SEARCH

SEARCH BY CITATION

ABSTRACT

  1. Top of page
  2. ABSTRACT
  3. I. Product Market Threats, Payouts, and Cash
  4. II. Understanding Fluidity
  5. III. Data and Summary Statistics
  6. IV. Payout Policy
  7. V. Conclusions
  8. Appendix A
  9. Appendix B
  10. REFERENCES
  11. Supporting Information

We examine how product market threats influence firm payout policy and cash holdings. Using firms' product text descriptions, we develop new measures of competitive threats. Our primary measure, product market fluidity, captures changes in rival firms' products relative to the firm's products. We show that fluidity decreases firm propensity to make payouts via dividends or repurchases and increases the cash held by firms, especially for firms with less access to financial markets. These results are consistent with the hypothesis that firms' financial policies are significantly shaped by product market threats and dynamics.

We examine how the product market threats faced by a firm shape its payout policy and cash holdings. Using computational linguistics, we analyze over 42,000 individual firm business descriptions from firm 10-Ks to construct new measures of the structure and evolution of the product space occupied by firms. These measures include product fluidity, a new measure of the competitive threats faced by a firm in its product market that captures changes in rival firms' products relative to the firm. We use these new text-based measures to test hypotheses about firms' financial policies including the dividend and repurchase payouts to shareholders and the cash balances retained by firms.

Payouts and cash balances are important because U.S. corporations hold and distribute considerable amounts of cash. In 2008, payouts through dividends and repurchases exceeded $450 billion, more than twice their level in 1997. Despite these large payouts, firms' cash balances have increased. Nonfirm nonfinancial businesses in the United States have doubled their cash balances over our sample period (1997 to 2008) to $1.3 trillion and these balances continue to grow.1 Payouts remain significant relative to cash balances. Between 1997 and 2008, annual payouts represent between 25% and 38% of cash on hand. Equivalently, cash balances represent 3 to 4 years of annual payouts.

Our focus on product market threats to explain payouts and cash balances is motivated by both evidence from the field and received theory. The survey evidence in Brav et al. (2005) indicates that 70% of managers perceive the “…stability and sustainability of future earnings” as central to their choice of payout policy. A stable environment conducive to dividend payouts is less likely when product markets are changing, particularly when changes are due to competitive threats by rival firms. Thus, fluidity should be negatively related to the propensity to pay dividends and positively related to holding higher cash balances. Paying lower dividends and repurchasing fewer shares while retaining cash and liquid assets can provide flexibility to firms in less stable markets, allowing them to react more aggressively to competitive threats when they materialize.

This role for product market fluidity is also consistent with the industrial organization literature on product life cycle, as in Abernathy and Utterback (1978) or Klepper (1996). Firms should be more willing to make a dividend commitment when their product has reached a stable point and when they foresee fewer competitive threats. Alternatively, DeAngelo, DeAngelo, and Skinner (2008) discuss a life cycle theory with managerial agency in which free cash flow creates a mindset of resource abundance. Payouts prevent managers from wasting cash flow. Payouts may be less necessary when firms are disciplined by competitive threats from product markets. If so, fluidity should be negatively associated with payouts.

We develop a new measure of product market threats, namely, fluidity, using firms' business descriptions provided in 10-Ks. While we formally define fluidity later, intuitively it measures the change in a firm's product space due to moves made by competitors in the firm's product markets. The focus on rivals is a distinguishing feature of fluidity. For instance, even if a company's current product mix is stable, entry by rivals can pose competitive threats to a firm. Our measure picks up such threats. Further supporting a link to competitive threats, we find that fluidity is positively correlated with the relation between a firm's product descriptions and the business descriptions of entrepreneurial firms receiving venture capital or undertaking initial public offerings (IPOs).

Our central finding is that product market fluidity has an economically significant relation to dividends, repurchases, and cash holdings. Firms facing changes in their product markets have a lower propensity to pay dividends or repurchase shares. These firms also pay lower dividend amounts. These results are most pronounced for dividends, which entail a long-term commitment to pay out cash flows that is not easily reversed. Firms in more fluid markets are also less likely to initiate and more likely to omit dividends. Fluidity is also associated with higher cash balances, especially for firms with less access to financial markets. Collectively, the findings suggest that firms facing fluid product markets and competitive threats adopt more conservative financial policies.

We emphasize that product market fluidity, which is an ex ante measure of threats, can be quite different from measured cash flow risk. For instance, managers of a firm facing competitive threats may change their products frequently to maintain ex post stable profits, so measured cash flow risk may be low although ex ante threats are high. For instance, Apple Inc., which has changed its product mix frequently over the last decade, has high profits and low measured cash flow risk yet pays no dividends. Empirically, product market fluidity and cash flow risk are only moderately correlated and both are independently negatively associated with the likelihood of paying dividends. Echoing the link between competitive threat and both payout and cash policies, a recent article in the Wall Street Journal on Hewlett-Packard (HP) also mentions the link between financial conservatism and product market threats.2

We conduct extensive robustness tests. First, we explicitly consider whether our results can be explained by growth options. We consider multiple firm-level controls, including market-to-book, asset growth, firm age, research and development (R&D) expenditure, and patenting intensity. We next verify that our product fluidity measure is distinct from cash flow risk and that product text descriptions capture changing threats in the product market in unique ways not reflected in firms' financial data. We further show that our results are robust to including controls for firm risk (Hoberg and Prabhala (2009)), firm maturity (DeAngelo, DeAngelo, and Stulz (2006)), and earnings losses (DeAngelo, DeAngelo, and Skinner (1992)), and are robust to excluding the largest dividend payers, which have been shown to pay a large fraction of dividends (DeAngelo, DeAngelo, and Skinner (2004)). Finally, we show that our results support the view of Gaspar and Massa (2006), Hou and Robinson (2006), and Peress (2010) that firms in concentrated industries have more stable cash flows and hence are more likely to make payouts.

Our study is the first to examine the extent to which some product markets are highly fluid over time while others are not. Our new text-based measures build upon Hoberg and Phillips (2010a, b), who introduce text-based industry classifications to test theories in corporate finance and industrial organization. We develop this research further by considering the dynamics of the product market space and using it to construct novel measures of the competitive threats in a firm's product markets. We show that our new threat measure can explain financial policies beyond static cross-sectional measures of competition.3

Our study also underscores the advantages of using product text descriptions over and above current methods of classifying firms. Product text contains novel information about firms' competitive environments for multiple reasons. First, product descriptions must meet regulatory standards. Item 101 in Regulation S-K specifies that the descriptions must be representative and significant. Thus, product descriptions contain timely information about a firm's products. Second, they improve on Standard Industry Classification (SIC) codes in that they offer extensive detail regarding firm products that can be compared directly to rival firms, allowing continuous measures of pairwise similarity between firms. Third, product descriptions are updated annually, allowing the measurement of how product markets change over time. In contrast, SIC codes are static over time regarding product location, and firm assignments rarely change.

Our study adds to the existing dividend and repurchase literature. In particular, it is related to Grullon, Michaely, and Swaminathan (2002) and DeAngelo, DeAngelo, and Skinner (2006), who argue that firm maturity is a key determinant of dividends.4 Our results suggest that firm-level maturity indeed matters, but is only one of two components of maturity. The other is a distinct product market component that reflects the extent to which a firm's products face competitive threats. Our evidence suggests that both aspects of maturity matter in setting payout policy.

We also contribute to the growing literature on cash holdings (e.g., Bates, Kahle, and Stulz (2009), Lins, Servaes, and Tufano (2010)). In particular, we introduce a new dimension, product market threats—and show that it explains cash holdings. The results also add to the literature that links product market interactions and predation to firms' financial policies theoretically by Telser (1966), Poitevin (1989), and Bolton and Scharfstein (1990), and empirically by Phillips (1995), Haushalter, Klasa, and Maxwell (2007), and Fresard (2010).

The remainder of the paper is organized as follows. In Section 'Product Market Threats, Payouts, and Cash' we develop and motivate the testable hypotheses. Section II introduces product market fluidity and provides examples and descriptive statistics that facilitate its interpretation. Section III discusses our data and methodology. Section IV presents the main empirical results, and Section 'Conclusions' concludes.

I. Product Market Threats, Payouts, and Cash

  1. Top of page
  2. ABSTRACT
  3. I. Product Market Threats, Payouts, and Cash
  4. II. Understanding Fluidity
  5. III. Data and Summary Statistics
  6. IV. Payout Policy
  7. V. Conclusions
  8. Appendix A
  9. Appendix B
  10. REFERENCES
  11. Supporting Information

In this section, we describe and develop new hypotheses on the role of product market characteristics in cash payout policy and the precautionary cash balances held by firms. Using information from product descriptions contained in firms' 10-Ks, we measure each firm's product market fluidity. We emphasize that it is a forward looking variable that is likely to have unique information incremental to that contained in other accounting numbers, such as firm R&D, historical cash flow risk, or the negative earnings indicator suggested by DeAngelo, DeAngelo, and Skinner (1992). Because the product descriptions in a firm's 10-Ks are legally required to be accurate and current, they are likely to reflect updated assessments of top management about the opportunities and threats that a firm faces in its product markets. Thus, product descriptions are likely to contain information beyond that in accounting numbers, which tend to be backward looking.

We postulate that product market fluidity, or instability, is important given that a firm's payout policy is based on expectations of the future market for their products. When product markets are changing rapidly, the future is less certain. It is unlikely that such firms would have expectations of the sustainable, stable earnings that 70% of the executives surveyed in Brav et al. (2005) cite as the most critical determinant in making payouts. While stability and risk matter for both repurchases and dividends, they are perhaps more critical for setting dividends whose relative rigidity and inflexibility make managers especially conservative with regard to dividend policy.

The importance of product fluidity is also suggested by the literature on product life cycles and the competitive threats that firms face. Prominent work in this area includes Abernathy and Utterback (1978) and Klepper (1996), who argue that products and their associated industries develop over time. These product changes give rise to a natural life cycle. Relevant to forming our hypotheses, firms selling newer products or products that have more technological risk face more fluid product markets and thus greater risk of future competitive threats. Such firms might refrain from making payouts until their product markets mature, as payout conservatism can strengthen the firm's competitive positioning should a threat materialize. As firms' product markets begin to stabilize, they may be more willing to make payouts. We expect a greater willingness to make payouts via repurchases, which entail less of a longer-term commitment. Note that product fluidity may, but need not be, related to firm maturity. For instance, while older firms and firms that have retained a large portion of their past earnings might be viewed as mature firms, some older firms may face renewed technological change or product innovation or new threats from rival firms, and thus be in less mature product markets. This discussion leads to the first testable hypothesis in our paper.

  • Hypothesis 1 (Product market fluidity and payout policy):
  • Competitive threats as reflected in product market fluidity are negatively relatedto the probability of repurchases and the probability of paying dividends.

Our second hypothesis is based on the effect of product market fluidity on a firm's cash balances. There is a large empirical literature on cash.5 Bates, Kahle, and Stulz (2009) discuss four motives for holding cash, namely, the transaction, agency, tax, and precautionary motives for holding cash. Most relevant for our purpose is the precautionary motive for holding cash, according to which firms face uncertain demands for liquidity in the future and hold cash as a hedge.6 The importance of the precautionary motive is stressed by Lins, Servaes, and Tufano (2010), who survey CFOs from 29 countries to understand why they store cash beyond operational needs. They find that cash in excess of operational needs is primarily used to meet unexpected demand for cash in bad times, consistent with the precautionary motive driving the demand for cash.

We add to the cash literature by examining the role played by competitive threats from a firm's product market. The relation can be motivated in a straightforward way. The precautionary motive suggests that firms hold cash as a hedge against future adverse cash flow shocks. The role of cash in combating product market threats is also stressed by the literature on predatory threats (Bolton and Scharfstein (1990), Haushalter, Klasa, and Maxwell (2007), Fresard (2010)). In our study, product market fluidity is a measure of the competitive threat faced by a firm, which threatens the stability of a company's future cash flows. Thus, firms facing more product market fluidity should hold greater cash balances. This discussion leads to our second hypothesis.

  • Hypothesis 2 (Product market fluidity and cash holdings):
  • Competitive threats facing a firm's products are positively related to the extentof a firm's cash holdings.

II. Understanding Fluidity

  1. Top of page
  2. ABSTRACT
  3. I. Product Market Threats, Payouts, and Cash
  4. II. Understanding Fluidity
  5. III. Data and Summary Statistics
  6. IV. Payout Policy
  7. V. Conclusions
  8. Appendix A
  9. Appendix B
  10. REFERENCES
  11. Supporting Information

In this section, we briefly define fluidity, our key measure of product market threats. We then provide examples and statistical evidence to illustrate its intuition and its ability to capture threats from a firm's product market environment. To more directly assess the interpretation of fluidity as a measure of competitive threat, we next examine the relation between fluidity and the business descriptions of entrepreneurial firms that raise venture capital and of firms that recently went public through IPOs. These firms represent emerging competitive threats to public firms.

A. Definition

Fluidity captures how rivals are changing the product words that overlap with firm i's vocabulary. Because fluidity focuses on product space dynamics and changes in products, it is an entirely new construct relative to the industry definitions and variables used in Hoberg and Phillips (2010a). Specifically, let inline image denote a scalar equal to the number of all unique words used in the product descriptions of all firms in year t. Let inline image denote an ordered Boolean vector of length inline image identifying which of the inline image words are used by firm i in year t. Element j of inline image equals one if firm i uses word j in its product description and zero otherwise. We normalize inline image to unit length and define the result as inline image.

To capture the changes in the overall use of a given word j in year t, we define the aggregate vector inline image as

  • display math(1)

A firm's product market fluidity is simply the dot product between its own word vector inline image and normalized inline image:

  • display math(2)

Intuitively, fluidity is a “cosine” similarity between a firm's own word usage vector inline image and the aggregate change vector inline image.7 Quantitatively, the dot product in equation (2) measures the cosine of the angle between the two vectors. Because the dot product is based on nonnegative vectors, fluidity is the cosine between vectors in the first quadrant. Thus, fluidity lies in the interval [0, 1].

Fluidity is greater when a firm's words overlap more with inline image, the vector that reflects rival actions. It is therefore the competitive threat that is higher. This intuition is illustrated in Appendix A, where we present an example of a hypothetical market with three firms and seven product words. The example highlights the simple but crucial point that fluidity reflects product market threats and instabilities arising out of competitor actions, not necessarily own-product instability. The notion that rival threats are important, perhaps even more so than static measures of market share, is consistent with theories of contestable markets in industrial organization (Baumol, Panzar, and Willig (1982)). Fluidity is an empirical construct that captures this intuition.

B. Examples

Table I reports specific examples of high and low fluidity firms for 1997 and 2008, the first and last years of our sample.

Table I. Examples of Firms with High and Low Product Market Fluidity
Panel A: Firms in 1997 with Low Product Market Fluidity (Stable Products)
WEIS MARKETS, GENLYTE GROUP, ACME ELECTRIC, ALBERTO CULVER, SWANK, PHOENIX
FOOTWEAR GROUP, BURKE MILLS, BALDOR ELECTRIC, FRANKLIN ELECTRIC, BUTLER
MANUFACTURING, ROYAL APPLIANCE MFG, WYANT, WOLOHAN LUMBER, EASTMAN KODAK, MACYS,
GOLD STANDARD, O SULLIVAN, LUFKIN INDUSTRIES, INTERNATIONAL ALUMINUM, MAY DEPARTMENT
STORES, HARLAND JOHN H, ILLINOIS TOOL WORKS, NATIONAL SERVICE INDUSTRIES, GENESIS
WORLDWIDE, SUPERIOR UNIFORM GROUP
Panel B: Firms in 1997 with High Product Market Fluidity (Fluid Products)
BOYD GAMING, COX COMMUNICATIONS, SHOWBOAT, COMCAST, TRUMP HOTELS & CASINO RESRTS,
NEXTEL COMMUNICATIONS, P D L BIOPHARMA, MILLENNIUM PHARMACEUTICALS, ONYX
PHARMACEUTICALS, HORIZON C M S HEALTHCARE, HARRAHS ENTERTAINMENT, LIGAND
PHARMACEUTICALS, AMERISTAR CASINOS, TRIANGLE PHARMACEUTICALS, I D T, IMMUNE
RESPONSE, SUN HEALTHCARE GROUP, HILTON HOTELS, MAGELLAN HEALTH SERVICES, NEXSTAR
PHARMACEUTICALS, PLAYERS INTERNATIONAL, NEUREX, AZTAR, CORVAS INTERNATIONAL, MCLEODUSA
Panel C: Firms in 2008 with Low Product Market Fluidity (Stable Products)
SHERWIN WILLIAMS, ALBERTO CULVER, AMPCO PITTSBURGH, SUPERIOR UNIFORM GROUP, CINTAS,
U S DATAWORKS, COLGATE PALMOLIVE, LIBERTY GLOBAL, COMPUTER SCIENCES, LAWSON
PRODUCTS, VALHI, LIMITED BRANDS, PEPSIAMERICAS, ANIXTER INTERNATIONAL, E D A C
TECHNOLOGIES, LANCE, FRIEDMAN INDUSTRIES, DECORATOR INDUSTRIES, MCCORMICK,
FLEXSTEEL INDUSTRIES, STEPAN, PACCAR, CASS INFORMATION SYSTEMS, BLYTH, MOD PAC
Panel D: Firms in 2008 with High Product Market Fluidity (Fluid Products)
VICAL, ALTUS PHARMACEUTICALS, ALNYLAM PHARMACEUTICALS, ENZO BIOCHEM, ZYMOGENETICS,
CYTOKINETICS, THRESHOLD PHARMACEUTICALS, ICAGEN, ANADYS PHARMACEUTICALS, INHIBITEX,
ABRAXIS BIOSCIENCE, EMERGENT BIOSOLUTIONS, RIGEL PHARMACEUTICALS, G T X, OREXIGEN
THERAPEUTICS, O S I PHARMACEUTICALS, AMGEN, UNITED THERAPEUTICS, ISIS PHARMACEUTICALS, IVIVI TECHNOLOGIES, BIOCRYST PHARMACEUTICALS, VERTEX PHARMACEUTICALS, ENZON PHARMACEUTICALS, NEUROGESX, I D M PHARMA

In all panels in Table I, firms are sorted from most extreme to least extreme. Panels A and C report firms that have the lowest local product fluidity based on 10-Ks in 1997, the beginning year of our data, and 2008, our ending year. Many of these firms are commodity firms as well as department stores. In Panels B and D, we report firms that have the highest local product fluidity.

Communications and media have been undergoing competitive changes due to ongoing battles between cable, television, print, and online media. In 1997, a considerable fraction of these firms were in the gaming and communications industries. The high fluidity of the gaming industry may come as a surprise to some readers. However, in 1997 this industry was in a high state of flux due to ongoing changes in the competitive landscape. The 10-Ks reveal that Native American casinos, riverboat casinos, and new casinos in previously disallowed locations went online, challenging existing operations. State laws were also changing, further increasing uncertainty about future market structure. In 2008, biotechnology firms faced fluid markets due to changes related to healthcare regulation and ongoing shocks related to innovation and government approval.

C. Dividends, Payouts, and Fluidity Transitions

Table II presents payout and cash holdings summary statistics for fluidity quintiles as well as transition matrices showing how fluidity changes over time. We present these statistics to illustrate the economic effects we describe and the interpretation of fluidity in terms of competitive threats and product life cycle.

Table II. Dividends, Repurchases, and Product Market Fluidity
 MostQuintileQuintileQuintileMost 
SampleStable234FluidObs.
Panel A: Payout Statistics
% dividend payer0.4870.3170.2100.1450.09142,999
Dividend yield0.0120.0060.0040.0030.00242,999
Dividend/assets0.0110.0060.0040.0030.00242,999
% repurchasers0.5050.4560.4150.3920.31042,999
Cash+equiv/assets0.1020.1390.1910.2480.37542,999
Panel B: 1-Year Transition Probabilities
Most stable product markets0.7780.1680.0400.0100.0054,691
Quintile 20.2360.5060.1910.0550.0134,698
Quintile 30.0340.2680.4600.1880.0504,697
Quintile 40.0100.0500.2720.4730.1944,698
Most fluid product markets0.0040.0090.0390.2370.7114,693
Panel C: 3-Year Transition Probabilities
Most stable product markets0.7790.1770.0320.0100.0031,597
Quintile 20.1970.5600.2050.0330.0051,598
Quintile 30.0310.2400.5060.2020.0221,599
Quintile 40.0100.0430.2580.5140.1751,598
Most fluid product markets0.0010.0040.0510.2230.7211,598
Panel D: 6-Year Transition Probabilities
Most stable product markets0.7450.1970.0510.0050.002589
Quintile 20.2220.5030.2000.0630.012590
Quintile 30.0480.2840.4480.1780.042589
Quintile 40.0050.0640.3190.4290.183590
Most fluid product markets0.0020.0150.0750.2840.625589

Panel A shows that 48.7% of firms with the lowest fluidity pay dividends. In contrast, only 9.1% of firms with the highest fluidity are payers. Dividend yields and repurchases exhibit similar patterns, declining as fluidity increases. Cash holdings also increase substantially from 8.2% of assets for the least fluid quintile to 21.5% for the most fluid quintile.

Panels B to D present transition matrices examining future product market fluidity as a function of initial product market fluidity. Panel B (C, D) displays 1-year (3-year, 6-year) transition probabilities based on nonoverlapping time periods. The rows indicate the initial fluidity quintile of the firm and the columns indicate which fluidity quintile the firm is in t years later. The result is a 5 × 5 grid containing the empirical distribution of transitions. For example, in Panel B, row 1, column 1, 77.8% of firms in the most stable product market fluidity quintile remain in this quintile 1 year later.

The results in Panels B to D show that fluidity is persistent. Low fluidity firms are very likely to be low fluidity firms ex post at 1-, 3-, and 6-year horizons. Interestingly, the persistence seems somewhat lower for high fluidity quintiles. For example, a firm in the most fluid quintile remains in this category 71.1% of the time, while firms in the most stable quintile retain this designation 77.8% of the time. More generally, the table shows that over time firms are more likely to “graduate” to a category with more stable product markets than they are to move to a category with less stability. For example, in Panel B, firms with the most stable product market move to quartile 2 with a 16.8% probability, while firms in the highest fluidity category move to a more stable quartile 4 with a 23.7% probability. These findings are consistent with product life cycle theories in which fluidity is likely to decrease over time.

D. Fluidity: Threats from VC-Backed and IPO Firms

In this section, we examine the relation between fluidity and the similarity of a firm's products to the business descriptions of IPOs and venture capital (VC)-backed firms. These tests assess whether there is a direct link between fluidity and the competitive threats from entrepreneurial firms. We also examine whether such a link to fluidity is distinct from risk variables.

We use VentureXpert to extract all business descriptions for funded ventures in each year of our sample. We then compute the average word usage frequency of venture firms using the same set of words in our 10-K universe. We denote the unit-normalized version of this aggregate word vector in a given year by inline image, the “venture vocabulary.” We next measure the degree to which a given public firm is threatened by venture-backed firms as the dot product between a firm's (unit-normalized) word description vector and inline image, which we define as the firm's “VC Score.” We similarly compute a firm's “IPO Score” based on product descriptions of firms going public in a given year, which we extract using SDC Platinum. The IPO Score measures threats from firms going public.

To examine potential links to competitive threats, we regress fluidity on both VC Score and IPO Score, as well as several controls. A positive coefficient would provide additional evidence that our measure of product market fluidity in equation (2) reflects competitive threats. We control for firm size and firm age, which is important because small and young firms may be more similar to VC-backed and IPO firms. We include the Herfindahl–Hirschman Index (HHI) to control for static competition levels, and we also include industry fixed effects. To be conservative, both HHI and industry controls are based on the text-based industry classifications in Hoberg and Phillips (2010a, b). Of particular interest is the control for risk, as this assesses whether the information contained in product market fluidity is similar to that of the risk variables. We measure risk using stock market volatility as in Hoberg and Prabhala (2009). Using cash flow volatility as in Bates, Kahle, and Stulz (2009) produces similar results.

Table III shows that product market fluidity is significantly and positively related to both VC Score and IPO Score in all specifications. Fluidity interacted with firm R&D is also negative, which is consistent with greater spending on R&D in fluid markets deterring potential entry by VC-backed and IPO firms. Interestingly, measured firm risk is insignificant in three of the four specifications. As all variables are standardized to unit standard deviation, we can compare the coefficients to assess economic magnitude. In all columns, the coefficient on risk is an order of magnitude smaller than that for product market fluidity. These results suggest that fluidity is distinct from cash flow risk. Given that IPO and VC-backed firms are likely in less mature products, the results are also consistent with the product life cycle interpretation of fluidity, in the spirit of Abernathy and Utterback (1978) or Klepper (1996).

Table III. Fluidity and IPO/Venture Capital Product Descriptions
 IPOscoreIPOscoreVCscoreVCscore
 (1)(2)(3)(4)
Total risk−0.0010.0020.0480.013
 (−1.249)(0.224)(4.506)(1.520)
Local product fluidity0.2000.2280.3820.432
 (28.715)(27.583)(32.998)(37.516)
R&D/sales0.0070.0030.0970.037
 (0.773)(0.341)(6.949)(2.995)
Local fluidity X R&D/sales−0.009−0.012−0.029−0.022
 (−2.227)(−3.132)(−4.634)(−3.991)
NYSE size percentile0.0660.0690.0750.083
 (8.769)(10.253)(6.159)(8.564)
Log firm age0.0230.0180.0200.011
 (3.229)(2.871)(1.726)(1.146)
HHI (TNIC)−0.033−0.001−0.033−0.033
 (−5.121)(−1.469)(−3.394)(−3.791)
Constant0.3800.7360.0780.076
 (10.380)(14.251)(2.000)(1.173)
Ind. Fixed EffectsNoYesNoYes
R20.2040.2620.1950.405
N35,48935,48935,48935,489

III. Data and Summary Statistics

  1. Top of page
  2. ABSTRACT
  3. I. Product Market Threats, Payouts, and Cash
  4. II. Understanding Fluidity
  5. III. Data and Summary Statistics
  6. IV. Payout Policy
  7. V. Conclusions
  8. Appendix A
  9. Appendix B
  10. REFERENCES
  11. Supporting Information

A. Data

We construct our Compustat–CRSP sample following Fama and French (2001) and Hoberg and Prabhala (2009). We start with 61,136 firm-years from 1997 to 2008 that have adequate Compustat and CRSP data. The years are chosen based on the availability of our text-based data. We then apply the same screens as Hoberg and Prabhala (2009). After discarding regulated utilities (SIC codes between 4900 and 4949) and financials (SIC codes between 6000 and 6999), we have 48,159 observations. We next screen out observations in which firms have book values of less than $250,000 or assets of less than $500,000. This leaves us with 45,631 observations.

Following Fama and French (2001), we identify companies that are dividend payers if their dividends per share (Compustat annual data item 26) is greater than zero. We identify companies that are share repurchasers in a given year using the method suggested in Grullon and Michaely (2002). In the Compustat universe, we define stock repurchases as annual data item 115 (purchase of common and preferred stock) less the reduction in the value of any preferred stock outstanding (annual data item 56). We label a firm as a repurchaser of shares if this difference is greater than zero. We also separately analyze firms that repurchase for 2 consecutive years, and large repurchasing firms, defined as those firms whose repurchases exceed 1% of total assets. For cash, we compute a firm's cash holdings as cash scaled by assets.

The data to construct the fluidity measure in equation (2) are derived from business descriptions in annual firm 10-Ks. Our sample of 10-Ks comes from Hoberg and Phillips (2010a), who extract business descriptions using PERL Web crawling scripts, Advanced Programming Language (APL) programming, and human intervention when documents are nonstandard. Our primary sample includes filings associated with firm fiscal years ending in calendar years 1997 to 2008. We also use 1996 data to compute text-based variables requiring lagged data, but do not use 1996 data otherwise.

We merge each firm's text product description to the CRSP/Compustat database using the central index key (CIK).8 Of the 45,631 observations available in CRSP and Compustat, we are left with 43,904 after requiring that same-year text data are also available. The final sample comprises 42,999 observations over the 1997 to 2008 period after requiring that lagged text-based data are also available. As discussed in Hoberg and Phillips (2010a), this text-based database generally has uniform coverage of the CRSP and Compustat sample during these years.

We use the product text description to construct our fluidity measure as in equation (2). In addition, for robustness, we consider variations of this fluidity measure based three dictionaries (see Appendix B in Hoberg and Phillips (2010b)): (i) a list of all words; (ii) a list of words from firms with a local clustering coefficient in the top two terciles (local fluidity), which reflects threats from nearby rivals; and (iii) a list of words from rivals with a local clustering coefficient in the lowest tercile (the “broad dictionary”). Although we report the results based on the local dictionary because nearby rivals likely pose more serious competitive threats, the results are similar across specifications. We also include self-fluidity, which compares a firm's year t and year inline image products in isolation (without regard to rival products). This variable is simply one minus the cosine similarity between the current and previous years' business descriptions.

In the industrial organization and corporate finance literatures, a potential endogeneity issue is that the agents (e.g., senior management), who set the financial policies, also choose the product market strategies (see Graham, Harvey, and Rajgopal (2005), for example). This issue is somewhat mitigated in our study. While a firm's own top management certainly sets the firm's payout policies, fluidity reflects moves by rival firms competing in a firm's product space. Going further, we also examine robustness to using the broad measure of product fluidity discussed above, which focuses on the broad set of words that all other firms in the Compustat–CRSP universe use. This broad measure is akin to a “market index” whose changes are likely to be exogenous from any one firm's perspective, just as the aggregate stock market return is considered exogenous to any single firm's return. These robustness results are reported in our Internet Appendix.9

B. Summary Statistics

We relate payout policy and the cash holdings of firms to fluidity and several control variables. The controls include variables suggested by the dividend literature. In addition, we include other controls to capture factors, such as growth options and specifications, that include industry fixed effects. We describe these variables and their construction in Appendix B.

Table IV presents summary statistics for the variables. Statistics for the payout policy and cash holding variables are reported in Panel A. The table shows that 24.5% of the firms are dividend payers while a larger set of firms, 41.2%, are repurchasers. The average firm also holds 20.9% of its assets in cash with a median holding of 10.8%.

Table IV. Payout Summary Statistics
VariableMeanSDMinimumMedianMaximum
Panel A: Data on Payout Status and Cash Holdings
Dividend payer0.2450.4300.0000.0001.000
Equity repurchaser0.4120.4920.0000.0001.000
Both payer & repurchaser0.1570.3640.0000.0001.000
Dividend yield0.0050.0140.0000.0000.237
Dividend/assets0.0050.0140.0000.0000.168
Dividend initiator0.0140.1190.0000.0001.000
Dividend increaser0.0890.2850.0000.0001.000
Dividend decreaser0.0070.0820.0000.0001.000
Cash+equiv/assets0.2090.2360.0000.1080.961
Panel B: Data from 10-K Text Analysis
Local product market fluidity6.9323.3621.3746.37920.628
Self product fluidity21.04314.9201.45717.15478.464
HHI0.2210.2310.0190.1271.000
Panel C: Data from the Existing Literature
Total risk0.0430.0250.0100.0370.168
Market-to-book2.0941.9280.3581.47925.424
Asset growth0.0350.329−3.1960.0550.875
Income/assets−0.0280.298−3.0010.0610.325
NYSE size percentile0.2600.2840.0000.1381.000
Log firm age2.9460.9880.6212.9444.974
Negative earnings dummy0.3410.4740.0000.0001.000
R&D/sales0.3291.6760.0000.00223.816
Retained earnings/assets−0.4511.776−15.4390.0800.914
Credit line dummy0.6120.4870.0001.0001.000
Cash flow risk0.3370.0530.2110.3290.615
3-year sales growth0.4180.797−2.4970.3065.324
Text+applied patents0.6980.4590.0001.0001.000

Statistics for the text-based variables are presented in Panel B and for the key control variables in Panel C. Our fluidity variables are multiplied by 100 for convenience. The statistics for self-fluidity suggest that some firms experience very little change from year to year, whereas others experience substantial changes. The average product market fluidity is 6.93%.

Table V reports the Pearson's correlations between our variables. Product market fluidity is −30.2% correlated with the static HHI measure using the text-based industry classifications of Hoberg and Phillips (2010a). Firms that are in more competitive industries have rivals that are more likely to change their products, and thus operate in markets with greater fluidity. Product fluidity is also −34.7% correlated with log firm age, so older firms tend to reside in more stable product markets. However, this correlation is far less than 100%, confirming that our product maturity variable is distinct from this measure of firm maturity. For example, older firms might experience technological shocks, bringing their product markets back to a more fluid state. Finally, fluidity is also modestly correlated with R&D (30.6%), consistent with the fact that product market fluidity requires at least some investment. Below we show that these modest correlations, while economically sensible, do not explain the links between product market fluidity and dividend or cash policy.

Table V. Pearson's Correlation Coefficients
  Local Product MarketSelf-Product TotalMarket-toAssetIncome/NYSELog FirmR&D/
RowVariableFluidityFluidityHHIRisk-BookGrowthAssetsPercentileAgeSales
(1)Self-fluidity0.200         
(2)HHI−0.302−0.021        
(3)Total risk0.1770.1910.045       
(4)Market-to-book0.2450.104−0.0780.040      
(5)Asset growth−0.029−0.018−0.023−0.2620.120     
(6)Income/assets−0.312−0.2040.029−0.470−0.1540.559    
(7)NYSE size percentile0.0250.028−0.191−0.4100.2020.2070.288   
(8)Log firm age−0.347−0.1360.031−0.327−0.1300.0250.2670.279  
(9)R&D/Sales0.2940.053−0.0760.1090.174−0.099−0.342−0.081−0.126 
(10)Neg. earn. dummy0.3060.194−0.0130.4850.068−0.334−0.620−0.341−0.2740.237

IV. Payout Policy

  1. Top of page
  2. ABSTRACT
  3. I. Product Market Threats, Payouts, and Cash
  4. II. Understanding Fluidity
  5. III. Data and Summary Statistics
  6. IV. Payout Policy
  7. V. Conclusions
  8. Appendix A
  9. Appendix B
  10. REFERENCES
  11. Supporting Information

A. The Propensity to Pay Dividends

In this section, we study the relation between fluidity and firms' payout decisions. We start with the propensity to pay dividends and then examine changes in payout policy, including dividend initiations, omissions, increases, and decreases. We standardize the independent variables so they have unit standard deviation. This scaling does not affect significance levels or the economic impact of our variables but facilitates economic interpretation of the results.

Rows 1 to 3 of Table VI, Panel A, present results of estimating a logit model in which the dependent variable is one if the firm pays dividends and zero otherwise. Rows 4 to 6 use a linear probability model. Each specification includes a different set of control variables as noted, and all specifications include time fixed effects and standard errors adjusted for clustering by firm. Panel B estimates panel ordinary least squares (OLS) regressions where the dependent variable in row 7 is the firm's dividend yield expressed as dividends divided by fiscal year-end price. Row 8 analogously examines each firm's dividend to assets ratio. Since we are interested in examining differences between dividend-paying firms in Panel B (Panel A examines the choice to pay or not pay), we restrict attention to the subsample of dividend-paying firms in Panel B.

Table VI. Dividend Policy and Product Fluidity
 LocalSelf-  Log   NYSE Neg.Ret3-YearExtraNum.
 ProductProduct TotalFirmMarket-AssetIncome/SizeR&D/Earn.Earnings/SalesControls+Obs/
RowFluidityFluidityHHIRiskAgeto-BookGrowthAssetsPercentileSalesDummyAssetsGrowthIndustryR2
Panel A: Dividend Payer vs. Nonpayer
Dependent variable: Dividend payer dummy (logistic model)
(1)−0.610−0.0510.068−1.3590.664−0.374−0.5541.3060.602    No42,999
 (−12.98)(−1.96)(2.00)(−18.83)(15.54)(−5.99)(−16.27)(12.85)(14.12)     0.360
(2)−0.416−0.0210.071−1.1170.566−0.322−0.3770.8260.652−8.036−0.0322.174−0.482Controls39,768
 (−8.23)(−0.78)(2.00)(−15.07)(12.62)(−4.46)(−10.60)(5.60)(14.46)(−4.55)(−0.36)(7.61)(−10.25)Only0.379
(3)−0.3630.0230.021−1.0380.538−0.161−0.3950.4730.704−1.125−0.1562.559−0.503Both39,768
 (−6.31)(0.82)(0.55)(−13.75)(11.54)(−2.37)(−10.21)(3.36)(14.55)(−0.95)(−1.77)(7.92)(−10.20) 0.419
Dependent variable: Dividend payer dummy (linear probability model)
(4)−0.076−0.0130.008−0.0790.101−0.022−0.035−0.0030.124    No42,999
 (−17.78)(−5.03)(1.94)(−21.79)(20.76)(−7.45)(−15.19)(−1.03)(22.14)     0.347
(5)−0.072−0.0110.010−0.0780.104−0.022−0.036−0.0120.1250.016−0.0580.010−0.018Controls39,768
 (−14.25)(−4.03)(2.30)(−20.06)(19.03)(−6.44)(−14.92)(−3.37)(21.09)(8.47)(−9.65)(3.33)(−8.25)Only0.354
(6)−0.055−0.002−0.006−0.0640.091−0.014−0.036−0.01330.1230.009−0.0240.015−0.021Both39,768
 (−10.44)(−0.68)(−1.41)(−17.76)(17.23)(−4.31)(−15.36)(−4.05)(21.35)(4.26)(−8.93)(5.17)(−10.04) 0.420
Panel B: Dividend Levels
Dependent variable: Dividend yield
(7)−0.2060.032−0.0070.1270.081−0.217−0.710−0.209−0.243−1.0520.504−1.174−0.472Both10,343
OLS(−3.73)(1.25)(−0.20)(1.31)(1.98)(−3.43)(−10.04)(−0.94)(−6.560)(−0.80)(3.30)(−4.06)(−6.42) 0.330
Dependent variable: Dividend/assets
(8)−0.2720.0320.027−0.400.011.209−0.6590.920−0.1411.5880.159−0.480−0.465Both10,343
OLS(−4.79)(1.19)(0.82)(−5.61)(0.37)(14.16)(−11.04)(4.28)(−3.54)(1.13)(1.274)(−2.23)(−6.04) 0.363

Both panels in Table VI show similar results. Firms in fluid product markets are less likely to pay dividends, pay lower dividend yields, and have lower dividend to assets ratios.10 This result confirms our first hypothesis that product market threats matter. Firms facing greater competitive threats in their product markets are less likely to pay dividends. These findings are also robust to including other text-based product market variables as well as firm risk, R&D expenditures, and firm maturity.

Table VI also reports results for the static HHI measure of competition based on the Hoberg and Phillips (2010b) industry classifications. Industry concentration is positively related to dividend propensity but its economic effects are not large. This result is different from that in Grullon and Michaely (2007). This is likely because our measure of competition more accurately reflects competition as shown by Hoberg and Phillips (2010b). In addition, our samples are different: our sample has better cross-sectional coverage as Herfindahl measures based on Census data are only available for manufacturing firms, but are more limited in the time series as our sample begins in 1997 due to the required availability of machine readable 10-Ks.

B. Economic Significance

To assess the economic significance of fluidity, we conduct a number of tests. We first estimate the propensity to pay dividends for our panel from 1997 to 2008 including the base Fama and French (2001) variables and risk variables but excluding our text variables. Holding the propensity to pay constant, we then examine whether the text variables are different for payers and nonpayers. We estimate a base specification without the text variables, split firms into propensity to pay quartiles, and then compare the mean and median text variables within each propensity quartile.

Table VII reports the results. Explanatory variables for the logit include: total risk, log firm age, firm size, M/B, asset growth, the negative earnings loss dummy of DeAngelo, DeAngelo, and Skinner (1992), R&D/sales, and year fixed effects. Within each quartile of the predicted propensity to pay, we report the mean and median of the three text-based product characteristics for payers and nonpayers.

Table VII. Text Characteristics for Propensity Matched Firms
 Quartile 1Quartile 2Quartile 3Quartile 4
VariablePayerNonpayerPayerNonpayerPayerNonpayerPayerNonpayer
Local fluidity6.695inline image9.0975.696inline image7.1575.493inline image6.4784.967inline image6.101
 6.234inline image8.7765.250inline image6.8184.996inline image5.9984.393inline image5.515
Self-fluidity24.50126.65218.035inline image21.61617.212inline image18.82517.77617.544
 18.622*22.56514.107inline image17.88413.542inline image15.48014.21914.469
HHI0.300inline image0.2050.286inline image0.2240.264inline image0.2110.2110.211
 0.176inline image0.1020.189inline image0.1320.172inline image0.1330.132*0.127
# observations8310,67371010,0452,4798,2767,2763,479

The results in Table VII show that there are significant differences in our text variables for each propensity quartile. Within all quartiles of predicted dividend propensity, firms that actually pay dividends have significantly lower values of local product market fluidity. The text variables appear to contain additional explanatory power over and above variables shown to affect payout propensity in past studies.11 DeAngelo, DeAngelo, and Skinner (2004) and DeAngelo, DeAngelo, and Skinner (2008)) argue that several large companies with long dividend paying histories continue to pay dividends. Time-series variation in dividend-paying propensity are thus driven by small to mid-size firms. If we exclude the beginning of period payers in the top size quartile, we find similar results, suggesting that our results are not driven by very large firms with little variation in their payouts.

Table VIII presents a second set of results to illustrate the economic significance of the results from Table VI for dividend propensity. Each column gives the propensity to pay dividends at the 10th, 25th, 50th, 75th, and 90th percentile of a variable holding the other variables at their median values. We display results for the logit model with and without Hoberg–Phillips FIC-300 industry controls.

Table VIII. Economic Significance
 LocalSelf-  Log    
 ProductProduct TotalFirmR&D/Ret.Cash FlowSales
PercentileFluidityFluidityHHIRiskAgeSalesEarningsRiskGrowth
Panel A: Propensity to Pay Dividends: Full Model, No FIC Fixed Effects
100.2090.1510.1450.2830.0750.1500.0150.1460.199
250.1820.1500.1460.2270.1080.1500.0920.1470.170
500.1490.1490.1490.1490.1490.1490.1490.1490.149
750.1130.1470.1550.0740.2020.1000.1890.1500.122
900.0830.1440.1650.0290.2800.0390.2320.1530.090
Panel B: Propensity to Pay Dividends + FIC Fixed Effects
100.2100.1510.1460.2860.0830.1490.0200.1460.200
250.1840.1500.1470.2320.1140.1490.1000.1470.171
500.1490.1490.1490.1490.1490.1490.1490.1490.149
750.1170.1470.1550.0800.2070.1020.1920.1510.123
900.0870.1450.1660.0330.2850.0450.2350.1540.091

Table VIII shows that the propensity to pay dividends has economically meaningful sensitivity to local product market fluidity. Varying the local product market fluidity variable from the 50th to the 90th percentile decreases the propensity to pay dividends from 14.9% to 8.3% in Panel A without industry fixed effects, and from 14.9% to 8.7% in Panel B with industry fixed effects.

C. Dividend Initiations and Omissions

Thus far, our analysis has been cross-sectional. We supplement this analysis with a time-series analysis that tests whether product characteristics are related to dividend initiations, omissions, increases, and decreases. Table IX reports the results.

Table IX. Dividend Initiations, Omissions, Increases, and Decreases
 LocalSelf-  Log   NYSE Neg.Retained3-YearExtra Controls+Obs./
 ProductProduct TotalFirmMarket-AssetIncome/SizeR&D/Earn.EarningsSalesIndustryPseudo-
RowFluidityFluidityHHIRiskAgeto-BookGrowthAssetsPercentileSalesDummy/AssetsGrowthEffectsR2
Panel A: Dividend Initiations
(1)−0.397−0.0140.034−0.6660.029−0.176−0.7771.6600.181    Neither32,658
 (−6.90)(−0.28)(0.79)(−6.73)(0.65)(−2.16)(−12.62)(8.32)(3.90)     0.028
(2)−0.2400.0170.022−0.5590.032−0.005−0.7171.0980.258−0.046−0.2690.161−0.385Both29,620
 (−3.19)(0.30)(0.41)(−5.33)(0.70)(−0.06)(−9.13)(3.91)(4.97)(−0.35)(−2.67)(1.26)(−4.28) 0.039
Panel B: Dividend Omissions
(3)0.3140.1640.0680.543−0.170−0.265−0.067−0.195−0.766    Neither10,341
 (5.19)(3.41)(1.18)(9.36)(−3.03)(−2.37)(−1.26)(−3.54)(−7.80)     0.075
(4)0.1900.1230.0350.493−0.171−0.319−0.0270.052−0.8930.2020.135−0.486−0.220Both10,148
 (2.12)(2.18)(0.51)(7.53)(−2.50)(−2.28)(−0.46)(0.65)(−7.79)(1.63)(2.27)(−3.82)(−3.55) 0.104
Panel C: Dividend Increases
(5)−0.208−0.165−0.033−0.5940.0860.2230.1380.2840.383    Neither10,341
 (−4.33)(−5.33)(−0.85)(−9.55)(2.06)(5.16)(4.85)(5.69)(8.31)     0.172
(6)−0.151−0.147−0.069−0.4880.0590.1920.0810.2080.520−0.280−0.1490.3360.094Both10,148
 (−2.72)(−4.68)(−1.54)(−7.68)(1.32)(3.78)(2.58)(3.15)(9.56)(−3.53)(−3.45)(5.30)(2.58) 0.241
Panel D: Dividend Decreases
(7)−0.1000.193−0.0990.1410.178−0.167−0.261−0.107−0.223    Neither10,341
 (−1.29)(3.73)(−1.73)(2.47)(2.20)(−0.99)(−3.72)(−1.54)(−2.27)     0.022
(8)−0.1630.215−0.1610.0550.173−0.067−0.151−0.012−0.152−0.1990.326−0.114−0.299Both10,148
 (−1.59)(3.51)(−2.00)(0.81)(2.19)(−0.53)(−2.07)(−0.16)(−1.49)(−1.66)(4.96)(−1.47)(−4.31) 0.047

Panel A of Table IX examines the decision to initiate dividends. We begin with the subsample of firms that do not pay a dividend in year inline image. The dependent variable is one if the firm begins paying dividends in year t, and is zero otherwise. The results show that dividend initiations are less likely in locally fluid product markets. In fact, fluidity is among the most significant variables in the model. Among the other control variables, dividend initiators are more likely to be less risky, older, larger, and more profitable.

Panel B of Table IX examines the decision to omit dividends. Here, we begin with the subsample of firms that do pay dividends in year inline image. The dependent variable is one if the given firm ceases dividend payments in year t. This is the mirror image of dividend initiators. The table shows that omitters have more fluid products, both local product market fluidity and self-fluidity. One interpretation of this result is that these firms might have experienced technological shocks, and their product markets moved from a stable state to a more fluid one. Product life cycle theories (Abernathy and Utterback (1978), Klepper (1996)) would suggest that such firms face more competitive risk as they restart the search for a dominant design, and hence they might cease dividend payments to preserve cash. Omitters are also riskier and younger firms, and they are more likely to have negative earnings.

Panels C and D of Table IX examine the decision to increase or decrease dividends. In both panels, we limit the subsample to firms that pay dividends in year inline image. In Panel C, the dependent variable is one if the firm increased its dividend in year t and zero otherwise. In Panel D, the dependent variable is one if the firm decreased its dividend in year t, and is zero otherwise. Consistent with our central hypothesis, firms that increase dividends are more likely to have less fluid products. Both local product market fluidity and self-fluidity are negative and highly significant. Regarding the control variables, increasers are also less risky, older, profitable, and more likely to have high market-to-book ratios and high asset growth. The results for decreasers in Panel D are insignificant.

The only text-based variable that is highly significant is self-product fluidity. Hence, firms that reduce dividends are likely to have products that change relative to the firm itself over time. Such firms might be reacting to technology shocks in their own product offerings. We also observe that firms in concentrated product markets are less likely to decrease dividends. Overall, our results are consistent with the view that firms with greater product market stability are more likely to pay dividends and initiate or increase dividends. These firms likely face fewer competitive threats, and are more willing to make payouts.

We also explore how dividends respond to changes in fluidity. The results, reported in the Internet Appendix, show that longer run changes in fluidity, but not short-run changes, are negatively associated with increased payouts. These results are consistent with managers responding to more permanent changes in the product market given the signaling costs of changing dividends.

Overall, our results are consistent with firms that have greater product market stability paying more dividends, initiating more often, and increasing dividends. These firms likely face fewer competitive threats, and are more willing to make payouts.

D. Repurchases

In this section, we examine the propensity to repurchase shares using multivariate logit regressions. As before, we include our text-based product characteristic variables in addition to the control variables used in Fama and French (2001) and Hoberg and Prabhala (2009). Table X presents the results of a logit specification where the dependent variable is one if the firm repurchases shares during the year (Panel A), if a firm repurchases shares at a level that is higher than 1% of its assets (Panel B), and if a firm is a repurchaser 2 years in a row (Panel C). As before, we estimate the logit regressions using a panel specification with time fixed effects, and standard errors that account for clustering by firm.

Table X. Repurchase Policy and Product Fluidity
 LocalSelf-  Log   NYSE Neg.Retained3-YearExtra Controls+Obs./
 ProductProductTNICTotalFirmMarket-AssetIncome/SizeR&D/Earn.EarningsSalesIndustryPseudo-
RowFluidityFluidityHHIRiskAgeto-BookGrowthAssetsPercentileSalesDummy/AssetsGrowthEffectsR2
Panel A: Positive Repurchase Dummy
(1)−0.1590.026−0.026−0.3770.062−0.083−0.4300.4080.386    Neither42,999
 (−7.79)(1.84)(−1.50)(−15.66)(3.27)(−4.40)(−21.77)(14.50)(17.51)     0.137
(2)−0.1150.028−0.026−0.3710.022−0.081−0.4200.2340.384−0.060−0.1720.152−0.107Controls39,768
 (−5.06)(1.91)(−1.48)(−14.11)(1.05)(−3.71)(−19.58)(7.21)(16.36)(−2.14)(−9.29)(5.36)(−6.11)Only0.141
(3)−0.098−0.007−0.014−0.3910.053−0.108−0.4130.2420.4070.011−0.1620.153−0.111Both39,768
 (−3.65)(−0.44)(−0.72)(−14.40)(2.46)(−4.74)(−18.96)(7.43)(17.08)(0.41)(−8.69)(5.48)(−6.17) 0.167
Panel B: Repurchase More than 1% of Assets Dummy
(4)−0.1310.018−0.029−0.429−0.0090.091−0.5690.5310.394    Neither42,999
 (−6.12)(1.13)(−1.52)(−15.07)(−0.47)(5.02)(−23.99)(15.06)(17.86)     0.124
(5)−0.1030.017−0.041−0.411−0.0610.086−0.5680.3410.407−0.056−0.2190.165−0.146Controls39,768
 (−4.32)(1.06)(−2.09)(−13.51)(−2.90)(4.07)(−21.61)(8.28)(17.39)(−1.58)(−10.18)(4.76)(−6.87)Only0.134
(6)−0.084−0.027−0.015−0.458−0.0190.022−0.5800.3800.4430.019−0.2100.152−0.140Both39,768
 (−3.00)(−1.62)(−0.75)(−14.00)(−0.87)(0.99)(−21.12)(8.96)(18.50)(0.54)(−9.62)(4.58)(−6.25) 0.162
Panel C: 2-Year Repurchase Dummy
(7)−0.201−0.005−0.026−0.5390.146−0.040−0.4470.4520.403    Neither42,999
 (−8.13)(−0.31)(−1.25)(−16.08)(6.72)(−1.76)(−20.05)(12.04)(16.18)     0.158
(8)−0.145−0.003−0.031−0.5130.075−0.012−0.3870.1980.411−0.081−0.1830.256−0.266Controls39,768
 (−5.29)(−0.16)(−1.48)(−14.47)(3.17)(−0.49)(−16.16)(4.61)(15.69)(−1.73)(−8.09)(5.19)(−10.62)Only0.165
(9)−0.114−0.038−0.025−0.5350.113−0.051−0.3820.2170.4400.001−0.1760.248−0.270Both39,768
 (−3.55)(−2.15)(−1.08)(−14.45)(4.53)(−1.92)(−15.69)(5.10)(16.39)(0.01)(−7.73)(5.36)(−10.53) 0.191

Panel A displays results for the positive repurchase dummy. Results are similar but generally weaker than those for dividend propensity presented in Table VI. Firms in more fluid product markets are less likely to repurchase shares. The results in Panel B for aggressive repurchasing and those in Panel C for 2-year repurchasers are similar to those in Panel A. We conclude that the negative relationship between product market fluidity and repurchasing is robust.12

E. Cash Holdings and Fluidity

We next examine if fluidity is positively related to the cash balances held by firms. Such evidence would suggest that competitive threats measured by fluidity are related to firms being more conservative on multiple financial policies—dividends, repurchases, and cash holdings.

Our specifications follow the cash literature (e.g., Bates, Kahle, and Stulz (2009)). The dependent variable is cash plus cash equivalents divided by firm assets and control variables include those suggested by the cash literature. The new controls include an estimate of the foreign tax burden and industry acquisition intensity. Foreign tax burden equals the maximum of zero or foreign income times a firm's marginal effective tax rate computed as in Graham (1996) minus foreign taxes paid as in Foley et al. (2007). The industry acquisition intensity is the total number of acquisitions divided by the number of firms in a given industry. As in previous tables, independent variables are standardized prior to fitting regressions to permit more intuitive comparisons across variables. All specifications include year fixed effects, while specification 2 includes FIC-300 industry fixed effects. Table XI presents the results.

Table XI. Cash Holdings and Fluidity
 (1)(2)
  With Industry
  Controls
Local product fluidity0.0450.030
 (20.378)(12.932)
TNIC HHI−0.023−0.011
 (−12.445)(−5.589)
Log firm age−0.011−0.009
 (−5.650)(−5.499)
Market-to-book0.0490.038
 (27.237)(21.707)
ln(book assets)−0.047−0.042
 (−23.282)(−20.729)
Earnings/assets−0.006−0.001
 (−2.985)(−0.559)
Ind. acq. intensity0.007−0.001
 (4.715)(−0.710)
Foreign tax burden0.0110.006
 (7.115)(4.505)
Cash flow risk0.0330.013
 (14.531)(5.831)
Capx/assets−0.034−0.024
 (−25.583)(−17.377)
Neg. earn. dummy0.0230.006
 (6.183)(1.666)
R&D/assets0.0560.043
 (20.151)(13.785)
Constant0.1700.182
 (23.062)(12.172)
R20.4410.525
N41,65541,637

We find that firms with high local product market fluidity maintain higher cash balances. This result is significant at the 1% level in all specifications. We conclude that our results for cash holdings echo the broad theme of financial conservatism observed in our dividend regressions. Firms facing competitive threats in their product markets adopt more conservative financial policies as they hold higher cash balances in addition to paying lower dividends and repurchasing less.

Following Bates, Kahle, and Stulz (2009), we also construct samples to examine the effect of fluidity on firms that are likely to have less access to external financing. We examine the effect of fluidity on cash holdings by young versus old firms (columns 1 and 2), loss-making versus profitable firms (columns 3 and 4), and noninvestment grade versus firms with investment grade bond ratings (columns 5 and 6). All specifications include year and FIC-300 industry fixed effects. Table XII presents the results.

Table XII. Cash Holdings and Fluidity: Age, Earnings, and Bond Ratings
 (1)(2)(3)(4)(5)(6)
 YoungOldNeg.Pos.NoninvestmentInvestment
 FirmsFirmsEarn.Earn.GradeGrade
Local product fluidity0.0480.0260.0330.0210.0340.010
 (8.228)(10.856)(8.633)(8.666)(12.557)(2.910)
TNIC HHI−0.019−0.010−0.008−0.009−0.009−0.009
 (−3.585)(−4.971)(−2.385)(−4.727)(−4.032)(−2.916)
Log firm age−0.011−0.014−0.023−0.004−0.010−0.004
 (−1.299)(−5.918)(−7.013)(−2.036)(−5.034)(−1.253)
Market-to-book0.0320.0390.0260.0470.0340.049
 (8.080)(20.576)(11.000)(18.604)(17.886)(12.699)
ln(book assets)−0.059−0.040−0.034−0.043−0.041−0.041
 (−10.481)(−18.450)(−9.394)(−20.844)(−16.065)(−11.209)
Earnings/assets−0.0010.001−0.0020.0150.001−0.001
 (−0.183)(0.232)(−0.742)(1.859)(0.225)(−1.039)
Ind. acq. intensity−0.0113−0.0004−0.00170.0012−0.00120.0029
 (−2.075)(−0.293)(−0.845)(0.747)(−0.794)(1.234)
Foreign tax burden0.0090.0060.0010.0060.0070.005
 (2.183)(4.165)(0.028)(4.620)(3.778)(3.680)
Cash flow risk0.0140.0130.0200.0080.0170.003
 (2.308)(5.428)(4.909)(3.541)(6.065)(0.921)
Capx/assets−0.021−0.025−0.028−0.022−0.024−0.023
 (−7.258)(−16.373)(−13.110)(−13.837)(−15.024)(−9.156)
Neg. earn. dummy0.0240.004  0.0020.004
 (2.642)(1.092)  (0.447)(0.387)
R&D/assets0.0210.0490.0320.3940.0390.292
 (4.147)(13.544)(11.181)(2.934)(12.965)(2.345)
Constant0.1430.1840.1540.2380.1810.225
 (2.853)(11.518)(5.966)(9.323)(10.785)(6.084)
R20.5480.5280.5370.4590.5320.493
N4,47737,16014,11227,52530,43811,199

We find that fluidity is significantly more important for the holding of cash balances for younger firms, loss-making firms, and noninvestment grade firms. These results are consistent with the effects of fluidity being more pronounced when firms with less access to capital markets face competitive threats. More broadly, our results are consistent with firms holding higher precautionary cash balances when they face more competitive threats from rival firms.

V. Conclusions

  1. Top of page
  2. ABSTRACT
  3. I. Product Market Threats, Payouts, and Cash
  4. II. Understanding Fluidity
  5. III. Data and Summary Statistics
  6. IV. Payout Policy
  7. V. Conclusions
  8. Appendix A
  9. Appendix B
  10. REFERENCES
  11. Supporting Information

Our paper examines how product market threats and underlying product dynamics impact firm payout policy and cash holdings. We use computational linguistics of the text in 42,000 firm business descriptions contained in 10-Ks to characterize the competitive threats faced by firms in their product markets. While managers frequently cite stability and risk as the most important determinants of payout policy, our results shed light on the underlying product-side mechanisms that cause these factors to affect payout and cash holding policy.

Our analysis shows two central ways through which product characteristics affect payout policy and cash balances. First, firms facing greater product market fluidity are less likely to pay dividends or repurchase shares. These firms also hold higher cash balances. Thus, fluidity captures threats to a firm's product market beyond measured risk. We also show that fluidity is significantly related to the text in business descriptions of IPO firms and firms newly raising venture capital, consistent with fluidity capturing forward-looking threats in a firm's competitive environment. These results are consistent with the hypothesis that ongoing competitive threats drive firms' payout and cash holding decisions. More broadly, firms with higher competitive threats adopt more conservative financial policies, in line with the deep pockets arguments of Telser (1966) and Bolton and Scharfstein (1990). The results also support the product life cycle theories of Abernathy and Utterback (1978) and Klepper (1996) and the life cycle perspective of payouts advocated by DeAngelo, DeAngelo, and Skinner (2006).

Overall, our evidence suggests that product characteristics affect payout polices and cash holdings along more than one dimension, and they also affect the choice of payout type. These results motivate the need for additional work linking product market competition and product life cycles to payout policy, cash holdings, and other financial and strategic decisions of corporations. In this regard, our results also highlight the advantages of using the dynamic aspects of product text to examine the structure of product markets and competitive threats. Product text descriptions offer not only a sharper characterization of the competitive structure of product markets, but also a more timely measure of its dynamics.

Editor: Campbell Harvey

Appendix A

  1. Top of page
  2. ABSTRACT
  3. I. Product Market Threats, Payouts, and Cash
  4. II. Understanding Fluidity
  5. III. Data and Summary Statistics
  6. IV. Payout Policy
  7. V. Conclusions
  8. Appendix A
  9. Appendix B
  10. REFERENCES
  11. Supporting Information

In this section, we present an example of how self-fluidity and product market fluidity are computed. Consider an industry with three firms in the portable telephone product market. Suppose, the product market vocabulary consists of the following words: telephone, cellular, digital, analog, Internet, iPhone, and Android. Further, suppose the three firms use the following subsets of this overall vocabulary in years inline image and t, respectively:

 Firm 1Firm 1Firm 2Firm 2Firm 3Firm 3
WordYear inline imageYear 0Year inline imageYear tYear inline imageYear t
TelephoneYesYesYesYesYesYes
CellularYesYesYesYesYesYes
DigitalNoNoYesYesNoNo
AnalogYesYesNoNoYesYes
InternetYesYesNoYesNoYes
iPhoneNoNoNoYesYesYes
AndroidYesYesNoYesNoNo

To compute firm 1's self-fluidity in year t, we simply take one minus the cosine similarity of firm 1's year inline image and year t normalized word vectors. This is equal to inline image, which is zero. This example is interesting because we observe that firm 1 did not change its own products and has zero self-fluidity, yet we will soon see that firm 1's surrounding product market (accounting for the changes of its rivals) has nontrivial fluidity levels. For firm 2, self-fluidity is inline image, which is 0.293. Hence, firm 2 has a nontrivial self-fluidity.

To compute firm 1's product market fluidity, we first need to compute the overall change in usage vector inline image as the difference in sums of the non-normalized word vectors. This is a property of all firms in the economy and is given by

  • display math(A1)

From equation (A2), firm 1's product market fluidity is thus (where 2.449 is the normalizing content for inline image)

  • math image(A2)

We conclude that firm 1 faces a substantial amount of product market fluidity despite the fact that firm 1 has not changed any of its own products. The main idea is that the changes in the smart phone vocabulary terms (Internet, iPhone, and Android) generate exposure to rival movements for firm 1 and the ability of firm 1's rivals to move in this space can be viewed as a competitive threat to firm 1.

Appendix B

  1. Top of page
  2. ABSTRACT
  3. I. Product Market Threats, Payouts, and Cash
  4. II. Understanding Fluidity
  5. III. Data and Summary Statistics
  6. IV. Payout Policy
  7. V. Conclusions
  8. Appendix A
  9. Appendix B
  10. REFERENCES
  11. Supporting Information

Following Fama and French (2001), we control for firm size using New York Stock Exchange (NYSE) market capitalization percentile, which is the fraction of NYSE firms having equal or smaller capitalization than firm i in year t. Other control variables include Asset growth, which is the percent growth in assets from year inline image to year t, and Earnings/Assets (profitability), which is earnings before extraordinary items plus interest expense plus income statement–deferred taxes divided by assets. We also control for market-to-book. Book Equity (BE) is stockholder's equity minus preferred stock plus balance sheet–deferred taxes and investment tax credit minus post retirement assets. If stockholder's equity is not available, we replace it by either common equity plus preferred stock par value, or assets minus liabilities. Preferred Stock is preferred stock liquidating value, or preferred stock redemption value, or preferred stock par value. Market Equity is the fiscal year closing price times shares outstanding. Following Hoberg and Prabhala (2009), we control for a firm's risk by including the standard deviation of its daily stock returns from CRSP in the given calendar year.

We include additional variables that could be correlated with our product text variables. We control for firm age by including the natural log of one plus firm age. We compute age for a given firm as the current year minus the firm's founding date. For the 15.2% of firms in our sample missing age data, we use the CRSP listing vintage as a substitute for the founding date.13 We also include a negative earnings dummy, which equals one when a firm reports negative earnings in a given year and is zero otherwise, as DeAngelo, DeAngelo, and Skinner (1992) show that this variable has a large impact on dividend payout. We control for R&D because firms spending more money on R&D may invest more in creating new products. We also consider whether our measure provides any information not already contained in R&D and thus include lagged R&D divided by sales as an additional control.

We also include retained earnings divided by total assets, following DeAngelo, DeAngelo, and Skinner (2006), as this variable is related to firm maturity. This retained earnings variable has a correlation with product fluidity of only −24.5%, confirming that product market fluidity (related to product life cycles) is indeed distinct. We also include log sales growth of the firm itself from year inline image to year t. We also consider firm patenting activity, as measured through a combination of two data sources: (i) the NBER patent citations file as extended by Bronwyn Hall and (ii) the words patent, patents in each firm's product description. The Text plus Applied Patents variable is one if the given firm mentions patents in its product description or if it applied for a patent in the most recent 3-year window.14 We also consider a dummy variable indicating whether the firm has a credit line (Chava and Roberts (2008)). Firms with credit lines may have better access to external credit and thus might be more willing to pay dividends. We also consider Hoberg and Phillips (2010a) FIC industry fixed effects.

Finally, we account for the current product market competition faced by firms. We consider each firm's HHI based on the “Textual Network Industry Classification” (TNIC) industries formed using firm-by-firm similarity measures as in Hoberg and Phillips (2010a).15 These measures are updated each year. Firm i's industry cluster comprises firms j whose product descriptions have similarity to i's products exceeding a threshold, as discussed in section 'Conclusions'. A of Hoberg and Phillips (2010a). This measure also excludes firm pairs in industries that are more than 1% vertically related based on Bureau of Economic Analysis (BEA) input–output tables.

  1. 1

    Cash and cash equivalents amount to $1.73 trillion as of the end of 2012. See Federal Reserve Release B.102 (http://www.federalreserve.gov/releases/z1/current/accessible/b102.htm).

  2. 2

    In the article, an analyst at Evercore Partners notes that low cash “absolutely limited its options” despite HP's relatively healthy balance sheet. The article goes on to say that this is because “the sector changes more rapidly” than other sectors. See Ben Worthen, “H-P's handling of cash in spotlight,” Wall Street Journal, November 21, 2011.

  3. 3

    Studies using text to analyze finance theories include Antweiler and Frank (2004), Boukus and Rosenberg (2006), Li (2006), Tetlock (2007), Tetlock, Saar-Tsechanksy, and Macskassy (2008), Hanley and Hoberg (2010), Loughran and McDonald (2011), and Hoberg and Phillips (2010b).

  4. 4

    Related work includes Venkatesh (1989), Chay and Suh (2009), and Fink et al. (2010).

  5. 5

    A partial list of recent work includes Opler et al. (1999), Faulkender and Wang (2006), Dittmar and Mahrt-Smith (2007), Han and Qiu (2007), Gamba and Triantis (2008), Bates, Kahle, and Stulz (2009), Foley et al. (2009), Riddick and Whited (2009), and Lins, Servaes, and Tufano (2010).

  6. 6

    This motive is often attributed to Keynes (1936).

  7. 7

    For a discussion of dot products and cosine similarities in text processing, see Sebastiani (2002).

  8. 8

    We thank the Wharton Research Data Service for providing us with an expanded historical mapping of SEC CIK to Compustat gvkey.

  9. 9

    The Internet Appendix may be found in the online version of this article.

  10. 10

    As discussed earlier, we base our main results on local product market fluidity, which restricts 10-K words to those that appear in localized product market groups. Our results continue to hold if we instead use broader words as shown in our Internet Appendix.

  11. 11

    These quartiles represent univariate comparisons between payers and nonpayers adjusted only for Fama–French characteristics, so quartiles pick up other factors that are omitted.

  12. 12

    We show in the Internet Appendix that fluidity is significant in explaining repurchases among firms that pay no dividends. Also, Table X is based on gross repurchases made by a firm, which is a good approximation of the dollar repurchases disclosed by firms in their 10(b)-18 quarterly disclosures (Banyi, Dyl, and Kahle (2008)). We further experimented with net repurchase measures, which subtract share issuances, for example, those due to option exercises, using the method suggested by DeAngelo, DeAngelo, and Skinner (2008) in their footnote 1 (see also Skinner (2008), footnote 7). Our results are robust to this alternative reported in the Internet Appendix.

  13. 13

    Our results are similar if we use listing vintage for all observations instead of the founding date. We thank Gustavo Grullon and James Weston for generously providing us with the data on firm founding dates used to construct this control variable.

  14. 14

    Our results change little if we instead use measures based on patent counts instead of a dummy variable, or if we separately control for text-based patent measures, patent applications, or patents granted.

  15. 15

    This industry classification is constructed to be as coarse as three-digit SIC code industries.

REFERENCES

  1. Top of page
  2. ABSTRACT
  3. I. Product Market Threats, Payouts, and Cash
  4. II. Understanding Fluidity
  5. III. Data and Summary Statistics
  6. IV. Payout Policy
  7. V. Conclusions
  8. Appendix A
  9. Appendix B
  10. REFERENCES
  11. Supporting Information

Supporting Information

  1. Top of page
  2. ABSTRACT
  3. I. Product Market Threats, Payouts, and Cash
  4. II. Understanding Fluidity
  5. III. Data and Summary Statistics
  6. IV. Payout Policy
  7. V. Conclusions
  8. Appendix A
  9. Appendix B
  10. REFERENCES
  11. Supporting Information

Disclaimer: Supplementary materials have been peer-reviewed but not copyedited.

FilenameFormatSizeDescription
jofi12050-sup-0001-Tables.pdf57K

Table S1: Local versus Broad Fluidity Measures

Table S2: Dividend Initiations, Omissions, Increases, and Decreases versusChanges in Fluidity

Table S3: Repurchase versus Net Repurchase Policy and Product Fluidity

Table S4: Repurchases by Nonpayers

Table S5: Repurchases by Nonpayers: Multinomial Logit

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.