Determinants of Corporate Cash Policy: Insights from Private Firms *

Transcription

1 Determinants of Corporate Cash Policy: Insights from Private Firms * Huasheng Gao Nanyang Business School Nanyang Technological University S3-B1A-06, 50 Nanyang Avenue, Singapore hsgao@ntu.edu.sg Jarrad Harford Foster School of Business University of Washington Box , Seattle, WA jarrad@uw.edu Kai Li Sauder School of Business University of British Columbia 2053 Main Mall, Vancouver, BC V6T 1Z kai.li@sauder.ubc.ca First version: November, 2010 This version: March, 2012 Abstract We provide one of the first large sample comparisons of cash policies in public and private U.S. firms. We first show that on average private firms hold less than half as much cash as public firms do. The higher cash holdings of public firms are partially caused by the fact that public firms add more to their cash reserves in a given year, even controlling for a number of spending and savings factors, than do similar private firms. Further, consistent with the presence of financing frictions, we find that private firms adjust much more slowly to their target cash levels when they have a cash deficit than public firms do, and private firms cash-to-cash flow sensitivity is higher than that of public firms. Overall, our evidence supports both the agency conflicts and the financing frictions views of corporate cash policy. Keywords: cash holdings; cash-to-cash flow sensitivity; financing frictions; agency conflicts; private firms; speed of adjustment JEL Classification: G30; G32 * We are grateful for helpful comments from Tom Bates, Mark Grinblatt, Rob Heinkel, Marc Martos-Vila, Rong Wang, Xueping Wu, Shan Zhao, seminar participants at City University of Hong Kong, Hong Kong Baptist University, Nanyang Technological University, UBC, UCLA, UIBE, Renmin University, Shanghai Advanced Institute of Finance, and Shanghai University of Finance and Economics, and conference participants at the Third Shanghai Winter Finance Conference, the Asian Finance Association Meetings, the China International Conference in Finance, the Tsinghua International Corporate Governance Conference, and the Northern Finance Association Meetings. Li acknowledges the financial support from the Social Sciences and Humanities Research Council of Canada. All errors are ours.

2 1. Introduction Corporate holdings of cash reserves have received increasing academic interest. 1 As of 2010, public firms in the U.S. held on average 18.8% of their assets in cash or near-cash instruments. 2 Work explaining cash holdings has focused primarily on financing frictions and agency conflicts. Financing frictions lead firms to have a precautionary demand for cash holdings, which has been studied as early as Keynes (1936). A specific form of financing frictions, the wedge between internal and external costs of capital created by information asymmetry, can lead firms with greater information asymmetry about their investment opportunities to hold more cash. Evidence in favor of this explanation has been found by Harford (1999) and Opler, Pinkowtiz, Stulz, and Williamson (1999). Agency conflicts should also affect cash policies. Dittmar, Mahrt-Smith, and Servaes (2003) study cash holdings across countries and conclude that in countries where investor protection is lower, firms hold more cash, while in countries where investors have more power, they use that power to force managers to disgorge the cash. More broadly, Nikolov and Whited (2010) estimate that typical agency problems increase cash holdings by 22%, resulting in a 6% drop in shareholder value. Alternatively, Harford, Mansi, and Maxwell (2008) show that firms with more entrenched managers actually hold less cash than otherwise similar firms and conclude that managers would prefer to overinvest rather than maintain observably high cash levels. In this study we exploit a database of private firms to help understand public firms cash policies. It is worth noting that the cash policy of private firms in itself is of great interest to financial economists due to a lack of data prior to our study. Further, the contrast between public and private firm behavior in cash management serves as cross-validation of prior research on cash policies using only public firms. We expect that the variation in agency conflicts across these two types of firms is likely to be at least as substantial as the variation within public firms. Further, differences across these two types of firms in financing frictions allow us to explore the relative importance of these two effects on cash levels and 1 Starting from the seminal work by Baumol (1952), Miller and Orr (1966), and Jensen (1986), there has been a recent surge of papers including Opler, Pinkowitz, Stulz, and Williamson (1999), Dittmar and Mahrt-Smith (2007), Foley, Hartzell, Titman, and Twite (2007), Harford, Mansi, and Maxwell (2008), Bates, Kahle, and Stulz (2009), Nikolov and Whited (2010), Acharya, Davydenko, and Strebulaev (2011), Gryglewicz (2011), and Dittmar and Duchin (2011), examining determinants of corporate cash policy. 2 This number is based on all public U.S. firms listed on the NYSE, AMEX, and NASDAQ in The corresponding numbers for 2007, 2008, and 2009 are 21.9%, 20.3%, and 19.7%, respectively. 1

3 changes, speed of adjustment to target cash levels, and cash-to-cash flow sensitivities. We construct tests to identify whether each effect matters as well as to measure their net effect on levels and changes of cash, speed of adjustment, and cash-to-cash flow sensitivities. Using a sample of public and private U.S. firms over the period , we first show that on average private firms hold less than half as much cash as public firms do. This is despite the fact that they arguably have less access to external financing and would be expected to have a stronger precautionary motive due to financing frictions. Even controlling for standard factors affecting cash reserves, we find that the effect of agency costs from being public, net of the reduced effect of financing frictions, still leads public firms to hold cash reserves that are approximately 4% of assets higher than are those of similar private firms. This key finding remains in a reduced form model of cash holdings that accounts for the joint determination of leverage, investment, dividends, and cash holdings, as well as accounting for the transitory component of cash holdings, and controlling for other sources of liquidity. We then show that consistent with the presence of financing frictions, private firms adjust much more slowly to their target cash levels when they have a cash deficit than public firms do. Conversely, due to the combination of both financing frictions and agency conflicts (primarily), we find that public firms with a cash surplus adjust much more slowly to their target levels of cash than do private firms. To clearly capture the effect of financing frictions on corporate cash policies, Almeida, Campello, and Weisbach (2004) show theoretically and empirically that one must examine the sensitivity of the firm s cash holdings to its cash flow unconstrained firms will display savings behavior that is much less sensitive to their cash flows than will constrained firms. Consistent with the presence of financing frictions, we find that private firms cash-to-cash flow sensitivity is higher than that of public firms. We apply a treatment regression approach to addressing the selection issues that companies may choose to go public or stay private. We find that the differences in the level and change in cash holdings are even greater between public and private firms after controlling for selecting into public status. Our study contributes to the literature by being one of the first to examine the cash policies of private firms and by establishing a conservative estimate of the effect of agency costs on cash holdings through the use of a sample of private U.S. companies. Our ability to speak to big picture questions of financing frictions and agency costs (with the somewhat surprising finding that private firms hold less 2

4 cash) is the real distinction of our paper from others. Previous investigations of the issue have been limited to using data on public companies only. For example, Dittmar et al. (2003) show that one would expect U.S. firms to hold less cash than firms in countries with weaker investor protection, and we extend that by showing that even given better investor protection, U.S. firms still hold more cash than they would if their agency costs were mitigated by being private. We also show that despite the evidence that financing frictions such as the cost of external financing are greater for private firms, the effect of agency conflicts is strong enough to lead to much higher cash holdings, as well as higher growth in those holdings, in public firms. In using private firms, we join a recent surge of papers using data on private companies to draw new insights into public company behavior. Bargeron, Schlingemann, Stulz, and Zutter (2008) show that private acquirers pay significantly less for targets than public acquirers which they attribute to lower agency costs stemming from more concentrated ownership structures. Brav (2009) examines the financial policies of private and public firms in the U.K. and shows that private firms tend to borrow more, resulting in higher leverage ratios. Maksimovic, Phillips, and Yang (2010) show that public firms participate more in mergers and acquisitions than do private firms, are more cyclical in their acquisitions, and are impacted more by macro factors, consistent with public firms having better access to capital. Asker, Farre-Mensa, and Ljungqvist (2011) contrast investment behavior of private firms with that of public firms and find that public firms invest less and are less responsive to changes in investment opportunities compared to observably similar private firms. They conclude that agency problems resulting from the separation of ownership and control in publicly listed firms distort investment incentives. Michaely and Roberts (2012) show that private firms smooth dividends significantly less than their public counterparts. They conclude that the scrutiny of public capital markets plays a central role in the propensity of firms to smooth dividends over time. In contemporaneous work, Farre-Mensa (2011) explores why public firms hold more cash. He concludes that the inability of public firms to disclose information selectively is an important driver of the difference in cash holdings between public and private firms. Farre-Mensa uses a sample of private firms that are contributed without firm identification by mid-market accounting firms (Sageworks, the data provider, collects the data). That sample of private firms does not necessarily have public debt. While selective disclosure is likely part of the explanation for higher cash holdings in public firms as compared 3

5 to private firms in his sample, it is less applicable in our sample, which consists mostly of firms that must file with the SEC due to meeting the size criterion for disclosure and/or being formally defined as issuers (of public debt). Securities law, including Regulation FD, generally makes no distinction between firms that issue public equity and firms that have private equity with public debt. Both are classified as filing issuers, limiting their ability to selectively disclose. In a nutshell, due to the very different private firm samples employed, Farre-Mensa (2011) and our paper offer complementary explanations for the robust phenomenon that public firms hold more cash than private firms. The plan of the paper is as follows. We review the literature and develop our hypotheses in the next section. We describe our sample and present summary statistics in Section III. We examine the difference in cash policies among public and private U.S. firms in Section IV. Sample selection issues are addressed in Section V. We conclude in Section VI with a brief summary. II. Literature Review and Hypothesis Development There is a substantial literature examining firms motives for holding cash (see Bates, Kahle, and Stulz (2009) for an excellent summary of the state of the literature). For our purpose, we only review papers directly related to our empirical investigation, namely the role of financing frictions and agency conflicts, then proceed to develop our hypotheses. II.A. Related Research Firms hold cash to protect themselves against adverse cash flow shocks that might force them to forgo valuable investment opportunities due to costly external financing. Empirical research on cash holdings has generally found support for the precautionary motive especially among firms with greater information asymmetry with external capital providers (for example, Opler et al. (1999)). Work by Bates et al. (2009) has provided partial explanations for the rising trend in cash holdings by public U.S. firms, finding support for precautionary motives, but not for agency-based explanations. Further support for the precautionary demand explanation is provided by Duchin (2010) showing that increasing cash flow uncertainty can help explain the build-up in cash holdings by public firms, and by McLean (2011) showing that share issuance has become an increasingly important source of cash for firms with high 4

6 precautionary motives as captured by large R&D expenditures and high cash flow volatility. From a more direct angle, several papers provide evidence of greater financing frictions for private firms. Brav (2009) shows that cash holdings of private U.K. firms are more sensitive to operating cash flows than those of public firms, and that the former do not increase their investments concurrent with an increase in performance. Bharath and Dittmar (2010) find that public firms with high cash holdings and few investment opportunities are more likely to go private. Saunders and Steffen (2011) compare borrowing costs for private and public U.K. firms, and show that private firms must pay higher borrowing costs, ceteris paribus. Using international surveys of public and private firms, Campello, Giambona, Graham, and Harvey (2011) show that lines of credit are more expensive for private firms, and Lins, Servaes, and Tufano (2010) show that private firms rely more on non-operational (excess) cash instead of lines of credit for their corporate liquidity. In addition to the precautionary motive of holding cash, Jensen (1986) argues that entrenched managers would rather retain cash than increase payouts to shareholders when their firms have poor investment opportunities. Stulz (1990) characterizes the shareholders problem as providing sufficient internal slack to avoid underinvestment while not providing too much so as to fund overinvestment. These discretionary cash holdings are typically estimated as the excess cash holdings derived from models controlling for the transaction and precautionary motives for holding cash. A number of recent papers by Dittmar et al. (2003), Pinkowitz, Stulz, and Williamson (2006), Dittmar and Mahrt-Smith (2007), and Harford et al. (2008) have provided support for the agency perspective of corporate cash policy: Excess cash reserves aggravate agency problems by providing a pool of accumulated free cash flow. Harford et al. (2008) find that firms with poor governance spend more cash than those with better governance, often to the effect that their accumulated cash reserves are actually lower. In contrast, studies such as Bertrand and Mullainathan (2003) suggest a slightly more benign form of agency problems the CEOs desire for a quiet life, would lead to higher-than-optimal buffer stock of cash holdings. II.B. Hypothesis Development Cash reserve policy should balance the precautionary demand identified in Baumol (1952), and Miller and Orr (1966) against agency problems highlighted in Jensen (1986), and Stulz (1990). 5

7 One of the primary reasons given for going public is to have lower-cost access to capital. Being listed provides liquidity and a market price for a firm s equity that substantially lowers its cost of equity capital (Bharath and Dittmar (2010), and Maksimovic et al. (2010)). The transparency provided by listing may also reduce its cost of debt (Campello et al. (2011), and Saunders and Steffen (2011)) and may increase access to public debt and other sources of liquidity (Faulkender and Petersen (2006), and Lins et al. (2010)). Given higher costs of accessing external capital, the precautionary motive should be stronger for private firms, leading to our first hypothesis: Hypothesis 1: Private firms hold higher cash reserves than otherwise similar public firms. The alternative, that public firms hold cash reserves that are equal to or greater than those of private firms, is motivated by the countervailing effect of agency problems. Private firms have much fewer agency problems than public firms. Private firms often have owner-managers and at a minimum have concentrated illiquid ownership and large private lenders providing greater monitoring incentives. The greater separation of ownership and control, along with the free-rider problem from dispersed highly liquid ownership, significantly increases agency problems in public firms ((Bhide (1993), Asker et al. (2011), and Gao, Lemmon, and Li (2012)). With reduced ability to overinvest, or enjoy the quiet life, private firm managers will have lower incentives to maintain a large supply of accessible funds. In summary, our prior discussion focuses on the tension between financing frictions and agency costs and their impact on the level of cash holdings in public and private firms, which is an important question given that holding excess cash is perceived to be potentially value destructive (see for example, Harford (1999)). Regardless of the relative average level of cash holdings by public and private firms from a static model, we also examine a dynamic model of cash holdings. Unlike the static model that implicitly assumes that firms can instantaneously adjust their cash holdings toward the target levels, the dynamic model recognizes that an adjustment process may take place and there are some lags for firms to adjust their cash holdings to their target levels. In general, we expect that the speed of adjustment to target levels would be faster for less financially constrained firms (public firms in our case) and slower for more financially constrained firms (private firms in our case). However, the speed of adjustment could be asymmetric due to the combination of agency conflicts and financing frictions. When they have insufficient cash holdings, public firms 6

8 adjustment speed will be faster from below than will private firms. In contrast, a priori, it is not clear how these two types of firms adjust when they have excess cash holdings the predicted effects of financing frictions and agency conflicts on the speed of adjustment are opposite to each other for public and private firms with a cash surplus. Specifically, due to more severe agency problems, managers of public firms may prefer excess cash, leading to a slower adjustment downward than private firms. On the other hand, financial slack should be more valuable for private firms, leading to a slower adjustment downward than public firms. However, assuming that we have unbiasedly characterized a model of optimal cash holdings, excess cash for private firms will still be in excess of their optimal financial slack. Therefore, they should not adjust truly excess cash more slowly than public firms. As a result, it remains an empirical question whether downward adjustment speed differs between private and public firms. Our second hypothesis is thus: Hypothesis 2: Conditioning on deficiency in cash, public firms adjust faster to their target levels of cash holdings than do private firms. Almeida et al. (2004) establish that constrained firms should and do show greater sensitivities of cash holdings to operating cash flows. Denis and Sibilkov (2010) further show that cash is associated with greater investment for constrained firms and that the marginal investment of constrained firms is associated with greater value increases than those of unconstrained firms. Based on the fact that private firms are more constrained than public firms, we have the following hypothesis: Hypothesis 3: Private firms exhibit higher cash-to-cash flow sensitivities than public firms. Riddick and Whited (2009) extend the model of Almeida et al. as well as make adjustments to reduce the measurement error in Tobin s q. They show that it is difficult to draw conclusions about financing constraints based on cash-to-cash flow sensitivities alone, without addressing other factors, such as the uncertainty of and persistence in cash flow shocks that affect these sensitivities. For example, the uncertainty of cash flows is correlated with, if not often the reason why, firms have higher external financing costs. However, in our setting, the source of external financing costs is the private status of the firm, which is not expected to be related to cash flow uncertainty. Further, we are making comparisons between similar public and private firms rather than making an absolute statement based solely on sensitivities. 7

9 In our empirical analysis, we test these hypotheses and also attempt to distinguish between some of the alternative explanations for the differences. In the next section we describe our data and present summary statistics. III. Our Sample III.A. Sample Formation Our primary data source is the Capital IQ database. Capital IQ is an affiliate of Standard & Poor s which produces the Compustat database. 3 We start with all public U.S. firms traded on the NYSE, AMEX, or NASDAQ, and all private U.S. firms filed Form 10-K (annual reports), Form 10-Q (quarterly reports), or Form S-1 (securities registration) with the SEC for the period Following prior work such as Opler et al. (1999), we remove financial and utilities firms. We further require that all sample firms have financial information for at least two consecutive years to estimate the annual changes in cash holdings; and our sample firm s operating cash flow over total assets is no less than 50%. Finally, we remove all firm-year observations associated with 1,105 IPOs and 124 going private transactions. Prior work including Miller and Orr (1966), Harford (1999), Opler et al. (1999), and Dittmar et al. (2003) has shown that cash holdings tend to vary systematically by industry and larger firms tend to have lower cash holdings due to economies of scale in the transaction motive for cash. These findings motivate us to industry- and size-match our sample of private firms. Specifically, we match each private firm in our sample with a public firm in the same industry (as defined by Fama and French (1997)) and closest in size (total assets) from the very beginning of our sample period, and only change the matched public firm when the initial match drops out of the Capital IQ database. The matching (with replacement) procedure helps mitigate the large difference in the size distribution between the two samples and the smaller, but potentially important difference in sample firm distribution across industries. 4 For most of 3 Starting from the late 1990s, Capital IQ provides data on some private U.S. firms. When available, Capital IQ provides data on firm accounting information with a similar level of detail as provided by Compustat for public U.S. firms. 4 The top five public firm industries are: Business Services (16.6%), Electronic Equipment (7.7%), Pharmaceutical Products (6.9%), Retail (6.1%), and Petroleum and Natural Gas (5.1%). The top five private firm industries are: 8

10 our analysis, we present results employing both the full sample and the matched sample of public firms for our sample of private firms. In our final sample, we have 6,676 private firm-year observations representing 2,574 unique private firms, 6,676 matched public firm-year observations representing 1,843 unique public firms, and 33,675 public firm-year observations representing 5,275 unique public firms, for the period Data for a vast majority (88%) of the private firm-year observations in our sample come from Form 10-K (annual reports), and the remainder (12%) comes from Form S-1 (and its supplemental Form 424B). The SEC requires firms with $10 million or more in total assets and 500 or more shareholders to file annual (10-K) and quarterly (10-Q) reports. 5 The SEC also requires firms to file Form S-1 (sometimes Form DEF 14A as well) associated with public debt issues. It is worth noting that the private firms in our sample are large firms with some access and/or intend to gain access to the public debt market, and are more comparable to public firms than are the private firms as examined by Brav (2009), Asker et al. (2011), and Farre-Mensa (2011), in terms of disclosure and information asymmetry. Nonetheless, while our firms are among the larger private firms with some access to bond markets, they are still more constrained than are public firms (they cannot easily raise new equity, for example) and have fewer agency conflicts. III.B. Summary Statistics Table 1 provides summary statistics for our sample. We have two samples of public firms. The first is all public firms for which we have data. The second is a sample of public firms matched to our private firms by industry and size. All dollar values are in 2010 dollars. All continuous variables are winsorized at the 2.5% and 97.5% levels. Cash is the ratio of cash and marketable securities to total assets. The first row shows that public firms hold substantially more cash. The mean (median) cash holdings is 20.13% (11.39%) for the all public firm sample, the mean (median) cash holdings is 15.49% (8.01%) for the matched public firm sample, while the mean (median) cash holdings is 9.45% (3.73%) for the private firm sample. The two- Business Services (12.8%), Communication (9.7%), Transportation (9.2%), Wholesale (7.9%), and Restaurants, Hotels, Motels (6.4%). 5 According to the SEC s website at it states as follows: Generally, smaller companies only have to file reports with the SEC if they have $10 million or more in assets and 500 or more shareholders, or list their securities on an exchange or Nasdaq. 9

11 sample t-test and Wilcoxon-test both reject the null that cash holdings in public firms (using either the full or matched public firm sample) is the same as that in private firms at the 1% level. On average, cash holdings in public firms are approximately twice that in private firms. One might argue that the difference in cash holdings between public and private firms is driven by the different industry representation across public and private firms. To mitigate that concern, we compute industry-adjusted cash holdings as the difference between firm-specific cash holdings and its industry median based on Fama and French s (1997) 48 industry classification involving all sample public and private firms. By using a uniform benchmark level of cash holdings across public and private firms for a particular industry, we emphasize the importance of common industry-specific investment opportunities to both public and private firms alike. We show that the contrast between public and private firms in terms of cash holdings is even more striking. The mean (median) industry-adjusted cash holdings is 4.58% (0.26%) for the all public firm sample, the mean (median) industry-adjusted cash holdings is 6.14% (1.20%) for the matched public firm sample, while the mean (median) industry-adjusted cash holdings is 0.33% ( 1.76%) for the private firm sample. The two-sample t-test and Wilcoxon-test both reject the null that industry-adjusted cash holdings in public firms (using either the full or matched public firm sample) is the same as that in private firms at the 1% level. Change in cash is simply the difference between this year s and last year s cash. We show that public firms change in cash is positive and three times as large as private firms, indicating that, on average, public firms add to their cash reserves each year and do so by significantly more than do private firms. The univariate statistics thus far are consistent with the agency conflicts hypothesis of cash policy whereby there are more serious agency problems in public firms compared to private firms. The mean (median) value of total assets is $1,746 million ($257 million) for the all public firm sample, the mean (median) value of total assets is $1,236 million ($249 million) for the matched public firm sample, and the mean (median) value of total assets is $1,307 million ($248 million) for the private 10

12 firm sample. 6 The two-sample t-test and Wilcoxon-test both reject the null that public firms using full public sample are of the same size as private firms at the 1% level, while public firms in the matched sample are of no significant difference in size compared to private firms. The fact that our private firm sample tends to consist of larger private firms actually makes our sample more comparable to public firms. The reader should bear in mind the sample selection criteria imposed on us by the data when deciding how our results might generalize. In terms of profitability, the two-sample t-test indicates that the average operating cash flow of public firms is significantly lower than that in private firms, while the Wilcoxon-test indicates that the median operating cash flow of public firms is significantly higher than that in private firms. The greater standard deviation of operating cash flow for private firms suggests greater positive skewness in that sample, explaining the differing mean and median results. We calculate cash flow volatility using the standard deviation of industry-median-adjusted quarterly operating cash flow over the previous two years. Using industry-adjusted cash flow controls for the differences across industries in the quarterly seasonality of cash flows and in the nature of firms operations. The two-sample t-test and Wilcoxon-test show that private firms have significantly higher cash flow volatility than public firms (and significantly higher median cash flow volatility, but significantly lower average cash flow volatility as compared to matched public firms). A standard precautionary demand model for cash holdings would predict a higher average level of cash holdings in the presence of greater cash flow volatility, but the univariate results from row 1 do not support that. Public firms sales growth is somewhat higher, while leverage is drastically higher in private firms, consistent with the fact that private firms must rely on debt and internally generated or privately placed equity, while public firms are able to tap the public equity market (as shown by Brav (2009), and Asker et al. (2011)). Our sample public and private firms have similar access to the public debt market: 22% of public firms and 23% of private firms have public debt outstanding. As with the greater cash flow volatility, greater leverage would increase demand for cash holdings both to reduce net debt and to 6 Using the Sageworks database, Asker et al. (2011) shows that the sample average total assets is $144.7 million and $120.0 million for their matched public sample and the private sample, respectively. The difference in firm size between the two samples is not statistically different at the 5% level. 11

13 provide a buffer to meet interest obligations. Our evidence thus far does not support either argument as private firms in our sample hold less cash than their public counterparts. Net working capital is defined as the difference between current assets and current liabilities excluding cash. Net working capital can be a substitute for cash (Opler et al. (1999)). Row 10 of the table shows that net working capital for public firms is significantly higher than private firms on average. Row 11 shows that public and private firms have similar amount of undrawn lines of credit relative to total assets, another important source of liquidity, in addition to cash and net working capital. Row 12 shows that public firms have a slightly higher tendency to have multiple segments. This result does not support the notion that selling non-core assets is another viable substitute to holding cash (Lang, Poulsen, and Stulz (1995)). We find that on average, public firms spend 4.58% on capital expenditure, and 2.07% on acquisitions, while private firms spend 5.19% of total assets on capital expenditures, and 2.68% of total assets on acquisitions. 7 These differences in capital expenditure and acquisitions are statistically significant at the 5% level. Private firms spend less on R&D, are more likely to pay dividends, and are younger. The former set of results is in mixed support of information asymmetries/transaction costs models of cash holdings, which predict that firms with lower R&D expenditures will hold less liquid assets, while firms with high dividend payouts will hold more liquid assets (Opler et al. (1999)). Foley, Hartzell, Titman, and Twite (2007) find that U.S. companies that would incur tax consequences associated with repatriating foreign earnings hold higher levels of cash. We define MNC, an indicator variable to take a value of one if the fraction of foreign sales to total sales of a firm exceeds 20%, and zero otherwise. We find that the fraction of multinational companies is highest among the all public firm sample (at 27%), the second highest among the matching public firm sample (at 18%), and the lowest among the private firm sample (9%). 7 Using the Sageworks database, Asker et al. (2011) shows that private firms invest significantly more, as captured by the annual change in either gross or net fixed assets, than do public firms. Using the plant level data, Maksimovic, Phillips, and Yang (2010) show that public firms are more acquisitive and more likely to sell assets than private firms. 12

14 Bates et al. (2009) note that the average cash ratio (relative to assets) for U.S. firms more than doubles from In Table 2, we present cash ratios over time for the all public firm sample, the matched public firm sample, and the private firm sample. For all public firms, the average (median) cash ratio increases from 18.89% (7.91%) in 2000 to 20.03% (14.21%) in For the matched public firms, the average (median) cash ratio increases from 14.30% (5.68%) in 2000 to 16.51% (10.75%) in In contrast, for the private firm sample, the average (median) cash ratio increases from 9.26% (2.63%) in 2000 to 11.04% (5.63%) in Despite having different degrees of agency conflicts, the public and private samples show similar rates of increase. Thus, we conclude, as do Bates et al. (2009), that the increase in cash ratios over time is not due to agency problems. Table 3 presents the correlation matrix for the variables used in this study. None of the correlations are high enough to present collinearity problems for our multivariate analyses. In the next section, we will run multiple regressions to test our hypotheses. IV. Main Results IV.A. Excess Cash Holdings Table 4 Panel A presents the regression results of a model for normal levels of cash holdings based on the extant literature (for example, Kim, Mauer, and Sherman (1998), Opler et al. (1999), Dittmar and Mahrt-Smith (2007), Foley et al. (2007), and Harford et al. (2008)): # &. (1) The dependent variable is the level of the cash ratio. In addition to firm-level determinants of cash, we also include the industry-year fixed effects to control for the effect of time-varying industry factors on cash policies in the regression. 8 The table shows that all firms have sharp increases in cash in 2009, remaining higher in In untabulated robustness checks, we confirm that all of our inferences hold after those two years are removed from the sample. 13

15 The results confirm the univariate findings from Table 1. Specifically, public firm cash holdings are still abnormally high controlling for a host of factors from the literature on cash holdings. We present results using the full public firm sample (Columns (1)-(2)) as well as the matched public firm sample (Columns (3)-(4)). The inferences are the same for the two samples: The coefficient on the public firm indicator variable is for the full sample of public firms together with the private firm sample (Column (2)); and the coefficient on the public firm indicator variable is for the matched sample of public firms together with the private firm sample (Column (4)). In brief, public firms hold cash reserves relative to total assets that are approximately 4% higher than are those of similar private firms. The coefficients on the control variables are consistent with prior findings: Large firms and more profitable firms hold lower cash reserves, while firms with greater cash flow volatility and sales growth hold more. Leverage, the public debt indicator variable, the number of segments, capital expenditures, acquisitions, and the dividend indicator variable have negative effects on cash reserves, while R&D expenditures have a positive effect on cash reserves. There is a substitution effect between non-cash working capital which can easily be converted into cash, different from the univariate statistics. Older firms hold less cash. Due to tax considerations, multinational firms hold more cash. We also present separate regressions for the private firms and matched public firms (Columns (5)-(6), respectively). For the most part, the explanatory variables have very different effects on the cash policies of private and public firms. For example, the effect of firm size on cash is about four times as large for private firms than for public firms. The effect of leverage is substantially greater for public firms. The reduced variation for some of the explanatory variables reduces their significance. For example, the matched public firms are less likely to be multinational, so the coefficient becomes only statistically significant at the 10% level, instead of the 1% level in Column (2). Since firms may choose leverage, cash holdings, payout policy, and investment policy simultaneously, following Opler et al. (1999), and Dittmar et al. (2003), we estimate a reduced form model of cash by removing leverage, dividends, capital expenditures, acquisitions, and R&D expenditures from the set of explanatory variables. Table 4 Panel B presents the results. We show that most of firm characteristics continue to have the same significant effects on the level of cash holdings. Importantly, the coefficient on the public firm indicator variable is for the 14

16 full sample of public firms together with private firms; and the coefficient on the public firm indicator variable is for the matched public firms together with private firms. Using the reduced form model of cash, public firms hold cash reserves that are 5.8% to 8.7% of assets higher than are those of similar private firms. Another robustness check we implement is to remove the transitory component of cash holdings, that is, the portion of cash holdings that will be spent in the near term, to see if our main findings on the excess cash holdings by public firms remain. To capture the transitory component of cash holdings, following Opler et al. (1999), we use the next year s cash spending, defined as the difference between normalized cash holdings in year t and year t+1, and add it to the set of explanatory variables. Table 4 Panel C presents the results. We show that most of firm characteristics continue to have the same significant effects on the level of cash holdings. Importantly, the coefficient on the public firm indicator variable is for the full sample of public firms together with the private firm sample; and the coefficient on the public firm indicator variable is for the matched sample of public firms together with the private firm sample. After controlling for the existence of the transitory component of cash holdings, public firms hold cash reserves that are 3.5% to 4.3% of assets higher than are those of similar private firms. Our final robustness check is to control for firms other sources of liquidity, by including undrawn lines of credit normalized by total assets. Table 4 Panel D presents the results. We show that firms with undrawn credit are less likely to hold more cash, consistent with recent work by Lins et al. (2010), and Campello et al. (2011) demonstrating that one important source of liquidity is unused credit lines a measure of external liquidity (vis-à-vis measures of internal liquidity in terms of cash holdings that we focus on in this paper and profitability). Nonetheless, after controlling for these other sources of liquidity, we still observe that public firms hold cash reserves that are 3.8% to 4.6% of assets higher than are those of similar private firms. In summary, our results reject Hypothesis 1 that financing frictions would lead private firms to hold more cash and support the alternative that reduced agency problems would lead private firms to hold less cash. In fact, given that it is unlikely that financing frictions are irrelevant, the results can be viewed as the net effect of the reduction in agency problems and the increase in financing frictions. Thus, the 15

17 conclusion that agency problems associated with the public status increases cash reserves relative to total assets by, on average, about 4% is conservative. IV.B. Changes in Cash After showing that public firms tend to hold more cash than private firms, it is important to show how this difference in cash holdings occurs. Table 5 presents the regression results of a model explaining annual changes in cash following Bates et al. (2009): # &. (2) The dependent variable is the change in the cash ratio. Consistent with the univariate results in Table 1, on average, in a given year public firms add to their cash reserves by more than do private firms: The coefficient on the public firm indicator variable is between to for the matched public firms and the full sample of public firms (together with the private firm sample). The change in firm size has a positive effect on the change in cash holdings. The lagged level of cash holdings is negatively associated with the change in cash holdings. Most of the coefficients on the change variables are consistent with our prior findings on the level variables. Overall, the regression results reveal that explaining the year-to-year change in cash is more difficult than explaining the level of cash. We conclude that public firms add more to their cash reserves in a given year, even controlling for spending and savings factors, than do similar private firms. Next, we examine to what extent public and private firms differ in their speeds of adjustment to their target levels of cash holdings. This together with our investigation of cash-to-cash flow sensitivities will help assess the relative effect of agency conflicts vis-à-vis financing frictions. IV.C. Speed of Adjustment of Cash Market imperfections such as transaction costs may prevent firms from rapidly adjusting their cash holding to the target levels. In Table 6, we use the partial adjustment model to estimate the speed of 16

18 adjustment (SOA) of cash holdings across public and private firms. Following Dittmar and Duchin (2011), we estimate the following model:. (3) The dependent variable is the change in the cash ratio, Cash* is the predicted cash ratio based on Column (2) of Table 4 Panel A, and thus (Cash* Lagged cash) measures the deviation of a firm s cash holdings from its target level of cash holdings. The coefficient captures SOA, and the coefficient captures the difference in SOA across public and private firms. We use the full sample in Column (1) and find that the coefficient on (Cash* Lagged cash) is positive and significant at the 1% level, while the coefficient on Public (Cash* Lagged cash) is not significantly different from zero. These results indicate that both public and private firms are actively adjusting their cash holdings towards target levels and that the speed of adjustment is not significantly different between a public and a private firm. However, the above analysis might hide heterogeneity between public and private firms when their actual levels of cash holdings fall above or below their target cash levels. As hypothesized, public and private firms can face asymmetric adjustment costs between building and depleting cash reserves. The cost of external financing may be particularly important for firms with a cash deficit to build their reserves. Therefore, when the firms actual levels of cash holdings are below their target cash levels, public firms should adjust their cash holdings to their target levels more rapidly than private firms. In Column (2) ((3)), we separately estimate Equation (3) using a sample of public and private firms with actual levels of cash falling above (and below) target levels. Using the subsample of firms with a cash surplus (Cash* Lagged cash 0) in Column (2), we find that the coefficient on Public (Cash* Lagged cash) is negative and significant at the 1% level. This result indicates that public firms are much slower in adjusting towards their target levels than private firms when more cash is held than the target level. In contrast, examining the subsample of firms with a cash deficit (Cash* Lagged cash > 0) in Column (3), we find a positive and significant coefficient on the interaction term Public (Cash* Lagged cash), suggesting that public firms adjust their cash holdings much faster toward their target levels than private firms when less cash is held than the target level. The evidence in Table 6 is consistent 17

19 with our Hypothesis 2 and suggests that between public and private firms, the former have fewer frictions in accumulating cash than the latter when it is low and are not as quick to reduce it when it is high, likely due to more severe agency problems in public firms. The relatively slow adjustment for public firms is consistent with extant results. Opler et al. (1999) show that cash-rich firms transition out very slowly. Harford et al. (2008) show that the worst-governed public firms spend the most cash on capital expenditures and acquisitions. Because these are large, lumpy investments, the average adjustment for public firms can appear slow, similar to the findings in the capital structure literature (see Strebulaev (2007) for example, who notes that infrequent large adjustments towards a target level of leverage can generate a slow adjustment result in panel data analyses). Overall, we conclude that both financing frictions and agency problem influence the adjustment speed of cash holdings: The public firms reduced financing frictions facilitate actions to reduce a cash deficit, while agency problems slow public firm managers desire to reduce a cash surplus. IV.D. Cash-to-Operating Cash Flow Sensitivities In Table 7, we estimate the cash-to-operating cash flow sensitivity for the sample firms following Almeida et al. (2004) (with the exception that we do not include M/B ratios given that private firms in our sample do not have market value of equity):. (4) The dependent variable is the change in the cash ratio. The key variable of interest is the interaction between the public firm indicator variable and operating cash flow. The first two rows show that public firms still have a regular tendency to add to their cash reserves, but that their sensitivity to operating cash flow the portion of operating cash flow that they save is significantly lower than it is for private firms. In both the full and matched public firm samples, the total effect is even negative, consistent with Riddick and Whited s (2009) theoretical arguments and 18

20 empirical findings. 9 Overall, the cash-to-cash flow sensitivities are quite similar in magnitude as shown in Almeida et al. (2004). On the other hand, the coefficients capturing cash-to-cash flow sensitivities for their constrained firms are substantially higher, suggesting that our sample of private firms do not behave as if they are as constrained as Almeida et al. s constrained firms. One explanation is that we have data on these firms because all our sample private firms file audited financial statements regularly with the SEC which would help facilitate their access to capital. We conclude that the results in Table 7 support the effect of financing frictions on cash policy as captured in Hypothesis 3: Private firms savings behavior will be more sensitive to operating cash flows. 10 V. Dealing with Endogeneity Going public, of course, is not an exogenous event: Most firms go public (or private) for reasons that correlate with their financing or investment decisions (see for example, Brav (2009), Asker et al. (2011), Bharath and Dittmar (2010), and Maksimovic et al. (2010)). To account for the possible selection effect, the processes of cash holdings and the public status of a firm can be modeled as follows: 11,, 1, 0; 0, (5) The dependent variable is the level of the cash ratio. X is a list of firm-level control variables. The coefficient of key interest is β 1, in front of the public firm indicator variable, Public. The variable Public* indicates the latent propensity of a firm becoming public. For the purpose of identification, we need 9 In Table 6, we have shown that public firms adjust upward more quickly than private firms when there is cash deficit, and now in Table 7, we show that public firms also save less of their operating cash flow. Both findings can be explained by the fact that public firms have higher baseline increases to cash (the coefficient on the standalone public firm indicator variable is positive) and the likelihood that public firms are turning to external financing to reduce cash deficits. 10 One might expect that agency problems would affect the degree to which managers choose to stockpile cash from cash flows as a large number of papers have shown. However, it is not obvious what the directional effect of agency problems on cash-to-cash flow sensitivities: Managers could prefer a stockpile of cash or immediate spending. As a result of this ambiguity, our interpretation of the results on cash-to-cash flow sensitivities is mainly from the financing frictions perspective. 11 See Li and Prabhala (2007) for an overview of dealing with selection effects versus treatment effects in corporate finance. 19