The value of financial flexibility and corporate financial

We find that firms for which shareholders consider financial flexibility more valuable have lower dividend payouts, prefer share repurchases to dividends, and exhibit lower leverage rati

<= |CORPORATE —= |FINANCE

Journal of Corporate Finance

Marc Steffen Rapp **, Thomas Schmid ”, Daniel Urban”

* Philipps-Universitat Marburg, School of Business and Economics, Accounting and Finance Group, Germany

> Technische Universitéit Mtinchen, Department of Financial Management and Capital Markets, Germany

Article history: We propose a novel approach to measure the value that shareholders assign to financial flexibility In

Received 15 December 2013 contrast to existing proxies for financial constraints, our measure is market-based, forward-looking Received in revised form 20 August 2014

Accepted 20 August 2014

Available online 28 August 2014

and not directly influenced by past financial decisions We find that firms for which shareholders consider financial flexibility more valuable have lower dividend payouts, prefer share repurchases

to dividends, and exhibit lower leverage ratios Moreover, these firms tend to accumulate more cash Our analysis contributes to the growing literature on financial flexibility and indicates that— _ ¬ in line with prior survey evidence—financial flexibility considerations shape corporate financial C35 policy

Keywords:

Financial flexibility

Payout policy

Capital structure

Cash holdings

© 2014 Elsevier B.V All rights reserved

1 Introduction

Understanding firms’ financial decisions is a key challenge in financial economics research Over the last decades, different ap- proaches such as agency cost or market imperfections have been put forward However, survey evidence among firms’ decision makers indicates that there is another factor that has gained only little attention in the academic literature so far: financial flexibility

In fact, CFOs claim that financial flexibility considerations are of first order importance with respect to firms’ financial policy deci- sions (Brounen et al., 2006; Graham and Harvey, 2001; Pinegar and Wilbricht, 1989) Gamba and Triantis (2008, p 2263) define finan- cial flexibility as “the ability of a firm to access and restructure its financing at a low cost” Adopting that view, there are two channels through which financial flexibility becomes valuable for firms First, financial flexibility can mitigate underinvestment problems in case of restricted access to capital Second, it can help to avoid costs associated with financial distress

One possible reason why empirical evidence in this context is sparse is because the value of financial flexibility for firms is not directly observable As a consequence, previous empirical literature focused mainly on empirical proxies for a firm's financial constraints.’ How- ever, these proxies measure the level, not the value of financial flexibility The level of financial flexibility is, however, endogenously determined by prior financial decisions Thus, such proxies cannot help in explaining why some firms choose financial policies

* We thank Lixiong Guo, Bernd Hayo, Sigitas Karpavicius, Christoph Kaserer, Sebastian Miller, Jan B Siewert, an anonymous referee, as well as participants of the 20th Annual Meeting of the German Finance Association in Wuppertal, the Frontiers in Finance 2012 conference in Warwick, the 2012 China International Conference in Finance

in Chongqing, and the 2012 Annual Meeting of the German Academic Association for Business Research in Bolzano for helpful comments This research was supported by the Deutsche Forschungsgemeinschaft through the Collaborative Research Center (CRC) 649 “Economic Risk” and the Research Data Center (RDC) at Humboldt University Berlin The usual caveat applies

* Corresponding author at: Am Plan 1, D-35037, Marburg Germany Tel.: + 49 6421 28 22019

E-mail addresses: msr@m-s-rapp.de (MLS Rapp), t.schmid@tum.de (T Schmid), daniel.urban@tum.de (D Urban)

| Measures for financial constraints are, among others, the investment-cash flow sensitivity (Fazzari et al., 1988), the cash-cash flow sensitivity (Almeida et al., 2004), the Kaplan-Zingales Index (Kaplan and Zingales, 1997), and the Whited—Wu Index (Whited and Wu, 2009)

http://dx.doi.org/10.1016/Jjcorpfin.2014.08.004

0929-1199/© 2014 Elsevier B.V All rights reserved.

that result in high or low levels of financial flexibility Furthermore, these measures are often not forward-looking This, for in- stance, may explain the low explanatory power of investment-cash flow sensitivity during the recent financial crisis (e.g., Chen and

Chen, 2012)

In this paper, we propose a novel empirical approach to estimate the value of financial flexibility from the perspective of a firm's shareholders Our measure, which we refer to as the value of financial flexibility (VOFF), aggregates empirical proxies of theoretically motivated determinants of the value of financial flexibility into a single measure Aggregation is done using capital market based, forward-looking weights based on the value-relevance of unexpected changes in cash holding, which we consider the most liquid means of a firm's financial flexibility Thus, the VOFF approximates the value a firm's shareholders assign to its financial flexibility, not the current level of a firm's financial flexibility The fact that our measure is market-based and forward-looking separates it from existing proxies in this context Also, unlike measures based on the level of financial flexibility, the VOFF is not directly influenced

by past financial decisions This allows us to use it in empirical models explaining corporate financial policy In the context of capital structure, for instance, today's leverage determines a firm's future borrowing capacity and thus its level of financial flexibility However, the question is why some firms preserve more of their debt capacity than others One possible explanation is that shareholders assign a higher VOFF to these firms

To empirically estimate the VOFF and to test its impact on firms’ financial policy decisions, we proceed in three steps First, we identify empirical proxies for the five factors that—according to the theoretical model by Gamba and Triantis (2008)—determine the value of financial flexibility The model of Gamba and Triantis (2008) indicates that besides a firm's growth opportunities and its profitability, the effective costs of holding cash, the cost of external financing, as well as the reversibility of capital determine the value of financial flexibility Thus, these factors reflect (i) the firm's business model and (ii) its external environment In contrast, they are independent of the firm's current financial policies

Second, we combine these five factors into a single measure To obtain their weights, we adopt the approach proposed by Faulkender and Wang (2006) to determine the marginal value of cash and analyze capital market reactions to (unexpected) changes

in a firm's cash holdings As cash is the most flexible financial means for a firm, we expect that this reaction depends on how valuable the shareholders consider financial flexibility for the specific firm Thus, we regress changes in market capitalization on the five factors that interacted with (unexpected) changes in cash Using the coefficients from that regression as a weighting-mechanism, we then aggregate the five factors into a single number for each firm year For example, if firms with high growth opportunities have high (low) positive abnormal returns when cash holdings increase, we would assign a high (low) weight to growth oppor- tunities Examining US listed firms from 1988 to 2010, we find substantial variation in the VOFF across firms We then challenge our measure in an event study setting Specifically, we test whether a firm's stock market return around the default of Lehman Brothers is sensitive to its VOFF As expected, firms with a high VOFF suffered more from the breakdown of (outside) financing opportunities

Third, we examine whether financial flexibility considerations help to explain corporate financial policies Therefore, we augment standard financial policy regressions with our VOFF measure We find that the VOFF plays a central role for payout, capital structure, and liquidity decisions In particular, firms with a higher VOFF exhibit (i) a lower propensity to pay dividends and lower dividend pay- out ratios, (ii) a higher propensity to omit a dividend, (iii) a preference for share repurchases over dividends when cash is distributed

to shareholders, (iv) lower leverage ratios, and (v) accumulate more cash All these results are of high statistical significance and in line with theoretical predictions Using the Jobs and Growth Tax Relief Reconciliation Act as an exogenous shock regarding a firm's payout policy allows us to provide evidence that causality runs from VOFF to financial policy decision, and not vice versa Overall, our results thus provide strong empirical evidence that—in line with survey evidence—financial flexibility considerations play an im- portant role for corporate financial policy

The idea to adopt the marginal value of cash measure proposed by Faulkender and Wang (2006) to study corporate behavior is not completely new Liu and Mauer (2011), for instance, use it to study CEO incentives Regarding corporate financial policies, Clark (2010) proposes to use the marginal value of cash to study corporate capital structure decisions Specifically, Clark (2010) shows that firms with a high marginal value of cash tend to have lower debt levels and—in case they raise external financing—are more likely

to issue equity While this is an important contribution, our approach differs with regard to two important aspects First, while Clark (2010) relies on the factors proposed by Faulkender and Wang (2006) for the marginal value of cash measure to proxy the (marginal) value of financial flexibility, we propose another measure that relies on theoretically motivated determinants of the value of financial flexibility following Gamba and Triantis (2008) This difference has important implications Faulkender and Wang (2006) define their marginal value of cash as a constant plus a weighted sum of the firm's changes in cash holdings and its leverage Thus, an approach to study corporate financial policies using the marginal value of cash measure is exposed to endogeneity issues.” Also, the Faulkender and Wang (2006) measure is a relative measure since it is sensitive to the level of cash holdings of a firm With our approach, which relies on five factors not influenced by current corporate financial policies, we aim to avoid both problems.’ Second,

we extend the analysis towards all three dimensions of corporate financial policies: payout decisions, capital structure decisions, as well

? Clark (2010) approaches this problem by using instruments obtained from Frank and Goyal (2009) for a firm's leverage

3 It is important to note, however, that while our theoretical arguments refer to the (absolute) value of financial flexibility, our empirical proxy indeed is a marginal measure More precisely, borrowing from the model of Gamba and Triantis (2008) we argue that the (overall) value that shareholders assign to a firm's financial flex- ibility is determined by (i) the firm's business model and (ii) its external environment As such, our measure is firm-specific but independent of the firm's financial pol- icies, which motivates our notion to call it the value of financial flexibility in order to separate it from the (with regard to the firm's financial policies) marginal value of financial flexibility, which reflects the current financial policy choice Still, our measure is—by construction—a marginal measure in the sense that it relies on the coef- ficient estimates based on a marginal increase in cash.

as cash accumulation decisions This allows us a more thorough understanding of the link between financial flexibility and corporate financial decisions

The remainder of this paper is structured as follows: Section 2 describes the sample construction In Section 3, we explain how we derive the VOFF, calculate its value for each firm year, and perform an event study around the collapse of Lehman Brothers In Section 4, we develop testable hypotheses for the impact of the VOFF on financial policy and empirically test these hypotheses The last section concludes; it summarizes the main results, discusses possible implications, and provides avenues for future research

2 Sample

To construct the sample, we obtain accounting information from Compustat and capital market data from CRSP We start with all ac- tive and inactive U.S firms that are traded at public stock exchanges in at least one year between 1988 and 2010 We drop firms not listed

at the NYSE, AMEX, or NASDAQ Financial firms (SIC code between 6000 and 6999) and firms with missing SIC code are also eliminated from the sample We further exclude firms whose shares are not traded as ordinary common stock by removing observations whose CSRP Share Code is missing or not equal to 10 or 11 Following Faulkender and Wang (2006), we remove observations with negative market capitalization, missing or zero total assets, negative dividends, and firms where cash holdings exceed the book value of assets

We convert the data to real values in 2005 dollars using the consumer price index obtained from the OECD Finally, we winsorize all ratios at the 1st and 99th percentiles in order to reduce the effect of outliers

3 The value of financial flexibility

Gamba and Triantis (2008) develop a theoretical model to study the value of financial flexibility Thereby, they define financial flexibility as “the ability of a firm to access and restructure its financing at a low cost” (p 2263) and argue that financial flexibility is valuable for firms since it mitigates underinvestment caused by lack of financing opportunities and even avoids financial distress In their analysis, the authors identify five determinants of the value of financial flexibility Specifically, their model indicates that besides

a firm's growth opportunities and its profitability, the effective costs of holding cash, the cost of external financing, as well as the re- versibility of capital determine the value of financial flexibility

In this section, we derive empirical proxies for these factors and explain our econometric approach to derive the VOFF empirically After that, we calculate the VOFF for each firm year and test our measure in an event study around the collapse of Lehman Brothers 3.1, Determinants of a firm's value of financial flexibility

In their theoretical model, Gamba and Triantis (2008) do not describe empirical measures for the five determinants of the VOFF Thus, we first identify empirical proxies for each determinant:

Growth opportunities Gamba and Triantis argue that growth opportunities affect the VOFF Higher growth opportunities

are expected to increase the VOFF since they are correlated with unexpected cash flow shocks Thus, financial flexibility is more valuable when growth opportunities are high In line with previous research, we approximate growth opportunities by a firm's one-year logarithmic sales growth rate, SGR¡¿

Profitability Gamba and Triantis also argue that firms with a higher profitability have a lower VOFF because they are

better able to rely on internally generated cash To measure the profitability of a firm, we employ the ratio of changes in firm earnings, AE;,, to its lagged market capitalization, Mj; — 1, where F;, is earnings before extraordinary items plus interest and deferred tax credits We rely on changes instead of absolute earnings levels to capture inter-temporary differences in a firm's operating health.*

Effective costs of holding cash In the model of Gamba and Triantis, the effective costs of holding cash, T, determined by the level of

corporate and personal tax rates, affect the VOFF These costs can be estimated by comparing the

taxation of interest at the individual level T; relative to the taxation at the corporate level Tc, We

thus define T;; = Tc,,/Ti,, where Te, indicates the cash effective tax rate on corporate level following Chen et al (2010) Specifically, we calculate Tc,, for firm i in year ¢ as the ratio of the firm's cash taxes paid to its pretax income in year t,° and T;, is the annual average federal income tax rates for the U.S median income household.°

If T;, were greater than one, for instance, this would imply that interest income is taxed more heavily

at the corporate level, resulting in higher effective costs of holding cash at the corporate level In other

4 Tn line with Faulkender and Wang (2006), we deflate changes in earnings by the lagged market value of equity Our results remain unchanged when we employ total assets in the denominator Our results are also robust to different empirical specifications of firm earnings

° The cash effective tax rate is set to zero when cash taxes paid are zero or negative Also, it is truncated to the range [0, 1]

® Data on individual taxation comes from the U.S Census Bureau Historical Income Tables (cf http://www.census.gov/hhes/www/income/data/historical/families/index html).

words, it is more beneficial to shareholders if they hold cash instead of the company Hence, higher effective costs of holding cash decrease a firm's VOFF.’

Cost of external financing Gamba and Triantis suggest that a firm's cost of external financing, i.e., flotation cost, also affects the

VOFF We approximate a firm's cost of external financing by the bid-ask spread of its stock Following

Anderson et al (2009) we calculate Spreadj as the average of all trades for a firm from the third Wednesday each month during a firm's fiscal year

There are two possible channels, one direct and one indirect, through which costs of external financing influence a firm's VOFF On the one hand, a firm with higher costs of external financing may have a higher VOFF because external financing is more difficult and/or time-consuming On the other hand, higher cost of external financing may reflect a higher level of agency problems due to the fear of managerial expropriation (e.g., Jensen, 1986; Rozeff, 1982) Thus, to reduce agency cost, shareholders may also attribute a lower value to financial slack when agency costs are high Our empirical analysis sheds light on the relative importance of these two possible channels

Reversibility of capital Finally, Gamba and Triantis argue that the VOFF is affected by the reversibility of a firm's capital Share-

holders of firms that are able to sell their assets quickly and with a low discount should attribute less value to financial flexibility We approximate the reversibility of capital by a firm's tangibility, Tang; ,, since tangible assets can be sold more easily than intangible assets.® This ratio is defined as property, plant, and equipment deflated by total assets Consequently, a higher reversibility of capital is expected

to decrease the VOFF

3.2 Construction of the value of financial flexibility

After identifying the five empirical proxies for the determinants of the VOFF, we now aggregate them into a single measure For this,

we proceed in two steps In step one, we analyze capital market reactions to changes in cash holdings, dependent on these factors Based

on that, we attribute weights to the five factors In step two, we calculate the final measure for the VOFF as the weighted sum of the five factors We calculate this measure separately for each firm and each year

In step one, we determine the weights to aggregate the five determinants Therefore, we analyze capital market reactions to changes

in a firm's cash holdings Cash holdings can be seen as the most flexible financial means for a firm Thus, we expect that the capital market reaction to changes in cash holdings depends on how valuable shareholders consider financial flexibility for a particular firm For example, if firms with high growth opportunities yield high (low) positive abnormal returns when cash holdings increase,

we assign a high (low) weight to growth opportunities In particular, we regress firm-specific annual cumulative abnormal returns

on the five determinants of a firm's VOFF that interacted with changes in cash controlling for firm characteristics as well as industry and year fixed effects

Similar to Faulkender and Wang (2006), we calculate a firm's abnormal return as the one-year stock return relative to the return of its benchmark portfolio Therefore, each stock is assigned to one of the 25 Fama and French (1993) value-weighted size and book-to- market portfolios based on its size and book-to-market ratio Thus, the annual cumulative abnormal return of firm i in fiscal year f is derived by subtracting annual return of the style-matched benchmark portfolio in year t (R?,) from the annual return the firm's stock

in year f (r¡¿).”

Since our dependent variable reflects unexpected changes in firms’ market values, we interact the five financial flexibility de- terminants with unexpected changes in cash Therefore, we follow an approach suggested by Almeida et al (2004) to calculate unexpected changes in cash, i.e., differences between realized and expected cash changes Assuming that the market has only access to information related to a firm's last fiscal year, expected cash holdings can then be estimated by the following regression equation:

,

where C;;is cash and marketable securities in year f M;; — ; is the lagged market value of equity TobQ;; _ + is lagged Tobin's Q defined

as the sum of total assets and market capitalization less the book value of common equity deflated by total assets, CFAL;; _ 1 is lagged earnings before extraordinary income and depreciation, but after dividends, and Logsize;, — ; is the natural logarithm of (inflation- adjusted) lagged total assets in millions of $US (2005 = 100)

7 when estimating the relative taxation of interest at the shareholder and the firm-level, we explicitly do not take personal tax rates on equity into account In doing

so, our approach approximates a lower bound for the taxation of interest at the firm-level because a firm's interest income is also subject to shareholder taxation in case

a firm distributes its earnings

Š In a related study, Campello and Hackbarth (2012) show that firms may benefit from a high fraction of tangible assets because it relaxes financial constraints and

allows for further investment

° Returns are calculated using monthly returns Portfolio break points and monthly portfolio returns are from Kenneth R French's web site (http://mba.tuck dartmouth.edu/pages/faculty/ken.french) Specifically, for each year, we assign every firm into one of the 25 size and book-to-market portfolios based on independent sorts on size and the book-to-market ratio at the beginning of a firm’s fiscal year Following Ince and Porter (2006), we drop observations with missing returns or ob- servations whose lagged stock price is less than $US 1 in order to mitigate the effects of penny stocks.

With AC indicating unexpected changes in cash holdings, i.e., the difference between actual and expected cash holdings, and a firm's abnormal stock return, r;; — R?,, our regression model that is used to study capital market reactions to changes in cash holdings

is as follows:

AC, AE rig Rit = Yo + V1 Mi, —+;SGR¡, + s1” + VAT; + YsSpTead,¿ + + Tang;,

-ÁC, AE;, ed ACi,

+;5GR;¿¡ x —+s —X + Vol it X —+ oSpread; tx

"Miya Mit-1) Mita Mit — (2)

AC, cy + Cita + + ANA\ ¢ 4 + ARD,, cy + Ali, L TAD,,

M '2Mjc + By, Mit 14 Mie 15 Mie 16 Mie 1

+ T¡¡1An8;¡ x ——

NF,

As previously stated, to study capital market reactions with respect to changes in cash holdings, we regress a firm's abnormal stock return on (i) unexpected changes in cash holdings, (ii) the empirical proxies for the five determinants suggested by Gamba and Triantis (2008), (iii) their interactions with unexpected changes in cash holdings, (iv) a set of firm-specific control variables, as well as (iv) industry and year dummies Firm-specific variables aim to control for sources of abnormal returns (other than changes

in cash) that may somehow be correlated with unexpected changes in cash These firm characteristics may be grouped into two categories

A firm's investment policy is approximated by past cash holdings, Cje@2q, as Well as changes in net assets (NAj), defined as total

assets minus cash holdings, and research and development expenses (RDj;), which are set to zero if missing Its financial structure

is covered by interest expense (/i,:), common dividends (Dj;), market leverage (L;,), defined as the ratio of long-term debt and short-term debt to the sum of long-term debt, short-term debt, and the market value of equity, as well as net financing (NF;,) defined

as total equity issuance minus repurchases plus debt issuance minus debt redemption To avoid that the results mainly are driven by large firms, all firm-specific control variable (with the exception of L;;) are deflated by the lagged market value of equity, Mic — 1, which is consistent with the denominator on the left-hand side Finally, Zi

In step two, we calculate our measure of interest, VOFF As previously stated, VOFF is determined as the weighted sum of the five determinants of the value of financial flexibility as suggested by Gamba and Triantis (2008), with the weights being extracted from the empirical implementation of Eq (2) Specifically, the weights are the coefficients of the corresponding interaction terms in Eq (2) Adding the coefficient for unexpected changes in cash, ‘yj, this results in the following calculation of the VOFF of firm i in fiscal year ¢:

AE it

VOFF; t= + Y7SGR; t + Ysa Mit nộ YaT it + 1oSpread; t + Yi1 Tang; t (3)

3.3 Calculation of the value of financial flexibility

Summary statistics for all variables involved in the calculation of the VOFF are provided in Table 1 For instance, the median cumu- lative abnormal return is — 6.75%, while the corresponding mean value amounts to 4.12%.'°

As a first step, we estimate Eq (2) to obtain the weights for the five determinants of the VOFF Results can be found in Table 2 Overall, the regression results are in line with our expectations and the theoretical considerations by Gamba and Triantis (2008)

In each specification, all interaction terms supposed to measure the VOFF exhibit the expected sign and are highly significant The positive coefficient for SGR;¢ x AC,; indicates that the VOFF is higher in firms with more growth opportunities, In addition, firms improving their profitability have a lower VOFF Furthermore, the negative and significant coefficient for T;; x AC;, indicates that the VOFF decreases as the effective costs of holding cash in the firm increase The VOFF is also lower for firms with higher costs of external financing, as shown by the negative coefficient for Spread; x AC;; This finding is in line with the view that higher costs of external financing reflect agency problems One reason for this may be that shareholders fear managerial expropriation (e.g., Jensen, 1986; Rozeff, 1982) Finally, firms with a higher reversibility of capital, as approximated by a firm's tangibility, Tang; ;, have a lower VOFF Based on the regression coefficients for the interaction terms, we now calculate the VOFF, as defined in Eq (3) The results are pro- vided in Table 3 The average VOFF is 1.04 There is considerable variation in the VOFF across firm, as indicated by the values for the 1st and 3rd quartiles and the standard deviation

Furthermore, we present the correlation coefficients between the VOFF and several known measures of financial constraints Mea- sures for financial constraints are (i) the investment-cash flow sensitivity ICFS;; (Fazzari et al, 1988), (ii) the cash-cash flow sensitivity CCFS;; (Almeida et al., 2004), the (iti) Kaplan—Zingales Index KZI;, (Kaplan and Zingales, 1997), and the Whited-Wu Index WWI, (Whited and Wu, 2009) Not surprisingly, we find in Panel B of Table 3 that the correlation coefficients between the VOFF and the mea- sures for financial constraints are small.'' This underlines that the concepts of the VOFF and measures for financial constraints differ considerably

10 The non-zero mean can be explained by sample restrictions, winsorization, and the non-value weighting scheme,

1" The low correlation between ICFS and CCFS has also been documented in Chen and Wang (2012).

Table 1

Summary statistics—VOFF calculation This table provides summary statistics for the variables used in the VOFF regressions r,, — R?, denotes the annual cumulative abnormal returns of firm in fiscal year t, AC;; measures unexpected changes in cash holdings (Section 3.2) SGR;; is the one-year logarithmic sales growth E;; is earnings before extraordinary items plus interest and deferred tax credits T;,; is the effective corporate tax rate divided by the federal income tax rate for the U.S median income

Tang;, is tangibility, defined as net property, plant, and equipment deflated by total assets NA;; is total assets minus cash holdings RD;, is research and development expense It is set to zero if missing J;; 1s interest expense D;, is common dividends, L;; is market leverage NF;, is total equity issuance minus repurchases plus debt is- suance minus debt redemption All variables except L;;, SGR;,, Tj, Spread;,, Tang;;, and excess stock return are deflated by the lagged market value of equity (Mj; —1).A denotes the one-year absolute change of a variable, Ratios are winsorized at the 1% level

3.4 The value of financial flexibility and the credit crunch

Although it is impossible to directly test the empirical validity of our measure for the VOFF, we argue that such proxy should help explaining firm performance if external financing opportunities suddenly deplete This is what happened after the largely unexpected collapse of Lehman Brothers on September 15, 2008.! Many financial institutions had significant exposures to Lehman Actual exposure,

however, was unknown, resulting in uncertainty and distrust in the U.S banking sector, a breakdown of the interbank market, and

severely dried capital markets Ultimately, corporate borrowing opportunities declined and led to a credit crunch.’*

Thus, we perform an event study around this date We expect that firms for which shareholders consider financial flexibility as highly valuable suffered more from the collapse of Lehman Brothers than those with a low VOFF For the former, external financing opportunities, which deteriorated as a consequence of the credit crunch, are of particular importance To test this, we calculate cumu- lative abnormal returns around September 15, 2008, for two event windows ([-1d; 1d] and |-1d; 2d]) using a Fama and French (1993) three factor model The estimation period is set to the 250 trading days ending 10 trading days before the collapse of Lehman Brothers The cumulative abnormal returns are then regressed on the VOFF as of the fiscal year ending before the collapse of Lehman Brothers, and on controls for firm size and leverage Regression results can be found in Table 4

In line with our expectations, we find that firms with a higher VOFF have lower cumulative abnormal returns This effect is of high

economic and statistical significance For the [-1d; 2d] window, for instance, firms with a VOFF one standard deviation above the

average firm incurred an additional loss of about 1% This finding suggests that firms for which shareholders attribute considerable value to their financial flexibility performed significantly worse during the collapse of Lehman Brothers Overall, this test provides evi- dence that our measure is able to distinguish between firms for which shareholders attribute a high or low value to the availability of external financing opportunities

4 VOFF and corporate financial policy

In this section we examine the impact of the VOFF on three financial decisions: payout policy, capital structure, and cash holdings After evolving testable hypotheses, we present descriptive statistics and results from multivariate regressions

4.1 Hypotheses

First, we focus on payout policy, noting that corporate payouts are associated with costs and benefits

On the one hand, the distribution of cash among shareholders may signal good earnings prospects to equity investors (e.g., Bhattacharya, 1979; Miller and Rock, 1985) Also, undistributed cash may be used by managers to increase their own utility, pos- sibly at the expense of the owners Thus, dividends may help to reduce agency conflicts of equity (e.g., Jensen, 1986) And indeed, Faulkender and Wang (2006) find that dividend paying firms have a lower marginal value of cash

12 On September 15, 2008, the Dow Jones Industrial Average closed 500 points or 4.4% lower—the biggest percentage loss since September 17, 2001, the first trading day after 9/11

13 Recent evidence for the effects of credit crunches on corporate lending can be found in Lin and Paravisini (2012).

Table 2 Regression results—VOFF This table presents the results of regressing annual cumulative abnormal returns on firm characteristics and (unexpected) changes in cash holdings The dependent variable

is ri, — R®,, defined as the cumulative abnormal return of firm ti in fiscal year t AC,, measures unexpected changes in cash holdings (Section 3.2) SGR;; is the one-year logarithmic sales growth E;, is earnings before extraordinary items plus interest and deferred tax credits T;,, is the effective corporate tax rate divided by the federal income tax rate for the U.S median in- come household Spread;; is the average bid-ask spread calculated as the average of all trades for each firm from the third Wednesday each month during a firm’s fiscal year Tang;, is tangi- bility, defined as net property, plant, and equipment deflated by total assets C;; is cash and short-term investments NA;,; is total assets minus cash holdings RD;; is research and develop- ment expense It is set to zero if missing Ï; is interest expense Dj, is common dividends Lj; is market leverage, NF; is total equity issuance minus repurchases plus debt issuance minus debt redemption A denotes the one-year absolute change of a variable All variables except L;,, SGR;.,,

Ti, Spread;,, Tang;,, and excess stock return are deflated by the lagged market value of equity (Miz — 1) All variables used as interaction terms are balanced at their means Ratios are winsorized at the 1% level White (1980) standard errors are clustered at the firm-level and given

*

in parentheses, Statistical significance at the 1%, 5%, or 10% level is indicated by ***, **, or *,

respectively

Dep variable Tic — RE

(0.047)

(0.019)

(0.024)

(0.001)

(0.098)

(0.017)

(0.130)

(0.078)

(0.012)

(0.784)

(0.179)

(0.017)

(0.273)

(0.241)

(0.420)

(0.024)

(0.015)

(0.032)

On the other hand, dividends reduce the firm's ability to internally fund future investments (e.g., Blau and Fuller, 2008) The firm then has to rely on external financing, which is costly and time-consuming (e.g., Jensen and Meckling, 1976; Ritter, 1987) Therefore, Myers and Majluf (1984) suggest that firms may find it valuable to build up financial slack through retention In line with this, Blau and Fuller (2008) find that firms with the low debt level have low dividend payouts This finding is consistent with the view that firms for which financial flexibility is valuable are likely to have low payouts Therefore, we expect:

Hypothesis H1 Firms with a high VOFF pay lower dividends (probability and amount)

Furthermore, we expect a positive relation between the VOFF and the likelihood of dividend omissions In general, firms try to pay stable or increasing dividends to signal positive firm outcomes, while dividend reductions or even omissions are only a means of last

Panel A: VOFF

Panel B: Correlation coefficients

resort because market performance declines sharply in the aftermath (e.g., Brav et al., 2005; Lintner, 1956; Michaely et al., 1995) However, as argued before, the ability to fund investments internally may be more important to shareholders of firms with a high VOFF than stable payouts Thus, shareholders of such firms may be more likely to consent to dividend omissions

Hypothesis H2 Firms with a high VOFF are more likely to omit a dividend

Previous literature suggests that financial flexibility considerations affect the decision to repurchase shares In particular, firms rely more on repurchases when cash flows are temporary and more volatile (Grullon and Michaely, 2002; Guay and Harford, 2000; Jagannathan et al., 2000; Lie, 2005) The reason for this is that share repurchases are more flexible than dividends Their omission leads to no or little negative consequences because equity investors do not perceive repurchases as an ongoing commitment Conse- quently, we expect shareholders of firms with a high VOFF to prefer share repurchases to dividends

Hypothesis H3 Firms with a high VOFF prefer share repurchases to dividends

According to the “classical” trade-off theory, firms balance tax benefits of debt against costs of bankruptcy when choosing their capital structure Empirical evidence, however, suggests that observed leverage ratios are too low compared to theoretical predictions It appears that firms deviate from optimal leverage, leaving substantial amounts of money on the table (e.g., Graham, 2000; Miller, 1977) Financial flexibility might explain debt conservatism (DeAngelo et al., 2011; Denis and McKeon, 2012; Graham and Harvey, 2001)

In this perspective, firms preserve parts of their debt capacity to be able to quickly fund future investment projects and/or unexpected cash flow shocks In contrast to equity financing, firms can raise debt more quickly In this context, (Denis and McKeon, 2012) show that firms reduce their debt levels to low, long-run target levels whenever possible to be able to raise debt in the future Thus, higher leverage reduces a firm's future ability to raise additional debt Consequently, we hypothesize that firms with a higher VOFF preserve higher fractions of their debt capacity for the future by choosing lower leverage ratios today

Hypothesis H4 Firms with a high VOFF have lower leverage

Cash holdings have both benefits and costs On the one hand, (high) cash holdings can foster agency problems between share- holders and managers (Jensen, 1986) On the other hand, cash holdings allow firms to quickly fund investment opportunities when external (debt or equity) financing is costly and/or time-consuming (e.g., Myers and Majluf, 1984) In contrast to debt financing, the funding of investments with cash holdings is quicker and less sensitive to potential market disturbances, as for example after the collapse of Lehman Brothers As argued before, equity financing is also affected by market sentiment and, more importantly, it is Table 4

'VOFF and the collapse of Lehman Brothers This table presents results of an event study around the collapse of Lehman Brothers (September 15, 2008) Cumulative ab- normal returns, calculated using a Fama and French (1993) three factor model, are based on event studies performed on all firms in the sample using an estimation window based on 250 trading days ending 10 trading days before September 15, 2008 VOFF;, represents the value of financial flexibility Logsize;, is the natural log- arithm of total assets in millions of $US (2005 = 100) L;; is market leverage White (1980) standard errors are given in parentheses Statistical significance at the 1%, 5%, or 10% level is indicated by *™*, *, or *, respectively

VOFFi x — 1 —0,019** — 0,028***

(0.008) (0.010)

(0.001) (0.001) Lie — 1 —0.040*** — 0,046***

(0.009) (0.012)

N 1960 1960

Table 5

Summary statistics—financial policies This table provides summary statistics for the variables used in the financial flexibility regressions, All ratios are winsorized at the

1% level

Panel A: Financial decisions

PayerD;, is a dummy variable that is set to one if firm i pays dividends in year t and zero otherwise Dl; is the ratio of cash dividends to net income It is set to one if dividends exceed net income or if the firm pays a dividend and net income is negative Omit;,; is set to one if the firm stops paying a dividend in year t and zero if the firm continues to pay a dividend RTP;, is share repurchases to total payout L;; is market leverage LT;,; is long-term debt market leverage defined as the ratio of long-term debt to the sum of long-term debt, short-term debt, and the market value of equity ACash;,; is changes in cash and short- term investments scaled by total assets

PayerDj,+ 93,770 0.3115 0.0000 0.0000 1.0000 0.4631

DỊ;; 93,769 0.1654 0.0000 0.0000 0.1584 0.3190

Omit,; 20,480 0.1229 0.0000 0.0000 0.0000 0.3283

RTPi¢ 44,798 0.5336 0.0000 0.6286 1.0000 0.4453

Lit 94,694 0.2191 0.0107 0.1343 0.3550 0.2408

LTit 94,694 0.1641 0.0009 0.0739 0.2631 0.2061

ACash;; 91,052 0.0102 —0.0378 0.0000 0.0385 0.1742

Panel B: Control variables

RE;¿ is retained earnings deflated by total assets TE;, is total shareholders' equity deflated by total assets ROA,; is net income divided by total assets, SGR;; is the one- year logarithmic sales growth Logsize;, is the natural logarithm of total assets in millions of $US (2005 = 100) Vol;, is the two-year volatility of monthly total shareholder returns IndLev;, is defined as the median leverage per industry and year TobQ;; is Tobin's Q defined as the sum of total assets and market capitalization less the book value of common equity deflated by total assets WC; is the ratio of net working capital minus cash plus short-term investments to total assets, Capex;;

is capital expenditures to total assets IndSigma,, is the mean of the standard deviations of cash flow/total assets over five years for firms in the same industry RDT;;

is research and development expense to total assets It is set to zero if missing Acq;, is the ratio of expenditures on acquisitions to total assets

further the most time-consuming way of funding an investment In this context, Denis and Sibilkov (2010) indicate that firm value is positively associated with changes in cash holdings This relation is stronger for firms for which external financing is difficult, e.g., because banks are not willing to lend them money Thus, we hypothesize that firms with a high VOFF increase their cash holdings more Hypothesis H5 Firms with a high VOFF accumulate more cash

4.2 Descriptive statistics

Table 5, Panel A, provides summary Statistics for the three financial decisions As can be seen, firms pay dividends in about one third of all years On average, 17% of net income is distributed among shareholders via dividends The mean (median) market leverage

of the sample firms is 22% (13%) Not surprisingly, long-term leverage is significantly lower with a mean value of 16% Panel B shows summary statistics for the control variables employed in the regressions described in the following

4.3 Payout policy

Now we investigate the impact of the VOFF on payout decisions In total, we use four dependent variables for the payout regressions PayerD;,; is a dummy variable that is set to one if firm i pays cash dividends in year t and zero otherwise D];, is the ratio of cash dividends

to net income as defined in Julio and Ikenberry (2004) and von Eije and Megginson (2008) It is set to one if dividends exceed net income

or if the firm pays a dividend and net income is negative '4

M4 We think that the payout ratio based on net income is a more appropriate measure for payouts than, for example, dividends to sales, because net income approx- imates roughly the cash a firm can distribute among shareholders.

OMIT;,, the dividend omission dummy, is set to one if a firm stops paying a dividend in year f and zero if the firm continues to pay a dividend RTP;,; is share repurchases to total payout The applied control variables in payout regressions follow Brockman and Unlu (2009) and Hoberg and Prabhala (2009)

Table 6 shows pooled logit and tobit regression results In the first two columns, the dependent variables are a dividend dummy, PayerD;,, and the ratio of dividends to net income, DI; ; We lag our financial flexibility measure to reduce potential concerns of reverse causality In accordance with Hypothesis H1, we find that both the likelihood and the amount of dividends are lower in firms with a high VOFF This effect is statistically significant at the 1% level Furthermore, it is also of high economic importance For instance, a one standard deviation increase in the (lagged) VOFF indicates that the dividend payout ratio decreases by 7%, which equals nearly 50% of the mean payout ratio in our sample Results for the control variables are in line with previous literature

After that, we investigate the decision to omit dividends For this, we regress the omission dummy on the (lagged) VOFF The coefficient for the VOFF is positive and significant at the 1% level, indicating that firms with a higher VOFF are more likely to omit dividend payments This finding provides support for Hypothesis H2

Lastly, we analyze only firms that distribute cash among shareholders For these firms, we calculate the ratio of share repurchases to total payout We find that firms with a high VOFF prefer share repurchases to dividends In particular, an increase in the (lagged) VOFF

by one standard deviation results in an increase in the ratio of share repurchases to total payout by about 0.09 Thus, if firms with a high VOFF decide to distribute cash among shareholders, they are much more likely to use share repurchases than dividends This provides support for Hypothesis H3 and is consistent with Brav et al (2005) In their paper, Brav et al show that firms that pay considerable attention to financial flexibility issues prefer repurchasing shares to paying cash dividends because they perceive repurchases as a more flexible means of payout In contrast to dividends, share repurchases are not regarded as an ongoing commitment by the capital market Overall, the findings in Table 6 suggest that financial flexibility is an important determinant of corporate payout behavior 4.4 Capital structure

In this section, we examine whether financial flexibility considerations help explaining leverage ratios As a measure of a firm's leverage, we employ market leverage L;; is defined as the ratio of long-term debt and short-term debt to the sum of long-term debt, short-term debt, and the market value of equity Alternatively, we consider long-term debt market leverage, LT;, For its calcu-

lation, we only consider long-term debt in the denominator Results for market leverage, L;;, and long-term market leverage, LT;,, are

presented in Table 7 In line with Frank and Goyal (2009), we include lagged firm-specific control variables and industry and year dummies We use pooled OLS and fixed effects regressions in this context

The coefficient for the (lagged) VOFF is negative and significant at the 1% level, both in the pooled OLS and firm-fixed effects model Furthermore, the negative impact of the VOFF holds true both for the leverage definition based on total and long-term debt This in- dicates that firms in which shareholders attribute considerable value to financial flexibility choose lower leverage ratios For example,

a one standard deviation increase of the VOFF leads to a drop in leverage of 0.02 in the pooled OLS models This economic impact is

Table 6

VOFF and payout policy This table presents the results of pooled logit and tobit regressions with payout decisions as dependent variables PayerDjy is a dummy Variable

insorized at the 1% level Statistical significance at the 1%, 5%, or 10% level is indicated

by ***, **, or *, respectively

VOFF;, —0.634*** —0,239*** 0.651*** 0.355***

(0.097) (0.035) (0.232) (0.058) RE¡¿ 0.583*** 0.118*** —0.105 —0.081***

(0.140) (0.027) (0.164) (0.029) TEit —0.03 —0.093* —1.452*** 0.166**

(0.165) (0.051) (0.306) (0.079) ROA\+ 0.898*** —0.498*** —0.12 0.4037”?

(0.299) (0.074) (0.506) (0.111) SGR¿¡¿ —0.626*** —0,284*** 0.221 —0.121***

(0.070) (0.027) (0.219) (0.041) Logsize; + 0.367*** 0.100*** —0.323*** —0.051***

(0.021) (0.007) (0.039) (0.010) Cash;; —0.304 —0.082 0.071 0.517?

(0.196) (0.070) (0.450) (0.116) Vol;¿ — 10.869*** —3.480*** 10.118*** 2.240***

(0.590) (0.203) (1.148) (0.268)

N 42,179 42,179 9,095 24,092