1
The Effects of Bank Monitoring on Firm’s Cash Holding
Policy and Value of Cash
Chien-Lin Lu* Department of Finance National Chengchi University
Yuanchen Chang Department of Finance National Chengchi University
ABSTRACT
This paper explores the role of creditors in firm’s cash holding policy. We test whether creditors affect firm’s cash reserves and the value of cash via covenants of bank loan contracts. The results show that the tightness of cash-related covenant limits firm’s attempt to accumulate cash, causing firms use their cash more efficiently. Borrowers tend to hold more cash reserves for precautionary motive since they might not be able to borrow from banks when they violate covenants. By separating the sources of cash, we show that firms liquidate net working-capital and reduce payout ratio to retain more cash after violating covenants. Finally, we find that the effects of covenant violations on cash are significant for high M/B firms, and the violation help low M/B firms increase the value of cash since firms save cash mainly for precautionary motive after covenant violations.
Key words: Bank loan; Cash holdings; Covenants; Value of cash
JEL codes: G21, G34, G39
*Corresponding author, PhD candidate, Department of Finance, National Chengchi University, Tel: +886-2-77030398, Email: 101357502@nccu.edu.tw
2 1. Introduction
Cash holdings usually play an important role affecting firm’s investment, capital
expenditure, and M&A decisions. There are a number of reasons for holding cash
proposed by prior research. Opler, Pinkowitz, Stultz, and Williams (1999) indicate
that firms with riskier cash flows and poor access to external capital hold more cash
for precautionary purposes.1 For agency concerns, Dittmar, Mahrt-Smith, and
Servaes (2003) find that, to reserve cash rather than to pay dividends to shareholders,
firms in countries where shareholders rights are not well protected hold more cash
than others. Dittmar and Mahrt-Smith (2007) and Pinkowitz, Stulz, and Williamson
(2006) show that cash is worth less when agency problem between insiders and
outside shareholders are greater.2
In this paper, we investigate the relationship between cash holdings and creditors’
actions. There are a few papers, discussing the effect of bank on firm’s cash policies.
However, numerous papers have shown that banks can influence firm’s decisions
through lending relationship. For example, Nini, Smith, and Sufi (2009) show that
bank are more likely to impose a capital restriction as a borrower’s credit quality
deteriorates, and the capital expenditure restrictions cause a reduction in firm
investment. Ahn and Choi (2009) indicate that borrowers’ earnings management
behavior generally decreases as the strength of bank monitoring increases. Chava
and Roberts (2008) show that capital investment declines sharply following a
financial covenant violation, when creditors use the threat of accelerating the loan to
intervene in management.
1 The precautionary motive for cash savings is also supported by Almeida, Campello and Weisbach
(2004), Bates, Kahle, and Stulz (2009), and McLean (2011).
2 The agency motive for cash savings is also supported by Harford (1999), Harford, Mansi, and
3
How creditors might affect borrowers’ cash holding policies? First, the
monitoring mechanism, induced by the conflict of interest between creditors and their
borrowers, provides incentives for creditors to monitor their borrowers with capital
restrictions or requirement of financial covenants (Diamend, 1984; Holmstrom and
Tirole, 1997). Through this channel, bank’s monitor should induce borrowers to
improve their performance to meet the requirement of financial covenants (Demiroglu
and James, 2010). It will also predict that firms are more likely to reduce their cash
reserves after the intervention of banks, and this incentive will be much stronger if
they receive more pressure from banks. On the other hand, the value of cash for the
firms monitored by banks should be higher than the value of cash hold by firms
without monitored by banks, since these firms should use cash in a more efficient way
to meet the financial requirements by banks (Faulkender and Wang, 2006).3
Second, bank may also encourage their borrowers to hold cash to reduce bank’s
exposure. Pinkowitz and Williamson (2001) argue that Japanese firms’ cash ratios
are positively related to bank power in Japan, conjecturing that banks with monopoly
power prefer firms holding more cash, which not only reduces the probability of
technical default but protects the right of creditors. Liu and Mauer (2011) find that
the positive relation between CEO compensation incentives and corporate cash
holdings is at least in part driven through the imposition of liquidity covenants in debt
contracts. They conjecture that this result is consistent with the concept that bank
encourages firms to hold cash to prevent their default risk. In the second channel,
banks might use the financial covenants to induce borrowers reserve their cash in
order to decrease the default risk which should hurt the right of creditor. It predicts
3 According to Faulkender and Wang (2006), the marginal value of cash should be positive for much of
firms, but it declines with larger cash holdings since firms might miss-use the amount of cash exceeding their need in the foreseeable future.
4
that firms are more likely to increase their cash reserves after the intervention of
banks, and the value of cash might decrease after firms stockpile their cash, since
firms might abuse the amount of cash exceeding their need in the foreseeable future.
Overall, the effect of bank loan covenant on firms’ cash policies is an unresolved issues and it is an empirical question that we try to address in this study.
To capture the effect of creditor’s monitoring, we use initial tightness of financial
covenants as our proxy of bank power. The initial tightness of financial covenants is
measured by the strictness of covenants while banks establish the loan agreement with
borrowers. Previous literature shows that borrowers will improve their performance
related to covenant, especially when they got tighter covenants (Demiroglu and James,
2010). We estimate covenants tightness following the procedures of both Demiroglu
and James (2010) and Murfin (2012) to calculate the loan contract strictness. We
also examine the effect of violation on firm’s cash holding policies. According to Sufi (2009), a covenant violation is associated with a 15 to 25% drop in the
availability of lines of credits, and it also means firms need to retain their cash or
cash-flow to prevent the unpredictable shocks in the future. Therefore, we expect
that violation should induce borrowers retain their cash-flow and increase the cash
holdings. We estimate loan’s violation using the data from Nini et al. (2009), which
identified 16,530 of loan’s violation event from 1996 to 2008.
In our empirical works, we first examine whether the tightness of loan and
violation alter firm’s cash reserves. Our results show that only the initial tightness of current ratio covenant, which is related to firm’s cash reserves, alters firm’s cash
holding policies, while the tightness of Debt/EBITDA do not change firm’s tendency
to hold cash. The results of tightness of current ratios are consistent with the
5
got tighter covenant. Loan’ violation, the signal of dropping in the availability of
credit lines, induces firms to increase their cash reserves. In order to investigate how
borrowers will act on their sources of cash after they got tight covenant or after they
violation, we further separate firms’ cash into three primary sources. The results
show that borrowers indeed reduce their borrowings from banks after loan’s violation,
and they try to liquidate their net working capital and reduce cash dividends to
increase their cash reserves, yet tight covenants do not have the similar effects.
Second, the initial tightness of covenants has no impact on the value of cash, and
violation effect the value of cash and excess cash in various ways. The initial
tightness of current ratio covenant significantly reduces firm’s value, but its tightness
does not alter the value of cash and excess cash. Similar with tightness of covenants,
loan’s violations are significantly and negatively related to firm’s value. For cash, the cash owned by borrowers is negatively related to firm’s value if firms violated in
the previous quarter, but the excess cash owned by borrowers is positively related to
firm’s value.
Third, as indicated by Demiroglu and James (2010), the tightness of covenants is
significantly altered by firm’s growth opportunities. We further split our sample into firms with high growth opportunities (M/B) and firms with low grow opportunities
(M/B), and we re-examine our tests to see whether our proxies have different impact
on firms that have different growth opportunities. The results show that the negative
relationship between tightness of covenants and cash policies are similar for different
groups of firms, but the positive effect of violation on cash policies only holds for
firms with high growth opportunities, which are more likely to save cash for future in
6
that also has tightness of current ratio covenant is negatively related to firm’s value, implying that these firms might miss-use their cash and decrease the value of cash.
Finally, there might be some omitted factors interacting between covenants
tightness, violation, and cash policies. To further check, we estimate the propensity
of covenants tightness as our instrumental variable and re-test all the regressions
regarding covenants tightness. Most of the results are consistent with our previous
findings, except that the tightness of current ratio covenant does not alter borrower’s
cash holding policies. However, we do not measure the propensity of violation since
we are unable to identify what determinants impact the propensity of violation.
This paper proceeds as follow. Section 2 introduces the data and the estimating
procedures of our measures. Section 3 examines the effect of the initial tightness of
covenants and loan’s violation on firm’s cash holdings policies. Section 4 tests the
effect of the initial tightness of covenants and loan’s violation on the value of cash,
and we present the results of robustness check. Section 5 provides conclusions.
2. Data and variables
2.1 Data and sample selection criteria
We construct the sample from three databases including Loan Pricing
Corporation’s (LPC) DealScan files from 1990 to 2010, Compustat quarterly files
from 1970 to 2010 and data of loan’s violation provided from Nini et al. (2009) from
1996 to 2008.4 LPC provides details of syndicated loan for distinguishing covenant
tightness and Compustat files contain financial statement items for estimating excess
cash and other primary measures in our paper. Our sample includes both survivor
and non-survivor firms, and we require that firms have positive assets (Compustat
4 Since we need the data of cash flow from year t-1 to t-10 to calculate industry sigma in each year,
and we also need the average coefficient of our equation (1) from year t-1 to t-10 to calculate the expected cash ratio, our Compustat data start from 1970 which prior to 1990 for 20 years.
7
quarterly data item #44) and positive sales (#2). Further, we exclude financial firms
(SIC codes between 6000 and 6999) and utility firms (SIC code between 4900 and
4999), and we restrict our sample to firms incorporated in the United States. We
exclude financial firms since they might carry cash for their capital requirement rather
than for economic reasons, and we exclude utility firms because their cash holdings
might be subject to regulatory requirements. To address the potential problem of
outliers, all accounting variables are winsorized at the 1st and 99th percentiles.
The sample period of LPC data is between 1987 and 2013. According to Carey
and Nini (2007), LPC’s data collection efforts focused primarily on the U.S. loan market until the 1990s, so we limit our sample period from 1990 to avoid potential
sample bias. To get the final sample, we first link Compustat database and LPC
database by the connecting files provided by Chava and Roberts (2008), which
provides the connecting information from 1982 to 2012. After deleting the
observation with missing value for any financial variable we used, our sample of
covenant tightness contains 13,847 facility level data and 190,937 firm observations.
To calculate the tightness of covenants, however, we might suffer from the
problem of lack of uniformity in how loan covenants are defined. As indicated by
previous papers, the definition of covenant variables should be different between
GAAP-based Compustat financials and covenant thresholds reported by DealScan
(Chava and Roberts, 2008; Demiroglu and James, 2010; Murfin, 2012). This
measurement error problem causes much of covenants violate immediately while loan
agreement established, and it also reduce the credibility of tightness measure. To
overcome this problem, we follow previous literature by limiting our facilities to
those with information in TearSheets, which provide detailed definitions for covenants.
8
used in loan agreement to minimize the measurement problem. Based on these
criteria, we choose current ratio covenant and Debt/EBITDA covenant as our targets.
However, the above procedure also limits our sample to 691 and 751 observations
with clearly identified tightness data of current ratio and Debt/EBITDA, respectively.
It also restrict our sample period to 1990 to 2010. On the other hand, to combine the
non-missing Compustat data with violation information in Nini et al. (2009), our data
with violation information contains 94,984 firm observations.
2.2 Excess cash measure
In this section, we briefly introduce the estimating procedure of how we
calculate the excess cash ratio, which is an important measure of cash policy
suggested by previous papers (Opler et al., 1999; Bates et al., 2009). To measure
excess cash, we follow Bates et al. (2009). First, we estimate the expected cash
ratios of each firm from the following Fama-MacBeth regression model:
𝐶𝑖,𝑡 = 𝛼 + 𝛽1(𝑀 𝐵⁄ )𝑖,𝑡+ 𝛽2𝑆𝐼𝑍𝐸𝑖,𝑡+ 𝛽3𝐶𝐹𝑖,𝑡+ 𝛽4𝑁𝑊𝐶𝑖,𝑡+ 𝛽5𝐶𝐴𝑃𝑋𝑖,𝑡+ 𝛽6𝐿𝐸𝑉𝑖,𝑡
+𝛽7𝐼𝑆𝑖,𝑡+ 𝛽8𝑅&𝐷𝑖,𝑡+ 𝛽9𝐷𝑖,𝑡+ 𝛽10𝐴𝑄𝑖,𝑡+ 𝛽11𝑁𝐸𝑖,𝑡+ 𝛽12𝑁𝐷𝑖,𝑡+ 𝜀𝑖,𝑡, (1)
where 𝐶 is cash ratio defined as cash and marketable securities (#36) divided by book assets (#44); 𝑀 𝐵⁄ is market-to-book ratios, defined as book assets (#44) minus book equity (#59) plus market equity (#12*#61) all divided by book assets (#44);
𝑆𝐼𝑍𝐸 is size defined as the logarithm of book assets (#44); 𝐶𝐹 is cash flow measured as earnings after interests, dividends, and taxes but before depreciation
(#21-#22-#6-#20); 𝑁𝑊𝐶 is net working-capital net of cash; 𝐶𝐴𝑃𝑋 is capital expenditures (#90); 𝐿𝐸𝑉 is leverage measured as long-term debt plus debt in current liabilities (#51+#45); 𝐼𝑆 is industry sigma measured as the standard deviation of industry cash flow to assets in the previous 10 years for each firm-year; 𝑅&𝐷 is
9
R&D measured as R&D (#4) / sales (#2), and is set equal to zero when R&D is
missing. Alternative measure is R&D/ assets; 𝐷 is a dummy variable which equals one in years in which a firm pays a common dividend (#20) and zero otherwise; 𝐴𝑄 is acquisitions (#94); 𝑁𝐸 is net equity measured as equity sales minus equity purchases (#84-#93); and 𝑁𝐷 is net debt measured as debt issuance minus debt retirement (#86-#92). All of the independent variables except for SIZE and M/B are
divided by book assets (#44). The definition of all variables used in this paper is
shown in Appendix.
Firms’ demand function of cash might be time-varying during our sample period, so we use a rolling window to estimate the cash demand function. At year t, we run
the annual cross-sectional cash demand regression from year t-1 to year t-10 and
calculate the time series average coefficients. Excess cash at year t is then calculated
as the difference between firms’ actual cash ratios and expected cash ratios predicted
by Equation (1), i.e., residuals from the equation.
2.3 Covenants tightness measure
Similar to the measure of proxy for cash holding policies, the measure of
covenants tightness is another important measure used in our empirical tests.
Demiroglu and James (2010), show that the initial ratio of financial covenant are
usually clustered at discrete levels from 0.25 from each other, and borrowers that have
similar financial ratios at the time of the loan agreement select from similar covenant
menus. Following Demiroglu and James (2010), we measure the tightness of
financial covenants by these discrete numbers to separate our covenant sample in
order to form clusters in which borrowers have the same covenant threshold choices.
10
(current ratio or Debt/EBITDA), then we further sort these borrowers by their
covenant choices, i.e., the covenant ratios. Finally, we define covenant as “tight”
when the covenant ratio is higher than the median of covenant ratio in each cluster
and loose otherwise.
Alternatively, we also follow the approach established by Murfin (2012) to
estimate the tightness of covenants. According to Murfin (2012), we derive the
slack for each covenant threshold which is the difference between the observed
accounting ratio and the covenant threshold that is specified in the loan contract. For
example, since current ratio (Debt/EBITDA) covenant gives the minimum (maximum)
threshold for accounting ratio, we use the accounting ratio (covenant threshold)
subtracted by covenant threshold (accounting ratio) as the slack of covenants. In
order to control for the different scale of each covenant slack, each slack is further
normalized by its respective standard deviation. This measure is calculated by the
slack of covenants, so a larger number of this measure indicates a loose covenant.
3. Empirical test
3.1 Summary statistics
Table 1 presents the summary statistics of our sample. Results of Panel A show
the distribution of borrower’s characteristics, indicating that some firms have negative
EBITDA and the range of Debt/EBITDA is widely. Since the covenant threshold of
Debt/EBITDA must be positive, we limit our sample only include the firms with
positive Debt/EBITDA. The average cash ratio is 15.8%, but in unreported results,
we find that the average cash ratio has a steady and increasing trend from 1990 to
2007, which is consistent with the results found in Bates et al. (2009). However, this
11
but in a slower increasing speed. The excess cash ratio on average is modest,
because there are few firms hold excess cash before 2001, and the excess cash ratio is
sharply increasing starting from 2003, but its increasing trend turns down after 2007.
Panel B of Table 1 presents the characteristics of loan. The average maturity of
loan in our sample is about four years, and the longest maturity is thirty-five years.
We require our sample has maturity that longer than a year, since we conjecture that
the tightness of loan that is shorter than a year should not has much power to restrict
firm’s decisions. Average covenant number is 2.48, which usually include the limit of cash flow or the minimum interest coverage.5 We do not consider interest
coverage as our target since the definition of interest might differ in different contracts,
and the concept of interest coverage is similar to Debt/EBITDA. Panel C of Table 1
further reports the distribution of our each proxy of covenant tightness and violation.
From the proxies estimated based on Demiroglu and James (2010), the ratio of tight
covenant is about 60% and 66% for current ratio covenant and Debt/EBITDA
covenant, respectively. On the other hand, about 7% of sample firms had violated
covenants during our sample period.
[Insert Table 1 here]
In order to test whether the initial tightness of covenants includes information
about borrowers, we next use the tightness of our two major covenants as cutoff point
to perform the univariate test. Results of Panel A of Table 2 show that borrowers
with tight current covenant on average have significantly lower cash reserves and
excess cash holdings. Although these borrowers have lower current ratio implying
worse quality of the borrowers with tight current covenant, the difference is
5 In our total sample of LPC data from 1987 to 2010, there are 10,728 of contracts include
Debt/EBITDA covenant, 44% of total facilities. On the other hand, there are 9,254 of contracts include interest coverage covenant, 38% of total facilities.
12
insignificant. In terms of loan characteristics, borrowers which have tight current
covenant on average have larger loan size, longer loan maturity, smaller lead arranger
shares, lower all-in-spreads and less covenants. We also compare the characteristics
of our alternative proxies for tightness, showing that firms got tight current ratio
covenant calculated by Demiroglu and James (2010) usually has tighter covenants of
current ratio that calculated by Murfin (2012).
From the results partitioned by Debt/EBITDA covenant, the firms with tight
Debt/EBITDA covenant on average are smaller, and they usually have higher ratio of
Debt/EBITDA. Further, firms with tight Debt/EBITDA covenant do not have
significantly different inclination of cash holding policies with firms with loose
Debt/EBITDA covenant. In the results of loan characteristics, the tight covenants on
average have much slack contract but shorter maturity. Similar to the results of
current ratio covenant, firms got tighter covenant defined from Demiroglu and James
(2010) also has tighter number defined by Murfin (2012).
In panel B of Table 2, we further examine the fraction of tightness by industry,
which is defined by two-digit SIC code. Results indicate that much of the industries
have more than forty percent of firms have loan including these two kinds of
covenants. Firms belonging to agriculture, minerals and construction industries are
most likely limited by tight covenants, and firms in manufacturing industry also on
average have tighter covenants. The distribution of the fraction of tightness of these
two covenants is similar to the trade-retail and services. Firms belonging to
trade-retail usually have more contracts with tight current ratio covenants but less
contracts with tight Debt/EBITDA covenants. In Panel C of Table 2, we also test the
13
tightness of current ratio covenant is irrelevant to the tightness of Debt/EBITDA
covenant, indicating the independent relationship between these two covenants.
[Insert Table 2 here]
3.2 Effect of covenant tightness and violation on the tendency to hold cash
In the multivariate test, we first explore the impact of covenant tightness on
tendency to hold cash. We employ following model adapted from Bates et al. (2009)
to examine the relationship between covenant tightness and cash holding inclination:
∆𝐶𝑖,𝑡 (∆𝑋𝐶𝑖,𝑡) = 𝛼 + 𝛽1𝐶𝑢𝑟𝑟𝑒𝑛𝑡_𝑡𝑖𝑔ℎ𝑡(𝐷/𝐸𝐵𝐼𝑇𝐷𝐴_𝑡𝑖𝑔ℎ𝑡)𝑖,𝑡−1+ 𝛽2∆𝐗𝑖,𝑡+ 𝜀𝑖,𝑡, (2)
where Current_tight and Debt/EBITDA_tight are defined in Section 2.3; ∆𝐶 is the change in cash ratio defined as cash and marketable securities divided by book assets;
∆X𝐶 is change in excess cash defined as Section 2.2. In our specification we control for the other variables X that have been found to affect cash holdings in the literature.
These variables include market-to-book ratios (𝑀 𝐵⁄ ), firm size, cash flow( 𝐶𝐹), net working capital (𝑁𝑊𝐶), capital expenditures (𝐶𝐴𝑃𝑋), leverage (𝐿𝐸𝑉), industry sigma (𝐼𝑆), R&D (𝑅𝐷), an indicator variable of dividend (𝐷), acquisitions( 𝐴𝑄), net equity (𝑁𝐸), and net debt (𝑁𝐷). All of these control variables, except industry sigma, are used as delta term to reflect the influence of change in financial condition
on cash reserves. We do not take the industry sigma as the delta term since industry
sigma is measured as standard deviation of industry cash flow to assets in the previous
ten years for each firm-year, which is a constant item within a year and should not be
taken as delta term.
The results of specification (2) in Table 3 show that tightness of current ratio is
14
by tightness of liquidity covenant force firms to reduce their inclination to hold cash.6
The tightness of Debt/EBITDA, however, does not alter firms’ decision of cash
policies. Thus, these results imply that only the tightness of current ratio covenant,
the covenant most related to firms’ cash holding policies, alters firms’ tendency to hold cash.
Additionally, we also test the regressions using firms’ excess cash as dependent
variable, showing the results in the right column of Table 3. Similarly, the results
indicate that only the tight current ratio covenant will impact firms’ cash holding
inclination, and the tightness of Debt/EBITDA has no effect on ratio of excess cash.
In sum, tight current ratio covenant not only induce firms to tilt toward holding less
cash, it also reduce stock piling of excess cash from firms. These results are
consistent with some findings of Demiroglu and James (2010). It shows that
tightness of covenant induce firms improve their covenant variable, and the
improvement is only induced by related covenant. In addition, these results also
support the monitoring hypothesis. The monitor provided by creditors will induce
firms to use their cash more efficiently, and they are more likely to reduce holding of
excess cash.
[Invert Table 3 here]
We next test the effect of violation on firms’ cash holding decisions. Different
to initial tightness, after loan’s violation, creditors might withdraw their existing capital for borrowers to avoid losses. After banks withdraw sources of capital, firms
should stock cash for precautionary motive in order to deal with unpredictable shocks
6 To save space, we do not report results estimated by the approach of Murfin (2010). The unreported
results are similar to the results calculated by Demiroglu and James (2010) and we are delighted to provide the empirical results if required.
15
in the future. Thus, we predict that violation of loans should induce borrowers
accumulate their cash to prevent any potential shocks in the future.
As shown in Table 4, after controlling the variables which are close to cash ratios,
violation is significantly and positively related to firms’ cash ratio. Similarly,
violation also induce firms to increase their holding of excess cash. Consistent with
our prediction, creditors might withdraw the capital after borrowers violate covenants.
In that case, borrowers need to retain their cash-flow and to reserve cash in order to
prevent unpredictable shocks in the future.
[Invert Table 4 here]
In order to see how borrowers change their distribution of sources to response to
tight covenant or violation, we next split sources of cash into three different parts.
We estimate the three major sources of cash following definition of McLean (2011),
and we also examine the distributions of other sources such as net working-capital
and cash dividends.7 As mentioned by DeAngelo, DeAngelo, and Wruck (2002), firms might liquidate their working-capital to meet requirement of covenant. We
review the change in net working-capital to test if firms liquidate their net
working-capital after receiving tighter covenant or violating covenants. Results
shown in Table 5 present that firms significantly reduce the source of debt after they
got a tight covenant of current ratio compared to the firms which have loose one, and
these firms also increase their net working-capital at the same time. The tightness of
Debt/EBITDA covenant, a limitation on cash-flow, does not alter firms’ behavior
except firms might save their operating income to meet the requirement of cash flow.
7 According to McLean (2011), the cash flow contributed by issue includes SEOs, private placements,
rights of offerings, stock sales through direct purchase plans, preferred stock, and employment options, grants, and benefit plans. The cash flow contributed by debt is the cash proceeds from debt sales. The cash contributed by cash flow is the net income plus amortization and depreciation.
16
Finally, consistent with our prediction, borrowers significantly reduce the
sources of debt and liquidate the net working-capital after violating covenants.
Borrowers also significantly reduce their cash dividends to reserve cash for future
needs. Thus, after the violation of loan, borrowers indeed try to liquidate their net
working-capital and reduce the payout ratio to retain their cash for precautionary
motive. The tightness of covenant, however, does not cause firms to liquidate the
net working-capital, which implies that borrowers might not face an immediately
demand to save cash since they still have not close to the edge of violation.
[Invert Table 5 here]
3.3 Effect of covenant tightness and violation on value of cash
We next examine the impact of tightness of covenants on value of cash. Based on the monitoring hypothesis, firms should use their cash in more efficient way after bank intervene their governance, otherwise banks may not have power to improve firms management. To test this hypothesis, we use the following equation as our model for testing the relationship between covenant tightness and value of cash according to Bates et al. (2009):
𝑀𝑉𝑖,𝑡 = 𝛼 + 𝛽1𝐶𝑢𝑟𝑟𝑒𝑛𝑡_𝑡𝑖𝑔ℎ𝑡(𝐸/𝐸𝐵𝐼𝑇𝐷𝐴_𝑡𝑖𝑔ℎ𝑡)𝑖,𝑡−1+ 𝛽2𝐶(𝑋𝐶)𝑖,𝑡
+𝛽3𝐶𝑢𝑟𝑟𝑒𝑛𝑡_𝑡𝑖𝑔ℎ𝑡(𝐸/𝐸𝐵𝐼𝑇𝐷𝐴_𝑡𝑖𝑔ℎ𝑡)𝑖,𝑡−1∗ 𝐶(𝑋𝐶)𝑖,𝑡+ 𝛽4𝐸𝑖,𝑡+ 𝛽5𝑑𝐸𝑖,𝑡 +𝛽6𝑑𝐸𝑖,𝑡+2+ 𝛽7𝑑𝑁𝐴𝑖,𝑡+ 𝛽8𝑑𝑁𝐴𝑖,𝑡+2+ 𝛽9𝑅𝐷𝑖,𝑡+ 𝛽10𝑑𝑅𝐷𝑖,𝑡+ 𝛽11𝑑𝑅𝐷𝑖,𝑡+2
+𝛽12𝐼𝑖,𝑡+ 𝛽13𝑑𝐼𝑖,𝑡+ 𝛽14𝑑𝐼𝑖,𝑡+2+ 𝛽15𝐷𝑖𝑣𝑖,𝑡+ 𝛽16𝑑𝐷𝑖𝑣𝑖,𝑡+ 𝛽17𝑑𝐷𝑖𝑣𝑖,𝑡+2 +𝛽18𝑑𝑀𝑉𝑖,𝑡+2+ 𝜀𝑖,𝑡, (3) where 𝑋𝑡 is level of variable X in year t divided by the level of total assets in year t;
𝑑𝑋𝑡 is the change in the level of X from year t-2 to year t, 𝑋𝑡− 𝑋𝑡−2; 𝑑𝑋𝑡+2 is the change in the level of X from year t to year t+2, 𝑋𝑡+2− 𝑋𝑡; Current_tight and Debt/EBITDA_tight are defined in Section 2.3, which taken as lag term to estimate the
impact of covenant on firm’s value; MV is market value defined as book assets minus book equity plus market equity ((#61*#12)+#51+#45); E is firm’s EBITDA (#21); NA
17
development (R&D) expense (#4); I is interest expense (#22); and Div is dividends
defined as common dividend paid (#20).
Modifying the specification of Bates et al. (2009), we include Current_tight and
Debt/EBITDA_tight to test how the tightness of different covenants affect value of
cash in the next quarter. We also use an interaction term 𝐶𝑢𝑟𝑟𝑒𝑛𝑡_𝑡𝑖𝑔ℎ𝑡(𝐸/ 𝐸𝐵𝐼𝑇𝐷𝐴_𝑡𝑖𝑔ℎ𝑡)𝑖,𝑡−1∗ 𝐶(𝑋𝐶)𝑖,𝑡 to test the difference between the value of cash
(excess cash) owned by firms with tight covenant last quarter and the value of cash
(excess cash) owned by firms with loose covenant last quarter.
According to the results of Table 6, tightness of current ratio covenant is
negatively related to firm’s market value, showing that additional limits on cash might shrink the value of firm. The interaction between tightness of current ratio covenant
and cash ratio does not changes firm’s value, indicating that initial tightness of covenant does not alter the value of cash. But, tightness of D/EBITDA covenant has
insignificant effect on firm’s value. In terms of excess cash, tightness of current
ratio covenant does not alter firm’s value neither by itself nor by its interaction with excess cash. Similarly, tightness of D/EBITDA covenant also have insignificant
effect on firm’s value after controlling for its interaction term with excess cash. To summarize, although the tightness of cash-related covenant significantly impact firm’s
cash holding policies, the tightness of covenant does not alter the value of cash.
Consistent with DeAngelo et al. (2002), even creditors use covenants to improve
firm’s management and to protect their right, borrowers still able to use some strategies like liquidating their working-capital to meet the requirement of covenant
but does not improve their management simultaneously.
18
Similarly, we also test the impact of loan’s violation using the same base model
as Equation (3). As shown in Table 7, violation of loan also has significantly
negative effect on firm’s value, implying that violation of loan is a bad signal for borrowers. In addition, the impact of violation on firm’s value is larger than
tightness of covenant, reflecting that violation is a worse signal which will
significantly hurt firm’s reputation. The value of cash decreased after violation of
covenant, but the value of excess cash will be increase at the same time. According
to the results of Table 4, firms usually stock their cash after covenant violations since
they probably loss capital source from banks. The decreased value of cash is
consistent with our prediction, implying that firms’ cash has less value when they retain much cash.
However, their excess cash becomes much valuable, which might be attributed to
two possible reasons. First, if excess cash is more likely be the cash saved as
precautionary motive against to agency motive, excess cash should be valuable for
firm. Second, the effect of loan’s violation will depends on types of firms, causing a
mixable result if we use specification supposed all firms have the same reaction after
violating covenants. Since it is difficult to test whether cash holdings by firms is
attributed to precautionary motive or agency motive, we test the second hypothesis by
partitioning all firms into different kinds of borrowers and examine our model on each
of them. The results are shown in the next Section.
[Invert Table 7 here]
3.4 The effect on firms with different growth opportunities
In this section, we further partition all firms into two subsamples using median of
19
market-to-book ratio significantly and consistently reduce the probability of tight
covenant, which implies that banks might give firms with high growth opportunities a
loose covenant to permit them a larger space for violation. Their findings
hypothesize that firms with different growth opportunities should be regarded as
different groups of borrowers by bank, and the firms with different growth
opportunities might act in a different way when they got pressure from covenant or
violation.
We show the results using base model Equations (3) and (4) in Table 8, which
only report the results of our central variables to save space. In Panel A of Table 8,
the results show that tightness of current ratio covenant significantly reduces the
firm’s incentives of holding cash no matter for firms with high growth opportunities or with low growth opportunities. The tightness of Debt/EBITDA covenant, in a
different way, causing high M/B firms to increase the cash reserves. Interestingly,
from the effect of loan’s violation on different types of firms, results indicate that the violation only induce high M/B firms increase their cash reserves, which result does
not be found from the low M/B firms. These results indicate that firms with high
growth opportunities are much care about their cash reserves and will be more likely
to stock cash for precautionary motive after losing the capital source from debt. In
contrary, the firms with low growth opportunities do not require such amount of cash
to prevent the unpredictable shocks.
In panel B of Table 8, we also test the effect of covenant tightness and covenant
violation on value of cash by different groups. The results show that the tightness of
current ratio covenant only signal a bad information for high M/B firms, implying that
tightness of covenant might be an alert for the firms with high growth opportunities.
20
cash for high M/B firms but further shrink its value. It seems that the cash-related
covenant has much more effect on the firms with high growth opportunities, and its
will be a bad signal for these firms that should much aware about their cash.
On the other hand, the violation of loan for high M/B firms has more serious
effect on their firms’ value than the effect on low M/B firms. The effect of violation of loan also is larger than the effect of tightness of current ratio covenant no matter we
focus on cash or excess cash. Finally, the results of Table 7 show that the value of
excess cash will be increase after violating covenant. After splitting firms into
different groups, results show that only the excess cash hold by firms with low growth
opportunities becomes much valuable after violation of loan, and this impact is
insignificant for the high M/B firms. Thus, although violation of covenant
significantly reduce the value of cash no matter hold by which types of firms, the
violation of covenant might induce low M/B firms decrease their excess cash reserves
that should further increase the value of excess cash.
[Invert Table 8 here]
3.5 Robustness check
The tests discussed thus far show the effects of loan tightness and loan violation
on firms’ cash policies and the value of cash. In this section we perform two
robustness checks on our results. First, we use some alternative proxies for covenant
tightness to test the sensitivity to measure of tightness. We use the approach of
Murfin (2012) to calculate a normalized distance between covenant threshold and
accounting ratio, and we also use the number of covenants in each contract as the
other alternative measure of tightness. All of the results are qualitatively similar to
21
Second, we cannot conclude that tightness of covenants does not altered by
borrowers’ cash reserves, i.e., creditors also might use firm’s cash reserves as their criteria for the tightness of covenants. In order to address this endogenous concern,
we estimate the propensity of covenant tightness by the specification of Table 3 of
Demiroglu and James (2010).8 For current ratio covenant, we employ the following
equation to estimate the propensity score of tightness of each covenant, then we use
these propensity tightness as instrumental variable to re-test our model.
𝐶𝑢𝑟𝑟𝑒𝑛𝑡𝑡𝑖𝑔ℎ𝑡𝑖,𝑡 = 0.8072 + 0.0117 𝑆𝑖𝑧𝑒𝑖,𝑡− 0.0329 𝐷𝑒𝑏𝑡𝑖,𝑡− 0.0625 𝐸𝐵𝐼𝑇𝐷𝐴𝑖,𝑡− 0.1734 𝐶𝐹 𝑣𝑜𝑙.𝑖,𝑡 −0.0509 𝑀/𝐵𝑖,𝑡− 0.0070 𝐶𝑢𝑟𝑟𝑒𝑛𝑡𝑖,𝑡− 0.0086 𝐶𝑢𝑟𝑟𝑒𝑛𝑡 𝑣𝑜𝑙.𝑖,𝑡− 0.0299 𝐴𝑔𝑒𝑖,𝑡 +0.0553 𝑆&𝑃 𝑟𝑎𝑡𝑒𝑑𝑖,𝑡− 0.0332 𝑁𝑢𝑚𝑒𝑏𝑒𝑟 𝑜𝑓 𝑝𝑎𝑠𝑡 𝑙𝑜𝑎𝑛𝑠 𝑤𝑖𝑡ℎ 𝑎𝑟𝑟𝑎𝑛𝑔𝑒𝑟𝑖,𝑡 −0.0018 𝑁𝑒𝑡 𝑝𝑒𝑟𝑐𝑒𝑛𝑡𝑎𝑔𝑒 𝑜𝑓 𝑏𝑎𝑛𝑘𝑠 𝑡𝑖𝑔ℎ𝑡𝑒𝑛𝑖𝑛𝑔 𝑠𝑡𝑎𝑛𝑑𝑎𝑟𝑑𝑠𝑖,𝑡 −0.0420 𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑙𝑒𝑛𝑑𝑒𝑟𝑠𝑖,𝑡− 0.0002 𝐿𝑜𝑎𝑛 𝑎𝑚𝑜𝑢𝑛𝑡𝑖,𝑡− 0.0053 𝑀𝑎𝑡𝑢𝑟𝑖𝑡𝑦𝑖,𝑡 +0.0648 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒 𝑝𝑟𝑖𝑐𝑖𝑛𝑔 𝑔𝑟𝑖𝑑𝑖,𝑡+ 𝜀𝑖,𝑡, (4) where 𝑆𝑖𝑧𝑒 is log of total assets; 𝐸𝐵𝐼𝑇𝐷𝐴 is EBITDA/Sales; 𝐶𝐹 𝑣𝑜𝑙. is cash flow volatility measured as standard deviation of annual cash flow; 𝐶𝑢𝑟𝑟𝑒𝑛𝑡 is current ratio; 𝐶𝑢𝑟𝑟𝑒𝑛𝑡 𝑣𝑜𝑙. is current ratio volatility measured as standard deviation of quarterly current ratio; 𝑆&𝑃 𝑟𝑎𝑡𝑒𝑑 is an indicator variable which equals one if firms are rated by S&P and zero otherwise; Net percentage of banks tightening standards is
quarterly data estimated by Federal reserve bank of ST. Louis;
𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒 𝑝𝑟𝑖𝑐𝑖𝑛𝑔 𝑔𝑟𝑖𝑑 is an indicator variable which equals to one if contract has performance pricing grid and zero otherwise. Age, Number of past loans with
arranger, Number of lenders, and Maturity are taken as log term, and Debt and Loan
amount are divided by assets. The fixed effect of covenant and collateral are
controlled in the model. The propensity of tightness of Debt/EBITDA covenant is
estimated by a similar model except including current ratio volatility.
8 We do not calculate the propensity of violation since we cannot identify major determinants of
22
Using propensity of tightness, results of Table 9 show that there has no
significant relationship between the propensity of covenant tightness and cash holding
policies except that the propensity Debt/EBITDA tightness will induce borrowers
increase cash reserves. However, the propensity of current ratio tightness still
retains a positive relationship with firms’ cash ratios. The insignificant results might
be driven by lack of observations or lack of representative of our instrumental
variable, which need further tests to make sure that we use a property measure.
Next, from the results of Table 10, the propensity of tightness of current ratio
covenant has significantly negative effect on firm’s value, which is consistent with
our previous findings. The excess cash hold by firms with tight covenant in the last
quarter will be much valuable, which has the same direction with previous findings
and it might be caused by the effect on low M/B firms according to Table 8. Finally,
the propensity of tightness of Debt/EBITDA covenant will reduce value of cash hold
by borrowers, which might primary be caused by the effect on high M/B firms. To
summarize, most of our results used propensity tightness as proxies are consistent
with the results found in the previous table.
4. Conclusion
In this paper, we test the impact of creditors on firms’ cash policies. We
employ initial tightness of covenants and loan’s violation as our two proxies for
intervene mechanisms of creditors, testing the impact of creditors’ intervention on
firms’ tendency of holding cash and value of cash. The results show that only the tightness of cash-related covenant limits firms’ inclination for cash stock. Violation
of loan greatly increase firm’s tendency to hold cash, which indicates that borrowers
23
separate sources of cash to see how firm will react after getting tight covenant or
violating covenant. Results indicate that, after violation of loan, borrowers lose
capital from banks, try to liquidate their net working-capital, and reduce payout ratio.
These findings are consistent with DeAngelo et al. (2002).
The tightness of tight cash-related covenant and violation are bad signals that
reduce firms’ value. However, the tightness of covenants does not alter the value of
cash, and violation of covenants will increase the value of excess cash that is
inconsistent with our prediction. We further split firms into two types of firms by
the median of M/B ratio, and we find that violation of covenant only induce high M/B
firms increase their cash reserves and it only improve the value of excess cash hold by
firms with low growth opportunities. In sum, our results show that tightness of
cash-related covenant significantly alter firm’s cash holding policies but its effect on
the value of cash differs for different types of firms. Violation of loan induce firms
to retain their cash, liquidate the net working-capital, and reduce the payout ratio for
24 Table 1 Summary statistics
This table reports the summary statistics of our sample from 1990 to 2010. Current ratio is defined as current assets divided by current liabilities. Debt/EBITDA is measured as long-term debt plus debt in current liabilities all divided by EBITDA. The estimating procedure of excess cash is shown in Section 2.2, and all definition of the other accounting variables are shown in Appendix. Lead arranger is the bank with ‘‘Yes’’ in ‘‘Lead Arranger Credit’’ item in LPC data. If lead banks have missing value, we equally distribute the remained shares to these banks. The definition of both Current_tight and D/EBITDA_tight are defined in 2.3, which are calculated by two approaches according to Demiroglu and James (2010) and Murfin (2012). Average_tight is an indicator variable which equals to one if the current_tight or Debt/EBITDA_tight is one and zero otherwise. All of the accounting variables are winsorized at the 1st and 99th percentiles to address potential problem of outliers.
Mean StD Min Max N
Panel A: Borrower characteristics
Total assets (millions) 1491.04 6056.19 0.0010 243,564 190,937
Current ratio 1.8786 1.2009 0.2369 8.4987 190,937
Debt/EBITDA 10.1931 24.5844 -97.8924 153.627 190,937 Market-to-Book ratio 2.6274 4.3426 0.5497 29.7747 190,937
Cash/Assets 0.1581 0.2115 0.0000 0.9220 190,937
Excess cash ratio 0.0007 0.0652 -0.1282 0.1897 190,937 Panel B: Loan characteristics
Loan amount (millions) 257.67 6563.36 0.05 30,000 13,847 Loan maturity (in months) 50.3420 29.3299 12.0000 420.000 13,847 Lead arranger shares 74.2872 33.8447 0.0000 100.000 13,847 All-in-drawn spreads (bps) 173.284 120.602 8.5000 1500.00 13,847
Covenant number 2.4754 1.0544 1.0000 8.0000 13,847
Panel C: Proxies for covenant tightness and violation
Current_tight (DJ) 0.6036 0.4895 0.0000 1.0000 691 D/EBITDA_tight (DJ) 0.6586 0.4745 0.0000 1.0000 751 Current_tight (Murfin) 0.0000 1.0000 -1.5391 1.8053 691 D/EBITDA_tight (Murfin) 0.0000 1.0000 -1.9758 2.1846 751 Average_tight (Murfin) -0.1306 1.0037 -5.5391 1.8053 1,344 Violation 0.0721 0.2587 0.0000 1.0000 94,984
25 Table 2 Summary statistics grouped by covenant tightness, correlation Panel A Summary statistics by covenant tightness (DJ)
Current ratio covenant Debt/EBITDA covenant
Variables Tight (1) Loose (2) Difference (1) - (2) Tight (1) Loose (2) Difference (1) - (2) Borrower characteristics Total assets 859.5 370.4 489.1 1093.1 1273.8 -180.7 Current ratio 2.1756 2.2900 -0.1144 2.2078 2.1370 0.0707 Debt/EBITDA 7.4194 9.1233 -1.7040 2.5748 2.3996 0.1752 Cash/Assets 0.0544 0.0848 -0.0304*** 0.0985 0.0950 0.0035 Excess cash/Assets -0.0452 -0.0270 -0.0182** -0.0505 -0.0448 -0.0057 Loan characteristics
Loan amount (millions) 107.37 106.32 1.05 180.7 258.3 -77.55
Loan maturity (in months) 49.8295 47.2785 2.5510 52.4510 63.2149 -10.7639***
Lead arranger shares 77.3166 77.7479 -0.4313 75.2242 73.1068 2.1174
All-in-drawn spreads (bps) 193.6 225.0 -31.4233* 178.7 220.3 -41.5706*** Covenant number 3.1186 3.3099 -0.1912 2.6138 2.9912 -0.3774*** Current_tight (Murfin) -0.7945 -0.0620 -0.7325*** -0.2815 -0.3691 0.0876 D/EBITDA_tight (Murfin) 0.0244 -0.4004 0.4248 -0.2179 0.7462 -0.9641*** Average_tight (Murfin) -0.7823 -0.0807 -0.7017*** -0.2157 0.7240 -0.9397*** N 417 274 493 258
26 Table 2 Continued
Panel B Fraction of tightness by industry (DJ)
Current ratio covenant Debt/EBITDA covenant
Agriculture, minerals, construction 0.720 0.796
Manufacturing 0.618 0.660
Transportation, communication 0.520 0.647
Trade-wholesale 0.628 0.722
Trade-retail 0.548 0.431
Services 0.406 0.669
Panel C Correlation of tightness
Current_tight (DJ) D/EBITDA_tight (DJ) Current_tight (Murfin) D/EBITDA_tight (Murfin)
Current_tight (DJ)
D/EBITDA_tight (DJ) 0.2067
Current_tight (Murfin) -0.3454*** -0.0846
D/EBITDA_tight (Murfin) 0.0219 -0.4652*** -0.1522
Average_tight (Murfin) -0.3457*** -0.4599*** 0.9747*** 0.9815***
This table reports the summary statistics grouped by covenant tightness. Panel A presents summary statistics by different criteria of tightness for all merged Compustat/LPC sample for 1990 to 2010. Current ratio is defined as current assets divided by current liabilities. Debt/EBITDA is measured as long-term debt plus debt in current liabilities all divided by EBITDA. Lead arranger is bank with ‘‘Yes’’ in the ‘‘Lead Arranger Credit’’ item in LPC data, while lead banks have missing value, we equally distribute the remained shares to these banks. The definition of both Current_tight and D/EBITDA_tight are defined in Section 2.3, which are calculated by two approaches according to Demiroglu and James (2010) and Murfin (2012). Average_tight is an indicator variable which equals to one if the current_tight or Debt/EBITDA_tight is one and zero otherwise. Panel B presents fraction of tightness by industry which category refers to 2-Degit SIC codes. Panel C presents correlation of our measures of covenant tightness. ***, **, and * stand for significance at the 1%, 5%, and 10% levels, respectively.
27 Table 3
Effect of covenant tightness on the tendency to hold cash
We explore the impact of covenant tightness on the tendency of holding cash in this set of regressions defined in Section 3.2. Dependent variable are the change in cash ratios (∆𝐶) and the change in excess cash (∆𝑋𝐶) which calculated procedure is shown in Section 2.2. Current_tight and D/EBITDA_tight are defined in Section 2.3, and all definition of the other variables are shown in Appendix. T-statistics employed robust individual standard errors are reported in parentheses. ***, **, and * stand for significance at the 1%, 5%, and 10% levels, respectively.
Dependent variable ∆𝐶𝑡 ∆𝑋𝐶𝑡 Intercept -0.0082** (-2.36) 0.0009 (0.32) -0.0045 (-1.39) 0.0004 (0.12) 𝐶𝑢𝑟𝑟𝑒𝑛𝑡_𝑡𝑖𝑔ℎ𝑡𝑡−1 -0.0124*** (-2.69) -0.0100** (-2.40) 𝐷/𝐸𝐵𝐼𝑇𝐷𝐴_𝑡𝑖𝑔ℎ𝑡𝑡−1 -0.0033 (-0.79) -0.0017 (-0.43) ∆𝑀/𝐵𝑡 0.0028 (0.25) 0.0149** (2.30) 0.0066 (0.68) 0.0087 (1.39) ∆𝑆𝐼𝑍𝐸𝑡 -0.0329 (-1.14) -0.0120 (-0.73) -0.0284 (-1.06) 0.0148 (0.94) ∆𝐶𝐹𝑡 -0.0145 (-0.15) 0.0281 (0.30) 0.0348 (0.39) 0.0426 (0.49) ∆𝑁𝑊𝐶𝑡 -0.0828 (-1.37) -0.1939*** (-3.67) -0.0318 (-0.60) -0.1702*** (-3.07) ∆𝐶𝐴𝑃𝑋𝑡 -0.0158 (-0.23) -0.0236 (-0.57) -0.2604*** (-0.2604) -0.4288*** (-8.33) ∆𝐿𝐸𝑉𝑡 -0.2248*** (-4.06) -0.1808*** (-4.33) -0.0129 (-0.23) -0.0514 (-1.36) ∆𝐼𝑆𝑡 0.1044 (0.05) 1.7247 (0.66) 2.1956 (1.24) 0.3994 (0.15) ∆𝑅&𝐷𝑡 0.0489 (0.19) 0.3431 (0.29) 0.2228 (1.30) 0.6787 (0.67) ∆𝐷𝑡 -0.0088 (-1.58) 0.0096*** (2.74) 0.0086 (1.64) 0.0080** (2.34) ∆𝑁𝐸𝑡 0.1324** (2.17) 0.1393* (1.91) 0.0288 (0.55) 0.0384 (0.83) ∆𝑁𝐷𝑡 0.0871 (1.57) -0.0066 (-0.99) 0.0714 (1.33) 0.0013 (0.34) N 669 742 669 742 Adj-R2 0.17 0.20 0.10 0.16
28 Table 4
Effect of violation on the tendency to hold cash
We show the impact of loan’s violation on the tendency of holding cash in this set of regressions defined in Section 3.2. Dependent variables are the change in cash ratios (∆𝐶) and the change in excess cash (∆𝑋𝐶) which calculated procedure is shown in Section 2.2. The information of violation is collected from Nini et al. (2009), and all definition of the other variables are shown in Appendix. T-statistics employed robust individual standard errors are reported in parentheses. ***, **, and * stand for significance at the 1%, 5%, and 10% levels, respectively.
Dependent variable ∆𝐶𝑡 ∆𝑋𝐶𝑡 Intercept -0.0025*** (-10.86) -0.0009*** (-3.86) 𝑉𝑖𝑜𝑙𝑎𝑡𝑖𝑜𝑛𝑡−1 0.0056*** (7.90) 0.0027*** (3.73) ∆𝑀/𝐵𝑡 -0.0009 (-1.64) 0.0078*** (14.27) ∆𝑆𝐼𝑍𝐸𝑡 0.0359*** (8.08) 0.0291*** (6.81) ∆𝐶𝐹𝑡 0.0137 (1.48) 0.0018 (0.18) ∆𝑁𝑊𝐶𝑡 -0.0582*** (-10.61) -0.0358*** (-7.00) ∆𝐶𝐴𝑃𝑋𝑡 -0.0470*** (-5.84) -0.3155*** (-32.20) ∆𝐿𝐸𝑉𝑡 -0.0929*** (-12.55) -0.1652*** (-22.06) ∆𝐼𝑆𝑡 0.5949** (2.39) 1.6361*** (6.82) ∆𝑅&𝐷𝑡 0.0001*** (2.84) 0.0001*** (-6.05) ∆𝐷𝑡 0.0012*** (4.37) 0.0015*** (-4.94) ∆𝑁𝐸𝑡 0.1339*** (18.63) 0.0781*** (10.27) ∆𝑁𝐷𝑡 0.0695*** (10.10) 0.0700*** (9.73) N 94,984 94,984 Adj-R2 0.04 0.12
29 Table 5
The trends of the primary sources of cash
This table presents the change in major sources of cash from year t-1 to year t. We follow the definition of McLean (2011) to separate different kinds of sources of cash, and we further estimate the net working-capital and cash dividends. All the results are divided by total assets. To save space, we only report the results which Current_tight and D/EBITDA_tight are defined as the procedure of Demiroglu and James (2010), since the results estimated by measures defined from Murfin (2012) is qualitatively similar. ***, **, and * stand for significance at the 1%, 5%, and 10% levels, respectively. Grouped by: 𝐶𝑢𝑟𝑟𝑒𝑛𝑡_𝑡𝑖𝑔ℎ𝑡𝑡−1 (DJ) Tight (1) Loose (2) Difference (1) – (2) Issuance -0.0013 -0.0001 -0.0012 Debt 0.0039 0.0246 -0.0207* Cash flow -0.0023 0.0473 -0.0065*
Net working capital -0.0057 -0.0283 0.0226***
Cash dividends -0.0004 0.0006 -0.0011 N 417 274 Grouped by: 𝐷/𝐸𝐵𝐼𝑇𝐷𝐴_𝑡𝑖𝑔ℎ𝑡𝑡−1 (DJ) Tight (1) Loose (2) Difference (1) - (2) Issuance 0.0004 0.0001 0.0003 Debt 0.0229 0.0174 0.0055 Cash flow -0.0068 -0.0080 0.0012
Net working capital -0.0111 -0.0231 0.0120*
Cash dividends 0.0006 -0.0002 0.0008 N 493 258 Grouped by: 𝑉𝑖𝑜𝑙𝑎𝑡𝑖𝑜𝑛𝑡−1 Violated (1) Not violated (2) Difference (1) – (2) Issuance -0.0001 0.0005 -0.0006 Debt -0.0132 0.0004 -0.0136*** Cash flow 0.0073 -0.0266 0.0340
Net working capital -0.0179 -0.0087 -0.0091**
Cash dividends -0.0005 -0.0002 -0.0003***
30 Table 6
Effect of covenant tightness on value of cash
In this table we examine the impact of covenant tightness on the value of cash and of excess cash by the specification defined in Section 3.3. Dependent variable of each regression is logarithm of market value, which is defined as book assets minus book equity plus market equity. Current_tight and D/EBITDA_tight are defined in Section 2.3, and all the definition of the other variables are shown in Appendix. 𝑋𝑡 is level of variable X in year t divided by level of total assets in year t; 𝑑𝑋𝑡 is the change in level of X from year t-2 to year t, 𝑋𝑡− 𝑋𝑡−2; 𝑑𝑋𝑡+2 is the change in the level of X from year t to year t+2, 𝑋𝑡+2− 𝑋𝑡. T-statistics employed robust individual standard errors are reported in parentheses. ***, **, and * stand for significance at the 1%, 5%, and 10% levels, respectively.
Dependent variable 𝐿𝑛(𝑀𝑉)𝑡 𝐿𝑛(𝑀𝑉)𝑡 Intercept 7.5369*** (6.99) 6.5232*** (6.77) 7.4878*** (6.53) 6.5138*** (6.72) 𝐶𝑢𝑟𝑟𝑒𝑛𝑡_𝑡𝑖𝑔ℎ𝑡𝑡−1 -0.1678 (-1.38) -0.0961** (-2.11) 𝐷/𝐸𝐵𝐼𝑇𝐷𝐴_𝑡𝑖𝑔ℎ𝑡𝑡−1 0.1012 (0.85) 0.0627 (1.03) 𝐶𝑢𝑟𝑟𝑒𝑛𝑡_𝑡𝑖𝑔ℎ𝑡𝑡−1 * 𝐶𝑡 -0.0190 (-0.65) 𝐷/𝐸𝐵𝐼𝑇𝐷𝐴_𝑡𝑖𝑔ℎ𝑡𝑡−1 * 𝐶𝑡 0.0091 (0.30) 𝐶𝑢𝑟𝑟𝑒𝑛𝑡_𝑡𝑖𝑔ℎ𝑡𝑡−1 * 𝑋𝐶𝑡 0.0873 (0.17) 𝐷/𝐸𝐵𝐼𝑇𝐷𝐴_𝑡𝑖𝑔ℎ𝑡𝑡−1 * 𝑋𝐶𝑡 -0.2622 (-0.52) 𝐿𝑛(𝐶)𝑡 0.0469*** (2.76) 12.9927 (0.31) 𝐿𝑛(𝑋𝐶)𝑡 0.7118** (2.02) 0.5427 (1.41) 𝐿𝑛(𝐶𝐹)𝑡 34.2288 (1.44) 3.5248 (0.11) 28.9944 (1.27) 11.5126 (0.28) ∆𝐶𝐹𝑡 2.6544 (0.16) 3.5248 (0.11) 0.2429 (0.02) 4.3438 (0.13) ∆𝐶𝐹𝑡+2 18.5789 (0.80) 6.0936 (0.23) 12.2208 (0.55) 5.6103 (0.22) ∆𝑁𝐴𝑡 0.7185*** (4.49) 0.3552** (2.19) 0.6947*** (4.34) 0.3045* (1.92) ∆𝑁𝐴𝑡+2 0.3482*** (4.46) 0.3120*** (4.02) 0.3575*** (4.48) 0.3175*** (4.06) 𝐿𝑛(𝑅&𝐷)𝑡 39.3477*** (2.70) 50.7051 (0.66) 39.8600*** (2.79) 66.0652 (0.85) ∆𝑅&𝐷𝑡 -69.5772 (-1.43) -76.7293 (-0.99) -63.0656 (-1.41) -83.7709 (-1.05) ∆𝑅&𝐷𝑡+2 -9.0100 (-0.54) -14.8461 (-0.88) -0.1909 (-0.01) -12.0554 (-0.73) 𝐿𝑛(𝑋𝐼𝑁𝑇)𝑡 -0.0240 (-1.02) -0.0164 (-0.68) -0.0543** (-2.26) -0.0317 (-1.36) ∆𝑋𝐼𝑁𝑇𝑡 -6.4610 (-1.01) -9.3151** (-2.02) -2.9626 (-0.46) -8.8872* (-1.94) ∆𝑋𝐼𝑁𝑇𝑡+2 -4.9814* (-1.95) -11.3448*** (-3.07) -4.6029* (-1.77) -11.0461*** (-3.06) 𝐿𝑛(𝐷𝑖𝑣)𝑡 2.8032 (1.41) 9.0199*** (2.94) 1.8682 (0.88) 8.2809*** (2.73) ∆𝐷𝑖𝑣𝑡 -0.6506*** (-4.03) -1.1069 (-1.07) -0.6631*** (-4.16) -0.8997 (-0.86)
31 ∆𝐷𝑖𝑣𝑡+2 1.1705 (0.85) 3.5247* (1.93) 0.7323 (0.52) 3.4619* (1.84) ∆𝑀𝑉𝑡+2 -3.2242*** (-7.00) -2.7134*** (-6.80) -3.3367*** (-6.90) -2.8203*** (-6.95) N 622 706 622 706 Adj-R2 0.33 0.25 0.33 0.24
32 Table 7
Effect of violation on value of cash
In this table we examine the impact of loan’s violation on the value of cash and of excess cash. Dependent variable is logarithm of market value, which is defined as book assets minus book equity plus market equity. The information of violation is collected from Nini et al. (2009), and all the definition of the other variables are shown in Appendix. 𝑋𝑡 is the level of variable X in year t divided by the level of total assets in year t; 𝑑𝑋𝑡 is the change in the level of X from year t-2 to year t, 𝑋𝑡− 𝑋𝑡−2; 𝑑𝑋𝑡+2 is the change in the level of X from year t to year t+2, 𝑋𝑡+2− 𝑋𝑡. T-statistics employed robust individual standard errors are reported in parentheses. ***, **, and * stand for significance at the 1%, 5%, and 10% levels, respectively.
Dependent variable 𝐿𝑛(𝑀𝑉)𝑡 𝐿𝑛(𝑀𝑉)𝑡 Intercept 2.0137*** (8.11) 8.6837*** (4.57) 𝑉𝑖𝑜𝑙𝑎𝑡𝑖𝑜𝑛𝑡−1 -0.4461*** (-9.31) -0.2488*** (-12.78) 𝑉𝑖𝑜𝑙𝑎𝑡𝑖𝑜𝑛𝑡−1 * 𝐶𝑡 -0.0519*** (-5.09) 𝑉𝑖𝑜𝑙𝑎𝑡𝑖𝑜𝑛𝑡−1 * 𝑋𝐶𝑡 0.5272*** (3.03) 𝐿𝑛(𝐶)𝑡 0.1007*** (19.59) 𝐿𝑛(𝑋𝐶)𝑡 0.7934*** (12.05) 𝐿𝑛(𝐶𝐹)𝑡 -1.1311*** (-7.95) -3.1507*** (-3.93) ∆𝐶𝐹𝑡 0.6377*** (6.12) 0.0156 (0.13) ∆𝐶𝐹𝑡+2 -0.7530*** (-6.43) -0.8679*** (-7.95) ∆𝑁𝐴𝑡 0.2355*** (12.38) 0.2047*** (10.22) ∆𝑁𝐴𝑡+2 0.3360*** (17.71) 0.3224*** (15.15) 𝐿𝑛(𝑅&𝐷)𝑡 2.6582*** (8.72) 3.2226*** (11.03) ∆𝑅&𝐷𝑡 -0.4396*** (-2.93) -0.7368*** (-4.31) ∆𝑅&𝐷𝑡+2 0.2920*** (2.64) 0.2261* (1.83) 𝐿𝑛(𝑋𝐼𝑁𝑇)𝑡 0.0017 (0.24) -0.0221*** (-3.21) ∆𝑋𝐼𝑁𝑇𝑡 1.0820*** (3.62) 0.7642** (2.45) ∆𝑋𝐼𝑁𝑇𝑡+2 0.0430 (0.21) -0.1727 (-0.79) 𝐿𝑛(𝐷𝑖𝑣)𝑡 0.1315 (0.81) -0.1231 (-0.86) ∆𝐷𝑖𝑣𝑡 -0.1121 (-1.40) -0.1184 (-1.30) ∆𝐷𝑖𝑣𝑡+2 -0.149 (-1.62) -0.3042*** (-3.30) ∆𝑀𝑉𝑡+2 -0.6566*** (-5.96) -0.5867*** (-5.95) N 78,502 78,502 Adj-R2 0.18 0.16
33 Table 8
Effect of covenant tightness and violation on cash policies and value of cash (By different groups)
This table presents the impact of covenant tightness and loan’s violation both on cash policies and on value of cash by different groups of firms. We partition firms into two groups by median of M/B, and re-run the Equation (3) and Equation (4) for different types of firms. To save space, we only report the results of central variables. T-statistics employed robust individual standard errors are reported in parentheses. ***, **, and * stand for significance at the 1%, 5%, and 10% levels, respectively.
Panel A:Effect of covenant tightness and violation on cash policies (by different groups)
Dependent variable ∆𝐶𝑡 ∆𝑋𝐶𝑡
High M/B Low M/B High M/B Low M/B
Panel A.1: Covenant tightness
𝐶𝑢𝑟𝑟𝑒𝑛𝑡_𝑡𝑖𝑔ℎ𝑡𝑡−1 -0.0186*** (-2.82) -0.0083** (-2.07) -0.0145** (-2.24) -0.0085** (-2.15) 𝐷/𝐸𝐵𝐼𝑇𝐷𝐴_𝑡𝑖𝑔ℎ𝑡𝑡−1 0.0056* (1.74) -0.0035 (-0.79) 0.0042 (1.28) 0.0002 (0.03)
Panel A.2: Covenant violation
𝑉𝑖𝑜𝑙𝑎𝑡𝑖𝑜𝑛𝑡−1 0.0127*** (7.16) 0.0008 (1.29) 0.0094*** (5.61) -0.0012* (-1.73)
Panel B:Effect of covenant tightness and violation on value of cash (by different groups)
Dependent variable 𝐿𝑛(𝑀𝑉)𝑡
(𝐶𝑡)
𝐿𝑛(𝑀𝑉)𝑡
(𝑋𝐶𝑡)
High M/B Low M/B High M/B Low M/B
Panel B.1: Covenant tightness
𝐶𝑢𝑟𝑟𝑒𝑛𝑡_𝑡𝑖𝑔ℎ𝑡𝑡−1 -0.1772* (-1.70) -0.0438 (-0.45) -0.1029** (-2.13) -0.0218 (-0.59) 𝐷/𝐸𝐵𝐼𝑇𝐷𝐴_𝑡𝑖𝑔ℎ𝑡𝑡−1 0.0246 (0.21) -0.0212 (-0.17) 0.0737 (1.18) 0.0025 (0.04) 𝐶𝑢𝑟𝑟𝑒𝑛𝑡_𝑡𝑖𝑔ℎ𝑡𝑡−1*𝐶𝑡 (𝑋𝐶𝑡) -0.0408* (-1.66) -0.0032 (-0.15) -1.2131** (-2.37) 0.2096 (0.45) 𝐷/𝐸𝐵𝐼𝑇𝐷𝐴_𝑡𝑖𝑔ℎ𝑡𝑡−1*𝐶𝑡 (𝑋𝐶𝑡) -0.0222 (-0.72) 0.0118 (0.39) -0.4259 (-0.82) 1.0464* (1.86)
Panel B.2: Covenant violation
𝑉𝑖𝑜𝑙𝑎𝑡𝑖𝑜𝑛𝑡−1 -0.4122*** (-6.64) -0.2024*** (-4.35) -0.2376*** (-6.95) -0.0872*** (-5.29) 𝑉𝑖𝑜𝑙𝑎𝑡𝑖𝑜𝑛𝑡−1 * 𝐶𝑡 (𝑋𝐶𝑡) -0.0540*** (-3.12) -0.0204** (-2.13) 0.2542 (1.09) 0.2899* (1.82)
34 Table 9
Robustness test - Effect of propensity of covenant tightness on cash policies In this table we estimate the effect of propensity of covenant tightness on cash policies following the procedure defined in Section 3.5. Dependent variable of each regression is the change in cash ratios (∆𝐶) and the change in excess cash (∆𝑋𝐶) which calculated procedure is shown in Section 2.2. All the definition of variables are shown in Appendix. T-statistics employed robust individual standard errors are reported in parentheses. ***, **, and * stand for significance at the 1%, 5%, and 10% levels, respectively. Dependent variable ∆𝐶𝑡 ∆𝑋𝐶𝑡 Intercept 0.0196 (0.74) 0.0009 (0.32) 0.0191 (0.64) -0.0034 (-0.42) 𝐶𝑢𝑟𝑟𝑒𝑛𝑡_𝑡𝑖𝑔ℎ𝑡𝑡−1 -0.00388 (-1.05) -0.0342 (-0.84) 𝐷/𝐸𝐵𝐼𝑇𝐷𝐴_𝑡𝑖𝑔ℎ𝑡𝑡−1 0.0218* (1.76) 0.0177 (1.41) ∆𝑀/𝐵𝑡 0.0032 (0.39) 0.0110* (1.69) -0.0087 (-1.17) 0.0158*** (2.65) ∆𝑆𝐼𝑍𝐸𝑡 -0.0321 (-0.86) -0.0677*** (-3.10) 0.0415 (1.23) -0.0634*** (-2.97) ∆𝐶𝐹𝑡 0.0547 (0.27) 0.1391 (1.43) -0.0588 (-0.30) 0.0479 (0.50) ∆𝑁𝑊𝐶𝑡 -0.1176 (-0.80) -0.4580*** (-6.34) -0.0122 (-0.08) -0.3891*** (-5.37) ∆𝐶𝐴𝑃𝑋𝑡 -0.1042** (-2.40) 0.0273 (0.58) -0.3542*** (8.21) -0.5474*** (-10.79) ∆𝐿𝐸𝑉𝑡 -0.1765* (-1.85) -0.3820*** (-5.77) 0.1151 (1.19) -0.1207* (-1.89) ∆𝐼𝑆𝑡 0.0775 (0.57) -0.1009 (-1.16) 0.0264 (0.17) -0.1094 (-1.27) ∆𝑅&𝐷𝑡 0.6267*** (3.35) 1.4225 (1.07) 0.2775 (1.50) 0.6831 (0.55) ∆𝐷𝑡 0.0036 (1.00) -0.0020 (-0.68) 0.0006 (0.16) -0.0036 (-1.17) ∆𝑁𝐸𝑡 0.1756** (2.23) 0.0844* (1.88) -0.0187 (-0.25) 0.1137* (1.77) ∆𝑁𝐷𝑡 0.1293** (2.13) 0.2431*** (3.81) -0.0342 (-0.84) 0.0177 (1.41) N 657 723 657 723 Adj-R2 0.17 0.33 0.18 0.37
