Financial Strength and Product Market Behavior: The Real Effects of Corporate Cash Holding

Financial Strength and Product Market Behavior: The Real Effects of Corporate Cash Holdings LAURENT FRESARD * ABSTRACT This paper shows that large cash reserves lead to systematic future market-share gains at the expense of industry rivals. Importantly, using shifts in import tariffs to identify exogenous intensification of competition, difference-in-difference estimations support the causal impact of cash on product market performance. Moreover, the analysis reveals that the “competitive” effect of cash is markedly distinct from the strategic effect of debt on product market outcomes. Also, this effect is stronger when rivals face tighter financing constraints and when the amount of interactions between competitors is large. Overall, the results suggest that cash policy encompasses a substantial strategic dimension. * Laurent Frésard is from HEC Paris. This paper is based on the first chapter of my dissertation at the University of Neuchatel. I am extremely grateful to Michel Dubois (Chair), François Degeorge, Michel Habib, Erwan Morellec and Milad Zarin for many helpful discussions. I also thank Tom Berglund, François Derrien, John Graham (the Co-Editor), Rachel Hayes, Campbell Harvey (the Editor), Uli Hege, Jean Imbs, Mesrop Janunts, Jens Martin, Sebastien Michenaud, Christophe Pérignon, Evgeny Plaksen, Gordon Phillips, Enrique Schroth, Linus Siming, René Stulz, Philip Valta, an anonymous referee, an anonymous associate editor and seminar participants at EPFL, Imperial College London, HEC Paris, the University of Lugano, the University of Lausanne, the 2009 Winter European Finance Conference in Klosters, the 2008 Frontiers of Finance Conference in Belize, the 2008 Chicago Quantitative Alliance (CQA) Meeting in Chicago, the 6 th Swiss Doctoral Workshop in Finance in Gerzensee, the first Swiss Corporate Finance Day in Neuchâtel, the 2008 EFA meeting in Athens, the 2008 EFMA meeting in Athens, and the 2007 French Finance Association Meeting in Paris for their valuable comments and suggestions. 2 Seldom has corporate strategy been turned on its head so quickly. Not long ago, cash holdings were considered a dangerous thing to accumulate and companies that hoarded large cash positions were viewed with a great deal of suspicion. 1 However, the recent market turmoil and the resultant worsening credit conditions have clearly emphasized the advantage of maintaining a liquid balance sheet, as many firms are desperately seeking to avoid a cash squeeze. 2 This rapidly changing perspective crucially underlines the need for a deeper understanding of what the implications of corporate cash policy really are. Indeed, although recent developments have considerably broadened our knowledge on the various determinants of corporate cash holdings, the literature has so far paid little attention to whether they have a material effect on firms’ day-to-day operations. This paper helps to bridge that gap by examining whether and how cash holdings include a strategic dimension that affect firms’ product market decisions. There are several reasons why cash holdings may influence a firm’s product market choices and those of its competitors. Primarily, a cash-rich firm can use its war chest to finance competitive strategies. For instance, a firm can rely on a strong balance sheet to hurt rivals’ bottom lines and prospects through aggressive pricing; see Bolton and Scharfstein (1990). More generally, a firm may use its cash reserves to fund a number of alternative competitive policies such as the location of stores or plants, the construction of efficient distribution networks, advertising targeted against rivals, or even the employment of more productive workers; see Campello (2006). From a different perspective, a firm’s stock of cash can signal the possibility of aggressive behavior, thereby distorting competitors’ actions in the product market. Accordingly, we can view cash holdings as a preemptive device that may affect, for instance, rivals’ entry or capacity expansion decisions; see Benoit (1984). Overall, theory predicts that cash holdings may have both direct and indirect effects on competitive outcomes. On this ground, I argue that irrespective of the mechanism at work, if 3 cash holdings really influence product market outcomes, we should observe cash-rich firms gaining market share at the expense of their competitors. However, because cash policy may be endogenous to product market performance, establishing a causal link going from cash holdings to product market outcomes is a difficult task. In this paper, I use two different empirical strategies to side-step this identification challenge. First, I use asset tangibility to force the exogenous portion of cash to explain market share growth. As a matter of fact, while a firm’s asset tangibility correlates with its cash reserves, there is little reason to believe that the tangible attributes of a firm’s assets has a direct influence on its product market performance other that through its association with financing ability. Based on firm-level data from a panel of 105 well defined product markets, instrumental variables estimates provide strong evidence that a firm’s stock of cash is associated with future market-share expansion at the expense of industry rivals. More specifically, firms with noticeably higher cash reserves expand their market shares relatively more than their competitors in future years. The estimates reveal an economically important “cash effect”. Across all industries, a one standard deviation increase in relative-to-rivals cash holdings enables the average firm to gain a share of 2.9% in its product market over the next two years. Second, to further mitigate endogeneity concerns that product market performance drives observed cash levels, I exploit exogenous variations of industry-level import tariffs as a quasi-natural experiment. Since the softening of trade barriers increases substantially the competitive pressure from foreign rivals, large reductions of import tariffs represent situations in which firms have to use their preexisting cash positions to compete in an unexpectedly modified product market environment. Difference-in-difference regressions confirm the positive impact of cash on market share growth. Firms with more cash on hand perform significantly better when their industry experiences an exogenous intensification of product market competition. 3 4 To cement the validity of my interpretation, I offer evidence that the estimates truly reflect the positive impact of cash reserves on product market performance rather than biases due to the potential unspecified effect of debt ratios. To do so, I use the methodology developed by Acharya, Almeida and Campello (2007) and identify when cash is equal to, and when it is different from, negative debt. Across various specifications, I uncover a substantial impact of cash on market share growth, even when cash is markedly distinct from negative debt. This analysis highlights that the competitive effect of cash is not the flip-side of the well documented effect of debt ratios. 4 This is an important result given the extensive literature documenting a connection between debt and product market, as well as the conventional view that cash is the negative of debt. Next, I take advantage of the cross-industry nature of the sample to investigate how the effect of cash holdings on competitive performance depends on industry characteristics. In particular, I explore how rivals’ financial status alters the competitive effect of cash holdings. Consistent with the idea that a surfeit of cash confers a strategic advantage over cash-poor rivals, I observe that the cash-performance sensitivity is magnified when rivals have weak financial positions. In a similar vein, I investigate to what extent the competitive effect of cash is determined by the quantity of strategic interactions between firms within their industry. The evidence points to noticeable differential effects. In particular, the effect of cash on market share growth turns out to be twice as large in competitive markets as in concentrated markets. Moreover, the larger the interdependence of a firm’s growth prospects with industry rivals, the greater the effect of cash. In the same way, product market performance is more sensitive to cash in sectors where foreign competition is substantial or when a firm operates in the technological core of its industry. Consistent with a strategic dimension, the impact of cash holdings on product market performance appears to be related to rivals’ financial condition as well as to the intensity of strategic interactions within product markets. 5 Finally, I also examine the impact of relative cash reserves on firm value and operating performance. Using various specifications, I show that firms with large cash reserves experience increases in both market value and return on assets in comparison with their cash-poor rivals. This result, which is robust to the inclusion of several control variables for investment opportunities, suggests that the competitive effect of cash is value enhancing. Putting all the findings together, this paper contributes in two main areas. First, by providing evidence that cash policy comprises a substantial product market dimension, it deepens our understanding of the implications of cash holdings. While previous studies provide evidence supporting a precautionary motive for holding liquid assets, my findings further reveal that such a precautionary behavior turns out to bring real benefits. Taken as a whole, the impact of cash reserves on market share growth appears substantial and depends on rivals’ financing and competitive conditions. As such, the strategic dimension of cash policy needs to be taken into account when assessing the soundness of observed cash levels and might have important implications for understanding how firms react to credit supply shocks that restrict their financing ability and/or governmental policies that modify the nature of their competitive environment. Second, this study complements recent papers that document a negative association between debt ratios and product market performance. This result is typically interpreted as evidence that highly indebted firms are financially fragile and thus can be severely affected by rivals’ competitive strategies. By establishing an independent link between cash holdings and product market outcomes, my results point to an additional channel through which finance affects product market behavior. Importantly, the results highlight that the connections between financial and product market decisions are multifaceted and suggest indirectly that cash and debt policy have distinct implications for product market conducts. In a related perspective, the study is consistent with firms integrating rivals’ financial conditions and 6 competitive positions in their decision processes. This latter point goes in the direction of recent theoretical models that endogenize product market effects into optimal financial decisions. 5 In the next section, I review the relevant literature and develop the main hypothesis. Section II describes the methodology and details the sample. Section III analyzes and characterizes the impact of cash holdings on firms’ product market performance. Section IV presents my conclusions. I. The Setting While much effort has recently been devoted to studying the determinants of firms’ cash policies, evidence on the real implications of firms’ cash reserves remains relatively scarce. 6 In particular, prior empirical work has paid little attention to the potential effects of firms’ cash holdings on their actions and performance in the product market. Yet, from an intuitive as well as a theoretical viewpoint, the idea that firms’ cash reserves might affect product market outcomes is of long standing. For instance, Tesler (1966) and later Bolton and Scharfstein (1990) argue that deep-pocketed firms may increase their output to drive down industry prices. To the extent that rivals face difficulties in accessing funds, the decrease in output price may induce losses for financially weak firms and may possibly drive them out of the market. Consequently, limited access to external funds can hinder a cash-poor firm’s ability to compete vigorously in the product market, which may in turn prompt financially strong rivals to adopt “predatory” behaviors. Chevalier and Scharfstein (1996) also suggest that cash-poor firms may be less inclined to invest in building market share. In their model, firms directly decrease product prices as a means to secure long-term market shares instead of maximizing short-term profits. More generally, cash holdings may be used to fund strategic practices other than predatory pricing. As pointed out by Campello (2006), examples of such 7 policies include decisions about capital outlays, research and development expenses, the location of stores or plants, distribution networks, advertising targeted against rivals, the recruitment of more productive workers, or the acquisition of key suppliers or business partners. Overall, this line of research suggests that cash-rich firms can use their war chests to finance competitive strategies that may, in turn, enhance their performance in the product market. From a related angle, a firm’s stock of cash may also influence other players’ actions indirectly. Certainly, one can view cash reserves as a preemptive weapon that may distort competitors’ strategies. For instance, Benoit (1984) formalizes this idea by showing that if a potential entrant faces financing constraints, the threat of competitive actions by cash-rich incumbents may be sufficient to prevent entry. Consequently, by limiting entry, incumbents’ cash holdings can be viewed as a potential driver of industry dynamics and hence affect firms’ competitive performance. 7 Similarly, cash holdings may act as a credible threat of competitive retaliation, that is, a “second strike” capability against potential capacity expansion by industry rivals. In this spirit, a firm’s cash holdings may affect rivals’ decisions to increase capacity and hence indirectly alter competitive outcomes. Surprisingly, while previous theoretical work suggests both direct and indirect links between a firm’s cash reserves and product market conduct, the empirical assessment of the interplay between finance and the product market concentrates mainly on linking firms’ competitive performance to some measure of debt financing. 8 In view of that, deep-pocketed firms are assumed to be those displaying low levels of leverage. Specifically, because of their limited capacity to raise additional funds, highly indebted firms are assumed to be financially fragile and thus can be severely affected by unlevered rivals’ competitive strategies. Yet, recent evidence challenges this unilateral focus in several dimensions and clearly suggests a potential role for cash holdings in explaining product market outcomes. First, Acharya, 8 Almeida, and Campello (2006) and Gamba and Triantis (2007) show that cash reserves and negative debt (debt capacity) are not equivalent when there is uncertainty about future cash flow. In this case, different combinations of cash and debt may have different effects on firm value and performance. This work draws attention to the fact that when external finance is costly, cash should not be considered as the opposite of debt. In such a context, it is likely that cash and debt play distinct roles in influencing competitive outcomes. Second, some recent works assert that the supply of capital has important implications for corporate capital structure. In particular, Faulkender and Petersen (2006) show that, all else being equal, firms that have access to the public bond market are more levered. Hence, their results suggest that a low level of leverage may not necessarily indicate high debt capacity but may instead be a sign of saturated debt capacity. The same intuition prevails in the work of Lemmon and Roberts (2007) and Sufi (2009). Under such circumstances, a low level of leverage may not be an accurate proxy for financial strength. Third, a number of studies show that corporate liquidity is empirically associated with business risk. In particular, Opler, Pinkowitz, Stulz and Williamson (1999) and Bates, Kahle and Stulz (2009) document that firms with riskier cash flow and limited access to external capital hold more cash. Looking at the influence of product market dynamics on cash policy Haushalter, Klasa, and Maxwell (2006) importantly document that the level of cash is positively associated with proxies for predation risk. Nevertheless, they remain silent on the product market consequences of cash policy. In all, theoretical works, together with evidence from different horizons clearly suggest that corporate cash holdings may play an important role in explaining observed product market outcomes. In this paper, I take a step in that direction by empirically examining whether and, if so, how cash holdings affect a firm and its rivals’ competitive behavior. Below, I provide compelling evidence of a causal link going from cash holdings to 9 product market performance and hence confirm that cash policy encompasses a strategic dimension by. II. Methods and Data A. Measuring the Impact of Cash on Product Market Outcomes To explore the interplay between cash holdings and product market outcomes, I investigate the link between cash and market share growth. As a matter of fact, I argue that irrespective of the mechanism at work, if cash holdings include a valuable strategic component, it will ultimately be reflected in firms’ performance in their product markets. Therefore, I examine whether firms with large cash reserves expand their market shares more than their industry rivals. To do so, I follow Campello (2003, 2006) and specify the following baseline model: 9 , , 2 , ( ) ' ? = + + + + MarketShares zCash X i t i t i t i t a ? ? ß e - i (1) where the subscripts i and t represent respectively the firm and the end of the year. The dependent variable, ?MarketShares, is sales growth minus its industry-year average, so that this variable measures a firm’s sales expansion in relation to that of its competitors or equivalently serves as a proxy for market share growth. 10 To reliably gauge the effects of cash holdings on market share dynamics, I need to characterize a firm’s cash position compared with those of its rivals. For that purpose, I follow MacKay and Phillips (2005) and “z-score” the ratio of cash to total assets within each industry-year. Specifically, I compute zCash by subtracting from the cash-to-asset ratio its industry-year mean and divide the difference by the industry-year standard deviation. The motivation for z-scoring cash can is the following. Imagine that a firm has 5% more cash than its average rival. Clearly, the competitive 10 advantage contained in this deviation is a function of the industry-year cash-to-assets dispersion. Indeed, a 5% cash deviation in an industry in which the standard deviation is 2% is likely to provide more strategic value than in an industry with a 15% standard deviation. Hence, I assume that the dispersion of liquid assets within an industry-year conditions the advantage provided by a firm’s cash reserves. The vector Xi includes control variables designed to capture other direct sources of product market performance that may directly correlate with firms’ cash positions. Specifically, I include firm size, past debt ratios and past market share growth. One- and twoyear lagged leverage mitigates the possibility that a correlation between cash and market share growth reflects the unspecified effect of capital structure, while one- and two-year lagged market share growth capture the influence of other firm characteristics that may have driven competitive performance in the recent past, such as change in store location or distribution network. 11 Finally, I account for time-invariant firm heterogeneity and time trend by including a vector of firm fixed effects as well as time dummies (ai and ?t ). I adjust the estimates’ standard errors for within-firm-period error clustering and heteroskedasticity; see Petersen (2009). In estimating equation (1), my primary interest is in the sensitivity of market share growth to relative-to-rivals lagged cash holdings (? ). Even though this measure is too general to pin down the specific channels through which cash holdings shape product market actions, it summarizes relevant information from the combination of direct and indirect strategic effects and is available for a large cross-section of industries. B. Identification There are two possible sources of endogeneity bias in specification (1) that make it challenging to identify a causal link going from cash holdings to product market performance. 11 First, cash policy may be endogenous to industry structure. Hence, while a positive association between cash and future market shares may indicate that high cash holdings enable firms to gain market shares, it may also arise if market share expansion drives observed cash levels. Second, a positive correlation between cash reserves and market share growth may arise even if there is no causal relation between them, simply because both a firm’s cash policy and its market share growth are affected by the same factors that are not observable to the econometrician. In this paper, I address both endogeneity concerns in several ways. Foremost, I use two identification strategies to tackle the potential endogeneity of cash policy to product market structure. First, I estimate specification (1) by using an instrumental variables (IV) approach. Specifically, I include in the set of instruments for cash two of its own lags as well as contemporaneous asset tangibility. The lags of cash capture systematic differences in the levels of cash whereas asset tangibility forces the exogenous portion of cash to explain product market performance. As recently reported by Capkun and Weiss (2007), cash holdings are negatively associated with “hard” assets such as inventory, receivables, or fixed capital. While a firm’s asset tangibility may correlate with its cash reserves, there is little reason to think that the tangible attributes of a firm’s assets has a direct influence on its product market performance other than through its association with financing ability. Hence asset tangibility can reasonably be regarded as exogenous to product market outcomes and thus represents a valid instrument for cash holdings. 12 Alternatively, I use reductions of industry-level import tariffs to further pin down the competitive effect of cash. As a matter of fact, reductions of import tariffs decrease substantially the cost of entering U.S. product markets and thus increase the competitive pressure on domestic producers. As a result, I use these events as a quasi-natural experiment to isolate the causal effect of cash on market share growth. Precisely, using a difference-in-12 difference methodology, I look at how ex-ante differences in predetermined cash holdings lead to differential market share responses following the exogenous increase of competition triggered by tariff reductions. Arguably, although the use of instrumental variables and quasi-natural experiment mitigate concerns about endogenous cash policy, it does not resolve by itself the possibility of spurious correlation due to unobservable characteristics. Note, however, that the test design addresses this problem in three ways. First, I include in equation (1) different control variables that should help capture a wide range of unobservable effects. In particular, the inclusion of firm-fixed effects removes omitted time-invariant firm factors that may lead to spurious correlations between cash holdings and product market performance. Second, as put forth by Campello (2006), the use of relative-to-industry variables minimizes the concern of spurious correlation driven by unobservable industry effects, since all industry-related factors are removed from the estimates. Third, I estimate the effect of cash across different groups that are sorted based on financial and competitive dimensions. As it is not obvious why potential omitted variables would have a stronger systematic effect on the cash-performance sensitivity across various groups, cross-sectional contrasts should further limit the risk of spurious correlation. 13 C. Sample Construction and Industry Definition I gather annual firm-level data from Compustat’s tapes over the period 1973 to 2006. I exclude firm-years for which information on sales, cash holdings, and total assets are not available. I also eliminate observations with negative equity, sales, or asset growth larger than 200%. I classify product markets (industries) at the four-digit SIC level and restrict my focus to manufacturing firms (2000-3999 SIC range). As pointed out by Clarke (1989) and Kahle and Walking (1996), some of the three- and four-digit codes may fail to define sound 13 economic markets. To minimize such concerns, I follow Clarke (1989) and exclude four-digit SIC codes ending with 0 and 9. Moreover, since the estimations use industry-adjusted data, I restrict the sample to include only industry-years with a minimum of ten firms with available information on sales, cash, and total assets. This selection procedure leaves me with a sample of 105 four-digit industries. In an online appendix, I provide the detailed definitions of the variables used in the following analysis and present descriptive statistics. Overall, they are comparable to those found in related studies, such as Campello (2006), Acharya, Almeida and Campello (2007) and Bates, Kahle and Stulz (2009). III. Results: The Effect of Cash Holdings on Market Share Growth A. Main Results To estimate the baseline specification (1), I proceed in two steps. First I obtain the exogenous portion of cash holdings by regressing them on their lagged values and asset tangibility where tangibility is a function of receivables, inventory, and fixed capital defined as in Berger, Ofek, and Swary (1996). In a second step, I z-score instrumented cash holdings and use them as a measure of financial strength in the market share equation (1). 14 Table I displays the instrumental variables estimates of the effect of relative-to-rivals cash holdings on market share growth. In column 1, the coefficient on zCasht-2 is significantly positive, suggesting that cash-rich firms outperform their more financially fragile rivals in the product market. In terms of economic magnitude, all else being equal, a one standard deviation increase in cash in relation to rivals in year t leads to a 2.9% (significant at 1%) gain in market share between years t+1 and t+2. In column 2, I perform a similar regression but consider a one-year lag of relative-to-rivals cash instead of the previous two-year lag. This change has no bearing on the conclusions. We continue to observe a positive and significant effect of cash on future market share expansion (coefficient of 0.03 with a t-statistic over 12). 14 [Table I about here] Note that the estimated coefficients of the control variables have the expected signs. In particular, we observe a negative association between two-year lagged leverage and future market-share development (-0.004 with a t-statistic of 3.52). In essence, this result corroborates the argument that excessive debt hurts product market performance. 15 We notice, however, that cash turns out to have a markedly larger impact on future market-share gains. Also, we observe that past performance explains a large portion of current performance. For completeness, column 3 reports the results of the first stage regression. Consistent with Capkun and Weiss (2007), asset tangibility is negatively related to cash holdings. Moreover, we see that lagged cash plays an important role in explaining observed cash levels. Also, the instruments have strong predictive power as the large R 2 suggests that the instrumental set explains a sizeable fraction of cash holdings variation and hence lessens the possibility that weak instruments contaminate the inference. In further support of the instruments, the test of overidentifying restrictions (J-statistics) cannot reject the joint null hypothesis that the instruments are uncorrelated with the error term and are correctly excluded from the second-stage regression. To give additional support for these first results, Table II presents additional versions of the baseline specification. In particular, I first control for past acquisition activity. As shown in Harford (1999), deep-pocketed firms are more likely to attempt acquisitions. Hence, the above results might simply translate the fact that cash-rich firms mechanically gain market share via external growth. Column 1 of Table II reveals that the competitive effect of cash is not altered by the inclusion of acquisition intensity (dollars spent on acquisitions scaled by assets). As expected, the one-year lagged acquisition intensity positively contributes to market share expansion, whereas the two-year lagged estimate shows a negative sign. Interestingly, this negative effect is in line with Harford (1999), who documents that acquisitions by cash-15 rich firms are followed by abnormal declines in operating performance. In column 2, I repeat this analysis by considering the sales contributions of acquisitions instead of the dollar amount spent on them. The results are virtually unchanged. I also take into account lagged market-to-book value to control for the residual effect of potential growth opportunities not captured by past sales growth and leverage. Column 3 reveals that this addition leads to similar results. [Table II about here] Another important issue relates to the use of z-scored cash to identify relative-to-rivals financial strength. Actually, the z-scoring procedure relies on estimates of the industry-year standard deviations of the cash-to-asset ratios. However, the requirement of a minimum of ten observations by industry-year induces a skewed distribution of cash-to-asset ratios for each industry-year that might twist the inference. I address this concern in three ways. First, in column 4, I restrict the sample to observations from industry-years with a minimum of 30 firms. In column 5, I avoid using standard deviation estimates and replace z-scored cash with its industry-adjusted value. Finally, following Campello (2006), I consider only observations from industry-years in which the skewness of cash-to-asset ratio is between -1 and 1 and report the results in column 6. Although these estimations lower the number of observations considerably, these changes have no bearing on the conclusions. B. Cash as Negative Debt? One potential important concern with the results in Tables I and II is that the effect of cash on product market performance might actually reflect the unspecified effect of debt ratios. As a matter of fact, the bulk of the literature that blends corporate financing to product market decisions concentrates on studying the impact of debt financing on competitive outcomes. In particular, several recent studies document that firms with low debt ratios, taken 16 as a proxy for their debt capacity, experience stronger product market performance than their highly indebted rivals. Combined with the common view of cash as “negative” debt, this evidence raises the possibility that my interpretation is erroneous. Note, however, that the baseline regression partly tackles this possibility by specifically controlling for lagged leverage. In this line, the results of the previous section suggest that the competitive effect of cash is not completely absorbed by that of debt ratios. 16 To further lessen the potential concern of misspecified strategic effects of debt, this section uses a simple argument. In essence, to the extent that the influence of cash on market share growth arises because cash is simply the negative of debt, there should be no competitive effect of cash when cash reserves are distinct from negative debt. 17 To empirically identify when cash reserves differ from negative debt, I draw from Acharya, Almeida and Campello (2007) and group firms according to their a priori debt capacity. Specifically, cash is not the same as negative debt when reducing debt today does not guarantee the ability to access similar debt conditions in the future, i.e., when debt capacity is saturated. Following their methodology, a firm has a saturated debt capacity when it faces financial constraints and when its investment opportunities tend to arrive when cash flows are low, i.e., when hedging needs are high. To measure financial constraints I rank firmyears based on their asset size and assign to the constrained (unconstrained) group those firms that are in the bottom (top) quartile of the size distribution. I apply a similar procedure to rank firms based on their age and payout ratio. Also, I classify firms as financially constrained if they never had their public debt rated during the sample period but report positive debt. 18 Next, as a proxy for hedging needs, I compute for each firm-year the median three-year-ahead industry sales growth, and then compute the correlation between this measure and the firm’s cash flow. 19 I thus assign to the group of “high hedging needs” firms for which the empirical 17 correlation between cash flow and industry future sales growth is below -0.2, and to the group of “low hedging needs” those firms for which this correlation is above 0.2. 20 [Table III about here] Table III compares the differential effect of cash on market share growth across groups of firms sorted both on measures of financial constraints and of hedging needs. To preserve space, this table only reports the coefficients on relative-to-rivals cash holdings. Of most interest are the coefficients obtained for firms that are financially constrained and have high hedging needs, i.e., firms with no debt capacity. Strikingly, we observe large, positive and significant cash-performance sensitivities across the four measures of financial constraints. The estimates range from 0.032 to 0.080 and point to a sizeable effect of cash on market share growth for firms characterized by saturated debt capacity. Reassuringly, since we observe a significant competitive effect of cash when cash is manifestly distinct from negative debt, the results reported in Table III dispel the concerns that the competitive impact of cash captures only the flip side of that of debt ratios. 21 C. Reverse Causality? To further alleviate potential concerns about reverse causality, this section exploits a quasi-natural experiment. Specifically, I examine how firms use their existing cash reserves to respond to unexpected variations of industry-level import tariffs. 22 According to the vast literature on barriers to trade, the globalization of economic activities and trade openness bring major changes in the competitive configuration of industries; see Tybout (2003) for a survey. In particular, as reported by Bernard, Jensen and Schott (2006), the lessening of trade barriers triggers significant intensifications of competitive pressures from foreign rivals. As such, tariff reductions represent real-side shocks that exogenously shift the competitive landscape of industries and hence modify the relative attractiveness of having cash on hand. 18 To measure reductions of import tariffs at the (four-digits SIC) industry level, I use product-level U.S. import data compiled by Feenstra (1996) and Feenstra, Romalis and Schott (2002). This data spans the period 1972 to 2001 and includes 67 of the 105 manufacturing industries (2000-3999 SIC range) of my initial Compustat sample. For each industry-year, I compute the ad valorem tariff rate as the duties collected at the U.S. custom divided by the Free-On-Board customs value of imports. Next, to identify sizeable variations of barriers to trade, I characterize tariff reductions in terms of the deviations of the yearly changes in tariffs from their median level. By this token, a tariff “cut” occurs in a specific industry-year when a negative change in the tariff rate is 2, 2.5, or respectively 3 times larger than its median change. Moreover, to make sure that large reductions of tariff truly reflect non-transitory changes in the competitive environment, I exclude tariff cuts that are followed by equivalently large increases in tariffs over the two subsequent years. The online appendix provides further details on tariff data and presents several tests that support the validity of this experiment. Next, to identify the effect of cash on product market performance, I estimate the following difference-in-difference regression: , , , , 2 , , , 2 , , , , ( ) ) ' i k t i k t k t i k t k t i t i k i k t MarketShares zCash CUT (zCash CUT X a ? ? f ? ß e ? = + + + + × - - + + (2) where i indexes firms, k indexes the industry in which firms operate and t indexes the time. Similar to specification (1), the dependent variable is a proxy for market share growth, zCash characterizes a firm’s cash position compared to that of its rivals, and the set of control variables (Xi,k) includes size, past performance and past leverage. CUTk,t is a dummy variable that equals one if the industry k has experienced a tariff cut over the last two years (t and t-1). Again, ai is a vector of firm fixed effects and ?t a vector a time fixed effects. Since all the variables in specification (2) are industry-adjusted, I do not include an industry fixed-effect. 19 The coefficient of interest in equation (2) is on the interaction between zCashi,k,t-2 with CUTk,t (?). Crucially, since tariff reductions occur in different industries in different periods, equation (2) effectively takes as a control group all firms operating in industries that do not experience a reduction in tariff in year t, even if they have already experienced one or will experience one later on. As a result, the coefficient ? measures the difference of cashperformance sensitivity between firms that experience an unanticipated competitive shock and firms that do not. To wit, following a reduction of tariff, firms have to use their predetermined cash holdings to compete in an exogenously modified product market environment. As such, if cash reserves really provide a competitive advantage, one should observe that the effect of cash on market share growth is magnified in the aftermath of tariff reductions (? > 0 ). In contrast, if the positive association between cash and product market performance only arises because firms can perfectly foresee the competitive outcomes related to their cash levels, the effect of cash on competitive performance should not be altered by tariff reductions. [Table IV about here] Table IV displays the estimates of the difference-in-difference regressions for the three definitions of tariff reduction (2, 2.5 and respectively 3 times the median of tariff changes). 23 Notably, columns 1 to 3 reveal that the estimates of ? are positive and largely significant across the three measures of tariff reduction. The estimated coefficients range between 0.013 and 0.034, with t-statistics between 2.41 and 2.66. Interestingly, if we compare column 1 and column 3, we note that the estimate of ? is about three times larger following very large cuts in tariff rates. In line with the idea that cash holdings provide a substantial product market advantage, the competitive effect of cash turns out to be amplified when bigger shocks hit firms’ competitive environment. Remarkably, there is virtually no change in the coefficients on the control variables. 24 Note that if firms optimally anticipate tariff reductions, we would 20 expect an “effect” of tariff reductions prior to their implementation. In columns 4 to 6, I replace CUT by its lagged value and find no evidence in favor of an anticipated effect. Taken together, this quasi-natural experiment mitigates concerns about potential reverse-causality and confirms that cash holdings are crucial for a product market advantage. Below, I characterize in more depth the nature of this competitive effect. D. Characterization: Inter-Industry Differences To further dissect the nature of the above results, I investigate how the impact of cash holdings on market share growth differs across and within industries. In particular, I ask whether the competitive effect of cash depends on rivals’ financial conditions as well as on the quantity of interactions between firms within their industries. D.1 The Effect of Rival’s Finance I start the inter-industry investigation by testing whether the impact of cash holdings on market shares gains is more pronounced in industries in which rivals have a harder time obtaining external funds. To examine this prediction, I measure the average rival’s financial strength using the average firm size, age, payout ratio and public bond rating across industries. 25 For each year and for each proxy, I rank the sample industries according to their average value and assign firms from industries in the bottom and top quartile to “low” and “high” industries respectively. Next, for each proxy, I estimate equation (1) across subgroups and compare the estimates of the cash-performance sensitivities (? ) across low and high industries. 26 [Table V about here] Panel A of Table V reports which firms benefit more from large cash reserves to boost their market shares. For brevity, I only display the cash-performance estimates. Across all 21 specifications, the effect of cash is larger when industry rivals have weaker financial positions. For instance, row 1 presents regression results for subgroups based on the average rivals’ size. A comparison of coefficients across subgroups shows that the sensitivity of market share growth to cash holdings is significantly larger in industries in which rivals are relatively small. The coefficient on cash decreases by 66% when one moves from small-size to large-size industries. A Wald test rejects the equality of the cash coefficient across subgroup estimations (p-value 0.001). In rows 2, 3 and 4, I obtain similar patterns when I split industry-years on the basis of firms’ age, payout ratio and the presence of a public bond rating. Overall, these results support the view that cash holdings enable firms to gain shares in their product market, and that the magnitude of these gains depends on competitors’ financial status. In other words, the competitive impact of cash holdings is determined jointly by the firm’s and its rivals’ financial strength. Interestingly, while the sensitivity of sales performance to cash reserves is larger when rivals are financially weak, it is also significantly positive in industry-years in which competitors turn out to be financially strong. Hence, cash holdings appear to play a systematic role in determining firms’ performance in their product market. D.2 The Effect of Industry Characteristics Now, I take a different perspective and analyze whether the competitive effect of cash holdings depends on the quantity of strategic interactions between firms within an industry. I use four schemes as proxies for the intensity with which firms interact in their product market. The first is the degree of industry concentration. Following MacKay and Phillips (2005), I collect four-digit SIC industry concentration ratios (Herfindahl-Hirschman Index, HHI) from the Census of Manufacturers for the years 1982, 1987, 1992 and 1997. Following the Department of Justice’s guidelines, I denote as “concentrated” those industries for which the HHI index is greater than 1,800, and as “competitive” those industries for which that index is 22 less than 1,000. 27 As a second proxy, I follow Haushalter, Klasa and Maxwell (2007) and employ the covariation between competitors’ growth opportunities. To compute this measure, I draw from Parrino (1997) and regress each firm’s monthly stock returns on the monthly equally weighted market return and an equally weighted portfolio containing firms in the same industry (excluding the firm itself). The regression coefficient on the industry portfolio return captures the interdependence of growth options. Acknowledging that such interdependence may change over time, I estimate this proxy using a 36-month rollingregression approach. The third proxy for the interdependence between firms is whether a firm operates at the technological core of its industry or on the fringe. As in MacKay and Phillips (2005), I define the typical technology as the median capital-labor ratio for a given industryyear and compute the similarity of operation as the absolute value of the difference between a firm’s capital-labor ratio and the industry-year median ratio. 28 A smaller value of this proxy reflects a greater similarity of a firm’s operation with the operations of industry rivals and, therefore, a higher risk of losing market share. The last proxy I use is industry-level import penetration, computed from the NBER-CES Manufacturing Industry Database for the period 1972 to 2001. Specifically, it is defined at the four-digit SIC level as the total value of import divided by imports plus domestic production (shipments minus exports). Irrespective of the proxy, Panel B confirms that relative-to-rivals cash holdings have a differential impact depending on the amount of strategic interactions. In particular, row 1 indicates that the importance of cash reserves to expanding sales more than rivals is almost twice as large in competitive markets as in concentrated markets. In row 2, we observe that the competitive effect of cash is much larger when the firm is close to the technological core of its product market than when it lies on the fringe. Row 3 reports similar conclusions when the correlation of a firm’s stock returns with the stock returns of its industry is used to measure the interdependence between competitors. Row 4 shows an analogous pattern. 23 While Haushalter, Klasa and Maxwell (2007) show that the average firm increases its holdings of cash when facing competitive risk, the analysis above provides evidence on the effectiveness of such a hoarding strategy. In particular, evidence reveals that holding more cash than competitors effectively translates into better product market performance when the interdependence among rivals is important. E. Impact on Firm Value and Operating Performance The results above suggest that relative-to-rivals cash holdings are positively related to future product market performance. In this context, how do the competitive effects of cash affect firm value? To provide some evidence on the valuation consequences of the cash effects, I examine how measures of market value and operating performance are related to relative-to-rivals cash. As a measure of market value, I use the market-to-book ratio. As a measure of operating performance, I use return on assets (ROA), defined as EBITDA divided by assets. Table VI presents regression results of industry-adjusted market-to-book ratio and return on assets on lagged z-scored cash holdings. To control for other sources of value besides relative-to-rivals cash, I include firm size, cash flow, investment, leverage and a dummy that equals one if the firm pays a dividend and zero otherwise. Since I explain relative-to-rivals valuation, I subtract from the control variables their industry means in each year. Given that payout policy and asset tangibility may directly affect firm value, I have to restrict the instrument set and include only two lags of cash to compute predetermined cash holdings. Moreover, I include firm’s fixed and time effects and adjust the estimates’ standard errors for within-firm-period error clustering and heteroskedasticity. In column 1, firm value increases significantly in lagged zCash (0.061 with a t-statistic of 7.18). All else being equal, financially strong firms have higher valuations than their industry rivals. Hence, the market places a 24 premium on firms that have more internal resources than their competitors. Noticeably, the economic magnitude of this premium is significant. A one standard deviation increase in cash relative to rivals translates into a 6% increase in the mean market-to-book ratio over the average competitor. Consistent with previous literature, the coefficients on size and leverage are negative, while those on investment, cash Flow, and the dividend dummy are positive. Similarly, column 4 reveals that relative-to-rivals cash also enhances operating performance. The estimates indicate a significant effect on ROA (0.004 with a t-statistic of 3.75). [Table VI about here] A potential concern with the results in Table VI is that a company with a lot of growth opportunities may hold much larger cash balances than rivals. To further limit the potential effect of endogeneity inherent in the level of cash, I include four additional variables as proxies for firms’ growth options. Specifically, I introduce lagged relative-to-rivals sales growth, market-to-book ratio and return on assets. In columns 2 and 4, although the magnitude of zCash coefficients declines slightly, they remain significantly positive. 29 In particular, with the additional control for growth options, a one standard deviation increase in cash relative-to-rivals yields a 1.8% value premium over the average rival. Overall, these findings are consistent with the hypothesis that financial strength contributes positively to firm value and operating performance. Alternatively, the results of this section support the idea that the market prices the expected market share gains associated with cash holdings. IV. Conclusion The main message of this paper is that firms’ cash holdings strategically influence product market outcomes. In particular, I provide evidence that larger relative-to-rivals cash reserves lead to systematic future market-share gains at the expense of industry rivals. 25 Importantly, this “competitive” effect of cash turns out to be magnified when rivals face tighter financing constraints and when the amount of strategic interactions between competitors is substantial. Also, the analysis reveals that the competitive effect of cash contributes to increase firm value and operating performance. In a nutshell, this paper nails down an important link between firms’ cash holdings and their future product market performance. As such, the findings point to several interesting avenues for future research, three of which I outline here. First, I do not explore the precise channel through which the extra product market performance is achieved. Such an analysis is beyond the scope of this paper. In preliminary results, however, I find evidence that incumbents’ stock of cash appears to significantly curb the entry of potential competitors, and distort rivals’ investment and acquisition policies. Taken at face value, these results suggest that cash policy comprises a preemptive dimension that impacts rivals’ actions. Second, my findings emphasize that firms do not operate in isolation but incorporate rivals’ financial status and competitive position in their decision process. Ultimately, this point calls our attention to the fact that firm’s product market interactions need to be considered when investigating corporate financial decisions. Although this idea has emerged in recent theoretical developments, it is fair to say that, so far, empirical evidence on product market feedbacks remains scarce. Third, the results may have implications in the context of the global financial crisis of 2008 and the associated subsequent recession. In two recent studies, Campello, Graham and Harvey (2009) and Duchin, Ozbas and Sensoy (2008) document that the cash-poor firms had to cut R&D, employment and capital spending to cope with the tightening credit conditions and avoid a cash squeeze. In the light of my results, firms that managed to hoard cash before the 2008 crisis might well benefit from the financial turmoil to gain a leading position in their 26 product market, eventually turning themselves into industry champions when the downturn ends. 30 I look forward to additional research on these and related questions. 27 References Acharya, Viral V., Heitor Almeida, and Murillo Campello, 2007, Is cash negative debt? A hedging perspective on corporate financial policies, Journal of Financial Intermediation 16, 515-554. Bates, Thomas W., Kathleen M. Kahle, and René M. Stulz, 2009, Why do U.S. firms hold so much more cash than they used to? Journal of Finance (forthcoming) Benoit, Jean-Pierre, 1984, Financially constrained entry in a game with incomplete information, Rand Journal of Economics 15, 490-499. Berger, Philip, Eli Ofek, and Itzhak Swary, 1996, Investor valuation and abandonment option, Journal of Financial Economics 42, 257-287. Bernard, Andrew B., Bradford J. Jensen, and Peter K. Schott, 2006. Trade costs, firms and productivity, Journal of Monetary Economics 53, 917-937. Blanchard, Olivier J., Florencio Lopez-de-Silanes, and Andrei Shleifer, 1994, What do firms do with cash windfalls? Journal of Financial Economics 36, 337-360. Bolton, Patrick, and David S. Scharfstein, 1990, A theory of predation based on agency problems in financial contracting, American Economic Review 80, 93-106. Campello, Murillo, 2003, Capital structure and product markets interactions: Evidence from the business cycles, Journal of Financial Economics 68, 353-378. Campello, Murillo, 2006, Debt financing: Does it boost or hurt firm performance in product markets? Journal of Financial Economics 82, 135-172. Campello, Murillo, and Zsuzsanna Fluck, 2006, Market share, financial leverage and the macroeconomy: Theory and empirical evidence, Working Paper, University of Illinois. Campello, Murillo, John R. Graham, and Campbell R. Harvey, 2009, The real effects of financial constraints: Evidence from a financial crisis, Working Paper, Duke University. 28 Capkun, Vedran, and Lawrence A. Weiss, 2007, A trade credit explanation for the increased level of corporate cash holdings, Working Paper, HEC Paris. Chevalier, Judith A., 1995, Do LBO Supermarkets charge more? An empirical analysis of the effects of LBOs on supermarkets pricing, Journal of Finance 50, 1095-1112. Chevalier, Judith A., and David S. Scharfstein, 1996, Capital-market imperfections and countercyclical markups: Theory and evidence, American Economic Review 86, 703- 715. Clarke, Richard N., 1989, SICs as delineators of economic markets, Journal of Business 62, 17-31. Dimitrov, Valentin, and Sheri Tice, 2006, Corporate diversification and credit constraints: Real effects across the business cycle, Review of Financial Studies 19, 1465-1498. Dittmar, Amy, and Jan Mahrt-Smith, 2007, Corporate governance and the value of cash holdings, Journal of Financial Economics 83, 599-634. Duchin, Ran, Ozbas Oguzhan, and Berk A. Sensoy, 2008, Costly External Finance, Corporate Investment, and the Subprime Mortgage Crisis, Working Paper, Marshall School of Business. Faulkender, Michael, and Mitchell A. Petersen, 2005, Does the source of capital affect capital structure? Review of Financial Studies 19, 45-79. Feenstra, Robert C., 1996, U.S. imports, 1972-1994, data and concordances, NBER Working Paper 5515. Feenstra, Robert C., John Romalis, and Peter K. Schott (2002), U.S. imports, exports and tariff data, 1989 to 2001, NBER Working Paper 9387. Gamba, Andrea, and Alexander Triantis, 2008, The value of financial flexibility, Journal of Finance 63, 2263-2296. 29 Grenadier, Steven R., 2002, Option exercise games: An application to the equilibrium investment strategy of firms, Review of Financial Studies 15, 691-721. Hadlock, Charles J., and Joshua R. Pierce, 2008, Is the KZ index Useful? New Evidence on Measuring Financial Constraints, Review of Financial Studies (forthcoming). Harford, Jarrad, 1999, Corporate cash reserves and acquisitions, Journal of Finance 54, 1969- 1997. Harford, Jarrad, Sattar Mansi, and William Maxwell, 2008, Corporate governance and firm cash holdings, Journal of Financial Economics 87, 535-555. Haushalter, David, Sandy Klasa, and William F. Maxwell, 2007, The influence of product market dynamics on the firm's cash holdings and hedging behavior, Journal of Financial Economics 84, 797-825. Hege, Ulrich, and Christopher Hennessy, 2007, Entry for buyout and optimal financial (in)flexibility, Working Paper, HEC Paris. Kahle, Kathleen M., and Ralph A. Walking, 1996, The impact of industry classification on finance research, Journal of Financial and Quantitative Analysis 31, 309-335. Khanna, Naveen, and Sheri Tice, 2000, Strategic response of incumbents to new entry: The effect of ownership structure, capital structure, and focus, Review of Financial Studies 13, 749-779. Kim, Chang-Soo, David C. Mauer, and Ann E. Sherman, 1998, The determinants of corporate liquidity: Theory and evidence, Journal of Financial and Quantitative Analysis 33, 335-359. Kovenock, Dan, and Gordon Phillips, 1997, Capital structure and product market behavior, Review of Financial Studies 10, 767-803. 30 Lemmon, Michael, and Michael R. Roberts, 2007, The response of corporate financing and investment to changes in the supply of credit: A natural experiment, Journal of Financial and Quantitative Analysis (forthcoming). Lyandres, Evgeny, 2006, Capital structure and interaction among firms in output market: Theory and evidence, Journal of Business 79, 2381-2421. MacKay, Peter, and Gordon M. Phillips, 2005, How does industry affect firm financial structure? Review of Financial Studies 18, 1433-1466. Morellec, Erwan and Alexei Zhdanov, 2008, Financing and takeovers, Journal of Financial Economics 87, 556-581. Novy-Marx, Robert, 2007, An equilibrium model of investment under uncertainty, Review of Financial Studies 20, 1461-1502. Opler, Tim, Lee Pinkowitz, René Stulz, and Rohan Williamson, 1999, The determinants and implications of corporate cash holdings, Journal of Financial Economics 52, 3-46. Opler, Tim and Sheridan Titman, 1994, Financial distress and corporate performance, Journal of Finance 49, 1015-1040. Parrino, Robert, 1997, CEO turnover and outside succession: A cross-sectional analysis, Journal of Financial Economics 46, 165-197. Parsons, Chris, and Sheridan Titman, Capital structure and corporate strategy, 2008, in Espen B. Eckbo, eds., Handbook of Empirical Corporate Finance, Volume 2, North-Holland. Petersen, Mitchell, A., 2009, Estimating standard errors in finance panel data sets: Comparing approaches, Review of Financial Studies 22, 435-480. Phillips, Gordon M., 1995, Inceased debt and industry product markets: An empirical analysis, Journal of Financial Economics 37, 189-238. Sufi, Amir, 2009, The real effects of debt certification: Evidence from the introduction of bank loan rating, Review of Financial Studies 22, 1659-1691. 31 Tesler, Lester G., 1966, Cutthroat competition and the long purse, Journal of Law and Economics 9, 259-277. Tybout, James, 2003, Plant- and firm-level evidence on “new” trade theories, in E.Kwan Choi and James Harrigan, eds., Handbook of Interntional Economics, Oxford: BasilBlackwell. Zingales, Luigi, 1998, Survival of the fittest or the fattest? Exit and financing in the trucking industry, Journal of Finance 53, 905-938. 32 Notes 1 See for instance “The Corporate Savings Glut”, The Economist, July 7, 2005, “Behind Those Stockpiles of Corporate Cash,” by Mark Hulbert, Wall Street Journal, October 22, 2006, or “Companies Are Piling Up Cash”, by Diana B. Henriques, New York Times, March 4, 2008. 2 See for example “All You Need Is Cash”, The Economist, November 20, 2008, “Desperately Seeking A Cash Cure”, The Economist, November 20, 2008 or the recent evidence reported by Campello, Graham and Harvey (2009). 3 Note that while these results confirm that cash holdings have a causal impact on market share growth, they cannot be interpreted as evidence that cash holdings were chosen optimally ex-ante. 4 See Parsons and Titman (2008) for a survey on the relation between debt financing and corporate strategy. 5 See for instance Grenadier (2002), Novy-Marx (2007), and Morellec and Zhdanov (2008). 6 See for instance Blanchard, Lopez-de-Silanes and Shleifer (1994), Harford (1999), Dittmar and Mahrt-Smith (2007), Harford, Mansi and Maxwell (2006), Kim, Mauer and Sherman (1998) or Opler, Pinkowitz, Stulz and Williamson (1999). Bates, Kahle and Stulz (2009) provide a comprehensive survey of this literature. 7 In a recent paper, Hege and Henessy (2007) suggest another channel through which cash holdings might affect entry decisions. They argue that deep-pocketed incumbents may actually prompt entry by increasing creditors’ recovery in liquidation, thereby providing potential entrants with funds. 8 While some studies report that high indebtedness leads to poor performance in the product market (for example Chevalier (1995), Phillips (1995), Kovenock and Phillips (1997), Zingales (1998), Khanna and Tice (2000), and Campello (2003)), others find that debt 33 increases firms’ aggressiveness in the product market competition (for example, Campello (2006), Lyandres (2006)). 9 Note that this empirical specification is very similar to those of Opler and Titman (1994), Campello (2003, 2006), Campello and Fluck (2006), and Dimitrov and Tice (2006). 10 Note that all the results in this paper hold if I consider changes in a firm's percentage sales of its total industry sales to measure market share growth. 11 In the following analysis, I further discuss and present evidence that the effect of cash is not a by-product of that of debt levels. 12 In the analysis below, I use detailed identification tests to show that the instruments succeed in identifying specification (1) parameters. 13 Zingales (1998) and Campello (2003) also highlight the use of cross-sectional contrasts in studying the interplay between capital structure and product market performance. 14 Note that the results in this paper are qualitatively similar if I directly instrument z-scored cash instead of using this two-step procedure. 15 See for instance Opler and Titman (1994), Campello (2003), Campello and Fluck (2006), and Dimitrov and Tice (2006). 16 Note that I obtain similar results if I use “net debt”, computed as total debt minus cash divided by total assets, as an alternative proxy for leverage. Consistent with the view that cash provide a material advantage in the product, the effect of net debt on market share growth is markedly less negative than the effect of debt. 17 Note that the inclusion of past leverage should limit this possibility if we consider that firms with large debt ratios are facing limited debt capacity. 18 The reasons to rank firm based on their size, payout and bond rating is carefully explained in Acharya, Almeida and Campello (2007). I further use firms’ age to identify external capital 34 access following Hadlock and Pierce (2008) who suggest that age is a particularly useful predictor of financial constraints. 19 As explained in Acharya, Almeida and Campello (2007), the premise of this measure is that firms perceived investment opportunities are related to estimates of future sales growth in their industries and that those estimates, on average, coincide with the ex-post observed data. 20 Notably, the results still hold if I use ± 0.1 or ± 0.3 cutoffs. 21 Note that as expected, the cash-performance sensitivity is largely reduced when cash is not different from negative debt. As such, the results of table III are in line with Acharya, Almeida and Campello (2007). While they identify when cash is not the negative of debt, the results in table III provide evidence that the inequality between cash and negative debt has real consequences for product market outcomes. 22 Several studies that examine the effect of financing on product market outcomes consider shocks to either financing conditions or competitive environment to deal with endogeneity concerns. Considering shocks to financing arrangements, Phillips (1995) and Chevalier (1995) examine competitive responses to a sharp exogenous increase in leverage ratios. Alternatively, other studies consider shocks to competitive environments. Zingales (1998) use the deregulation in the trucking industries, Khanna and Tice (2000) use the entry of Wal-Mart in local markets, and Campello (2003) uses change in macroeconomic conditions. 23 Again, I adjust the estimates’ standard errors for within-firm-period error clustering and heteroskedasticity using the approach suggested by Petersen (2009). 24 Noteworthy, we observe that the coefficient on CUT is not significant. This is consistent with the recent findings in the literature on barriers to trade. After a reduction of tariff some firms will expand their domestic market shares while others will lose some. In line with Bernard, Jensen and Schott (2006), the results indicate that the overall effect is neutral. 35 However, the findings in this section indicate that the reallocation of market shares following a reduction of import tariff depends on firms’ cash position prior to the tariff reduction. 25 The results remains the same if I focus only on the period 1986-2006 where firms’ bond rating are observed every year. 26 Specification (1) is estimated via a seemingly unrelated regression (SUR) system combining the two subgroups. The Wald test of the differences between the two subgroups is obtained using the standard errors provided by the SUR estimation. 27 In assigning firms to either concentrated or competitive markets I use the 1982 census data for Compustat firm-fiscal years in the 1980-1984 period, and the 1997 census data for firms in the 1995-1999 period. Hence the use of this variable considerably restricts the size of the sample. 28 Note that the industry-year median is weighted by each firm’s share of industry sales and excludes the firm itself. Furthermore, to make this variable comparable across industries, I divide it by the industry-year range of the capital-labor ratio. 29 Although I make cash instrumental by its own two lags and control for growth opportunities (and find it to be positively and significantly related to market-to-book), it is also possible that I do not fully capture growth opportunities. Thus, we could expect to observe a positive correlation between firm value and relative-to-rivals cash balances, but it does not necessarily follow that the financial strength causes the higher firm value. 30 Some anecdotical evidence seems to corroborate this tendency. For instance, John Chambers, the CEO of Cisco Systems recently declared in The Harvard Business Review (November 2008) that his firm tended to make more aggressive investment during bad times than good ones: “When rivals pulled back from Asia during the region’s financial crisis, Cisco deliberately increased its presence, gaining a leading position it has never relinquished”. 36 Table I The Impact of Cash on Market Share Growth (Baseline Estimation) This table presents results of panel regressions examining the effect of relative-to-rivals cash holdings on market share growth (specification (1)). Columns (1) and (2) reports the instrumental variables estimates where cash holdings are instrumented by their lagged values and asset tangibility. The dependent variable is ?MarketShares, the annual market-share growth given by industry-adjusted sales growth at time t [(Salest - Salest-1)/Salest-1]. Cash is the ratio of cash and marketable securities divided by total assets. Size is the natural logarithm of assets. Leverage is long-term debt over assets. All variables are adjusted for their four-digit SIC industry-year means, with instrumented Cash is further standardized (i.e., z-scored) within each industry-year. Instrumental variables estimations display diagnostic statistics for instrument over-identification restrictions (p-values of J-statistics reported) and exogeneity conditions (p-values for Durbin-Hausman-Wu reported). All variables are defined in Appendix A. The sample period is 1973 through 2006. Column (3) report the coefficients of the first-step estimation of cash on lagged cash and tangibility defined as in Berger, Ofek and Swary (1996). The estimations correct the error structure for heteroskedasticity and within-firm error clustering. I report t-statistics in brackets. ** and * denote statistical significance at the 1% and 5% level, respectively. Instrumental Variables First-stage Estimation (1) (2) (3) zCasht-2 0.029** Tangibilityt -0.468** [12.42] [12.55] zCasht-1 0.030** Casht-1 0.550** [12.59] [21.22] Sizet-1 0.042** 0.041** Casht-2 0.077** [16.21] [15.70] [4.65] Leveraget-1 0.006** 0.007** [2.78] [3.06] Leveraget-2 -0.004** -0.003** [3.52] [3.16] ?MarketSharest-1 0.011* 0.002 [2.05] [1.41] ?MarketSharest-2 -0.073** -0.069** [14.16] [13.50] Firm fixed-effects Yes Yes Yes Time fixed-effects Yes Yes Yes # Obs 36794 36808 43492 R 2 0.24 0.24 0.78 J-statistic (p-value) 0.26 0.22 Durbin-Hausman-Wu 0.01 0.03 37 Table II The Impact of Cash on Market Share Growth (Robustness) This table presents additional results of IV regressions examining the effect of relative-to-rivals cash holdings on market share growth (specification (1)). The depende i nt variable is ?MarketShares, the annual market-share growth given by industry-adjusted sales growth at time t [(Salest - Salest-1)/Salest-1]. Cash is the ratio of cash and marketable securities divided by total assets. Size is the natural logarithm of assets. Leverage is long-term debt over assets. All variables are adjusted for their four-digit SIC industry-year means, with instrumented Cash further standardized (i.e., z-scored) within each industry-year. Acquisitions is the amount spent on acquisitions over assets. SalesAcquisitions is the sales contribution of acquisitions. Market-to-Book is the market value of equity plus the book value of assets minus the book value of equity minus deferred taxes, scaled by total assets. All variables are defined in Appendix A. The sample period is 1973 through 2006. Instrumental variables estimations display diagnostic statistics for instrument over-identification restrictions (p-values of J-statistics reported) and exogeneity conditions (p-values for Durbin-Hausman-Wu reported). The estimations correct the error structure for heteroskedasticity and within-firm error clustering. I report t-statistics in brackets. ** and * denote statistical significance at the 1% and 5% level, respectively. Additional control variables #firms>30 Adj.Cash Skewness (1) (2) (3) (4) (5) (6) zCasht-2 0.027** 0.028** 0.028** 0.038** 0.273** 0.029** [11.45] [10.09] [12.01] [9.55] [14.72] [8.28] Sizet-1 0.039** 0.042** 0.042** 0.048** 0.043** 0.044** [14.65] [13.27] [16.00] [11.18] [16.46] [11.13] Leveraget-1 -0.001 -0.002 0.011** 0.009* 0.007** 0.006 [1.32] [0.96] [4.66] [2.36] [2.80] [1.69] Leveraget-2 -0.005** -0.005* -0.004** -0.005** -0.003** -0.009** [2.74] [2.15] [3.67] [3.46] [3.30] [3.49] ?MarketSharest-1 0.019** 0.034** 0.037** 0.034** 0.013** 0.035** [3.38] [5.30] [6.85] [4.35] [2.53] [4.70] ?MarketSharest-2 -0.061** -0.064** -0.081** -0.067** -0.075** -0.070** [11.49] [10.43] [15.40] [8.96] [14.61] [9.78] Acquisitions t-1 0.531** [11.66] Acquisitions t-2 -0.227** [4.91] SalesAcquisitions t-1 0.311** [10.45] SalesAcquisitions t-2 -0.059* [1.99] Market-to-Bookt-1 0.047** [8.57] Market-to-Bookt-2 0.005** [3.45] Firm fixed-effects Yes Yes Yes Yes Yes Yes Time fixed-effects Yes Yes Yes Yes Yes Yes # Obs 35452 27872 35272 17257 36793 18894 R 2 0.25 0.26 0.26 0.26 0.24 0.29 J-statistic (p-value) 0.19 0.32 0.27 0.33 0.21 0.18 Durbin-Hausman-Wu 0.01 0 0.01 0.01 0.02 0.02 38 Table III The Impact of Cash on Market Share Growth: Cash or Negative Debt (Debt Capacity) This table reports the estimates of zCash (? ) from a series of IV estimations of specification (1) across different groups. Firms are classified on the basis of their debt capacity identified using proxies for financing constraints and hedging needs as in Acharya, Almeida and Campello (2007). The dependent variable is ?MarketShares, the annual market-share growth given by industry-adjusted sales growth at time t [(Salest - Salest-1)/Salest-1]. All specifications include the same set of control variables as in Table I. All variables are defined in Appendix A. The sample period is 1973 through 2006. The estimations correct the error structure for heteroskedasticity and within-firm error clustering. The number of firm-year observations in each group is in italic. t-statistics in brackets. The standard errors for the differences between High and Low (and Constrained and Unconstrained) are computed with a SUR system that estimates industry group jointly. The last column (row) presents the pvalue associated with the F-tests that compare coefficients between High and Low (Constrained and Unconstrained) sub-groups. ** and * denote statistical significance at the 1% and 5% level, respectively. Financial constraints High Low Diff. High-Low criteria hedging needs hedging needs hedging needs Firm size Constrained firms (Cons.) 0.056** 0.044** 0.00** [5.31] [3.56] 2636 2151 Unconstrained firms (Uncons.) 0.014* 0.014 0.15 [2.46] [1.93] 3890 2481 Diff. Cons. –Unconst. 0.00** 0.01** Firm Age Constrained firms (Cons.) 0.080** 0.035** 0.00** [4.99] [2.84] 2393 3315 Unconstrained firms (Uncons.) 0.026** 0.010* 0.02** [3.96] [1.99] 3374 1910 Diff. Cons. –Unconst. 0.00** 0.00** Payout policy Constrained firms (Cons.) 0.038** 0.029** 0.02** [6.18] [3.60] 7279 4965 Unconstrained firms (Uncons.) 0.018** 0.011 0.01** [3.58] [1.82] 3825 2236 Diff. Cons. –Unconst. 0.01** 0.02** Bond ratings Constrained firms (Cons.) 0.032** 0.031** 0.17 [7.81] [5.08] 11,618 7329 Unconstrained firms (Uncons.) 0.018* 0.002 0.00** [2.43] [1.32] 2086 1797 Diff. Cons. –Unconst. 0.03** 0.00** 39 Table IV The Impact of Cash on Market Share Growth: Difference-in-Difference Estimations This table presents results of OLS difference-in-difference regressions examining the effect of relative-to-rivals cash holdings on market share growth (specification (1)) following large reductions of import tariffs. The dependent variable is ?MarketShares, the annual market-share growth given by industry-adjusted sales growth at time t [(Salest - Salest-1)/Salest-1]. Cash is the ratio of cash and marketable securities divided by total assets. Size is the natural logarithm of assets. Leverage is long-term debt over assets. All variables are adjusted for their four-digit SIC industry-year means, with Cash further standardized (i.e., z-scored) within each industry-year. Tariff reductions (CUT) are defined using three different cut-offs. Specifically a tariff cut occurs when industryyear change in tariff rate (?T) are negative and 2 (columns 1 and 4), 2.5 (columns 2 and 5) and respectively 3 (columns 3 and 6) times larger than its median value. For these three definitions, CUT=1 if an industry had experience a tariff cut in the last two years (t and t-1). The sample period is 1973 through 2001. The estimations correct the error structure for heteroskedasticity and within-firm error clustering. I report t-statistics in brackets. ** and * denote statistical significance at the 1% and 5% level, respectively. CUT CUT CUT CUT CUT CUT 2×med(?T) 2.5×med(?T) 3×med(?T) 2×med(?T) 2.5×med(?T) 3×med(?T) (1) (2) (3) (4) (5) (6) zCasht-2 0.033** 0.034** 0.035** 0.033** 0.033** 0.036** [9.91] [10.41] [11.07] [8.72] [9.19] [10.41] zCasht-2 × CUTt 0.013* 0.018** 0.034** [2.41] [2.63] [2.66] CUTt -0.001 -0.009 -0.017 [1.20] [1.39] [1.78] Sizet-1 0.054** 0.054** 0.055** 0.054** 0.054*3 0.055** [14.16] [14.19] [12.71] [13.27] [13.30] [13.45] Leveraget-1 0.009* 0.008** 0.009** 0.011** 0.01** 0.011** [2.06] [2.54] [2.60] [2.94] [2.93] [2.99] Leveraget-2 -0.008** -0.008** -0.008** -0.007** -0.007** -0.007** [5.11] [5.07] [5.08] [4.65] [4.61] [4.60] Sales Growtht-1 0.043** 0.049** 0.043** 0.042** 0.042** 0.042** [5.69] [5.72] [5.69] [5.33] [5.36] [5.35] Sales Growtht-2 -0.081** -0.081** -0.080** -0.082** -0.081** -0.080** [10.88] [10.90] [10.87] [10.49] [10.51] [10.48] zCasht-2 × CUTt-1 0.001 0.008 0.013 [0.61] [1.16] [1.06] CUTt-1 0.001 0.006 -0.014 [0.26] [0.94] [1.13] Firm fixed-effects Yes Yes Yes Yes Yes Yes Time fixed-effects Yes Yes Yes Yes Yes Yes # Obs 20514 20514 20514 17648 17648 17648 R 2 0.28 0.29 0.29 0.3 0.3 0.31 # Tariff cut (CUT=1) 317 180 43 317 180 43 40 Table V. Cross-Industries Impact of Cash on Market Share Growth This table reports the estimates for zCash (? ) from a series of IV estimations of specification (1). The dependent variable is ?MarketShares, the annual market-share growth given by industry-adjusted sales growth at time t [(Salest - Salest-1)/Salest-1]. I classify industries on the basis of different proxies for the average rival’s financial status (Panel A) and a firm’s interaction with its rivals (panel B). All specifications include the same set of control variables as in Table I. In classifying industries according to Size, Age, Payout policy, Bond rating, Correlation with industry and Import Penetration, I compute the industry-year average of those variables. Then, Low industries are those ranked in the bottom quartile of the respective distribution and High industries are those ranked in the top quartile of the same distribution. Concentration data (Herfindhal index) are from the Census of Manufacturers. Low concentration corresponds to a Herfindhal index below 1,000 (“competitive industry”) while High concentration corresponds to a Herfindhal index above 1,800 (“concentrated industry”). Concerning the classification based on Similarity of Operations, in each industry-year, I assign firms in the Low (High) group, those for which Similarity of Operations is below their industry-year median value. All regressions contain firm and time fixed effects. The standard errors for the differences between High and Low are computed with a SUR system that estimates industry group jointly. The last column presents the p-value associated with the F-tests that compare coefficients between High and Low sub-groups. The number of firm-year observations in each group is in italic and t-statistics are in brackets. ** and * denote statistical significance at the 1% and 5% level, respectively. Panel A Low High Diff. Low-High Rivals Size 0.047** 0.016** 0.01** [7.85] [4.35] #8880 #9988 Rivals Age 0.039** 0.013** 0.00** [5.80] [3.85] #8659 #10158 Rivals Payout Policy 0.033** 0.022** 0.03** [5.77] [5.43] #9358 #9807 Rivals Bond Rating 0.040** 0.015** 0.00** [7.16] [4.03] #9382 #9882 Panel B Low High Diff. Low-High Industry Concentration (HHI) 0.045** 0.021* 0.00** [8.78] [1.98] #9006 #1672 Similarity of Operations 0.018** 0.028** 0.02** [3.64] [5.36] #8793 #8894 Correlation with Industry 0.027** 0.037** 0.03** [6.40] [5.68] #8815 #7188 Import Penetration 0.034** 0.041** 0.06* [4.51] [4.70] #4128 #3653 41 Table VI The Impact of Cash on Firm Value and Operating Performance This table presents results of panel regressions examining the effect of relative-to-rivals cash holdings on firm value and operating performance. In columns (1) and (2), the dependent variable is the (industry-adjusted) Market-to-Book ratio at time t. In columns (3) and (4), the dependent variable is the (industry-adjusted) return on assets (ROA) at time t. Cash is the ratio of cash and marketable securities divided by total assets. Size is the natural logarithm of assets. Investment is given by (PPEt - PPEt-1)/PPEt-1. Leverage is the ratio of long-term debt over assets. Cash Flow is net operating income divided by assets. Dividend is a dummy that equals one if the firm pays dividend and zero otherwise. Sales Growth at time t are given by [(Salest - Salest-1)/Salest-1]. All variables are adjusted for their four-digit SIC industry-year means, with Cash further standardized (i.e., z-scored) within each industry-year, zCash. The sample period is 1973 through 2006. IV estimations display diagnostic statistics for instrument overindentification restrictions (p-values for J-statistics reported). All regressions contain firm and time fixed effects. The estimations correct the error structure for heteroskedasticity and withinfirm error clustering. I report t-statistics in brackets. ** and * denote statistical significance at the 1% and 5% level, respectively. Market-to-Book ROA (1) (2) (3) (4) zCasht-1 0.061** 0.018* 0.004** 0.001* [7.18] [2.40] [3.74] [2.01] Sizet-1 -0.276** -0.199** -0.018** -0.021** [27.42] [21.95] [12.08] [17.91] Investment t-1 0.977** 0.106 -0.003 -0.024 [6.27] [0.76] [0.17] [1.13] Leveraget-1 -0.173** 0.037 -0.022* -0.008 [2.79] [0.68] [2.38] [0.97] Cash Flowt-1 0.112** 0.135** 0.232** 0.290** [2.3] [3.22] [33.47] [12.49] Dividendt-1 0.04 0.013 0.015** 0.012** [1.87] [0.70] [4.79] [3.85] Sales Growtht-1 0.048 [2.53] Market-to-Bookt-1 0.379** [30.01] ROAt-1 0.277** [12.23] # Obs 33813 32983 34613 34517 R 2 0.5 0.58 0.61 0.62 J-statistic (p-value) 0.21 0.16 0.23 0.22 1 Internet Appendix to “Financial Strength and Product Market Behavior: The Real Effect of Corporate Cash Holdings” * This appendix outlines the construction of the variables used in the analysis and shows some descriptive statistics. In addition, it provides further information on the quasi-natural experiment used in section III.C and presents a detailed analysis supporting the validity of tariff reductions to identify the direction of the causality between cash and product market performance. Finally, it presents several additional tests that show that cash holdings and their effect on market share growth is persistent through time. A. Data Definition and Descriptive Statistics Table IA.I details the construction and source of the variables used in the analysis. Table IA.II presents the summary statistics of the main variables of interest. Overall, they are comparable to those found in related studies, such as Campello (2006), Acharya, Almeida and Campello (2007) or Bates, Kahle and Stulz (2009). B. Quasi-Natural Experiment: Variations of Import Tariffs B.1. Discussion of the Tariff Data The product-level U.S. import data used in the analysis are compiled by Feenstra (1996) and Feenstra, Romalis and Schott (2002). This dataset spans the period 1972 to 2001. The matching with my sample of Compustat firms leaves 67 industries with available information on imports, collected duties, exports and domestic production. For each industry- * Citation format: Laurent Frésard, 2009, Internet Appendix to “Financial Strength and Product Market Behavior: The Real Effects of Corporate Cash Holdings”, Journal of Finance xx(x), xxxx-xxxx, http://afajof.org/IA/2009.asp. Please note: Wiley-Blackwell is not responsible for the content of functionality of any supporting information supplied by the author. Any queries (other than missing material) should be directed to the author of the article. 2 year, I define the ad valorem tariff rate as the duties collected at the U.S. custom divided by the Free-On-Board customs value of imports. To measure reductions of import tariff, I compute the annual change of ad-valorem tariff rates. Since the coding of imports changed in 1989, I do not use the yearly changes between 1988 and 1989 and put them equal to zero. Then, to identify sizeable changes in tariff rates, I characterize tariff reductions in terms of the deviations of the yearly tariff changes from their median level. Precisely, I define that a tariff “cut” occurs in a specific industry-year when a negative change in tariff rate is 2, 2.5, or respectively 3 times larger than its median value. Moreover, to make sure that large reductions of tariff truly reflect nontransitory changes in trade policy, I exclude tariff cuts that are followed by equivalently large increases in tariffs over the two subsequent years. Figure IA.1 plots the annual tariff rate around the identified tariff cuts for the third (tightest) definition of tariff reduction. We observe a substantial reduction in tariff rate, indicating that three definitions of large tariff reductions consistently pin down sizeable changes in trade policy. Note that I obtain similar patterns for the two other definitions of tariff cut. Figure IA.2 displays the repartition of tariff reductions over time. Noticeably, tariff reductions are not clustered in specific time period but occur in different industries at different times. The key advantage of using tariff data is that they provide enough time-series and cross-industry variations suitable to identify the competitive effect of cash. Moreover, they are derived directly from product-level trade data collected at the border. Nevertheless, one caveat should be noted. The changes in tariff that I use are effective changes for a given industry. Hence, changes in the composition of products or importers within industries can induce variations in effective tariffs even if the statutory tariffs remain constant. Since I am interested in changes in competitive pressures induced by trade openness, this should not have any material effect on the analysis. 3 B.2. Validity of the Quasi-Natural Experiment To be considered as a valid quasi-natural experiment, reductions of import tariff have to fulfill three requirements. First, they should bring real-side changes in the competitive nature of the product market. Second, they should be exogenous to industry performance and financing. Third, they should be partly unanticipated. Given that the literature on international trade is relatively silent on the potential links between trade policy and industry-level financing, I use a combination of descriptive figures and reduced-form statistical evidence to support the validity of this quasi-natural experiment. The crux for using tariff reductions rests in the idea that lower tariffs make it less costly for foreign rivals to compete on domestic markets, thereby putting competitive pressure on U.S. firms. To verify this conjecture, I first examine whether reductions of import tariffs are associated with changes in the level of import penetration. Following Bertrand (2004) and Irvine and Pontiff (2009), I define Import Penetration as the total value of import divided by imports plus domestic production. This variable can be interpreted as the market shares of foreign competitors. Figure IA.1 (right axe) displays the evolution of the average import penetration in the years surrounding tariff reductions. Strikingly, we observe a substantial increase of import penetration after tariffs have been cut. The economic magnitude is large. We note that import penetration increases from 12% the year before the cut to above 15% one year after the cut. This event-time patterns support the intuition that reductions of import tariffs effectively breed competitive pressure on domestic firms and are in line with evidence from the trade literature (e.g. Bernard, Jensen and Schott (2006), Lee and Swagel (1997) or Trefler (1993)). Second, to be a valid experiment, the source of variation that shifts the competitive environment has to be exogenous with respect to firms’ cash policy and performance. 4 Arguably, a skeptic could contend that tariff levels are driven by political factors associated with financial outcomes. For instance, trade protection may be granted to industries with particular financing and/or performance profiles. Table IA.III reports various validity checks. First, Panel A compares the averages of four financial variables between firms in industries that will experience a tariff cut one year ahead and firms in industries that never experience a tariff reduction over the sample period. Notably, Panel A suggests that industries experiencing tariff reductions are generally comparable with industries that are not affected by tariff changes. Indeed, we do not observe any systematic difference in their average levels of cash, debt or in their average performance. To provide further support for the exogeneity of tariff reductions, I estimate various specifications linking tariff reductions to industries’ (median) past financing conditions, performance as well as macroeconomic factors. Results are presented in Panel B. Columns 1 to 3 report logistic estimations where the dependent variables are the three definitions of tariff reduction (tariff cut # 1 to tariff cut # 3). Although coefficients have generally the expected sign, past financing choices and performance do not seem to correlate with tariff reductions. Column 4 further reports results from an OLS regression where the dependent variable is the annual change in industry-level tariff rates. Again, we note no systematic ability of industry variables to predict trade policy. Across all specifications, only the annual changes in GDP and the number of firms within the industry seem to predict future tariff changes. Note that I obtain equivalent results if I use industry averages instead of medians. By and large, these results dispel the potential concerns about the endogeneity of tariff reductions to the major variables used in the analysis. Finally, the competitive changes triggered by tariff reductions should allow for unanticipated effects. More precisely, reductions of import tariff should make it difficult for firms to fully endogeneize its consequences in their ex-ante financial choices. Figures IA.3 plots the evolution of the average cash levels, debt ratios and market value in the years 5 surrounding tariff reductions for the third (tightest) definition of tariff reduction. Interestingly, we observe no systematic change in cash and debt levels prior to the tariff cuts. However, firms seem to alter significantly their financial choices in the aftermath of tariff changes. These event-time patterns corroborate the results of table IV (columns 4 to 6) that reveal no anticipation behavior prior to tariff reductions. Interestingly, we also note a sharp decline in the average firm value during the year that follows tariff reductions. This value shortfall is suggestive that, over the sample period, large tariff reductions were not fully anticipated by market participants. In light of these results, there is little reason to believe that cash holdings were chosen optimally beforehand to deal with the consequences of the increased product market competition. C. Persistence of Cash Holdings and their Competitive Component One important question that arises from the analysis is whether cash holdings and that their competitive effect on market share growth are persistent through time. To shed light on this issue, I conduct several additional tests. Specifically, to examine the persistence of firms’ financial strength, Table IA.IV presents the empirical transition probabilities for relative-torivals (z-scored) cash as well as for cash-to-asset ratios. For both variables, I first rank firms into annual quartiles over the sample period. In then use the quartile ranks to estimate transition probabilities as the empirical probability of a firm moving from one quartile during year t to another quartile in year t+1, t+2 or t+3. For both variables, this table suggests a high level of persistency. Consider for instance firms in the fourth quartile of z-scored cash, that is, entries in the last rows of the Panel A. Those are cash-rich firms. The empirical probabilities that a cash-rich firm in year t will remain a cash-rich firm (fourth quartile) in years t+1, t+2 or t+3 are respectively 0.645, 0.563 and 0.421. Notably, we observe similar patterns for cashpoor firms. 6 Next, Table IA.V reports the results of various specifications that gauge whether the effect of cash on market share growth persists over time. In the first three columns, I introduce additional lags of (instrumented) z-scored cash in the baseline specification (1) of table I to assess the inter-temporal impact of cash on market share growth. The coefficients on the lagged z-scored cash (t-3, t-4 and t-5) measure the lagged performance-cash sensitivities, which effectively indicate how an additional dollar of cash (relative-to-rivals) today impacts, ceteris paribus, market share growth two, three and four years later. Interestingly, even though the competitive effect of cash tends to decrease over time, we observe a positive and significant coefficient for zCasht-5. Alternatively, in columns 4 to 6, I estimate the effect of cash on multiple-years market share growth. In these regressions, the dependent variables are the growth in market shares obtained over the period t-1 to t+1, t-1 to t+2, and respectively t- 1 to t+3. For the three horizons, we notice positive and significant coefficients on cash, confirming that having more cash than rivals enable firms to expand market share over longer periods. On the whole, these additional results reveal an important degree of persistence of both cash holdings and their impact on product market performance, and hence provide additional support for my interpretation. 7 References Bertrand, Marianne, 2004, From the invisible handshake to the invisible hand? How import competition changes the employment relationship, Journal of Labor Economics 22, 723-765. Lee, Jong-Wha, and Phillip Swagel, 1997, Trade barriers and trade flows across countries and industries, The Review of Economics and Statistics 79, 372-382. Irvine, Paul J., and Jeffrey Pontiff, 2009, Idiosyncratic return volatility, cash flows, and product market competition, Review of Financial Studies 22, 1149-1177. Trefler, Daniel, 1993, Trade liberalization and the theory of endogenous protection: An econometric study of U.S. import policy. Journal of Political Economy 101, 138-160. 8 Figure IA.1 Evolution of Tariff Rate and Import Penetration around Tariff Reductions This figure displays the average tariff rates and import penetration surrounding years of tariff reductions. Tariff reductions are defined when industry-year change in tariff rate (?T) are negative and 3 times larger than its median value. The dashed line represent tariff rate while the solid line represents import penetration. The sample period is 1973 through 2001. 0% 1% 2% 3% 4% 5% 6% 7% 8% -2 -1 0 1 2 years before and after tariff reductions Rate of import tariff 10% 11% 12% 13% 14% 15% 16% 17% 18% Import penetration Figure IA.2 Repartition of Tariff Reductions over Time This figure displays the repartition of tariff reductions over time. Tariff reductions are defined using three different cut-offs. Specifically a tariff cut occurs when industry-year change in tariff rate (?T) are negative and 2 (blue bars), 2.5 (red bar) and respectively 3 (green bar) times larger than its median value. The sample period is 1973 through 2001. 0 5 10 15 20 25 30 35 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 year # of tariff cuts Tariff cut #1 Tariff cut #2 Tariff cut #39 Figure IA.3 Evolution of Firm Valuation around Tariff Reductions This figure displays the average cash-to-asset ratio (dashed line), debt-to-asset ratio (solid line) and book-tomarket ratio (dotted line) surrounding years of tariff reductions. Tariff reductions are defined when industry-year change in tariff rate (?T) are negative and 3 times larger than its median value. The sample period is 1973 through 2001. 0.12 0.14 0.16 0.18 0.2 0.22 0.24 0.26 0.28 0.3 -2 -1 0 1 2 years before and after tariff reductions cash and debt ratios 1 1.5 2 2.5 3 3.5 market-to-book ratio10 Table IA.I Definition of the main variables Total Assets Total assets (Compustat item 6) (in million USD) Sales Sales (item 12) Size Logarithm of total assets (item 6) ?MarketShares Growth in sales computed as Salest minus Salest-1 divided by Salest-1 minus industryyear average Cash Cash and short-term investment (item 1) scaled by total assets Investment Growth in property, plant, and equipment (PPE) (item 7) computed as PPEt minus PPEt-1 divided by PPEt-1 Age Number of years preceding the observation year that the firm has a non-missing stock price in the CRSP-Compustat merged database. Leverage Long-term debt (item 9) scaled by total assets Bond Rating Dummy variables that equals one if a firm has a public bond rated during the sample period. Tangibility 0.715*Receivables (item 2) plus 0.547*Inventories (item 3) plus 0.535* Fixed capital (item 8) [see Berger et al. (1996)] Cash flow Sum of net income before extraordinary items (item 18) and depreciation and amortization (item 14) scaled by total assets Market-to-Book Market value of equity (item 24 multiplied by item 25) plus book value of assets minus book value of equity minus deferred taxes (item 6 – item 60 – item 74), scaled by total assets Capital-Labor ratio Gross property, plant, and equipment (item 7) divided by the number of employees (item 29) multiplied by 1,000 ROA Ratio of operating income before depreciation and amortization expenses (item 13) to total assets Acquisitions Amount spent in acquisitions (cash) (item 129) scaled by total assets SalesAcquisitions Sales contribution of acquisition (item 249) scaled by total assets Herfindhal index (HHI) Four-digits SIC industry concentration ratios gathered in the Census of Manufacturers (1982, 1987, 1992, and 1997 editions) Import Penetration Total value of annual imports divided by the sum of total import and domestic production. Data are defined in Feenstra (1996) ?GDP Annual change in real GDP from the Federal Reserve Bank of St.Louis ?IndPro Annual change in industrial production from the Federal Reserve Bank of St.Louis11 Table IA.II Summary Statistics This appendix reports summary statistics for the main variables used in the empirical analysis. The final sample has statistical properties that are very similar to those reported in comparable studies that use Compustat (see, for example, Campello (2006)). The sample period is 1973 through 2006. Included firms are from industries selected at the four-digit SIC level following Clarke (1989). #Obs Mean Median Std.Dev Pct. 25 Pct. 75 Cash 54346 0.186 0.092 0.218 0.030 0.265 Sales Growth 47424 0.136 0.098 0.331 -0.026 0.245 Assets ($Million) 54347 687 59 2289 16 280 Investment 53845 0.055 0.042 0.049 0.021 0.075 Leverage 54809 0.139 0.100 0.146 0.007 0.226 12 Table IA.III Differences between Industries that Experience or Not a Reduction of Tariff This table presents univariate and multivariate comparisons between firms in industries that experience a reduction of import tariff are firms in industries that do not. Tariff reductions are defined using three different cut-offs. Specifically a tariff cut occurs when industry-year change in tariff rate (?T) are negative and 2 (tariff cut#1), 2.5 (tariff cut#2) and respectively 3 (tariff cut#3) times larger than its median value. The sample period is 1973 through 2001. Panel A reports the means and the number of firm-year observations in industries that will experience a tariff reduction one-year ahead and those that do not. Panel B reports results from logistic and OLS regressions that explains variations in trade policy as a function of lagged industry (median) variables and lagged macroeconomic variables. In the logistic estimations, the dependent variable is a dummy that equals if the industry experiences a tariff reduction and zero otherwise. In the OLS regression the dependent is the annual variation of import tariff. I report t-statistics in brackets. ** and * denote statistical significance at the 1% and 5% level, respectively. Panel A : Desciptive statistics Variables tariff cut#1 tariff cut#2 tariff cut#3 Non-Affected Cash 0.183 0.190 0.199 0.196 5697 3321 751 21608 Leverage 0.193 0.191 0.186 0.198 5763 3352 758 21884 Market-to-Book 2.056 2.056 2.011 2.105 5610 3259 724 21025 ROA 0.050 0.048 0.037* 0.053 5709 3318 749 21618 Panel B : Multivariate analysis Logistic regressions OLS Variables tariff cut#1 tariff cut#2 tariff cut#3 ?Tariff Ind.Casht-1 -3.978 -3.059 -3.432 0.216 [1.62] [1.13] [0.67] [0.38] Ind. Leverage t-1 0.418 0.936 0.664 0.238 [0.25] [0.45] [0.45] [0.59] Ind. Market-to-Book t-1 0.571 0.27 0.356 -0.014 [1.10] [0.82] [1.06] [0.19] Ind.ROA t-1 0.217 0.578 0.962 -0.508 [1.09] [1.15] [1.14] [0.69] Ind. Size t-1 -0.045 -0.023 -0.398 0.025 [1.08] [1.12] [1.11] [0.65] Ind. #firms t-1 0.016* 0.023* 0.019** -0.002 [1.65] [1.92] [1.98] [0.69] ?GDP t-1 5.216 3.385 5.528** -2.119 * [1.20] [1.07] [2.87] [1.69] ?IndPro t-1 -1.643 -0.974 -3.525 0.011 [1.71] [1.33] [1.49] [1.02] Log Likelihood [R 2 ] -420.04 -286.78 -86.82 [0.06] #Obs 1001 931 600 1072 13 Table IA.IV Empirical Cash Holdings Transition Probabilities This table displays the empirical transition probabilities for z-scored relative-to-rivals cash-to-asset (zCash) as well as for cash-to-asset (Cash). The transition probability for cell (i,j) is the probability of a firm moving from zCash (Cash) quartile i during year t to zCash (Cash) quartile j in year t+1 (or t+2 and t+3). The sample period is 1973 through 2006. The probabilities do not sum to one because of rounding errors. Numbers in brackets are the actual number of firms in each cell. Panel I. Relative-to-Rival Cash (zCash) Panel II. Cash-to-Asset (Cash) t/t+1 1st quar. 2nd quar. 3rd quar. 4th quar. t/t+1 1st quar. 2nd quar. 3rd quar. 4th quar. 1st quar. 0.707 0.203 0.056 0.033 1st quar. 0.786 0.167 0.03 0.016 (7821) (2247) (620) (367) (8689) (7849) (332) (185) 2nd quar. 0.206 0.475 0.227 0.09 2nd quar. 0.169 0.556 0.206 0.068 (2288) (5254) (2515) (999) (1870) (6147) (2280) (759) 3rd quar. 0.055 0.227 0.485 0.231 3rd quar. 0.031 0.212 0.524 0.231 (618) (2514) (5364) (2559) (344) (2345) (5802) (2564) 4th quar. 0.029 0.094 0.231 0.645 4th quar. 0.013 0.0647 0.238 0.682 (328) (1041) (2556) (7131) (152) (715) (2641) (7548) t/t+2 1st quar. 2nd quar. 3rd quar. 4th quar. t/t+2 1st quar. 2nd quar. 3rd quar. 4th quar. 1st quar. 0.608 0.24 0.091 0.059 1st quar. 0.717 0.198 0.053 0.03 (6721) (2661) (1011) (662) (7929) (2193) (596) (337) 2nd quar. 0.239 0.386 0.251 0.122 2nd quar. 0.196 0.464 0.231 0.107 (2651) (4272) (2783) (1350) (2173) (5137) (2554) (1192) 3rd quar. 0.092 0.242 0.41 0.254 3rd quar. 0.056 0.232 0.448 0.261 (1022) (2675) (4540) (2818) (628) (2571) (4963) (2893) 4th quar. 0.059 0.131 0.246 0.563 4th quar. 0.029 0.104 0.266 0.6 (661) (1448) (2721) (6226) (325) (1155) (2942) (6634) t/t+3 1st quar. 2nd quar. 3rd quar. 4th quar. t/t+3 1st quar. 2nd quar. 3rd quar. 4th quar. 1st quar. 0.541 0.24 0.11 0.154 1st quar. 0.655 0.205 0.068 0.129 (5333) (2372) (1084) (2266) (6455) (2027) (670) (1903) 2nd quar. 0.258 0.349 0.248 0.178 2nd quar. 0.229 0.411 0.24 0.161 (2549) (3442) (2443) (2662) (2257) (4053) (2372) (2374) 3rd quar. 0.12 0.252 0.384 0.245 3rd quar. 0.073 0.255 0.409 0.257 (1183) (2488) (3783) (3601) (726) (2517) (4027) (3785) 4th quar. 0.079 0.156 0.257 0.421 4th quar. 0.041 0.126 0.282 0.45 (781) (1545) (2536) (6194) (408) (1250) (2777) (6621) 14 Table IA.V Persistence of the Cash-Performance Sensitivities This table presents the results of IV regressions examining the effect of relative-to-rivals cash holdings on market share growth (specification (1)). In columns (1) to (3) the dependent variable is ?MSt , the annual marketshare growth given by industry-adjusted sales growth at time t [(Salest - Salest-1)/Salest-1]. In columns (4) to (6) the dependent variables are multiple years market-share growth given by industry-adjusted multiple years sales growth at time t+1, t+2 and t+3 [(Salest+k – Salest-1)/Salest-1 for k=1,2 and 3]. zCash is the z-scored ratio of cash and marketable securities divided by total assets. Size is the natural logarithm of assets. Leverage is long-term debt over assets. All control variables are adjusted for their four-digit SIC industry-year means. The sample period is 1973 through 2006. The estimations correct the error structure for heteroskedasticity and within-firm error clustering. I report t-statistics in brackets. ** and * denote statistical significance at the 1% and 5% level, respectively. (1) (2) (3) (4) (5) (6) Variables ?MSt ?MSt ?MSt ?MSt,t+1 ?MSt,t+2 ?MSt,t+3 zCasht-2 0.027** 0.026** 0.026** 0.069** 0.097** 0.118** [10.41] [9.89] [9.32] [13.71] [12.33] [10.77] zCasht-3 0.011** 0.012** 0.010** [4.75] [3.86] [3.37] zCasht-4 0.008* 0.006* [2.33] [2.12] zCasht-5 0.005* [1.98] Sizet-1 0.043** 0.043** 0.043** 0.019** 0.017** 0.012** [16.54] [15.94] [15.17] [3.38] [4.41] [4.72] Leveraget-1 0.008* 0.008** 0.009** -0.012* -0.026** -0.032** [2.46] [2.91] [2.43] [2.24] [3.07] [2.72] Leveraget-2 -0.001** -0.006** -0.009** -0.014** -0.020** -0.030* [3.42] [3.66] [3.04] [3.09] [3.78] [2.54] ?MarketSharest-1 0.008 0.008 0.005 -0.095** -0.166** -0.195** [1.60] [1.50] [1.23] [8.08] [9.05] [7.59] ?MarketSharest-2 -0.073** -0.076** -0.083** -0.124** -0.149** -0.191** [14.15] [13.79] [14.11] [10.98] [8.85] [7.70] Firm fixed-effects Yes Yes Yes Yes Yes Yes Time fixed-effects Yes Yes Yes Yes Yes Yes # Obs 34774 31558 28447 33687 31089 27556 R 2 0.24 0.24 0.25 0.32 0.41 0.45