in net working capital.
Regression model 2:
Our first regression model follows the DCF valuation approach taken in Kaplan
and Ruback (1995). However, the analysis ignores incremental investment in cash and
marketable securities. Whether such incremental investment should be included or
excluded is unclear because some formulations of the DCF valuation framework use
definitions of working capital that includes cash and marketable securities and some do
not. Consequently, to address formulations that do, we modify equation (2) as follows:
(5) ( )( ) ( ) ( )
( )
( 1) 1 1 1( )
L S C
F
OCF t INV t INV t INV t PV t CASH t
r g
+ + + + = +
Basically, we have taken the broader definition of investment in working capital and
broken it into investment in cash and investment in working capital, as we define it. We
can re-write this expression as:
( ) ( ) ( ) ( )1 1 1 1
( ) ( 1)F L S CPV t CASH t OCF t INV t INV t INV t r g r g r g r g
= + +
We can re-express this relationship as the following regression model:
(6) ( ) ( ) ( ) ( )* * * *0 1 2 3 4 5( | ) ( ) ( 1) 1F L S CE PV t X CASH t OCF t INV t INV t INV t = + + + + + + +
where:
( ) ( )
( )
* * *
*
1 1 1( ) ( ) ; ( ) ; ( ) ;
1( ) .
L L C C
S S
OCF t OCF t INV t INV t INV t INV t r g r g r g
and INV t INV t
r g
= = =
=
For regression model 2 we again interpret 4 < 0 and 5 < 0 as evidence of over-
investment in net working capital and cash, respectively.
To estimate these two regression models, we use Compustat data. Selected
summary statistics for each variable based on these data are provided in Table 2. The
first variable, MVF(t), represents the market value of the firm computed as in Fama and
7/31/2019 Determinants and Consequences
10/35
8
French (2002). Specifically, we start with total assets, subtract the book value of equity
and add back the market value of equity as of the end of fiscal year.
To estimate, CASH(t), we use the cash and marketable securities balance of the
firm at the same point in time as we measure its market value. We do this to be
consistent with the way that Kaplan and Stein measured cash balances in their valuation
model.
To estimate OCF(t+1), we use the two approaches described in Kaplan and Stein
(1995), and like their paper, we only report the results based on the second approach as
the results are similar. Specifically, we start with net income; add back depreciation and
amortization expense, interest expense, and the proceeds from the sale of fixed assets.
To estimateINVL(t+1), we use the firms investment in long-term assets (PPE)
from its cash flow statement. Using changes in PPE as an alternative measure does not
change our conclusions and so we only report results using this measure.
To estimateINVS(t+1), we use a definition of net working capital that is consistent
with the one in Kaplan and Steins paper. Specifically, we use current assets minus cash
and marketable securities, minus accounts payable, and minus accrued expenses. There
are two important points to note about this definition. First, we are separating out
investment in cash and marketable securities. Second, we are focused on the investment
in current assets that must be financed with non-spontaneous or outside sources of
financing. This definition is consistent with our valuation model.
To estimateINVC(t+1), we compute the change in the balance of cash and
marketable securities between fiscal years. It is important to note that we estimate
OCF(t+1),INVL(t+1),INVS(t+1), andINVC(t+1) for the fiscal year subsequent to the date
on which we measure the value of a firm and its cash balances. We do this to be
consistent with our valuation model.
The next issue that we have to confront is how to specify the data generating
process for our regression models. It should be fairly obvious that MVF(t) is a non-
negative random variable. While some researchers have scaled MVF(t) by the book
value of assets to create an estimate of Tobins q. We do not take this approach, though
we will see that it does not later, as it introduces additional problems when there is more
variation in book values of assets than in any of the explanatory variables. One
7/31/2019 Determinants and Consequences
11/35
9
alternative is to take the logarithm of the dependent variable and use OLS to estimate a
linear regression on it. Unfortunately, as Manning (1997) explains, this is not always
appropriate, and can lead to biased estimates of the marginal effects of the explanatory
variables.9 Consequently, we follow the recommendation of Hardin and Hilbe (2001)
and use a generalized linear model approach with a log link assumption. Specifically, we
adopt a general estimating equation approach (a GMM approach) using the logarithm link
function, ln(E(y|x)), and estimate the standard errors using the Rogers/Huber/White
estimators adjusted for clustering at the firm level.10
The results of estimating our two regression models using this estimation
approach are reported in Table 3. Both models give fairly similar results. Current cash
balances, operating cash flow, and investment in fixed assets are all positively priced.
The last inference suggests that additional investment in fixed assets for most firms
increases their value. Interestingly, all these inferences are consistent with those that
would be derived from estimates reported in either Faulkender and Wang (2005) or
Pinkowitz and Williamson (2005). More importantly, for our study, we find that the
coefficient on the investment in working capital variable is significantly negative.
Following the interpretation of equation (4), this result implies that at the margin, firms
tend to over-invest in working capital on average. Such a conclusion is consistent with
both the ITworld.com survey mentioned earlier, as well as various annual REL Working
Capital Surveys.
Apparently, the market recognizes this over-investment and discounts firms for it.
According to our estimated equations, evaluated at the mean values of the explanatory
variables, an additional $1 million investment in working capital is associated with a
roughly $129 thousand reduction in firm value. To put this in perspective, a firm that
under-utilizes debt by $1 million, can increase firm value by roughly $140 thousand at
current rates by increasing its interest tax shield.11 Consequently, it is clear that working
9 Nevertheless, it is worth pointing out that our subsequent conclusions do not change if we use thisapproach.10 We should note that we did use the transformed dependent variable method to test for whether fixedeffects were important and fail to reject the use of the random effects model for these data. Later, in ouranalysis of the cash conversion cycles, we face this problem and so use a different specification.11 This estimate assumes new 10 year, AAA rated debt and a 34% tax rate, which is likely an over-estimategiven actually corporate tax rates are significantly less.
7/31/2019 Determinants and Consequences
12/35
7/31/2019 Determinants and Consequences
13/35
11
with a basis for comparison. While Pinkowitz and Williamson uses the methodology of
Fama and MacBeth (1973) to estimate their regression models, we use the panel data
approach advocated in Peterson (2005) as it appears superior to the Fama and MacBeth
methodology. Thus all our standard errors are estimated using Roger estimators adjusted
for clustering on the firm level. Our results for their base specification are somewhat
different than theirs as our estimated coefficient on the level of cash variable is 0.701,
rather than 0.97 as in their study. This difference may be due to the difference in time
period studied: their regression covers 1950 to 1999, while ours covers 1990 through
2004. Other than differences in numerical values, our estimates share the same signs as
their estimates.
Building on this base specification, we next estimate a regression model with net
assets reduced by investment in accounts receivable and inventory and then add a
variable for investment in net working capital, defined by accounts receivable plus
inventory minus accounts payable as this mimics our prior definition of net working
capital. The results of this estimated regression model are reported in column 3 of Table
5. The negative and significant coefficient on the level of net working capital investment
is consistent with our prior estimated valuation model result in that it suggests that firms
on average over-invest in net working capital.
Before reaching a conclusion on how much, we next estimate a specification that
includes prior and future changes in net working capital investment and report the results
in column 4 of Table 5. The reported results suggest that prior and future investment in
net working capital increase firm value. Such estimated coefficients suggest that
Pinkowitz and Williamson interpretation of their change variables is somewhat
questionable as we should not observe a negative sign on the coefficient associated with
the current level of investment in net working capital if their interpretation was correct.
Nevertheless, we can estimate the total effect of investment in net working capital
on firm value by evaluating its effect through current, past and future investment in net
working capital. Evaluated at our sample averages, the total effect is that an additional
$1 million investment in net working capital overall reduces firm value by roughly
$119,326. What is striking about this estimate is that it is close to the $129,000 estimate
that we derive from our prior valuation analysis. Given this consistency, we conclude
7/31/2019 Determinants and Consequences
14/35
12
that our valuation analysis suggests that the market views firms as over-investing in net
working capital on average.
5. What influences the efficiency of a firms working capital management?
Given the above evidence that corporations over-invest in net working capital, the
next question is why. One obvious possibility is that managers do not expend the effort
necessary to minimize net working capital because of incentive compatibility problems,
or agency problems. Prior literature suggests that there are three likely sources of
misalignment: (1) CEO incentives, (2) board incentives, and (3) the structure of corporate
governance.
We will explore each of these possibilities, but before we do we must first
develop a basic model to identify potential control variables. Thus, we conduct this
analysis in a series of steps. We first develop a core model, and then we explore the
influence of board characteristics, CEO compensation and ownership, and finally
corporate charter provisions on a corporations efficiency in managing its working
capital.
As our dependent variable in these regressions, we use a firms cash conversion
cycle (i.e., the inventory conversion period plus the receivables collection period minus
the payables deferral period using Compustat data) as our measure of the efficiency of its
working capital management. While there are alternatives, such as Shin and Soenens
NTC measure, the cash conversion cycle measure (CCC) is standard in many corporate
finance textbooks and is used in theREL Working Capital Surveys as a summary
measure. Consequently, we follow industry and textbook practice and use this measure
for the efficiency of a firms overall working capital management. Summary statistics for
this variable, and all our other variables used in this analysis are reported in Table 5.
Note that we winsorize this and all of our accounting and compensation variables at the
1% level to avoid distortions due to outliers.
For our core model, we conjecture that the following factors are significant
influences on a firms working capital management. First, prior research such as
Hawawini, Viallet, and Vora (1986) suggests that industry practices are significant
determinants of a firms working capital management practices. The working capital
7/31/2019 Determinants and Consequences
15/35
13
policies of say a software company are going to be quite different from those of a retail
shoe company. Consequently, it is important to control for the influence of industry
practices on a firms working capital practices. To do this, we use the median cash
conversion cycle of firms within a firms industry, CCCM, to proxy for the typical
working capital practices within such industry. For our identification of a firms industry
we use the Fama and French (1997) 48 industry delineations set out in Appendix B.
Second, firm size may influence the efficiency of a firms working capital
management. Larger firms may require larger investments in working capital because of
their larger sales levels. Or, alternatively, larger firms may be able to use their size to
forge relationships with suppliers that are necessary for reductions in investments in
working capital. Current supply chain management practices require a lot of coordination
between companies and are typically easier for a larger firm to implement than for a
smaller firm to implement. Thus, firm size is likely to influence the efficiency of a firms
working capital management, though the direction of the effect is an open question. We
use a firms total assets, TA, as our proxy for its size.
Third, the proportion of a firms assets accounted for by fixed assets might
exercise an influence on a firms working capital performance. For example, the
inventory problems of an automobile parts manufacturer are likely to be quite different
from that of a software manufacturer. Further, the receivables problems of these types of
companies are also likely to be different. To measure this variable, we take the ratio of a
firms property, plant and equipment to its total assets, and name it PTA.
Fourth, based upon Nunns (1981) evidence, we expect firm sales to influence a
firms working capital management. In this connection, and consistent with our earlier
regression results, we expect a firms expected future sales to influence its working
capital investment, and so its cash conversion cycle. For example, a firm might build up
inventories in anticipation of future sales growth, and as a result, may also increase its
use of trade credit. To proxy for such growth, we use the firms percentage sales growth
over the future two years and name this variable, FSG.
Finally, some might argue that firms with some degree of market power are able
to work deals with suppliers and customers that give them an advantage over competitors.
To capture this possibility, we compute the Herfindahl-Hirschman index using sales data
7/31/2019 Determinants and Consequences
16/35
14
for each firms industry, again using Fama and Frenchs (1997) 48 industry delineations.
The more concentrated the industry the more likely this will influence the cash
conversion cycles of firms within it. We denote this variable asHHI.
To determine the relevance of the above core factors to the efficiency of a firms
working capital management, we regress the firms cash conversion cycle, CCC, on these
above factors. Before conducting this analysis, we must address the specification of the
data generating process as CCCis a non-negative random variable. While we would
prefer to use the same data generating process specification used in our valuation
analysis, it does not appear appropriate for these data. A Hausman type test indicates that
a random effects model is inappropriate in this case, and so we use a fixed effect model
on a logarithmic transformed dependent variable and estimate Rogers/Huber/White
standard errors adjusted for firm level clustering.
The results of estimating our core regression model,Model 1, for corporate cash
conversion cycles are reported in column 2 of Table 6. These results suggest that firms
do not use their size or market power to reduce their cash conversion cycle. If anything,
they use their position to relax their efforts. Of the factors examined, industry practices
are the main determinant of a firms working capital practices. In addition, positive
future sales growth is associated with increased investment in net working capital. And
finally, firms with more tangible long-term assets reduce their investment in net working
capital. Interestingly, this result appears to mirror the conclusion of the ITworld.com
survey mentioned earlier as firms with more intangible long-term assets appear more lax
in their management of their working capital.
Given these results, we now add the board characteristics of a firm to our core
regression model to extend it. We use two characteristics to capture the essential features
of a corporations board: its size measured by the number of directors (DIR), and its
proportion of outsiders on the board (POD). Prior literature leads us to expect that larger
boards might be lax in monitoring management and so be associated with longer cash
conversion cycles than other firms in their industry. Conversely, prior literature suggests
that more outsiders on the board lead to greater monitoring of management, which we
expect will result in shorter cash conversion cycles for these firms. The results of
estimating this expanded regression model,Model 2, are reported in column 3 of Table 6.
7/31/2019 Determinants and Consequences
17/35
15
These results suggest that board size is not a significant influence, but that board
composition is. The greater the proportion of outsiders on the board, the better the
performance of the firms working capital management. This result is consistent with the
monitoring role of outsider directors.
Continuing this line of inquiry, we now include the compensation and share
ownership of the CEO in our expanded regression model. The more the CEO is paid, the
more likely they will have incentives to reduce a firms cash conversion cycle.
Consequently, we expect the firms cash conversion cycle to be negatively correlated
with the CEOs total current compensation. We measure such compensation that
comprises of CEOs salary and bonus using the Execucomp database and denote it as
TCCOMP. Note that we exclude their current period stock option grants from this
measure and only focus on their current non-stock compensation.
We exclude their current stock option grants because we instead focus on their
total unexercised stock option holdings. Stock options granted in the past might be just
as important an influence as current stock option grants in our attempt to capture
managerial incentive alignment with shareholder interests. Consequently, it might be
better to recognize a CEOs total unexercised stock option position. To estimate this
quantity, we use the Execucomp database to estimate the dollar value of the CEOs
unexercised stock options and denote this variable as TUO.
Finally, we can expect the CEOs current shareholdings to influence the
management of the firms cash conversion cycle. For this reason, we construct the
proportion of stocks held by the CEO and call it CEOPS. Unfortunately, the effect of this
variable is less clear as it could either create incentives for the manager to tightly control
this cycle, or if it could create incentives for managers to expend less effort on this
activity if they have the power to avoid the expenditure of such effort (Morck, Shleifer,
and Vishny (1988)).
The results of estimating this further expanded regression model,Model 3, are
reported in column 4 of Table 6. While both CEO compensation components have a
negative influence on their firms cash conversion cycle, only the total current
compensation component has a statistically significant effect. In some ways this is
consistent with our earlier expectation that a firms investment in working capital
7/31/2019 Determinants and Consequences
18/35
16
primarily influences its performance in the current and near future periods.
Consequently, we should expect current CEO compensation to have a greater influence
on the firms cash conversion cycle, while we might expect their unexercised stock
options to influence their long-term investment decisions. Interestingly, the CEO share
ownership is significantly positively related to their firms cash conversion cycle. So, the
incentive effect of stock ownership appears to be dominated by other effects of CEO
stock ownership.
Expanding our regression model further, we now include a consideration of the
firms corporate charter provisions. Such provisions have figured prominently in recent
literature on cash management (e.g., Dittmar and Mahrt-Smith (2003) and Harford,
Mansi, and Maxwell (2004)) as result of Gompers, Ishii, and Metricks (2003) evidence
on the relationship between these corporate characteristics and equity returns. To begin
this analysis, we follow Harford, Mansi, and Maxwell and simply include Gompers, Ishii
and Metricks governance index, GINDEX, as an additional regressor and report the
results in column 5 of Table 6 (i.e.,Model 4). The reported evidence does not suggest
that these firm characteristics are significant influences on a corporations working
capital performance.
Whether this conclusion is correct is somewhat unclear as the GINDEX assumes
that all of charter provisions have the same influence on a firms cash conversion cycle
and that assumption has been subject to criticism in recent governance literature
(Bebchuk, Cohen and Ferrell (2004)). For example, executive severance agreements such
as golden parachutes can give management an incentive to agree to a takeover, while
poison pills ostensibly are intended to deter takeovers. More importantly, some
provisions (e.g., advance notice requirements) are intended to primarily influence internal
changes in corporate governance, while other provisions (e.g., supermajority
requirements for a merger) are intended to solely deter external changes in corporate
control without any impact on internal governance. Consequently, we create several
indices which group governance features by intended purpose.
Our component indices are: internal provisions, external provisions, compensation
and liability provisions, minority voting provisions, and state laws. The components of
each of these indices are presented in Appendix A. The rationale for each is as follows.
7/31/2019 Determinants and Consequences
19/35
17
The internal provisions index,INTERNAL, includes provisions that limit the
constitutional rights of shareholders, like staggered boards, limitations on shareholder
rights to amend charter and bylaws, and advance notice requirements. Therefore, the
internal provisions index focuses on provisions that primarily influence internal
governance or changes in the internal control of a firm. The external provisions index,
EXTERNAL, which is constituted of provisions like poison pills, supermajority
requirements to approve mergers, fair price, and anti-greenmail, focuses on provisions
that are primarily used to thwart external control contests (i.e., takeover bids). The
compensation and liability provisions index, C&L, focuses on provisions that primarily
influence directors legal liability costs, or compensation received by officers and
directors in the event of a control change. The minority voting provisions index,MVP,
focuses on shareholder voting rules, mainly cumulative and confidential voting rules.
The state laws index, SLAWS, focuses on antitakeover provisions endorsed state law.
Because these antitakeover statutes are often implemented by default in all firms
incorporated in a particular state, and are sometimes redundant with the presence of firm-
level antitakeover defenses, it is not clear that they add much.
We report in column 6 of Table 6 (i.e.,Model 5) the regression results from
substituting these component indices for the GINDEX. While the negative sign on both
the internal provisions index and the compensation and liabilities index are consistent
with the arguments in Baranchuk, Kieschnick, and Moussawi (2005) in that such
provisions help managers to maximize the value of potential growth options in their
establishments, neither coefficient is statistically significant. Further, none of the
coefficients of the different corporate charter indices are statistically significant and so
we conclude that the corporate charter characteristics of a corporation do not significantly
influence its working capital management performance.
Overall, our evidence for the compensation and governance variables suggest that
monitoring of management and managerial compensation are more important influences
on a firms management of its working capital.
7/31/2019 Determinants and Consequences
20/35
18
6. Summary and Conclusions
Several recent business studies suggest that US corporations, on average, over-
invest in working capital. If correct, and if recognized by the market, then one should
observe a negative relation between investment in working capital and firm value. We
address this issue by examining data on samples of U.S. corporations from 1990 through
2004. We find that on average firms have over-invested in their working capital in the
sense that additional investment in working capital is associated with a reduction in firm
value. Such a conclusion appears consistent with the various annual surveys by REL
Consultancy for the CFO Magazine on corporate working capital performance, and the
recent ITworld.com survey of IT firms working capital practices. Apparently the market
recognizes this over-investment and discounts firms for it. However, one can also view
the flip side of this evidence and explain why firms like Wal-Mart suggest that their
working capital management practices are a source of their value.
Given this evidence, we then turn to the question of what factors influence the
efficiency of a corporations working capital management. We find that the inefficiency
of a firms working capital management is positively correlated with firm size and
uncorrelated with its industrys concentration. We interpret these results as suggesting
that firms are not using their market power at the margin to improve the efficiency of
their working capital management practices. Instead, they tend to follow the practices of
their industry. Further, they tend to invest in working capital in anticipation of future
sales growth. Expanding this analysis to include different firm governance features, we
find evidence that the larger the proportion of outsiders on a firms board, the better its
working capital management performance. Such evidence is consistent with the
monitoring of management role of outside directors. Taking the CEOs compensation
and stock ownership also proves important. The larger the CEOs current compensation
the better the firms working capital management performance. However, the larger the
CEOs share of the firms stock, the contrary behavior is shown. Finally, taking
corporate charter features in account, we find no evidence that any such features are
significant influences on a corporations working capital management practices.
Consequently our evidence appears to emphasize the role of board monitoring of
management and managements compensation in its control of the firms working capital.
7/31/2019 Determinants and Consequences
21/35
19
One question that is raise by our study is what determines industry practices, as it
is clear from our firm level analyses that industry practices are a critical determinant of
firm practices. We defer this issue to future research.
7/31/2019 Determinants and Consequences
22/35
20
References
Baranchuk, N., R. Moussawi, and R. Kieschnick, 2005, Antitakeover Provisions inIPOs: A Growth Options Perspective, University of Texas at Dallas working paper.
Bebchuk, L., 2003, Why Firms Adopt Antitakeover Arrangements, University ofPennsylvania Law Review 152, 713-753.
Bebchuk L., A. Cohen, and A. Ferrell, 2004, What Matters in Corporate Governance?,Harvard John Olin Discussion Paper No 491, Harvard Law School.
Bebchuk L., and L. Stole, 1993, Do Short-Term Objectives Lead to Under- orOverinvestment in Long-Term Projects?Journal of Finance, vol. 48-2, 719-729.
Becht, Marco, Bolton, Patrick and Roell, Ailsa A., 2002, "Corporate Governance andControl", ECGI - Finance Working Paper No. 02/2002, http://ssrn.com/abstract=343461 .
Bushman, R., A. Smith, and F. Zhang, 2005, Investment-Cash Flow Sensitivities areReally Capital Investment-Working Capital Investment Sensitivities,http://ssrn.com/abstract=842085.
Cremers, K.J. Martijn, Nair, Vinay B. and Wei, Chenyang, 2004, "The Congruence ofShareholder and Bondholder Governance" NYU, Law and Economics Research PaperNo. 04-007. http://ssrn.com/abstract=516564 .
Dittmar, A., J. Mahrt-Smith, and H. Servaes, 2002, Corporate Liquidity, SSRNworking paper, http://ssrn.com/abstract=319445.
Dittmar, A. and J. Mahrt-Smith, 2005, Corporate Governance and the Value of CashHoldings, SSRN working paper, http://ssrn.com/abstract=687464.
Fama, E., 1980, Agency problems and the theory of the firm,Journal of PoliticalEconomy 88, 288-307.
Fama, E. and J. MacBeth, 1973, Risk, return, and equilibrium: Empirical Tests,Journalof Political Economy 81, 607-636.
Fama, E.F. and K. French, 1997, Industry Costs of Equity,Journal of FinancialEconomics 93, 153-194.
Fama, E. and K. French, 2002, Testing Trade-Off and Pecking Order Predictions AboutDividends and Debt,Review of Financial Studies 15, 1-33.
Faulkender, M. and M. Peterson, 2005, Does the Source of Capital Affect CapitalStructure?, forthcoming in theReview of Financial Studies.
7/31/2019 Determinants and Consequences
23/35
21
Faulkender, M. and R. Wang, 2005, Corporate Financial Policy and the Value of Cash,forthcoming in theJournal of Finance.
Gompers, P., J. Ishii, and A. Metrick, 2003, Corporate Governance and Equity Prices,Quarterly Journal of Economics 118, 107-155.
Gompers, P., J. Ishii, and A. Metrick, 2004, Incentive versus Control: An Analysis ofDual Class Companies, NBER Working Paper No. 10240, January 2004.
Hardin, J. and J. Hilbe, 2001, Generalized Linear Models and Extensions, CollegeStation: Stata Press.
Harford, J., S. Mansi, and W. Maxwell, 2004, Corporate governance and firm cashholdings, SSRN working paper, http://ssrn.com/abstract=595150.
Hawawini, G., C. Viallet, and A. Vora, 1986, Industry Influence on Corporation
Working Capital Decisions, Sloan Management Review 27, 15-24.
Heij, B., P. Boer, P. Franses, T. Kloek, and H. de Dijk, 2004,Econometric Methods withApplications in Business and Economics, New York: Oxford University Press.
ITworld.com, 2002, Poor capital management costs industry billions, ITworld.com,http://www.itworld.com/Man/4215/020614capitalmgmt/pfindex.htm (6/14/02).
Jensen, M., 1986, Agency Costs of Free Cash Flow, Corporate Finance and Takeovers,American Economic Review 76, 323-329.
Kaplan, S. and R. Ruback, 1995, The Valuation of Cash Flow Forecasts, Journal ofFinance 50, 1059-1094.
Kim, Y. and K. Chung, 1990, An Integrated Evaluation of Investment in Inventory andCredit: A Cash Flow Approach,Journal of Business Finance and Accounting 17, 381-390.
Litov, Lubomir P., 2005, Corporate Governance and Financing Policy: New EvidenceSSRN working paper, http://ssrn.com/abstract=637341 .
Malekzadeh A., V. McWilliams, and N. Sen, 1998, Antitakeover Amendments,Ownership Structure, and Managerial Decisions: Effects on R&D Expenditure, WorkingPaper.
Manne, H., 1965, Mergers and the Market for Corporate Control,Journal of PoliticalEconomy73, 110.
Manning, W., 1997, The logged dependent variable, heteroscedasticity, and theretransformation problem,Journal of Health Economics 17, 283-295.
Edited by Foxit ReaderCopyright(C) by Foxit Software Company,2005-2008For Evaluation Only.
7/31/2019 Determinants and Consequences
24/35
22
Mauer, D., A. Sherman, and C. Kim, 1998, The Determinants of Corporate Liquidity:Theory and Evidence,Journal of Financial and Quantitative Analysis 33, 335-350.
Mikkelson, W. and M. Partch, 2002, Do Persistent Large Cash Reserves Hinder
Performance?, SSRN working paper, http://ssrn.com/abstract=186950
.
Modigliani, F. and M. Miller, 1963, Corporate taxes and the Cost of Capital: ACorrection,American Economic Review 53, 433-443.
Molina, C., 2005, Are Firms Underleveraged? An Examination of the Effect ofLeverage on Default Probabilities,Journal of Finance 60, 1427-1459.
Morck, R., A. Shleifer, and R. Vishny, 1988, Management Ownership and MarketValuation: An Empirical Analysis. Journal of Financial Economics, Vol. 20 (1988), pp.293-316.
Nunn, K., 1981, The Strategic Determinants of Working Capital: A Product-LinePerspective, Journal of Financial Research 4, 207-219.
Pinkowitz, L. and R. Williamson, 2005, What is a Dollar Worth? The Market Value ofCash Holdings, http://ssrn.com/abstract=355840.
Rajan, R. and M. Peterson, 1997, Trade Credit: Theories and Evidence,Review ofFinancial Studies 10, 661-692.
REL Consultancy, 2005, REL 2005 Working Capital Survey,http://www.relconsult.com/CFO;jsessionid=CA3EDDA21398FB627617D2345C115D07
Richards, V. and E. Laughlin, 1980, A Cash Conversion Cycle Approach to LiquidityAnalysis, Financial Management9, 32-38.
Ruback, R. and A. Sesia, 2000, "Dell's Working Capital." Harvard Business School Case201-029.
Sartoris, W. and N. Hill, 1983, A Generalized Cash Flow Approach to Short-TermFinancial Decisions,Journal of Finance 38, 349-360.
Schiff, M. and Z. Lieber, 1974, A Model for the Integration of Credit and InventoryManagement,Journal of Finance 29, 133-140.
Shin, H. and L. Soenen, 1998, Efficiency of Working Capital Management andCorporate Profitability, Financial Practice and Education 8, 37-45.
Shleifer A and Vishny R. W., 1997, A Survey of Corporate Governance, The Journalof Finance, 52, 737-783.
7/31/2019 Determinants and Consequences
25/35
23
7/31/2019 Determinants and Consequences
26/35
24
Table 1
Number of Sample Firms by Year and by Analysis
The number of firms in our full sample of Compustat and CRSP firms, which is used inour valuation analysis in Section 4, is reported in column 2. After merging with IRRC
Governance, IRRC Directors, and with Execucomp datasets, the maximum number offirms in our subsample, which is used in our cash conversion analysis in Section 5, isreported in column 3.
YearNumber of Firms in the
Valuation AnalysisMaximum Number of Firms
in the CCC Analysis
1990 5,969 1,361
1991 6,071 1,327
1992 6,377 1,319
1993 6,664 1,306
1994 6,945 1,3721995 7,657 1,422
1996 7,780 1,371
1997 7,584 1,359
1998 7,713 1,734
1999 7,627 1,592
2000 7,205 1,585
2001 6,701 1,520
2002 6,311 1,810
2003 5,874 1,748
2004 5,043 1,800
7/31/2019 Determinants and Consequences
27/35
25
Table 2
Basic Summary Statistics
MVF(t) represents the market value of a firm, as defined by Fama and French (2002), attime t. CASH(t) represents the amount of cash and marketable securities reported for thefirm at time t. Time t is the end of the fiscal year prior to the fiscal year (t+1). OCF(t+1)
represents net income plus depreciation and amortization expense plus interest expenseplus sales of fixed assets. INVL(t+1) represents investment in fixed assets (property,plant and equipment) during year t+1. INVS(t+1) represents investment in net workingcapital, where net working capital is defined as current assets minus cash minus accountspayable minus accrued expenses. INVC(t+1) represents investment in cash andmarketable securities during year t+1. All variables are in millions, and were winsorizedat the 1% level.
Mean Median Std Deviation
MVF(t) 1416.184 114.270 4501.54
CASH(t) 84.400 5.59 662.209
OCF(t+1) 102.126 4.908 345.019
INVL(t+1) 61.095 3.334 202.212
INVS(t+1) 3.929 0.249 64.417
INVC(t+1) 7.801 0.037 62.691
7/31/2019 Determinants and Consequences
28/35
26
Table 3
Analysis of Firm Valuation: Capital Cash Flow ModelMVF(t) represents the market value of a firm, as defined by Fama and French (2002), attime t. CASH(t) represents the amount of cash and marketable securities reported for the
firm at time t. Time t is the end of the fiscal year prior to the fiscal year (t+1). OCF(t+1)represents net income plus depreciation and amortization expense plus interest expenseplus sales of fixed assets. INVL(t+1) represents investment in fixed assets (property,plant and equipment) during year t+1. INVS(t+1) represents investment in net workingcapital, where net working capital is defined as current assets minus cash minus accountspayable minus accrued expenses. INVC(t+1) represents investment in cash andmarketable securities during year t+1. All variables based upon Compustat data werewinsorized at the 1% and 99% levels. The regression model is a general estimatingequations model with a log link (i.e., ln(E(y|x))) with standard errors estimated usingRogers/Huber-White estimators adjusted for firm level clustering. P-values associatedwith the null hypothesis that the coefficient equals zero are reported within parentheses.
Model 1 Model 2
Constant 7.964791(0.00)
7.987754(0.00)
CASH(t) 0.0005372(0.00)
0.000567(0.00)
OCF(t+1) 0.0003641(0.00)
0.0003309(0.00)
INVL(t+1) 0.0004833
(0.00)
0.0004894
(0.00)INVS(t+1) -0.0000405
(0.01)
-0.0000101
(0.02)
INVC(t+1) 0.0001917(0.00)
2 1210.44
(0.00)1202.87(0.00)
7/31/2019 Determinants and Consequences
29/35
27
Table 4
Basic Summary Statistics for Fama-French Valuation Analysis
The definitions of most of these variables follow those in Pinkowitz and Williamson(2005). X(t) is the level of variable X in year t divided by the level of assets in year t.dX(t) is the change in the level of X from year t-2 to year t divided by total assets in year
t. dX(t+2) is the change in the level of X from year t to year t+2 divided by total assets inyear t. M is the market value of equity. E is earnings before extraordinary tems, plusinterest, deferred tax credits and investment tax credits. NA is assets minus cash. NNAis NA minus accounts receivable and inventory. RD is research and developmentexpense. I is interest expense. DIV is common dividends. C is cash and marketablesecurities. NWC is account receivable plus inventory minus accounts payable. Allvariables are in millions, and were winsorized at the 1% level.
Mean Median Std Deviation
M(t) 1.714 0.939 2.318
E(t) -0.117 0.042 0.477
dE(t) 0.008 0.009 0.353dE(t+2) -0.023 0.013 0.467
dNA(t) 0.501 0.493 0.252
dNA(t+2) 0.138 0.067 0.898
dNNA(t) 0.181 0.168 0.423
dNNA(t+2) 0.172 0.030 0.710
RD(t) 0.062 0 0.129
dRD(t) 0.003 0 0.062
dRD(t+2) 0.025 0 0.119
I(t) 0.032 0.018 0.043
dI(t) -0.0005 0 0.033
dI(t+2) 0.008 0 0.048DIV(t) 0.005 0 0.017
dDIV(t) 0.00005 0 0.007
dDIV(t+2) 0.00017 0 0.009
C(t) 0.176 0.076 0.220
NWC(t) 0.207 0.191 0.211
dNWC(t) -0.0003 0.016 0.201
dNWC(t+2) 0.089 0.01 0.308
7/31/2019 Determinants and Consequences
30/35
28
Table 5
Analysis of Firm Valuation: Fama-French ApproachThis analysis follows Pinkowitz and Williamsons (2005) variation of Fama and Frenchs (1998) valuationmodel. X(t) is the level of variable X in year t divided by the level of assets in year t. dX(t) is the change inthe level of X from year t-2 to year t divided by total assets in year t. dX(t+2) is the change in the level ofX from year t to year t+2 divided by total assets in year t. M is the market value of equity. E is earnings
before extraordinary tems, plus interest, deferred tax credits and investment tax credits. NA is assets minuscash. NNA is NA minus accounts receivable and inventory. RD is research and development expense. I isinterest expense. DIV is common dividends. C is cash and marketable securities. NWC is accountreceivable plus inventory minus accounts payable. Following Peterson (2005), each regression model is afixed effects model with standard errors estimated using Rogers estimators adjusted for firm levelclustering. P-values associated with the null hypothesis that the coefficient equals zero are reported withinparentheses.
M(t) M(t) M(t)
Constant 1.453(0.00)
1.565(0.00)
1.422(0.00)
E(t) -0.754(0.00)
-0.546(0.00)
-0.800(0.00)
dE(t) 0.522(0.00)
0.470(0.00)
0.405(0.00)
dE(t+2) -0.202
(0.00)
-0.168
(0.00)
-0.344
(0.00)dNA(t) -0.862
(0.00)
dNA(t+2) 0.276(0.00)
dNNA(t) -0.955(0.00)
-0.854(0.00)
dNNA(t+2) 0.343(0.00)
0.500(0.00)
RD(t) 3.847(0.00)
3.89(0.00)
3.696(0.00)
dRD(t) 1.757(0.00)
2.086(0.00)
1.477(0.00)
dRD(t+2) 4.128(0.00)
4.569(0.00)
3.897(0.00)
I(t) 0.144
(0.87)
-0.146
(0.86)
-0.011
(0.99)dI(t) -3.572
(0.00)-2.399(0.00)
-3.885(0.00)
dI(t+2) -0.132(0.81)
2.478(0.00)
-0.277(0.00)
DIV(t) 9.509(0.00)
10.376(0.00)
9.256(0.00)
dDIV(t) -1.202(0.21)
-1.172(0.23)
-1.436(0.14)
dDIV(t+2) 7.988(0.00)
9.361(0.00)
7.762(0.00)
dM(t+2) 0.203(0.00)
.183(0.00)
.208(0.00)
C(t) 0.701(0.00)
.773(0.00)
.709(0.00)
NWC(t) -1.439
(0.00)
-1.110
(0.00)dNWC(t) 0.907
(0.00)
dNWC(t+2) 1.244(0.00)
R2 0.27 0.29 0.30
7/31/2019 Determinants and Consequences
31/35
29
Table 6
Basic Summary Statistics
CCC represents the cash conversion cycle for a firm, defined as the inventory conversionperiod plus the receivables collection period minus the payables deferral period usingCompustat data. CCCM represents the median cash conversion cycle for a firms
industry using Fama and French (1997) 48 industry classification where firm is allocatedto one of the industries identified in Appendix B. TA represents the total assets (bookvalue) of a firm. PTA is the ratio or the firms property, plant and equipment, to its totalassets. FSG is the firms sales growth over the future two years, computed usingCompustat data. HHI represents the Herfindahl-Hirschman index for a firms industry(using the Fama and French delineations) using Compustat sales data. DIR represents thefirms number of directors, collected from IRRC Directors database. POD is theproportion of board members who are outsiders, collected from IRRCs Directorsdatabase. TCCOMP represents the CEOs total current compensation, excluding stockoptions. TUO represents the CEOs total unexercised stock options. CEOPS representsthe proportion of stock held by the CEO. GINDEX represents Gompers, Ishii and
Metrick (2003) governance index. INTERNAL, EXTERNAL, MVP, C&L, and SLAWSrepresent governance indices identified in Appendix A.
Mean Median Std Deviation
CCC 915.678 143.494 2807.534
CCCM 948.266 149.435 2950.86
TA 7656.707 1427.94 20990.76
PTA 0.310 0.257 0.242
FSG 0.078 0.061 0.174
HHI 0.068 0.051 0.070
DIR 9.745 9 3.085
POD 0.625 0.666 0.186
TCCOMP 1522138 1102150 1109268
TUO 15300000 2361523 39400000
CEOPS 0.027 0.002 0.078
GINDEX 9.065 9 2.785
INTERNAL 1.958 2 1.545
EXTERNAL 1.963 2 1.092
MVP 0.221 0 0.437
C&L 2.269 2 1.569
SLAWS 1.743 1 1.307
7/31/2019 Determinants and Consequences
32/35
30
Table 7
Analysis of Factors Influencing Management of Working CapitalThe dependent variable in each regression is Ln(CCC), which represents the logarithm of the cash conversion cycle fora firm, defined as the inventory conversion period plus the receivables collection period minus the payables deferralperiod using Compustat data. Ln(CCCM) represents the logarithm of the median cash conversion cycle for a firmsindustry using Fama and French (1997) 48 industry classification where firm is allocated to one of the industriesidentified in Appendix B. TA represents the total assets (book value) of a firm. PTA is the ratio or the firms property,plant and equipment, to its total assets. FSG is the firms sales growth over the future two years, computed usingCompustat data. HHI represents the Herfindahl-Hirschman index for a firms industry (using the Fama and Frenchdelineations) using Compustat sales data. DIR represents the firms number of directors, collected from IRRCDirectors database. POD is the proportion of board members who are outsiders, collected from IRRCs Directorsdatabase. Ln(TCCOMP) represents the logarithm of the CEOs total current compensation, excluding stock options.Ln(TUO) represents the logarithm of the CEOs total unexercised stock options. CEOPS represents the proportion ofstock held by the CEO. GINDEX represents Gompers, Ishii and Metrick (2003) governance index. INTERNAL,
EXTERNAL,MVP, C&L, and SLAWS represent governance indices identified in Appendix A. Note that all accountingvariables were winsorized at the 1% level. The below regression is estimated using a fixed effects model with standarderrors estimated using Rogers/Huber/White estimators corrected for firm level clustering. P-values for the nullhypothesis that the coefficient equals zero are reported within parentheses.
Model 1 Model 2 Model 3 Model 4 Model 5
Constant 1.623859
(0.00)
0.687534
(0.01)
0.918074
(0.00)
0.918085
(0.00)
0.899097
(0.00)Ln(CCCM) 0.603557
(0.00)0.662138
(0.00)0.676244
(0.00)0.675400
(0.00)0.673348
(0.00)
Ln(TA) 0.068978(0.00)
0.1437(0.00)
0.142941(0.00)
0.143153(0.00)
0.145330(0.00)
PTA -0.302807(0.00)
-0.079926(0.31)
-0.119592(0.17)
-0.122012(0.17)
-0.125612(0.16)
FSG 0.276770(0.00)
0.316564(0.00)
0.271339(0.00)
0.271929(0.00)
0.271927(0.00)
HHI 0.064759(0.62)
DIR 0.000529(0.79)
POD -0.066063(0.02)
-0.084497(0.00)
-0.081428(0.01)
-0.081808(0.01)
Ln(TCCOMP) -0.016490(0.00)
-0.016399(0.00)
-0.016260(0.00)
Ln(TUO) -0.002773(0.29)
-0.002767(0029)
-0.002768(0029)
CEOPS 0.093137(0.01)
0.091170(0.02)
0.088009(0.02)
GINDEX -0.001766(0.66)
INTERNAL -0.004930(0.42)
EXTERNAL 0.004064
(0.64)MVP 0.043138
(0.11)
C&L -0.007786(0.27)
SLAWS 0.003331(0.78)
F statistic 202.10(0.00)
62.52(0.00)
33.40(0.00)
33.40(0.00)
23.40(0.00)
7/31/2019 Determinants and Consequences
33/35
31
Appendix A: Classification of Antitakeover ProvisionsThis table describes our index building methodology and classification of antitakeover provisions, which issymmetric to IRRC provisions in Gompers, Ishii and Metrick (2003), and is outlined in Bebchuk Cohen andFerrell (2004). The classifications are used to create explanatory variables that count the number ofprovisions within each classification that are in place.
Provision Class and Index IRRC Provision Notes
Internal Provisions Index INTERNAL
1 Unequal voting provisionsUnequal Voting (Dual Classstructures, Time phased, Others)
2 Classified Board with Staggered Terms
3 Limits to Amend Charter
4Limits on Shareholder Rights
Limits to Amend Bylaws
5 Limits to Call Special Meeting
6 Limits for Written Consent
7
Limits on Internal Governance
Advance Notice Requirements
External Provisions Index EXTERNAL1 Poison Pill
Does not require shareholderratification
2 Super Majority to Approve Merger
3 Antigreenmail
4 Director Duties (Stakeholder Clause)
5 Fair Price
Compensation and Liability Provisions Index C&L
1 Director Liability
2 Director Indemnification
3
Director Liability & Indemnification
Indemnification Contracts
4 Severance Agreements w/Change in Control
5Severance Agreements (Empl.Contracts)
6
Executive Severances (GoldenParachutes)
Compensation Plans withChange in Control
(Some) Do not require shareholderratification
Minority Voting Provisions Index MVP
1 Confidential Voting
2 Cumulative Voting [or not(Eliminate Cumulative Voting)] Default rule in only 5 states
State Laws Index SLAWS
1 Business Combination (Freezeout) Law 33 States
2 Fair Price Law 27 States - Similar to fair price3 Control Share Acquisition Law 27 States - Similar to supermajority
4 Recapture of profits Law 7 States - Similar to anti-greenmail
5 Control Share Cash out Law 3 States - Similar to fair price
6 Director Duties (Stakeholder Clause) Law 2 States - Similar to director duties
7/31/2019 Determinants and Consequences
34/35
32
Appendix B: Industrial Classifications
These industrial classifications are similar to those used in Fama and French (1997).More documentation is available athttp://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html .
Code DescriptionCorresponding 4-Digit SIC Code (Compustat's
DNUM)
1 Agriculture 0100-0799, 2048-2048
2 Food Products2000-2046, 2050-2063, 2070-2079, 2090-2095, 2098-2099
3 Candy & Soda 2064-2068, 2086-2087, 2096-2097
4 Beer & Liquor 2080-2085
5 Tobacco Products 2100-2199
6 Recreation 0900-0999, 3650-3652, 3732-3732, 3930-3949
7 Entertainment 7800-7841, 7900-7999
8 Printing and Publishing 2700-2749, 2770-2799
9 Consumer Goods
2047-2047, 2391-2392, 2510-2519, 2590-2599, 2840-2844, 3160-3199, 3229-3231, 3260-3260, 3262-3263,3269-3269, 3630-3639, 3750-3751, 3800-3800, 3860-3879, 3910-3919, 3960-3961, 3991-3991, 3995-3995
10 Apparel2300-2390, 3020-3021, 3100-3111, 3130-3159, 3965-3965
11 Healthcare 8000-8099
12 Medical Equipment 3693-3693, 3840-3851
13 Pharmaceutical Products 2830-2836
14 Chemicals 2800-2829, 2850-2899
15 Rubber and Plastic Products 2830-2836
16 Textiles 2200-2295, 2297-2299, 2393-2395, 2397-2399
17 Construction Materials
0800-0899, 2400-2439, 2450-2459, 2490-2499, 2950-2952, 3200-3219, 3240-3259, 3261-3261, 3264-3264,3270-3299, 3420-3442, 3446-3452, 3490-3499, 3996-3996
18 Construction 1500-1549, 1600-1699, 1700-1799
19 Steel Works Etc 3300-3369, 3390-3399
20 Fabricated Products 3400-3400, 3443-3444, 3460-3479
21 Machinery 3510-3536, 3540-3569, 3580-3599
22 Electrical Equipment3600-3621, 3623-3629, 3640-3646, 3648-3649, 3660-3660, 3691-3692,3699-3699
23 Automobiles and Trucks2296-2296, 2396-2396, 3010-3011, 3537-3537, 3647-3647, 3694-3694, 3700-3716, 3790-3792, 3799-3799
24 Aircraft 3720-372925 Shipbuilding, Railroad Equip. 3730-3731, 3740-3743
7/31/2019 Determinants and Consequences
35/35
33
Appendix B: Industrial Classification (continued)
Code DescriptionCorresponding 4-Digit SIC Code (Compustat's
DNUM)
26 Defense 3480-3489, 3760-3769, 3795-379527 Precious Metals 1040-1049
28 Non-Metal & Ind. Metal Mining 1000-1039, 1060-1099, 1400-1499
29 Coal 1200-1299
30 Petroleum and Natural Gas 1310-1389, 2900-2911, 2990-2999
31 Utilities 4900-4999
32 Communication 4800-4899
33 Personal Services
7020-7021, 7030-7039, 7200-7212, 7215-7299, 7395-7395, 7500-7500, 7520-7549, 7600-7699, 8100-8199,8200-8299, 8300-8399, 8400-8499, 8600-8699, 8800-8899
34 Business Services
2750-2759, 3993-3993, 7300-7372, 7215-7299, 7395-
7395, 7500-7500, 7520-7549, 7600-7699, 8100-8199,8200-8299, 8300-8399, 8400-8499, 8600-8699, 8800-8899
35 Computers 3570-3579, 3680-3689, 3695-3695, 7373-7373
36 Electronic Equipment 3622-3622, 3661-3679, 3810-3810, 3812-3812
37 Measuring and Control Equip. 3811-3811, 3820-3830
38 Business Supplies2520-2549, 2600-2639, 2670-2699, 2760-2761, 3950-3955
39 Shipping Containers 2440-2449, 2640-2659, 3210-3221, 3410-3412
40 Transportation4000-4099, 4100-4199, 4200-4299, 4400-4499, 4500-4599, 4600-4699, 4700-4799
41 Wholesale 5000-5099, 5100-5199
42 Retail 5200-5299, 5300-5399, 5400-5499, 5500-5599, 5600-5699, 5700-5736, 5900-5999
43 Restaurants, Hotels, Motels5800-5813, 5890-5890, 7000-7019, 7040-7049, 7213-7213
44 Banking 6000-6099, 6100-6199
45 Insurance 6300-6399, 6400-6411
46 Real Estate 6500-6553
47 Trading 6200-6299, 6700-6799
48 Miscellaneous 3900-3900, 3990-3990, 3999-3999, 9900-9999