Agency costs of free cash flow and the market for corporate control Suzanne Ching-Fang Lin BCom (University of Auckland), MCom (Hons) (University of Sydney) This thesis is presented for the degree of Doctor of Philosophy of The University of Western Australia School of Economics and Commerce 2006
Abstract This thesis investigates the relevance of Jensen s (1986) free cash flow theory to the market for corporate control in Australia. Jensen posits that firms generating cash in excess of that required to fund positive NPV projects face greater agency problems as the free cash flow exacerbates the conflict of interest between shareholders and managers. One implication from Jensen s free cash flow theory is that firms with high levels of free cash flow are more likely to initiate takeovers that are value-decreasing. There are two practical issues in testing Jensen s theory; first, constructing an appropriate proxy for free cash flow and secondly, identifying firms with free cash flow. These issues are addressed directly in the first of the two essays that comprise this thesis. The first essay develops and assesses the merits of four operational measures for free cash. One of them is a stock measure while the others are flow measures. The stock measure is included because previous studies have mostly used the stock measure of cash when identifying firms rich in free cash (henceforth, cash rich firms), despite that Jensen (1986) has made explicit reference to free cash flow. We test the validity of this approach by investigating whether stock measures of free cash coincide with flow measures. Our results reveal that the stock and flow measures of free cash give rise to quite different lists of cash rich firms. This is an important empirical contribution of the thesis. Given the lack of definitive criteria for deciding which operational measure of free cash flow is most appropriate, we identify multiple sets of free cash flow firms based on the different operational measures developed. For each operational definition, two methods are used to identify cash rich firms. The first method defines a firm as cash rich if its cash variable ranks in the tenth percentile. The second method defines firms as cash rich if their cash variable value is greater than one and a half standard deviations of the value predicted by a model. ii
In view of both the sensitivity to the cash measures used to proxy for Jensen s definition of free cash flow and the lack of compelling reasons to favour any one of the cash measures over the others, an investigation of the characteristics of free cash flow firms are conducted on those firms identified as robustly cash rich. Robustly cash rich firms are those whose four measures of cash all fall in the top ten percentile of values in a given year and those firms in which all four measures of cash are greater than one and a half standard deviations away from their predicted values. A total of 26 firms or 33 firm-years are identified as robustly cash rich from the two methods. We find that less than half of the firm-years (i.e., 39% of the combined total of 33 firm-years) are robustly cash rich under both methods. Robustly cash rich firms have significantly stronger growth opportunities, proxied by market-to-book ratio, compared to the rest of sample firms, indicating that the cash is likely to have been reserved for financing future investment opportunities. This argument is further supported by the finding that most robustly cash rich firms show an increase in capital expenditures in the subsequent year to being identified as robustly cash rich, even though the level of capital expenditures in the year identified as robustly cash rich are not significantly different from other sample firms. Moreover, we find that nearly half of the robustly cash rich firms have made a takeover attempt within three years of being identified as robustly cash rich. While the evidence from robustly cash rich firms appears to provide support for the free cash flow theory, further tests are required in order to determine if subsequent investments made are value-decreasing as suggested by Jensen (1986). Thus, the second essay tests the free cash flow theory directly by examining the association between free iii
cash flow and the takeover activity in Australia. We measure firms free cash flow with excess cash which adjusts for firms normal cash levels and is closer to Jensen s (1986) definition of free cash flow, that is, the second method that has been used to identify cash rich firms in essay one. Two measures of excess cash are used, including excess cash holdings and excess accounting cash flow. This study finds that the likelihood of a takeover attempt is, on the one hand, negatively related to excess cash holdings and, on the other hand, positively related to the level of excess accounting cash flow. However, we find that takeovers initiated by firms with excess accounting cash flow are not value-decreasing. In fact, bidders with high levels of excess accounting cash flow perform better in the long-run. Therefore, evidence based on the excess accounting cash flow measure is inconsistent with the implication from Jensen s (1986) free cash flow hypothesis, which suggests that firms endowed with free cash flow are not only more likely to make acquisitions but the acquisitions made are value-decreasing. The second essay also examines the governance structures of cash rich bidders. Cash rich bidders, based on the accounting cash flow measure, with higher proportions of non-executive directors on the board are found to have better long-run returns, showing support for the monitoring role of non-executive directors. Overall, the results from the second essay run counter to the free cash flow theory of takeover as no evidence is found to support the argument that cash rich bidders carry out value-decreasing acquisitions. iv
Table of Contents 1. Introduction 1 2. Essay 1: The cash hoarding behaviour of Australian firms 8 2.1 Introduction 8 2.2 Literature review: What explains differences in cash holdings? 9 2.2.1 Asymmetric information and financial distress costs 11 2.2.2 Cash flow variability and cash substitutes 15 2.2.3 Other considerations 17 2.3 Cash measures 20 2.3.1 Cash holdings (CASH) 20 2.3.2 Free cash flow (FCF) 20 2.3.3 Modified free cash flow (MFCF) 21 2.3.4 Accounting cash flow (ACCCF) 21 2.4 Sample selection 23 2.5 Cash hoarding behaviour of Australian companies 25 2.5.1 Aggregate level summary data 26 2.5.2 Industry level summary data 31 2.6 Identification of cash rich firms 37 2.6.1 Method 1: Firms in the top ten percent of any measures of cash 37 2.6.1.1 Characteristics of cash rich firms based on Method 1 38 2.6.2 Method 2: Level of excess cash from multivariate regressions 40 2.6.2.1 Development and estimation of the cash normal model 40 2.6.2.2 Identification of cash rich firms 57 2.6.2.3 Characteristics of cash rich firms 60 2.7 Identification and characteristics of robustly cash rich firms 74 2.7.1 Method 1: Firms in the top ten percent for all measures of cash 74 2.7.2 Method 2: Firms whose cash measures are all greater than 1.5 standard deviations of the predicted values 75 2.7.3 Characteristics of robustly cash rich firms 77 2.8 Conclusion 91 3. Essay 2: Cash richness, acquisitions, and internal governance 96 3.1 Introduction 96 3.2 Literature review 100 3.2.1 Free cash flow hypothesis and acquisitions 101 3.2.2 Bidder long-run takeover performance 108 3.2.3 Bidder takeover performance and internal governance 116 v
3.3 Hypotheses development 123 3.3.1 Free cash flow hypothesis and acquisitions 124 3.3.2 Cash rich bidders long-run performance and internal governance 127 3.4 Data sources and sample selection 131 3.5 Likelihood of takeover attempts and the level of excess cash 135 3.5.1 Methodology and variable definitions 136 3.5.2 Univariate results 139 3.5.3 Multivariate results 148 3.6 Bidder long-run takeover performance and the level of excess cash 156 3.6.1 Methodology and variable definitions 156 3.6.2 Univariate results 162 3.6.3 Multivariate results 170 3.7 Cash rich bidders long-run performance and internal governance 180 3.7.1 Methodology and variable definitions 180 3.7.2 Univariate results 185 3.7.3 Multivariate results 195 3.8 Conclusion 202 4. Conclusion 206 References 211 vi
List of Tables Table 2.1: Summary of prior studies on the determinants of cash holdings 10 Table 2.2: Definitions of operating cash flow and three proxies of free cash flow adopted in this thesis 23 Table 2.3: Impact of applying the sample selection criteria 25 Table 2.4: Table 2.5: Table 2.6: Table 2.7: Table 2.8: Table 2.9: Descriptive statistics for CASH, FCF, MFCF and ACCCF over the sample period 1992-1999 27 Comparison of sample descriptive statistics with those from three related studies 28 Correlation matrix of cash holdings and three measures of cash flow (Lower triangle: Pearson / Upper triangle: Spearman-rank) 30 Cash and cash flow measures by industry sector over the sample period 1992-1999 36 Number of firms in the top 10% of each cash measure by industry sectors over the sample period 1992-1999 39 Specification of the cash normal model for four different measures of cash richness 44 Table 2.10: Sample selection after removing financial services industries 45 Table 2.11: Sample descriptive statistics by industry sectors 48 Table 2.12: Table 2.13: Table 2.14: Pooled time-series cross-sectional regression of CASH on explanatory variables for each industry over the sample period 1992-1999 50 Pooled time-series cross-sectional regression of FCF on explanatory variables for each industry over the sample period 1992-1999 54 Pooled time-series cross-sectional regression of MFCF on explanatory variables for each industry over the sample period 1992-1999 55 Table 2.15: Pooled time-series cross-sectional regression of ACCCF on explanatory variables for each industry over the sample period 1992-1999 56 Table 2.16: Number and percentage of cash rich firm-years and number of cash rich firms based on four measures of excess cash 57 Table 2.17: Descriptive statistics for cash rich firm-years over the sample period 1992-1999 62 Table 2.18: Summary statistics for cash rich firm-years and the rest of firm-years 66 Table 2.19: Number of times a firm is identified as cash rich over the period 1992-1999 68 vii
Table 2.20: Table 2.21: Number and percentage of cash rich firms that remain cash rich after being first identified as cash rich in year 1992 70 Frequency of cash rich firms and cash rich firm-years by industry sectors for four measures of cash richness 73 Table 2.22: Number of times a robustly cash rich firm stays in the top ten percent of all four measures of cash over the sample period 1992-1999 75 Table 2.23: Degree of overlapping between four measures of cash for cash rich firms identified from the second method (i.e., the level of excess cash) 76 Table 2.24: A comparison of two groups of robustly cash rich firms 78 Table 2.25: Descriptive statistics of four measures of cash for robustly cash rich firm-years over the sample period 1992-1999 80 Table 2.26: Descriptive statistics of robustly cash rich firm-years across the sample period 1992-1999 81 Table 2.27: Descriptive statistics for robustly cash rich firm-years and the rest of sample firm-years 83 Table 2.28: Prior equity issuance by robustly cash rich firm-years and their delisting years 85 Table 2.29: The subsequent spending behaviour of robustly cash rich firm-years 88 Table 2.30: The merger and acquisition activities of robustly cash rich firm-years within 3 years after being identified as robustly cash rich 90 Table 3.1: Sample selection criteria for the analysis on long-run post-acquisition performance 132 Table 3.2: Sample selection criteria for the analysis on cash rich firms long-run performance and internal governance 135 Table 3.3: Number of acquiring firms and number of takeover announcements by industry sectors over fiscal years 1993-2000 141 Table 3.4: Table 3.5: Table 3.6: Table 3.7: Table 3.8: Summary statistics for bidder characteristics and the rest of sample firms 144 Summary statistics for deal characteristics over fiscal years 1993-2000 147 Correlation matrix (Lower triangle: Pearson / Upper triangle: Spearman-rank) 149 Logistic regression results where excess cash is based on cash holdings 152 Logistic regression results where excess cash is based on accounting cash flow 155 viii
Table 3.9: Table 3.10: Table 3.11: Summary statistics for the sample acquiring firms used in the long-run post-acquisition performance regression 163 Summary statistics for bidders with positive and negative excess cash 166 Cross-tabulations for method of payment and excess cash, and for single vs. multiple acquirers and excess cash 169 Table 3.12: Correlation matrix (Lower triangle: Pearson / Upper triangle: Spearman-rank) 170 Table 3.13: Multivariate regressions of bidder long-run post-acquisition performance on excess cash holdings (Excash) and other explanatory variables 172 Table 3.14: Relationship between method of payment and firm size 175 Table 3.15: Multivariate regressions of bidder long-run post-acquisition performance on excess accounting cash flow (Exacccf) and other explanatory variables 178 Table 3.16: Summary statistics for a sample of Excash rich bidders 187 Table 3.17: Summary statistics for a sample of Exacccf rich bidders 191 Table 3.18: Summary statistics for cash rich bidders and the rest of cash rich firms 192 Table 3.19: Correlation matrix for a sample of cash rich bidders (Lower triangle: Pearson / Upper triangle: Spearman-rank) 194 Table 3.20: Multivariate regressions of Excash rich bidders long-run post-acquisition performance and internal governance mechanisms 197 Table 3.21: Multivariate relation between focused acquisitions and internal governance variables 198 Table 3.22: Multivariate regressions of Exacccf rich bidders long-run post-acquisition performance and internal governance mechanisms 201 ix
List of Figures Figure 2.1: Median cash and cash flow measures over the period 1992-1999 27 Figure 2.2: Distribution of CASH for the sample period 1992-1999 58 Figure 2.3: Distribution of FCF for the sample period 1992-1999 59 Figure 2.4: Distribution of MFCF for the sample period 1992-1999 59 Figure 2.5: Distribution of ACCCF for the sample period 1992-1999 60 x
Acknowledgements In this thesis, I have gained valuable insights, direction and encouragement from several people whom I would like to thank. First, I thank Buddha. With His blessing, I am able to complete this thesis, which I never thought was possible. I also extend deep gratitude to Master Yuan Lai for his teaching and encouragement and for bringing happiness to my life. I am extremely grateful to my supervisors, Professor H. Y. Izan and Associate Professor Ray da Silva Rosa, for their guidance and tremendous dedication throughout the writing of this thesis. Their careful readings of numerous drafts of the thesis and detailed comments have immeasurably improved the quality of my work. I also wish to thank The University of Western Australia for the research support. I am especially indebted to my parents for their love and support. They have been instrumental in helping me complete this thesis. My father gave me the inspiration to pursue a PhD because of his love of learning. My mother also encouraged me to keep going when I encountered obstacles during the process of writing the thesis. I am also thankful to my sister, Michelle, with whom I have shared numerous joyful and sad moments. xi
CHAPTER 1 INTRODUCTION The modern corporation is characterised by a separation of ownership and control that generates potential conflict of interest between managers and shareholders. Berle and Means (1932, p. 40) argue that managers who make various investment and financing decisions on a day-to-day basis do not bear the full consequences and therefore are prone to make decisions that serve their own interests rather than shareholders interests. The consequent wealth losses suffered by shareholders are termed agency costs in the financial economics literature and they are a function of shareholders inability to monitor their agents (i.e., managers) perfectly. In their seminal paper on principal-agency theory, Jensen and Meckling (1976) contend that agency costs are particularly high in firms with free cash flow, defined as the cash flow left after the firm has invested in all available positive net present value (NPV) projects. Agency costs become more acute in firms with free cash flow because the excess cash shields managers from internal and external governance mechanisms that would otherwise constrain their freedom to act in ways contrary to their shareholders interests. For instance, having excess cash means that managers do not have to go to the debt or equity market for funds to finance new projects and so the discipline imposed by this requirement is loosened. Jensen (1986) builds on Jensen and Meckling (1976) by observing that the takeover market is a natural arena for the agency costs associated with free cash flow to manifest. Jensen s premise is that management value investments in operations more than 1
investments in financial assets (Lang et al., 1991, p. 317) because their perquisites increase more with the former type of investments. As a result, when all positive NPV projects have been exhausted, managers prefer to invest the free cash flow in negative NPV projects rather than to distribute the cash to shareholders. By keeping the free cash flow in the firm, managers can ensure that their power and control of resources are not under threat. Moreover, as managers compensation and perquisites are often an increasing function of firm size 1, managers have the incentives to expand the firm beyond its value-maximising size. It is this last point that is the basis for Jensen s argument that firms with free cash flow are more likely to make value-decreasing takeovers. The aim of this thesis is to test Jensen s free cash flow theory on the market for corporate control in Australia. The free cash flow hypothesis suggests that firms with cash flow in excess of that required to fund positive NPV projects face greater conflict of interest between shareholders and managers, thereby leading to higher agency costs. Accordingly, firms with high levels of free cash flow are more likely to make value-decreasing acquisitions. To test this, we need to first identify firms that are cash rich. Thus, in the first essay (Chapter 2), we identify four different ways in which firms cash richness can be measured and employ two approaches to identify cash rich firms, which are deemed to be firms with free cash flow. This essay is motivated by the fact that one difficulty in testing the free cash flow theory is the development of a measure that corresponds with Jensen s (1986) free cash flow definition. Thus, in addition to the cash holdings used by most of previous studies in the examination of free cash flow theory [e.g. Harford (1999); Pinkowitz (1998)], this study 1 Managerial compensation has been widely documented to be positively associated with firm size. See Ciscel and Carroll (1980) and Agarwal (1981). 2
employs three flow measures based on operating cash flows. The three flow measures are free cash flow, modified free cash flow, and accounting cash flow. The free cash flow (FCF) is defined as the operating cash flow less ordinary dividends, preference dividends, and capital expenditures divided by total assets. The modified free cash flow is defined as operating cash flow less preference dividends divided by total assets; that is, FCF with ordinary dividends and capital expenditures added back and divided by total assets. The accounting cash flow is defined as earnings after interest paid, tax paid and dividend paid but before depreciation, divided by total assets. Using multiple measures allows us to assess the robustness of our findings and indeed the different measures. There is no consensus on the most appropriate measure so an important contribution of this thesis is documenting whether the choice of measure of free cash flow is a significant issue. Another motivation for the first essay is that there is a lack of understanding about the cash hoarding behaviour of Australian firms and the extent to which the agency costs of free cash flow are prevalent in Australian companies. Thus, we first present the level of cash holdings and cash flow held by ASX-listed companies. From this, we identify companies that are cash rich and provide in-depth analyses of the aforementioned issues on firms identified as robustly cash rich. Robustly cash rich firms are firms that are cash rich under all four measures of cash. The second essay (in Chapter 3) tests the free cash flow theory directly by focusing on firms takeover decisions to see if agency costs of free cash flow are a problem in Australian companies. According to Jensen s (1986) free cash flow hypothesis, firms endowed with free cash flow are not only more likely to make acquisitions but the acquisitions made are more likely to be value-decreasing. Therefore, the first two 3
research questions of the second essay examine the relationship between free cash flow and the likelihood of a takeover attempt by publicly listed companies and the relationship between free cash flow and long-run takeover performance of bidders. In this essay, we adopt two measures of cash richness, excess cash holdings and excess accounting cash flow, as the analyses from the first essay show that these two measures are most different; specifically, the comparison between the lists of cash rich firms identified using the four different cash measures shows that cash holdings and accounting cash flow measures have the lowest number of the same cash rich firms. The context of takeover also provides a good setting for investigating the impact of different governance arrangements on the wealth of shareholders in firms with large free cash flow. Thus, to provide a stronger test on the effectiveness of internal governance in mitigating the conflict of interest between managers and shareholders, the third research question of the second essay tests if the existing governance structures of firms with large free cash flow are effective in protecting shareholders interests. Specifically, we examine the relationship between corporate governance structures and the long-run takeover performance of cash rich bidders, which arguably face greater agency costs of free cash flow compared to other bidders. The third research question is motivated by the fact that apart from the studies by Gibbs (1993) and Harford (1999), the relationship between corporate governance and the free cash flow hypothesis has not been extensively researched. To extend on Harford s (1999) study that considers only the level of insider ownership, additional internal governance variables, including board composition and board size, are examined in the second essay. This study finds that based on the cash holdings reported at the balance date, Australian firms hold, on average, 14.2% of their total assets as cash over the period between 1992 4
and 1999. This result is much higher compared to US companies. Kim, Mauer and Sherman (1998) report that US companies between 1975 and 1994 hold on average 8.1% of total assets in cash. Thus, Australian companies appear to hold more cash on average. While most of prior studies focus on the stock measure of cash only, our analyses show that the stock and flow measures of cash are in fact negatively correlated and give quite different lists of cash rich firms. This result thus provides an important empirical contribution to the literature as it suggests that the results of prior studies may not be robust. Further, our analyses show that cash rich firms tend to cluster in certain industries depending on the measure of cash. For example, the gold industry, which represents the largest proportion of Australian firms, is found to have the highest percentage of cash rich firms based on the cash holdings measure while having a relatively low percentage of cash rich firms based on the free cash flow measure. Based on the two methods used to identify robustly cash rich firms, a total of 33 firm-years with overlapping firm-years counted just once (or 26 firms) are found to be robustly cash rich over the sample period 1992-1999, and 13 firm-years (or 39% of the combined total of 33 firm-years) are robustly cash rich under both methods. Over half of the robustly cash rich firms are found to have issued equity in the previous three years to being identified as robustly cash rich, indicating that prior equity issuance may explain why these firms are robustly cash rich. These 33 robustly cash rich firm-years also show significantly stronger growth opportunities, proxied by the market-to-book ratio, compared to the rest of sample firms, indicating that the cash is likely to have been reserved for financing future investment opportunities. An examination of the spending behaviour of robustly cash rich firms shows support for the prediction. Most 5
of these firms increase their capital expenditures in the subsequent year to identified as robustly cash rich and about half of the robustly cash rich firms have made takeover attempts within three years after being identified as robustly cash rich. Thus, the results appear to provide support for the free cash flow hypothesis. However, it is necessary to investigate if the subsequent investments made are value-decreasing as suggested by the free cash flow theory. Direct tests of the free cash flow theory are thus carried out in the takeover market. The test on the relationship between the level of excess cash and firms takeover decisions shows that the level of excess cash holdings is significantly negatively related with the probability of a takeover attempt but has no significant explanatory power for the cross-sectional variation in long-run post-acquisition returns. On the other hand, we find that firms possessing high levels of excess accounting cash flow are more likely to make takeover attempts and that bidders with higher excess accounting cash flow have better long-run post-acquisition performance. Therefore, contrary to Jensen s (1986) free cash flow hypothesis, acquisitions made by bidders with high levels of excess accounting cash flow are not value-decreasing. The test on the internal governance structures of cash rich bidders show that based on the cash holdings measure, none of the internal governance variables have any explanatory power for the cross-sectional variation in the long-run performance of cash rich bidders. However, when the level of cash is measured by the accounting cash flow, the proportion of non-executive directors on a board is significant and positively related with cash rich bidders long-run returns, showing support for the monitoring role of non-executive directors. 6
Overall, although Australian firms have, on average, higher levels of cash holdings compared to US companies, the evidence from this thesis shows no support for the free cash flow theory of takeover in the Australian market. The organisation for the remaining chapters of this thesis is as follows. Chapter 2 contains the first essay on the cash hoarding behaviour of Australian firms, where we adopt four different measures of cash and two approaches to identify cash rich firms and robustly cash rich firms. In Chapter 3, we present the second essay cash richness, acquisitions, and internal governance, where we test the free cash flow theory in the takeover context. Specifically, the likelihood of takeover attempts and bidder long-run performance in relation with firms cash richness are tested. The effectiveness of cash rich bidders internal governance is also examined. Lastly, in Chapter 4, a summary of results from the two essays is drawn. Conclusions and implications from our findings are also provided. 7
CHAPTER 2 ESSAY 1: THE CASH HOARDING BEHAVIOUR OF AUSTRALIAN FIRMS 2.1 Introduction The discretion that managers have in employing liquid assets has been the subject of much debate as liquid assets held by corporations are large in total and often constitute a substantial portion of a firm s total assets. Schwetzler and Reimund (2004) report that in 1998, the world s largest firms held 1.5 trillion USD in cash and marketable securities. Harford (1999) also finds that in many well-known companies such as IBM and Chrysler, 20 percent or more of their equity value is represented by cash. The world s largest software company, Microsoft, held 70% of its total assets (or 9 billion US dollars) in cash and short-term investments in 1997 (Bank, 1997). We find that over the period between 1992 and 1999, Australian companies held, on average, 14.2% of their total assets as cash. One explanation for the observed high level of corporate cash holdings in Australia is the high proportion of resources firms. Firms in the resources sector, including gold, other metals and diversified resources industries, represent 26.2% of all Australian listed firms over the sample period. As resource firms often have high start-up and development costs, they have greater incentives to hold more cash. Apart from the differences in industry mix, why do Australian firms hold such large amounts of cash? From a shareholder s perspective, are large cash balances value-adding or value-destroying? As discussed earlier, agency theory implies that cash held in excess of normal requirements is value-decreasing; however, an alternative 8
perspective is that excess cash is a value-increasing response to capital market imperfections, in particular, the information asymmetry between the market and the firm (Myers & Majluf, 1984). In order to avoid high transaction costs associated with raising external capital (Myers, 1977) and the costs of financial distress, firms may hold excess cash so that projects may be funded internally. To test which of the two perspectives is most relevant to the Australian market requires that we determine the cash generating and hoarding behaviour of Australian companies and review their possible reasons. This is the first objective of this essay. The second objective is to identify those firms that are cash rich and thus most likely to be free cash flow firms in Jensen s sense, i.e., firms that are likely to make value-decreasing capital investments for shareholders. This essay proceeds as follows. First, the literature on why firms have different cash holdings is reviewed with a particular focus on the market imperfections-based explanations. These explanations motivate our choice of variables when developing our models to predict normal cash holdings and cash flow. Secondly, four measures of cash are developed and the sample selection procedures are discussed. Thirdly, descriptive statistics on aggregate and industry levels of cash holdings and cash flow are presented. Finally, cash rich firms and firms that are robustly cash rich are identified and analysed. 2.2 Literature review: What explains differences in cash holdings? This review begins with a brief elaboration of the market imperfections theory on firms cash holdings. We then provide a review of the variables that have empirically been found to be significant in explaining firms cash holdings, with a view to developing a model for determining the level of cash held by firms in Section 2.6.2. Table 2.1 summarise the discussion that follows. 9
Table 2.1: Summary of prior studies on the determinants of cash holdings Results of prior studies investigating the determinants of cash holdings are summarised below. A tick,, under the Use column indicates that a particular variable has been incorporated in the study. Asterisks indicate significance at the 1% (***), 5% (**) and 10% (*) levels. I represents statistical insignificance. S means statistically significant. The signs, + or -, are not necessarily predicted signs; they are actual results. Variables 1 Kim et al. (1998) (2003) Ozkan & Ozkan Faulkender (2003) 6 Harford (1999) Opler et al. (1999) 10 Use Sig Sign Use Sig Sign Use Sig Sign Use Sig Sign Use Sig Sign Information asymmetry Firm size I - I - *** - S - Firm age ** + Growth opportunities *** + *** + S + (MTB) Anticipate future credit I + problem Supply of cash Firm is publicly traded I + Firm didn t apply for *** - the credit in the past when needed Bad credit marks *** - Cash flow variability Cash flow variability 2 I + I - S + Cash cycle variability I - Cash flow Cash flow 3 *** - *** - S + Cash flow t+1 Cash flow t+2 Average cash cycle *** - Investment opportunities Return spread *** - Forecast of Future *** + Economic Conditions Recession Financial distress costs Total debt ratio *** - *** - S - Bank debt *** - *** + Bankruptcy predictor I - ΔRisk premium t+1 R&D expenditures * 8 / + S 12 + I 9 Capital expenditures S + Total Expenses *** + Ownership structure Managerial ownership 4 I - Controlling shareholder 5 ** + 7 I - Largest shareholder is I - the manager Cash substitutes Net working capital ** - *** + S - (Liquidity) Firm has a line of credit *** - Dividends * + S - Taxes Firm is a C-Corp I + Industry Industry classification *** 11 S - 1 Only results from pooled time-series cross-sectional regression are reported here. 2 Two alternative measures are used. One is cash flow variability and the other is free cash flow variability. 3 Two alternative measures are used. One is cash flow and the other is free cash flow. 4 Ozkan and Ozkan (2003) have also considered the non-monotonic relationship between managerial ownership and cash holdings by incorporating the square and cube of percentage of directors shareholdings. The square of percentage of directors shareholdings is found to be significantly positively related to firms cash holdings at the 5% level. On the other hand, the cube of percentage of directors shareholdings is found to be negative and significant at the 5% level. 5 Dummy variable Presence of a controlling shareholder. 6 The results reported here are based on the ordinary least squares regression with the ratio of cash-to-sales as the dependent variable. 7 Percentage of shares owned by the controlling shareholder. 8 Percentage of employees doing R&D 9 Dummy variable whether the firm does R&D. 10 The results reported here are based on the Fama-MacBeth (1973) model. 11 Opler et al. (1999) control for the industry effect using a dummy variable, one if the firm is in a regulated industry and zero otherwise. 12 R&D expenditures-to-sales. 10
2.2.1 Asymmetric information and financial distress costs In their seminal paper, Myers and Majluf (1984) note that asymmetric information between firms and investors makes it costly for firms to raise external finance because investors are not able to distinguish high quality firms from low quality firms. One consequence of this is that investors demand a higher return to protect against investing in lemons. This theory underpins Myers (1977) pecking order hypothesis which states that firms prefer to use internal cash rather than external finance when funding new projects. The preference for using internal cash is stronger for firms that face higher information asymmetry between themselves and the investors. If the information asymmetry problem is very severe, firms may choose to forgo positive NPV projects if they cannot finance them internally. In these circumstances, firms that hold more cash than is required for their current needs may enhance the value of the firm since they run less risk of passing up positive NPV projects. Financial distress costs such as bankruptcy costs and losses in asset values are another source of market imperfections that reduce the attractiveness of external financing via the debt market and give firms incentives to hold more cash. Higher levels of cash holdings reduce the likelihood of missing a debt repayment or experiencing financial distress. Thus, firms with high financial distress costs are more likely to increase their cash holdings (Faulkender, 2003). How valid is the market imperfections explanation for firms cash holdings? The theory cannot be tested directly since we lack direct measures of information asymmetry and the costs of financial distress. However, researchers have used proxies for these constructs and their findings are reviewed next. 11
Firm size Large firms are argued to have less information asymmetry problems than small firms (Brennan & Hughes, 1991) in part because there are more analysts following large firms than small firms. As large firms have more borrowing capacity relative to small firms and enjoy economies of scale in raising external capital, the cost of external financing is smaller for large firms. Large firms are also less likely to face financial distress because they are likely to be diversified and can sell off non-core segments when they face a liquidity shortage. Hence, they are expected to hold lower levels of cash than small firms. On the other hand, agency theory predicts that firm size is positively related to managers propensity to hold excess cash because the excess cash allows managers to avoid external funding and market scrutiny (Opler et al., 1999). Further, since large firms tend to have dispersed ownership and the large firm size can deter unwanted takeovers, agency theory predicts a positive relationship between firm size and firms cash holdings. However, contrary to this prediction, both Faulkender (2003) and Opler et al. (1999) find a negative relationship between firm size and the level of firms cash holdings. Growth opportunities Myers and Majluf (1984) argue that firms whose value depends on growth opportunities are more vulnerable to asymmetric information problems. This is because growth firms constantly need cash injections to fund profitable projects. Since information asymmetry makes external financing more expensive than internal financing, firms with greater investment opportunities are expected to hold more cash to avoid missing out on valuable investment opportunities. In line with Myers and Majluf (1984), Kim et al. 12
(1998), Ozkan and Ozkan (2003), Faulkender (2003) and Opler et al. (1999) all find a positive relation between growth opportunities and cash holdings. That is, firms with higher growth opportunities hold more liquid assets. Total debt ratio The ratio of total debt to total assets is sometimes used as a proxy for the cost of financial distress because when the debt ratio is large, it implies that further debt financing is expensive (Baskin, 1987) and that the firm has accessed the debt markets already (Faulkender, 2003). For both reasons, firms are likely to hold less liquid assets. By the same argument, a firm with a low level of debt may hold more cash because it is less subject to the discipline of financial markets. On the other hand, firms leverage and cash holdings may be positively related because firms with high levels of debt may hold more cash to reduce financial distress (Faulkender, 2003). This positive relationship is also expected because of the asymmetric information between creditors and managers regarding firms activities. Firms with a large amount of debt may find it difficult to raise additional funds from the debt market as creditors will demand higher compensation for the perceived higher default risk involved. Thus, firms with high agency costs of debt may in fact hold more liquid assets to finance value-enhancing projects. Using total debt ratio as a measure of financial distress costs, Kim et al. (1998), Ozkan and Ozkan (2003) and Opler et al. (1999) all find support for the first view that leverage and level of cash holdings are negatively related. That is, firms with high levels of debt tend to hold less cash. Ozkan and Ozkan (2003) also incorporate another measure, total bank borrowings to total debts, in the model and again find a negative relation. 13
However, in contrast to the results from other studies [e.g. Kim et al. (1998), Ozkan and Ozkan (2003) and Opler et al. (1999)], Faulkender (2003) whose study focuses on small firms finds a positive relation between leverage and cash holdings using the ratio of outstanding loan balance to sales as a proxy. The study by Faulkender (2003) therefore supports the second view instead; that is, firms will increase their cash holdings as their borrowing increases to avoid financial distress. One explanation for the divergent results is that small firms have smaller borrowing capacity and are more susceptible to financial distress than large firms and thus are more inclined to increase their cash holdings as their debt increases. R&D expenditures R&D expenditures have also been used as a measure of firms financial distress costs because R&D typically involves large cash outflows (Faulkender, 2003). R&D can be an important part of investment expenditures, and funds are likely to be tied up when firms are engaged in large scale R&D projects. Thus, firms with large future R&D expenses are likely to hold more liquid assets to avoid financial distress. Although Faulkender (2003) and Opler et al. (1999) adopt different measures of R&D expenditures, both report a positive relationship between R&D and firms cash holdings, consistent with expectation. In the former study, two proxies, including the percentage of employees doing R&D and whether the firm has R&D, are used to measure R&D expenditures while the latter study uses the ratio of R&D expenditures to sales. However, there is a potential multicollinearity problem in these two studies. In Faulkender s (2003) study, both R&D expenditures and total expenses are included in the same model to proxy for financial distress costs. Similarly, in Opler et al. (1999), 14
they incorporate both R&D expenditures and capital expenditures in the same model. As in Australia R&D expenditures are not identified in the notes of the financial statements, in this essay, we adopt only capital expenditures in the model, which also avoids the multicollinearity problem. 2.2.2 Cash flow variability and cash substitutes Cash flow variability measures how likely firms will experience a shortage of cash, which can be costly as they may have to forgo profitable investments. Minton and Schrand (1999) argue firms that constantly experience cash flow shortfalls face higher costs of external financing because of market imperfections. Hence, firms with greater cash flow variability are expected to hold more cash to avoid unexpected cash constraints. The cash flow itself provides a ready source of liquidity. Thus, firms with a high level of cash flow can afford to hold less cash. Moreover, firms that pay dividends may need to maintain a higher level of cash in order to meet dividend payments in times of liquidity constraints. The evidence on whether cash flow variability is associated with cash holdings is reasonably strong though not overwhelmingly so. Kim et al. (1998) and Opler et al. (1999) find that cash flow uncertainty is significantly associated with larger liquid assets holdings by firms. However, Ozkan and Ozkan (2003) report a negative but insignificant relation between cash flow variability and cash holdings. One explanation for Ozkan and Ozkan s (2003) finding is that firms may use alternative methods to raise funds when facing cash shortfalls. These alternatives include borrowing and converting non-cash liquid assets into cash, for example, by liquidating receivables. To the extent that non-cash liquid assets can provide another source of fund, 15
firms with high levels of liquid assets besides cash can afford to hold less cash. The evidence on this is discussed next. Cash flow from operations Cash flow from operations provides a source of liquidity for firms to meet their debt payments and operating expenditures (Kim et al., 1998). Therefore, firms with high levels of operating cash flow can hold a lower level of cash holdings. Consistent with this proposition, both Kim et al. (1998) and Ozkan and Ozkan (2003) report a significant negative relationship between the size of operating cash flow and cash holdings. Dividends The inclusion of dividends controls for firms varying dividend policies. Dividend payouts are expected to have a positive relation with cash holdings because dividend-paying firms are likely to reserve high levels of cash in case of cash shortfalls that would cause them to miss out on dividend payments. One possible counter argument is that firms may cut back dividends when they are cash deficient, thereby suggesting a negative relationship between dividend payout and cash holdings. However, it is well-known that dividend payout is sticky because firms are reluctant to cut back dividends and the share-markets tend to react negatively to firms announcements of a dividend cutback. In any event, Ozkan and Ozkan (2003) report a marginally significant positive relation between dividends and cash holdings. Net working capital Net working capital excluding cash provides a measure of the ease of converting non-cash liquid assets into cash. Firms with more non-cash liquid assets can convert 16
them into cash when there is a shortage of cash and thus can hold less cash. Accordingly, a negative relationship between the balance of working capital and the level of cash holdings is expected. Studies of large public firms by Opler et al. (1999) and Ozkan and Ozkan (2003) confirm the above prediction. However, Faulkender (2003), who focuses on small firms, reports a positive relationship between the balance of net working capital and the level of cash holdings. Faulkender (2003) argues that for small firms, a high working capital may indicate a low cash conversion cycle and hence small firms are likely to hold more cash to avoid a cash shortage when working capital is large. 2.2.3 Other considerations Other potential determinants of corporate cash holdings identified in the literature include the presence of controlling shareholder, the nature of a firm s industry effect, whether the firm is publicly traded, and the firms past credit history. Each of these is discussed below. Controlling shareholder Shleifer and Vishny (1986) argue that large shareholders are more likely to exert control over managers than dispersed small shareholders because the former have economies of scale in monitoring. 2 Accordingly, large shareholders can monitor managers more effectively than smaller shareholders and force managers to distribute excess cash. As for small shareholders, the costs of monitoring are likely to outweigh the benefits. Therefore, the presence of large shareholders is likely to reduce the agency costs of free cash flow. One implication is that firms with large shareholders may be perceived as 2 La Porta et al. (1999) document that Australia is not characterised by controlling structures. As a result, this variable will not be included in the cash normal model for identifying cash rich firms. 17
higher quality firms, thereby lowering the costs of external financing and reducing the need to hold large amounts of liquid assets. In sum, a negative relationship between the largest shareholder ownership and cash holdings is expected. However, one counter argument is that controlling shareholders may use their ownership rights to enjoy private benefits of market power at the expense of minority shareholders wealth (Faccio et al., 2001). For example, controlling shareholders may prefer stockpiling a large amount of cash to expand the firm beyond its optimal size because large firms are associated with more market power. If this is the case, a positive relationship between the presence of large shareholders and cash holdings is expected. Ozkan and Ozkan (2003) use a dummy variable to measure the presence of an ultimate controller with at least ten percent of share ownership in the firm and report a significant positive relation between the presence of controlling shareholders and the level of firms cash holdings. Ozkan and Ozkan (2003) interpret their results as supporting the argument that controlling shareholders sustain their market power by accumulating cash. In contrast, Faulkender (2003) uses the percentage of shares held by the largest shareholder to measure controlling shareholder ownership and finds an insignificant negative relationship with firms cash holdings. The differences in results may be due to the fact that different proxies for controlling shareholder have been used. Industry Damodaran (1997) argues that differences in firms demand for cash across industries reflect their differing nature of business operations and financing demands. Harford (1999) documents a substantial variation in the level of cash across industries; the cash to sales ratio ranges from 0.031 in the wholesale industry to 0.325 in the financial 18
services industry. Opler et al. (1999) find that companies with strong growth opportunity industries tend to hold more cash than those in low growth industries, consistent with the proposition that firms hold cash to invest in positive NPV projects as they arise. Several studies, including Kim et al. (1998), Faulkender (2003), Opler et al. (1999) and Harford (1999), have controlled for the industry effect. The first two studies group firms according to their industry categories and use industry dummy variables to control for industry-specific effects on cash holdings. Opler et al. (1999), on the other hand, control for the industry effect by employing a regulation dummy variable one if the firm is in a regulated industry and zero otherwise. In Harford s (1999) study, the industry effect is taken into consideration by dividing firms into industry groups so that the baseline cash holdings for each industry can be estimated separately. Public status and past credit history In addition to examining firms demand for cash that most studies focus on, Faulkender (2003) also examines factors that affect firms supply of cash. Three additional variables are incorporated in the model to proxy for the supply effect of cash. These variables include whether the firm is publicly traded, whether the firm applies for credit in the past when needed, and whether the firm has past credit problems. These variables capture different dimensions of asymmetric information problems. Results from the supply side of cash holdings show that firms with poor access to capital either because of poor credit history or the fear of being rejected hold relatively less cash. 19