Do Firms Set Free Cash Free? Misallocation vs. Opportunism in Acquisitions Lee Pinkowitz, Jason Sturgess, and Rohan Williamson* March 2010 Abstract We examine the impact of excess cash on the decision and method of payment in acquisitions. Consistent with prior literature, we find that firms that have excess cash do make more acquisitions. However, those acquisitions frequently use stock as the method of payment. We investigate this finding further and show that cash-rich firms use overvalued equity to make acquisitions. Moreover, there is no evidence that firms waste excess cash on acquisitions; cashrich firms that pay with cash actually acquire undervalued firms. Our results suggest that with respect to acquisitions, managers are opportunistic with the firm’s financial resources; both its equity and its cash. Keywords: Cash holdings, method of payment, acquisitions ______________________ *Respectively, Associate Professor and Beit Research Fellow, Georgetown University, Assistant Professor, Georgetown University, and Associate Professor and Stallkamp Research Fellow, Georgetown University. We thank seminar participants at George Mason University and Georgetown University. We also thank the Capital Markets Research Center for Research Support. Do Firms Set Free Cash Free? Misallocation vs. Opportunism in Acquisitions One of the primary questions addressed in the finance literature is whether or not managers act in the best interest of their shareholders. In particular, mergers and acquisitions are a key avenue for exploring Jensen’s (1986) free cash flow problem, which argues that firms with free cash flow will make poor investments. Harford (1999) finds evidence that appears consistent with this argument by showing that high-cash firms make more acquisitions than other firms, and that those acquisitions have lower announcement returns. Perhaps naturally, it is assumed that acquisitions of the type discussed by Harford are made using the firm’s cash. Yet we observe firms that have high levels of cash holdings, but use stock as the method of payment. For instance, on August 8th, 2003, USA Interactive successfully acquired Lending Tree Inc for nearly $730 million. Despite having almost $4 billion in cash and marketable securities, USA Interactive paid for the deal solely with equity. Using a sample of 58,117 firm years from 1984-2005, we find 5,968 firm years where at least one bid occurred. We confirm the results in the extant literature by showing that cash-rich firms are more likely to make acquisitions. However, we document that being cash-rich does not make a firm more likely to use cash as the method of payment. In fact, we find higher levels of cash in the 1,354 bidders who paid solely with equity, than the 3,517 bidders who paid solely with cash. This suggests that the USA Interactive example is more the rule than the exception and raises the question: Why might highcash firms resort to using stock in the acquisition? In this study we address this question and, in so doing, revisit whether the acquisition decisions of high-cash firms destroy shareholder value. The finding that high-cash firms use equity as the method of payment can be consistent with either self-interested managers squandering shareholder value, or managers acting in the best interests of existing shareholders. If one believes the former, as implied by Jensen (1986), firms with excess cash flow will waste the cash on negative NPV investments; which naturally extends to investment in negative NPV acquisitions. However, Jensen’s argument need not imply that cash-rich firms use cash to pay for acquisitions. Instead, a self-interested manager may choose to use stock because she values continued 1 flexibility and freedom from capital market discipline (Easterbrook (1984) and Jensen (1986)). Thus, a self-interested manager may choose to use stock to make acquisitions and reserve the cash for other value destroying investments. Alternatively, if managers act in the interest of shareholders, they should opportunistically use equity to make acquisitions when the stock is overvalued.1 This line of argument borrows from a rich literature examining the role of overvalued stock in acquisitions.2 Particularly relevant to this evaluation are Rhodes-Kropf and Viswanathan (2004) and Shleifer and Vishny (2003) who argue that overvaluation is one of the key drivers in merger waves. The use of “cheap” stock when cash is available is consistent with the mechanisms described in Stulz (1990) and Myers and Majluf (1984), where managers stockpile excess cash to prevent future underinvestment due to costly external funds. Survey responses in Graham and Harvey (2001) support this, while Mikkelson and Partch (2002) provide empirical evidence that managers hold excess cash while acting in the best interest of shareholders. They find that the operating performance of high-cash firms is comparable to or greater than the performance of firms matched by size, industry, or prior performance. If such managers are better able to identify acquisition opportunities, we should observe the empirical association between high cash and takeover activity. The two explanations have distinct empirical predictions. The first – which we label the misallocation hypothesis – predicts the likelihood of cash bidding increases with cash levels and acquisitions will be negative-NPV investments. It also predicts that managers will acquire overvalued targets and are just as likely to pay with undervalued equity as overvalued equity. As such, cash-rich firms will have lower announcement returns, regardless of their method of payment. The second hypothesis – the opportunism hypothesis – implies that managers act opportunistically with both the firm’s cash and its equity. Under the opportunism hypothesis, even if the likelihood of bidding increases with cash levels, acquisitions will be positive NPV investments. 1 This While the existing discussion focuses on overvalued equity, there are other reasons using stock as the method of payment can benefit existing shareholders. For instance, Hansen (1987) argues that equity is beneficial when information asymmetry is high as it leads to risk sharing between the bidder and target. 2 See Rhodes-Kropf, Robinson, and Viswanathan (2005), Dong, Hirshleifer, Richardson, and Teoh (2006), Ang and Cheng (2006), and Fu, Lin, and Officer (2008) for example. 2 hypothesis also predicts that the use of stock increases with the overvaluation of the acquirer. Lower announcement returns will be limited to deals where the firm uses equity as the method of payment, thus signaling overvaluation. To determine which mechanism is at play, this paper uses the methodologies employed in Harford (1999) and Opler et al (1999) to identify firms with excess cash, and combines these with the methodology in Rhodes-Kropf, Robinson and Viswanathan (RKRV) (2005) to identify overvalued firms. Specifically, we examine the valuations of bidders and targets, for both cash-rich and non-cash-rich firms. With this approach, we are able to determine whether acquisitions by cash-rich firms are in the best interests of shareholders, or if they decrease shareholder value. Consistent with the prediction of managerial opportunism, firms with excess cash use stock to make acquisitions when their equity is overvalued - rather than spending their excess cash. Though announcement returns are negative for acquisitions by cash-rich firms, the negative returns are isolated to the deals where equity is the method of payment. Further, we document that when cash-rich firms pay with cash, they actually pursue undervalued targets, and do not overpay. The findings in this study are consistent with managers being opportunistic with the firm’s financial resources; both its equity and its cash. Managers appear to act in the best interest of shareholders by using the cheapest source of capital to fund acquisitions. Our results suggest that acquisitions are not a mechanism that managers use to waste firm cash holdings.3 The rest of the paper is organized as follows. Section 1 discusses the firm’s excess cash holdings and the role it plays in acquisitions along with a motivation of the hypotheses to be tested. Section 2 describes the data. Section 3 discusses the results on the impact of liquid assets on acquisitions. Section 4 discusses and tests possible explanations. Section 5 examines robustness, and section 6 concludes. 1. Cash Holdings and Acquisitions The finance literature has generally approached the impact of cash holdings on acquisitions and the method of payment from two perspectives. Jensen (1986) argues that firms with free cash flow will tend 3 However, we do not entirely rule out over-investment in the spirit of Jensen (1986). Though managers do not use acquisitions as a channel for inefficient behavior, they could waste free cash flow elsewhere (i.e. instance capital expenditures, research and development, perquisite consumption, or simply inefficient operation of the firm). 3 to waste it, and acquisitions are one mechanism for doing so. The assumption in Jensen (1986) is that self-interested managers will potentially undertake acquisitions which are negative NPV investments. The stylized assumption of this literature is that these acquisitions will be made with the excess or free cash flow of the firm; hence, limiting cash available to managers protects shareholders. Supporting this idea, Harford (1999) shows that firms with excess cash tend to make more acquisitions, and those acquisitions have lower announcement returns. Additionally, Harford et al (2008) show that managers of firms with weaker corporate governance use cash to overinvest. An alternate stream of literature is ambivalent on whether managers act in the interest of shareholders; rather it examines the role of method of payment in acquisitions. Specifically, the choice of method of payment will be determined by transaction characteristics related to investment opportunities, market valuations, and risk sharing.4 Myers’ (1984) pecking order theory argues that firms will finance investments in a manner that is cheapest and which maintains the financial flexibility of the firm. Focusing on acquisitions, Hansen (1987) shows that a bidder may prefer to pay with stock in place of cash if the target has better information about its investment opportunities than does the bidder. In this case, using stock would force the target shareholders to share the risk of a revaluation of its investment opportunities after the acquisition has been made. Martin (1996) empirically examines the method of payment in acquisitions and finds that higher acquirer growth opportunities, greater pre-acquisition acquirer returns, and greater pre-acquisition market returns increase the likelihood of stock being used as the method of payment. Martin also examines whether ownership structure affects method of payment. More recently, studies have claimed that the value of both acquirer and target equity are important. Both Rhodes-Kropf and Viswanathan (2004) and Shleifer and Vishny (2003) argue that the method of payment in mergers is driven by the relative valuations of the stock of the acquirer and the target. RhodesKropf and Viswanathan (2004) appeal to incomplete information, while Shleifer and Vishny (2003) apply a more behavioral approach; but each reaches the conclusion that acquirer overvaluation drives the 4 A small literature has also examined the influence of ownership. Stulz (1988) and Jung, Kim, and Stulz (1995) argue that managers will be reluctant to use stock to finance acquisitions if doing so will dilute their control and lead to outside intervention. Martin (1996) finds weak supportive evidence. He shows that acquirer managerial ownership is (negatively) related to the probability of stock over a middle range of ownership only. 4 decision to bid as well as the method of payment. In both models, acquirers with overvalued equity will more likely pay with stock and will likely make acquisitions of overvalued targets. The main difference in the two streams of the literature is the perceived motivation of management. One takes an agency cost approach where management is acting in its own best interest while the second assumes that managers are acting in shareholders’ best interest. We take these two perspectives to motivate the hypothesis to be investigated in this paper. 1.1. Method of Payment and Agency The arguments in Jensen (1986) and Harford (1999) imply an agency problem where managers engage in empire building (Jensen and Meckling (1976)) at the expense of investing in positive-NPV projects. While this suggests that managers would use their free cash as the method of payment in an acquisition, management may have other objectives which can play a role in their decision on how to pay. Self-interested managers would prefer to maintain freedom from future capital market discipline. If management’s objective were to maintain financial independence, in addition to empire building, then this could be met through using stock for acquisitions and reserving cash to spend post acquisition or on perquisite consumption, as in Harford et al (2008). As such, high cash would be a predictor of acquisitions even if it were not the method of payment. These arguments result in our first hypothesis. Misallocation Hypothesis: Self-interested managers have an incentive to misallocate the financial resources of the firm; both its cash and its equity. The likelihood of cash bidding increases with cash levels and acquisitions will be negative-NPV investments. Managers will acquire overvalued targets and are just as likely to pay with undervalued equity as overvalued equity. Cash-rich firms will have lower announcement returns, regardless of their method of payment. 1.2. Method of Payment, Opportunistic Managers and Stock Acquisitions If managers act in the best interest of shareholders they should choose the method of payment opportunistically. Consistent with this, both Rhodes-Kropf and Viswanathan (2004) and Shleifer and 5 Vishny (2003) argue that managers acting in the best interest of shareholders should prefer to use overvalued stock as the method of payment. Analogously, the arguments in Baker and Wurgler (2002) on market timing and firm capital structure claim managers use the cheapest funds for financing investments. Further, as predicted Myers and Majluf (1984) and by pecking order, financial flexibility allows the firm to take advantage of merger synergies, and fund future positive NPV projects without accessing the capital markets. Graham and Harvey (2001) report that managers prefer financial flexibility as well as attempt to time the market with security issues. Hence, a firm where management acts in the interest of shareholders by using overvalued stock to make acquisitions, while maintaining financial flexibility, will exhibit both high levels of cash and stock acquisitions. Finally, as argued in Shleifer and Vishny (2003), target owners may prefer overvalued equity since they may receive more for the firm than they might receive in a cash offer and thus gain from the transaction. This implies that even where the target is overvalued, bidder shareholders can still benefit by using relatively more overvalued stock for the acquisition. These arguments lead to our next hypothesis. Opportunism Hypothesis: Managers acting in the best interests of shareholders are opportunistic with both the firm’s cash and its equity. Managers make positive-NPV acquisitions and hold excess cash to achieve financial flexibility for the firm. Managers prefer to use stock as the method of payment only when the firm’s stock is overvalued, or relatively overvalued. Lower announcement returns will be limited to deals where the firm uses equity as the method of payment. 2. Data 2.1. Financial Data Financial data are drawn from the Compustat database. All publicly traded firms with sales, assets, and market capitalization (in year 2000 dollars) greater than $5 million, stock price greater than $1, and non-missing data in the period 1984 to 2005 are included in the sample. As our focus is on the cash holdings of firms, we construct several measures of corporate liquidity. First, we use raw cash holdings deflated by assets. Second, we construct three different measures of 6 abnormal cash holdings: residuals from regressions similar to those in Harford (1999), residuals from regressions similar to those in Opler, et al (1999), and deviation from the industry average. The appendix describes the construction of the excess cash variables. For our empirical analyses, we segment the universe of firms based on whether firms have high levels of (excess) cash. As such, we are more interested with relative liquidity among firms each year rather than the absolute level of liquidity. We assess this relative liquidity level in two ways. First, we construct dummy variables if a firm is in the highest quartile of (excess) cash in a given year. Second, we standardize all four of our liquidity measures to have zero mean and unit standard deviation. The reason for this transformation is that we attempt to remain agnostic as to which model of excess cash, if any, is better. Standardizing all our measures allows us to more directly compare across the liquidity variables. In addition to our cash variables, we include control variables to capture the impact of other firm characteristics on the bidder’s decision to acquire. These variables, suggested by the literature on acquisitions (Ambrose and Megginson (1992), Comment and Schwert (1996), Harford (1999), among others), include the abnormal return prior to the bid, sales growth, non-cash net working capital, priceearnings ratio, market to book ratio, leverage, and firm size. Following Harford (1999), all control variables are measured as four year averages from year t-4 to year t-1. Variable definitions and Compustat data item codes are reported in Table 1. To mitigate the effect of outliers, all our variables are winsorized, by year, at the 1% tails. 2.2. Acquisition Data Our sample of acquisition attempts is drawn from Thompson’s Securities Data Corporation (SDC) data set. We examine bids from 1984 to 2005 subject to the following criteria: 1. 2. 3. 4. 5. 6. The bidder owns less than 50% of the target prior to the bid and seeks to own 100%. The bidder is classified by SDC as a public corporation, but is not a financial company or utility. Both the bidder and the target are US firms. The deal type is classified as a disclosed value merger, an LBO, or a tender offer. The form of the deal is listed as merger, acquisition of assets, or acquisition of minority interest. All repurchases, equity carve outs, and limited partnerships are excluded. Because SDC contains data on very small acquisitions, we impose constraints on deal size. We identify bidding firms as any firm making one or more bids within a year, where the sum of all bids in a 7 year is at least: one percent of the bidder’s prior-year assets, one percent of the bidder’s prior-year market capitalization, and ten million dollars in size (deflated to year 2000 dollars using the CPI). The resulting data set includes 10,820 bidder years. To determine the method of payment for the bid years we confirm the SDC classification and combine this with hand-collected data. This is necessary because Faccio and Masulis (2005) report that there are frequent discrepancies within the SDC data field CONSID_STRUCTURE, which simply classifies deals as cash, stock or hybrid. To ensure we have classified the method of payment as accurately as possible, we examine our data in two ways. First, we use SDC’s data fields describing the value of consideration offered based on consideration type (i.e. cash, common stock, convertible debt, etc.)5 Second, we rely on SDC’s text field describing the consideration offered (data item CONSIDOC). We classify cash consideration as payment offered in the forms of cash, debt (including convertible debt), assets, or capital infusions.6 We classify equity consideration as payment offered in the forms of common or preferred stock (including convertible preferred), warrants, rights or options.7 We ignore consideration offered in the form of earnouts or royalties as they are contingent future payments. We then classify bids as cash, stock, or mixed based on both of the above methods and examine by hand those cases where there is a discrepancy. When possible, we go to the financial statements and news stories to determine the composition of consideration listed as “other” or “undisclosed” by SDC, as well as to hand-correct errors in the method of payment data.8 When we are unable to specifically identify the composition of consideration, we assume it is paid in an equal mixture of cash and stock. Ultimately, we classify firm years as a Stock Bid (Cash Bid) if all of the consideration offered within that year is comprised of equity (cash) components. Bids are classified as Mixed if both equity and cash are offered in acquisitions bids.9 This categorization serves as the most conservative way to segment method of payment to isolate cases where firms used equity exclusively. 5 The SDC data items are: VCASH, VCOM, VCDEBT, VCPFD, VPFD, VWAR, VOTH, VUNDIS The following codes are classified as cash consideration: CASH, A, NOTE, BOND, DEB, DBTRED, and LOAN 7 The following codes are classified as stock consideration: COM, STK, PFD, WAR, OPT, UNIT, PREF, SEC, ORD, TRACKING, TSH, ADS, and, APPRECRTS 8 A list of data corrections is available from the authors. 9 This includes firms who make a single bid in a year and offer some combination of cash and equity as well as firms which make multiple bids in a year where some of the bids were entirely equity while others were entirely cash. 6 8 We merge our SDC acquisition data with Compustat data and make two additional changes based on industry, defined using the Fama-French 30 industry definitions. First, the Tobacco Industry (Industry 3) has 64 firm year observations, but zero bids so we eliminate those firm years because the industry effect perfectly predicts no bids will occur.10 We also drop the coal industry with 78 firm years but only a single bid. Our final sample includes 58,117 firm years comprised of 5,968 firm years where at least one bid occurred and 52,149 firm years where no bid was recorded. For the 5,968 firm years with at least one acquisition attempt, approximately one quarter are classified as financed entirely with equity. Our categorization results in 1,354 stock bids, 3,547 cash bids, and 1,067 mixed bids. Table 2 shows the distribution of our sample across time and method of payment. We have a large sample of bids in each year from 1984-2005, ranging from a low of 131 bids in 1990 to a high of 478 bids in 1997. We also have a reasonable sample of bids based on method of payment. We classify 22.7% of our bids as all stock, ranging from a low of 6.2% in 2004 to a high of 38.9% in 1999. Our proportions of all equity bids are slightly lower than those reported in Betton, Eckbo, and Thorburn (2008) owing to the conservative nature by which we classify stock bids. 3. Liquidity and Acquisition Activity 3.1. Summary Statistics The summary statistics for our full sample are shown in Table 3. Of particular importance are the measures of liquidity. On average, firms hold 14% of their assets in the form of cash, while the median firm has 6.7%. The different excess cash measures show a similar right skew in the data with means between 3 and 6 percent of assets. The median firm has negative excess cash according to the Harford measure, positive excess cash according to the OPSW measure, and approximately zero excess cash when measured relative to the industry. However, because our purpose is not to address the magnitude of excess cash holdings across firms, but rather to examine relative liquidity, we focus on two measures to describe high-cash firms. First, we create dummy variables signifying a firm as having high (excess) cash if it is in the top quartile 10 Our results are nearly identical if we include these firm years. 9 of firm liquidity in a given year. As must be the case, one quarter of our firm years are defined as being high cash across all the measures. Second, we standardize our variables such that each year they have zero mean and unit variance. The benefit of these measures is that we can remain agnostic as to which of the models employed, if any, accurately predicts the level of excess cash. We need only to rely on the assumption that a model will provide a relatively accurate ranking of firms based on (excess) liquidity. Correlation coefficients for our liquidity measures suggest that using our measures makes sense. For instance, the Pearson correlation coefficients between the raw OPSW excess cash measure and the raw Harford excess cash measure is actually negative at -0.10. The correlation between the OPSW measure and the industry excess cash measure is only 0.09. Standardizing our variables increases those correlations to 0.07 and 0.24. More importantly, the Spearman rank correlations are 0.59 and 0.73, respectively.11 The correlations suggest that while using different models of “normal” cash holdings allow us to control for a variety of factors which may impact cash holdings, using the top quartile of each measure should provide us with a sample of firms which tend to have high levels of cash, regardless of one’s priors about the levels of cash holdings predicted by the models. Table 4 provides the first evidence on the relation between liquidity and acquisition activity. Panel A segments our sample by whether or not the firm launched a bid in a given year, without regard to method of payment. In the 5,968 firm years where at least one bid occurred, the level of cash holdings are significantly higher at both the mean and median than for the 52,149 firm-years where no bid was launched. The results are similar for the excess cash measures. While those results rely on the magnitude of our liquidity measures, we also focus on firms in the highest quartile. Of the bidders, 27.3% were ranked in the highest quartile of cash holdings, while only 24.7% of the non-bidders were in the top quartile. The proportions are significantly different at the 1% level. We find similar results for our rankings based on the excess cash measures. There are also significant differences in our control variables, providing support for including them in additional tests. Bidders are more likely to have greater abnormal stock returns and sales growth in the years leading up to the takeover attempt. Bidders also have significantly higher Price to Earnings 11 Correlation tables are available upon request from the authors. 10 (PE) and market to book (MB) ratios, which can be indicative of equity overvaluation as an incentive to undertake acquisitions. Lastly, bidding firms have significantly lower levels of non-cash net working capital and leverage, while also being larger in size. Overall, our summary statistics confirm Harford’s (1999) result that cash-rich firms are more likely to undertake acquisition activity. However, when we separate our sample on the basis of method of payment, the impact of cash holdings on acquisition activity becomes more interesting. Panel B of Table 4 examines only the 5,968 firm years in which at least one bid occurred and documents that high cash firms appear more likely to undertake stock acquisitions rather than use their cash. In the 1,354 cases where equity was the sole consideration offered, bidders had an average of 19.8% of their assets in cash. In contrast, for the 3,547 years where cash was the sole consideration offered, bidders held an average of 13.0% of their assets in the form of cash. This difference is statistically significant at the 1% level and indicates that while the presence of cash makes firms more likely to bid, it may not mean that they are spending that cash. We find similar results using medians and for our excess cash measures.12 When examining our ordinal rankings, we find that stock bidders are nearly twice as likely to be in the top quartile of cash holdings as firms which pay with cash. Roughly 40% of the stock bidders were in the highest quartile of cash holdings, while only 21.5% of cash bidders were. This difference between the proportions is significant at the 1% level and is similar using all of our excess cash measures. While our main comparisons are between all equity and all cash bids, it is also interesting to notice that liquidity levels are higher in mixed bids than in the all cash bids. Interestingly, while cash-rich firms are more likely to be involved in acquisition activity, it does not appear that they use cash to pay for the acquisitions. In fact, firms which bid completely with cash have less liquidity than firms which use solely equity, or a mixture of equity and cash. We also observe significant differences in control variables across sample splits by method of payments. Cash bidders tend to be larger and have greater leverage than those which pay with equity. 12 With the OPSW measure, cash bids have more liquidity at the mean, but not the median suggesting there are some observations with very high positive levels of excess cash. This provides further argument for using our ranking of cash-rich firms rather than the raw level. 11 Conversely, firms that bid with stock have higher prior abnormal returns, sales growth, PE, and MB ratios indicating that the value of bidder shares may be an important determinant of method of payment. 3.2. Cash Holdings and the Probability of Bidding The prior evidence suggests that while higher liquidity is associated with frequency of acquisition activity, it is driven by equity acquisitions. However, the prior evidence is univariate and it is important to control for other firm attributes as well. Since bidders have several choices in paying for acquisitions, we use multinomial logistic regressions to evaluate the choices.13 Table 5 documents the results from multinomial logistic regressions to determine not just if a firm is likely to be a bidder, but how it is likely to pay. Our dependent variable is set to zero if the firm does not undertake a bid in the following fiscal year, set to 1 if they launch an all stock bid, set to 2 if they launch an all cash bid, and set to 3 if they undertake a mixed acquisition attempt. The regressions are estimated controlling for industry and year effects. All p-values are based on standard errors clustered at the firm level (see Petersen (2009)). Regression (1) confirms that being in the top quartile of cash to assets makes a firm significantly more likely to bid than not. The p-values in brackets are based on a test of the difference in the coefficients against the outcome of not bidding. Not surprisingly, the coefficients for all three bidding outcomes: stock, cash, or mixed are significantly positive. Interestingly, the coefficient of 0.096 on cash bids shows that high cash firms are only marginally more likely to undertake a cash bid than to not bid at all. Moreover, the coefficient is 0.345 for all-stock bids, which is significantly larger than the coefficient on cash bids at better than the 1% level. The interpretation is that while being in the highest quartile of cash holdings makes a firm more likely to bid with cash than not bid at all, they are actually more likely to bid with stock than with cash. Marginal effects imply that the economic significance is large as well.14 The unconditional probability of a firm being a stock bidder is 139 basis points. Being in the top quartile of cash holdings increases the unconditional probability of being a stock bidder by 50 basis points, an increase of nearly 40%. 13 In unreported results, we also estimate simple logistic regressions. Consistent with prior literature, we find that high (excess) cash holdings significantly increase the likelihood of being a bidder. Results are available upon request. 14 Marginal effects are calculated in Stata using the prchange command. See Long and Freese (2006) for details. 12 We observe similar results in Regression (2) using our standardized cash variable to measure liquidity. Using the continuous measure, not only does an increase in cash make a firm more likely to bid with stock than bid with cash, but the coefficient on all cash bids is insignificant. This indicates that high cash firms are actually no more likely to spend their cash on an acquisition than not bid at all. Increasing cash holdings by one standard deviation around the mean increases the likelihood of undertaking a stock bid by roughly 20% (an increase of 27 basis points from a base of 139 basis points). As in the summary statistics, we see that in both regressions the likelihood of making a mixed bid is greater than that of paying with cash as liquidity increases. The coefficients on the control variables are similar across both specifications and mostly consistent with the evidence in Harford (1999). Bidders tend to be larger, have higher abnormal returns, and greater sales growth than non-bidders. The positive relation between working capital that Harford (1999) demonstrates seems to be driven by the cash bids, while the impact of market to book is only clear when segmented by method of payment. Harford found that MB was insignificant, but this is likely driven by the fact that it is significantly positive for stock bids, negative for cash bids, and insignificant for mixed bids. When the method of payment is not broken out, those effects work in opposite directions resulting in an insignificant estimate. From an overvaluation viewpoint, it is interesting to note that the coefficients on the average abnormal returns, while always significantly positive, do not differ across all three methods of payment. On the other hand, the coefficient on market to book are positive for stock bids and negative for cash bids, with mixed bids falling in the middle. These results suggest that firm-specific overvaluation is more likely to be a determinant of method of payment than an industry or sector run up over the past few years. Section 4 examines this more closely. While the first two specifications use raw liquidity, Panel B of Table 5 examines our excess measures of liquidity. For brevity, we only report the coefficients on the liquidity measures and omit the control variables. For both the Harford and industry adjusted measures, we find the same results as in Panel A. High liquidity makes firms more likely to be bidders, but the results are driven by firms which use equity as the method of payment. The OPSW measure does not provide the same conclusions. While 13 the stock coefficient is higher than that on cash for firms in the top quartile, the difference is not significant. In addition, using the continuous measure, the coefficient on cash bids is marginally larger than that for stock acquisitions. Nonetheless, the results in Table 5 indicate that the increase in takeover probability for cash-rich firms is not driven by firms using their cash. Rather, the results indicate that cash-rich firms are actually more likely to use equity as a method of payment. While our results are completely consistent with those in Harford (1999), segmenting by method of payment allows for a different interpretation of the empirical evidence. Since cash-rich firms are not using cash as the method of payment, the evidence is more supportive of the opportunism hypothesis rather than the misallocation hypothesis. 3.3. Cash Holdings and the Method of Payment The results in Table 5 show that cash-rich firms are more likely to be bidders than other firms, but they tend to use equity as the method of payment. However, it is possible that the results are affected by the exclusion of deal specific characteristics which are likely correlated with the choice of method of payment. For instance, Martin (1996) shows that higher investment opportunities of the bidder and target lead to stock financing of acquisitions, while bidder size is unimportant. In an effort to examine the robustness of our results, we estimate conditional multinomial logistic regressions using only the 5,968 firm years where a bid occurred. By excluding firm years where no bid was attempted, we can include not only bidder characteristics, but also deal specific characteristics in our regressions. Because many bidders in our sample launched multiple bids in a single year, we focus on the first bid in a year when examining deal specific characteristics. Table 6 shows our results. In addition to the bidder characteristics we previously examined, we also include variables which measure: the size of the deal, the degree to which it is met with resistance, and the identity of the target. Specifically, we construct the following variables. Relative value is defined as (deal value / (bidder market capitalization + deal value)), Deal Size is the natural logarithm of deal value deflated to year 2000 dollars using the CPI, Unsolicited is an indicator variable which is set to one if SDC describes the deal attitude as starting off unsolicited, Auction is an indicator variable set to one if there is more than one bidder in the control contest, Defense is set to one if the target uses some sort of takeover defense, Private 14 is set to one if SDC indicates that the target is private, Subsidiary is set to one if SDC indicates that the target is a subsidiary, and Outside Industry is set to one if the bidder and target are from different industries as defined by the Fama-French 30 classification. Regression (1) shows our results when we measure liquidity using the top quartile of raw cash holdings. The first column shows the comparison of stock bids versus cash bids, while for completeness, we also report the comparison between mixed bids and cash bids in the second column. The coefficient on High Cash is 0.258 with a p-value below the 1% level. In Table 6, the p-values in brackets are the relevant measure since we are addressing the difference between stock and cash deals, and are not concerned with a difference between stock and mixed deals. The coefficient on High Cash indicates that even conditioning on a bid occurring, firms in the top quartile of cash holdings are more likely to pay with equity rather than cash. Additionally, the marginal effect is large. Being in the top quartile of cash increases the probability of paying with equity by 282 basis points from an initial probability of 15.40% It is reassuring that the coefficients on the firm-level control variables are consistent with those from the unconditional regressions in Table 5 even with the additional deal-specific control variables. For instance, stock deals have higher sales growth and MB, while having lower NWC and leverage than cash deals. Results in Table 6 show that the coefficients on growth and MB are both positively significant, while that on leverage is significantly negative. While the coefficient on NWC is negative, it is not significant when deal variables are included. We also notice that the coefficient on prior abnormal returns is significantly positive, indicating that bidders who offer stock had a larger run-up in their stock prices in the recent past. Regarding the deal-specific variables, we find that firms tend to pay stock when the deal is larger both in absolute and relative size.15 Bidders are more likely to offer cash than equity if the deal is unsolicited, if there are multiple bidders for the target, or if the target undertakes defensive action. These are all consistent with the idea that when there is competition or target shareholders do not want to sell, they are more likely to be persuaded by offering cash rather than equity. 15 Bidder size is omitted from the Table 6 regression due to the inclusion of deal size and relative size. 15 One of the main reasons that a bidder might use equity is to avoid overpayment through risk sharing as argued in Hansen (1987). This is more likely to be the case when information asymmetries are relatively large. We empirically control for information asymmetries by looking at whether the target was privately held and whether it is in a different industry than the bidder. We expect that if information asymmetries are larger in private targets and those outside of the bidder’s industry, equity would be more likely to be the method of payment. Surprisingly, we find that the coefficient on Private is negative and significant, indicating that controlling for other factors firms are more likely to offer cash for private targets. While we do find a positive coefficient on Outside Industry, it is statistically insignificant. We find similar results when we use our standardized measure of cash holdings in regression (2). Firms with higher cash are more likely to choose equity as the method of payment. Panel B shows the results of regressions using our alternative measures of liquidity. For the Harford and Industry excess cash measures, we find identical results. For the OPSW excess cash measure, we again see insignificant results using the indicator variable, and significantly negative results with the continuous measure. While the majority of the results are consistent with the high cash firms using stock for acquisitions, the inconsistency of the OPSW measure requires further investigation. If the misallocation hypothesis was dominant, we should see firms with positive excess cash holdings using that cash as the method of payment. However, finding that a continuous measure of cash holdings is related to the propensity to use cash doesn’t necessarily imply that. While the result could be driven by high cash firms using their cash; alternatively, it could be evidence that low cash firms have to use equity. To distinguish between these explanations, we segment our standardized (excess) cash measures by whether they are positive or negative. We create two mutually exclusive variables: Standardized Positive cash equals the level of standardized cash if greater than zero, and zero otherwise; Standardized Negative cash equals the absolute value of standardized cash if less than zero, and zero otherwise. If firms were using excess cash to finance acquisitions, we would expect to see negative and significant coefficients on the Positive cash variables, with no prediction regarding how firms with little cash would behave. We estimate our multinomial logistics using these breakpoints and find coefficients greater than zero across all four measures of Positive Cash, indicating that when firms have excess cash 16 holdings, they are more likely to use equity, not their cash. This result is inconsistent with firms misallocating cash due to agency problems. Instead, we notice that the reason for the flip in sign with the OPSW continuous measure relates to firms with negative excess cash. The significantly positive coefficient on Standardized Neg. OPSW indicates that firms with little OPSW excess cash are more likely to use equity – possibly because they are cash constrained. Conversely, we find significantly negative coefficients for our other measures of Negative Cash, indicating that the firms that pay with cash actually hold little cash in the year before the bid. This is consistent with the idea that many cash deals are actually financed by issuing debt or equity (Schlingemann (2004)). 3.4. Cash Holdings and Announcement Returns We have demonstrated that while cash-rich firms are more likely to be bidders, in fact, they are significantly more likely to pay equity for the target. While these results are consistent with the opportunism hypothesis, they are not necessarily inconsistent with the misallocation hypothesis if selfinterested managers simply retain their cash to maintain freedom from future capital market discipline. The cleanest way to distinguish between the two hypotheses is to examine the bidder announcement returns. Harford documents that bidder announcement returns are lower for cash-rich bidders which seems supportive of the misallocation hypothesis. Once again, segmenting by method of payment allows for further insights. Table 7 shows regressions using cumulative abnormal returns from day -1 to day 1 as the dependent variable. Abnormal returns are calculated using the market model with parameters estimated over the period -370 to -253 as in Harford (1999). Our control variables include both bidder-specific and deal-specific characteristics. With our data requirements, we have 5,900 observations in the full sample. The first column shows the results of estimating our regression on the full sample of bids, without regard to method of payment. The coefficient on High Cash is negative and significant at better than the one percent level confirming Harford’s results that the market reaction to a bid announcement is worse for high cash bidders. Moreover, it is reassuring that the signs on our control variables are consistent with 17 those found in the prior literature. For instance, as in Moeller, Schlingemann and Stulz (2004), bidder announcement returns are negatively related to the Q of the bidder and the size of the deal. Conversely, as in (Chang (1998), Fuller, Netter and Stegemoller (2002), Faccio, McConnell and Stolin (2006)), bidder returns are higher if the target is a private company or a subsidiary. The lower market reaction to high cash firms could be explained by either the misallocation or opportunism hypothesis. The decision to pay for an acquisition using equity signals to the market that the bidder’s stock is overvalued (Travlos (1987) and Amihud, Lev, and Travlos (1990)). Because we have shown that high cash firms tend to pay with equity, it is possible that the negative announcement effect for high cash firms is really capturing the method of payment. To examine this possibility, the remaining three columns show the results of estimating the regression separately based on the method of payment. When we segment based on method of payment, the results clearly favor the opportunism hypothesis. Those firms that pay with equity wholly drive the negative announcement effect for high cash bidders. The coefficient is -0.021 for the stock bids while essentially zero for the cash bids. Not only is the coefficient in the stock regression significantly negative, but it is significantly more negative than the coefficient in the cash regression. It is important to note that this finding is not simply capturing lower announcement returns which are associated with equity offers, since that main effect will be embedded in the constants. That suggests that the market does not unconditionally attribute equity overvaluation as the reason for an equity bid; however, when a firm has high cash, but still chooses to pay equity, it serves as a strong signal that equity is overvalued. We also find evidence of the risk-sharing hypothesis in our abnormal return results. The announcement returns are higher for stock bids than cash bids if the target is outside the bidder’s industry or either a private firm or subsidiary. Targets with those characteristics likely have higher information asymmetries and thus using equity would be better for bidder shareholders (Martin (1996)). Panel B provides evidence for our other liquidity measures and demonstrates the same pattern. Again, for brevity, we omit the control variables, which, in all cases are similar to those reported in Panel A. For each of our liquidity measures we find that announcement returns are significantly negative only 18 for the stock bids and the difference between stock and cash bids is always significant. When we use our continuous measures, we find that for all but the OPSW measure, the negative announcement returns are limited to the stock acquisitions. Overall, the results in Table 7 support the opportunism hypothesis rather than the misallocation explanation. Essentially, it appears that if a firm offers to pay with equity by choice, it is a far worse signal than paying equity because you have no other option.16 4. Cash Holdings and Equity Overvaluation 4.1 Do Cash-rich Bidders have Overvalued Equity? The evidence so far demonstrates that while high cash firms are more likely to be bidders, they tend to use equity as the method of payment. Additionally, the results suggest there may be a link between high cash holdings and equity overvaluation. However the literature has shown that in many cases, the market values cash holdings less than dollar for dollar (Faulkender and Wang (2006), Pinkowitz, Stulz and Williamson (2006), Pinkowitz and Williamson (2007), Dittmar and Mahrt-Smith (2007), among others). Thus it appears an inconsistency remains which encourages us to more directly examine the relationship between cash holdings and overvalued equity. While many papers attempt to measure equity overvaluation, we choose to use the empirical methodology of Rhodes-Kropf, Robinson, and Viswanathan (2005), hereafter RKRV, that similarly examines equity overvaluation in acquisitions.17 Their method decomposes the market to book ratio into three components, a firm-specific error, a time-series sector error, and a long-run value to book ratio. We follow RKRV’s methodology and employ the regression for firm market value: mit = α 0 jt + α1 jt bit + α 2 jt ln( NI + )it + α 3 jt I ( < 0) ln( NI + )it + α 4 jt Levit + ε it The variables m and b are the logarithms of market value and book value of equity, respectively. NI+ represents the absolute value of net income, I(<0) is an indicator variable set to one if the firm has a negative net income in a given year, and Lev is the market leverage of the firm defined as {1 - (market 16 This is analogous to Cornett and Tehranaian (1994) who find that banks which raise equity voluntarily have lower announcement returns than those which raise equity to maintain their capital requirement. 17 We are not the first to employ the RRKV methodology. For example, Fu, Lin and Officer (2008) use it to study value creation in acquisitions. 19 value of equity/(market value of equity + book value of assets – deferred taxes – book value of equity))}. The subscripts represent observations at the firm (i), industry (j), and year (t) level. We segment firms into the Fama-French 12 industries defined as on Ken French’s web site. We estimate the above regression cross-sectionally for each industry year. We then calculate the firm-specific error, time-series sector error, and long-run value to book following RKRV (see Table 5 of their paper, page 579). While our sample differs, both in size and time frame, from RKRV, we find qualitatively similar parameter estimates as they report. We also find similar relationships as RKRV with respect to splits on the basis of bidder versus non-bidder, and segmentation based on method of payment.18 In order to examine whether our results are simply driven by a consistent correlation of high cash firms being overvalued, in Table 8, we report the MB decomposition based on our liquidity breakpoints. For firms in the top quartile of cash holdings, average firm-specific error is -0.0692 which is significantly negative. Conversely, firms in the other three quartiles of cash holdings have a significantly positive firm-specific error of 0.0234. The nearly 900 basis point difference is highly significant as well. We find similar results using our other measures of firm liquidity with the difference ranging from 900 basis points to 1900 basis points using the OPSW categorization. Overall, it appears that cash-rich firms do not have overvalued equity. Rather, the market seems to discount the equity of firms holding high amounts of cash. This is consistent with existing evidence in Faulkender and Wang (2006), among others. In contrast to the firm-specific errors, Table 8 indicates that firms are more likely to be in the top quartile of cash holdings when their industry is overvalued relative to its long run fundamentals. It is reasonable to believe that firm overvaluation should reflect not only the firm-specific error component, but the impact of a hot industry. Hence, we define total error to be the sum of both the firm-specific error and the time-series sector error. We find that even accounting for the overvalued sectors, cash-rich firms are undervalued by the market relative to similar firms with lower levels of cash. This result holds across all our liquidity measures with differences between 700 and 1900 basis points. 18 The notable exception is that we do not find evidence of their low buys high result. We find higher long-run value to book for bidders than non-bidders. It is possible that the timing difference in our sample is a major factor in this because our sample runs from 1984-2005, while RKRV use data from 1977-2000. While this suggests that more research should be done on that phenomenon, our focus is on the firm-specific overvaluation so we leave that to future research 20 Finally, we note that the long-run value-to-book estimates are considerably larger for high cash firms. This is reasonable given the characteristics of firms which are known to hold high levels of (excess) cash; namely smaller firms with greater growth opportunities (see Kim, Mauer and Sherman (1998), OPSW (1999), and others). The consistency with prior literature is reassuring and suggests that using the RKRV decomposition to examine liquidity is valid. If high cash firms are not systematically overvalued, then what explains the link between high cash firms and equity bidding? It is possible that while high cash firms are valued at a discount, on average, we might still see acquisition activity from those which happen to have overvalued equity. Table 9 addresses this issue by segmenting our sample not only by level of liquidity, but also by bidding activity. Panel A shows this two-way split. The first row provides the number of firm-year observations in each cell. For instance, of the 5,867 bid years, 4,254 are from bidders classified as not being in the top quartile of cash holdings, while 1,613 are in the top quartile. Segmenting by liquidity, we notice that non-bidders have similar firm-specific errors as the whole sample. Non-bidders in the top quartile of cash holdings are significantly undervalued with a mean of -0.0852, while those not in the top quartile have a significantly positive firm-specific error of 0.0155. The difference between the two is roughly 1,000 basis points which is highly significant; as shown by the p-value in the last column, which allows for comparisons across levels of liquidity holding bidding activity constant. Conversely, the pvalues beneath the rows allow for comparison across bidding activity, holding liquidity constant. In contrast to the full sample and the non-bidders, we find that the difference in firm-specific errors for bidders is less than 400 basis points. Not only is the difference in magnitudes considerably smaller, but the average firm-specific error of the high-cash bidders is actually positive at 0.0567. It appears that cash-rich firms which undertake bids are in fact overvalued rather than being discounted by the market. This difference becomes more apparent when we examine the total errors segmented by bidding activity. Accounting for both the firm and sector errors, we find that the large discount placed on cash-rich firms completely disappears for our bidders. The cash-rich bidders are as overvalued as bidders not in the top 21 quartile of cash holdings. While not reported, we find similar results when segmenting firms using our measures of excess cash.19 Panel B of Table 9 repeats the analysis using just bidders, segmented by method of payment. We focus only on firms which offered stock or cash exclusively and eliminate the mixed deals. We find substantial overvaluation for stock bidders, regardless of whether they are in the top quartile of cash holdings. In fact, it appears that high-cash stock bidders are actually more overvalued than stock bidders with less cash. However, the difference for both the firm-specific and total error components is insignificant. By contrast, for cash bidders, we again see a roughly 900 basis point discount for cash-rich firms, which is highly significant. The results once again support the opportunism hypothesis rather than the misallocation hypothesis. 4.2 Do Cash-rich Bidders Overpay for Targets? The combined results in Tables 7 through 9 strongly support the argument that cash-rich firms undertake acquisitions when their equity is overvalued. There appears to be little support for the idea that agency costs of free cash flow are a motivating factor in acquisitions. However, it may be the case that while overvaluation explains the reason for stock bids, high-cash firms may still be wasting their cash when they use it as the method of payment. While there was no evidence of negative announcement returns for cash-rich firms undertaking cash bids, we attempt to examine whether those firms overpay for their targets. It is difficult to determine whether the offer price for a target is “too much” because the true value of the target is unobservable. Nonetheless, we attempt to address this issue in two separate ways. First, we use the MB decomposition to determine whether the targets are overvalued. The methodology is exactly the same as in the prior section. Unfortunately, we can only decompose MB for publicly traded targets as there is no available market price for private firms. However, as shown in Chang (1998), announcement abnormal returns are positive when the target is private; therefore it seems 19 With the OPSW measure, we still find that cash-rich bidders are overvalued, but they are statistically less overvalued than bidders not in the top quartile. 22 reasonable to assume that using only public targets should, if anything, bias us towards finding support for the misallocation hypothesis. Panel A of Table 10 shows the results of the MB decomposition for 672 publicly traded targets, split almost evenly across stock and cash bids. The columns indicate a split based on the liquidity of the bidder in the takeover contest. We find that for stock bidders which are not in the top quartile of cash holdings, the average firm-specific error of the target is 0.0347. For stock bidders which are cash-rich, the average firm-specific error of the target is 0.1526. This implies that stock bidders are pursuing targets which are somewhat overvalued. However, Panel B of Table 9 shows that bidders are even more overvalued than their targets. Combined, these results are consistent with the opportunism hypothesis and the results in RKRV. For the misallocation hypothesis to be supported, cash-rich bidders who use cash should target overvalued firms. However, the average firm-specific error for the targets of those firms is actually negative and marginally significant at -0.1281. This indicates that the cash-rich bidders appear to be spending their cash on undervalued targets. Additionally, the estimate is significantly lower than the valuation of the targets by the bidders who are not in the top quartile of cash holdings. The results are even more robust when combining firm-specific and sector errors. We also use bid premiums to examine whether cash-rich bidders are overpaying for their targets. We define bid premium as (bid offer price / target stock price at time t), where we measure the premium relative to the target price one day, one week, and four weeks prior to the initial announcement. Panel B of Table 10 shows the bid premiums offered segmented by bidder liquidity and method of payment. While it is always the case that bid premiums are higher for cash bids, the important comparison is across the level of liquidity. We see that for cash bids, there is no difference in bid premium based on the cash-richness of the bidder. Cash-rich firms do not overpay for targets, controlling for method of payment. There is weak evidence that cash-rich firms offer higher bid premiums when they pay with equity, but since they are not spending their cash, it becomes difficult to argue that is supportive of the misallocation hypothesis. 23 For robustness, we repeat the analyses in Table 10 with all measures of excess cash and find similar results. Instead of wasting cash on overvalued targets, cash-rich bidders appear to act opportunistically in an effort to acquire undervalued assets. Further, this opportunism holds for cash-rich bidders regardless of whether they pay stock or cash. Cash-rich bidders with overvalued equity choose to use it as cheap currency in acquisitions. Cash-rich bidders with fairly valued equity use their liquidity to target relatively undervalued firms. In both cases, the actions of management are in the best interests of existing shareholders, and corroborate the opportunism hypothesis. 5. Robustness 5.1 Pooling of Interests Accounting Prior to June 30, 2001, bidding firms could use pooling of interests (hereafter pooling) to account for stock acquisitions.20 If bidders with high cash systematically prefer to use pooling and pooling also is characterized by lower acquirer announcement returns, then our results may be explained by the option to use pooling. While there is some evidence that pooling is valuable (see Robinson and Shane (1990) and Ayers, Lefanowicz and Robinson (2002)), there is no direct evidence that high cash firms systematically choose to use pooling. We nonetheless examine whether our results are robust to controlling for pooling. We present three pieces of evidence that our findings our robust to pooling. First, our multinomial logistic regressions are already robust since they account for the availability of pooling via the inclusion of yearly dummy variables. Second, we estimate the abnormal return regressions from Table 7 including a dummy variable if the year was prior to 2001, and hence pooling was available. Interestingly, we find that the coefficient on the pooling available dummy is positive for stock bids. This indicates that acquirer announcement returns were higher for stock bidders when pooling was available, without regard to whether they actually chose to use pooling. This seems to support the evidence in prior literature that the option to use pooling accounting could be valuable. More importantly though, our results regarding the 20 Bidders could account for acquisitions using either pooling of interests or the purchase method. While there were a dozen conditions that needed to be met to qualify for pooling of interests, a key one is that the method of payment had to be equity. 24 negative bidder announcement returns are unchanged; the lower returns for high cash firms are limited to those transactions where stock was the method of payment. Third, if pooling provided an incentive for firms to use equity as the method of payment, and managers act opportunistically, then to use stock post pooling cash-rich bidders would have to be even more overvalued. We examine our RKRV decomposition from Table 9, both pre- and post-2001 and find this to be the case. When pooling was available, there is no significant difference between the overvaluation of stock bidders based on liquidity (firm-specific error of 0.21 for cash-rich versus 0.21 for not cash-rich). However, after pooling was disallowed, in order for a cash-rich bidder to use stock, they needed to be far more overvalued (firm-specific error of 0.32 versus 0.12). We further find that our results from Table 10 showing that cash-rich firms which pay with cash do not seem to overpay for targets holds across both sub-periods. Overall, while the distribution of bids in Table 2 seems to suggest that the availability of pooling of interests accounting had an impact on method of payment; our results are robust to controlling for it. 5.2 Do High-Cash Firms have Enough Cash to Finance Acquisitions? Our primary analyses have examined the relative level of liquidity across firms so as to be comparable with prior literature. However, two issues potentially arise using that characterization. First, if in anticipation of undertaking an acquisition, a firm raised the cash through a debt offering, then monitoring through the debt channel would arise (Schlingemann (2004)), which would bias against finding support for the misallocation hypothesis.21 Second, even though a firm might be cash-rich, if the target was large enough, the bidder might still be unable to finance it completely through internal funds and thus paying through equity might not represent a choice. To account for both issues, we examine the subset of acquisitions where the bidder had, in the prior year, at least three times the amount of cash as the value of the acquisition. For simplicity, we refer to these as the 3X bidders. Focusing on the 3X bidders relieves concerns that debt was issued to fund the 21 It seems unlikely that this would be the impetus for our results though as they are driven by firms which pay with equity, not with cash. It is hard to believe that firms would raise cash through a debt offering and then pay for the transaction solely with equity. 25 transaction, and that the firm’s high cash levels are not high enough to finance the transaction. There are 655 bid years that qualify as 3X, segmented into 130 stock bids and 525 cash bids. While we require the firms to have a minimum of 3 times the cash to bid value, the 3X firms have a mean of almost 7 times and a median of 5 times. The mere fact that 20% of firms with roughly 5 times the amount of cash needed to pay, choose to solely use equity seems counter to the misallocation hypothesis. Nonetheless, we repeat the tests in Table 6 and once again find that these firms are no more likely to use cash as the method of payment.22 Next, we find that 3X firms which choose to pay with cash do not have negative announcement returns. Finally, we examine the RKRV overvaluation measures using the 3X sample. Our results are also similar: cash-rich firms which choose to pay with stock are significantly overvalued, and more so than the firms which pay with cash. Overall, the 3X subsample results tend to strongly support our main findings and suggest that the opportunism hypothesis fits the data considerably better than the misallocation hypothesis. 6. Conclusions In this paper we show that excess cash impacts the acquisition decision, but not necessarily the method of payment in the acquisition. The results have important implications for the impact of cash holdings on investment decisions. That is, though cash holdings do influence the acquisition decision, managers do act in the best interest of shareholders when deciding on the method of payment. We start by building on Jensen (1986) and Harford (1999). Jensen (1986) argues that firms should disgorge free cash flow since managers will waste excess cash on negative NPV investments. Indeed, Harford (1999) shows that firms with excess cash tend to make more acquisitions than firms without excess cash; moreover, acquisitions by cash-rich firms have lower bidder returns. The prior interpretation of this result was that it reinforces the agency arguments made by Jensen. This study first confirms Harford’s findings, and then attempts to understand the inconsistency between this interpretation and the observation that firms with excess cash frequently pay with stock. 22 Interestingly, with the 3X subsample, the coefficient on outside industry, which was insignificant in Table 6, becomes significantly positive. This suggests that for firms which are relatively unconstrained in their choice of using cash, equity is more likely when information asymmetries are larger, which supports Hansen (1987). 26 The paper examines two competing arguments to explain this inconsistency. The first is the misallocation hypothesis, in which managers act in their own self-interest and misallocate capital to negative NPV investments. The competing hypothesis is the opportunism hypothesis, where managers act in the best interest of shareholders by opportunistically using the firm’s cash and equity. To answer which is at work, we present three main findings. First, we show that cash-rich firms are more likely to be bidders, but that the method of payment is more likely to be stock. Second, controlling for acquirer-target characteristics we find that bidder returns are negative for high-cash firms only when they pay with stock. These results suggest that firms use stock only when it is overvalued – signaling this to the market in doing so – and use cash otherwise. Third, using the methodology in RKRV (2005) we identify over- and undervaluation in both the acquirer and target for deals involving high-cash bidders. Firms with excess cash use stock as the method of payment rather than using their excess cash when their own stock is overvalued. Additionally, overvalued stock is used to acquire less-overvalued targets. Finally, cash acquisitions by cash-rich firms tend to be for undervalued targets. The results strongly support the opportunism hypothesis. The managers of cash-rich firms act in the interest of their shareholders when making acquisitions. These cash-rich firms are more likely to make acquisitions, but they do not appear to be negative-NPV investments. Consistent with managers using the cheapest source of capital, stock is used only when it is overvalued or relatively overvalued, and cash is used to target undervalued targets. In providing evidence for the opportunism hypothesis, this paper focuses strictly on acquisitions as the sole mechanism for firm investment of excess cash. Though it is possible that managers of highcash firms are wasting excess cash in other ways, they do not appear to do so through acquisitions. The results in this paper add to the literature on cash holdings and the behavior of managers of firms with excess cash. Future studies can examine the role of governance and other firm specific characteristics and its impact on managers’ use of excess cash. Additionally, further examination of the other channels in which managers may use excess cash would add to our understanding of managerial behavior and firm cash holdings. 27 References Ambrose, Brent and William Megginson, 1992, The role of asset structure, ownership structure, and takeover defenses in determining acquisition likelihood, Journal of Financial and Quantitative Analysis 27, 575-589. Amihud, Yakov, Baruch Lev, and Nickolaos Travlos, 1990, Corporate control and the choice of investment financing: The case of corporate acquisitions, Journal of Finance 45, 603-616. Ang, James S. and Yingmei Chen, 2006, Direct evidence on the market-driven acquisition theory, Journal of Financial Research 29, 199-216. Ayers, Benjamin, Craig Lefanowicz, and John Robinson, 2002, Do firms purchase the pooling method? Review of Accounting Studies 7, 5-32. Baker Malcolm, and Jeffrey Wurgler, 2002, Market timing and capital structure, Journal of Finance 57, 1-32. Betton, S., B. E. Eckbo, and K. Thorburn, 2008, Corporate Takeovers, working paper, Dartmouth University. Chang, S., 1998, Takeovers of privately held targets, methods of payment, and bidder returns, Journal of Finance 53, 773-784. Comment, Robert and G. William Schwert, 1996, Poison or Placebo? Evidence on the deterrence and wealth effects of modern antitakeover measures, Journal of Financial Economics 39, 3-43. Cornett, Marcia Millon and Hassan Tehranian, 1994, An examination of voluntary versus involuntary security issuances by commercial banks: The impact of capital regulations on common stock returns, Journal of Financial Economics 35. 99-122. Dittmar, Amy and Jan Mahrt-Smith, 2007, Corporate governance and the value of cash holdings, Journal of Financial Economics 83, 599-634. Dong, Ming, David Hirshleifer, Scott Richardson, and Siew Hong Teoh, 2006, Does investor misevaluation drive the takeover market? Journal of Finance 61, 725-762. Easterbrook, Frank, 1984, Two agency-cost explanations of dividends, The American Economic Review 74, 650-659. Faccio, Mara, and Ronald V. Masulis, 2005, The choice of payment method in European mergers and acquisitions, Journal of Finance 60, 1345-1388. Faccio, Mara, John J. McConnell, and David Stolin, 2006, Returns to acquirers of listed and unlisted targets, Journal of Financial and Quantitative Analysis 41, 197-220. Faulkender, Michael and Rong Wang, 2006, Corporate financial policy and the value of cash, Journal of Finance 51, 1957-1990. Fu, Fanjian, Leming Lin and Micah Officer, 2008, Acquisitions driven by stock overvaluation, working paper, Loyola Marymount University. Fuller, Kathleen, Jeffrey Netter, and Mike Stegemoller, 2002, What do returns to acquiring firms tell us? Evidence from firms that make many acquisitions, Journal of Finance 57, 1763-1793. 28 Graham, John R., and Campbell R. Harvey, 2001, The theory and practice of corporate finance: Evidence from the field, Journal of Financial Economics 60, 187-243. Hansen, Robert G., 1987, A theory for the choice of exchange medium in the market for corporate control, Journal of Business 60, 75-95. Harford, Jarrad, 1999. Corporate cash reserves and acquisitions. Journal of Finance 54, 1969-1997. Harford, Jarrad, Sattar A. Mansi, and William F. Maxwell, 2008, Corporate governance and firm cash holdings in the US, Journal of Financial Economics 87, 535-555. Jensen, M. C., 1986, Agency costs of free cash flow, corporate finance and takeovers, American Economic Review, 323-329 Jensen, M. C. and W. H. Meckling, 1976, Theory of the firm: Managerial behavior, agency costs, and ownership structure, Journal of Financial Economics 3, 305-360. Jung Kooyul, Yong-Cheol Kim, and Rene Stulz, 1996, Timing, investment opportunities, managerial discretion, and the security issue decision, Journal of Financial Economics 42, 159-185 Kim, Chang-Soo, David C. Mauer and Ann E. Sherman, 1998, The determinants of corporate liquidity, Journal of Financial and Quantitative Analysis 33, 335-359. Long, J. Scott and Jeremy Freese, 2006, Regression Models for Categorical Dependent Variables Using Stata, 2nd edition, Stata Press, College Station, Texas. Martin, K, 1996, The method of payment in corporate acquisitions, investment opportunities, and management ownership, Journal of Finance 51, 1227-1246. Mikkelson, W. H. and M. Partch, 2002, Do persistent large cash reserves hinder performance, Journal of Financial and Quantitative Analysis, forthcoming. Moeller, Sara B., Frederik P. Schlingemann, and Rene M. Stulz, 2004, Firm size and the gains from acquisitions, Journal of Financial Economics 73, 201-228. Myers, Stewart C., 1984, The capital structure puzzle, Journal of Finance 39, 575-592. Myers, S. and N. Majluf, 1984, Corporate financing and investment decisions when firms have information that investors do not have, Journal of Financial Economics 13, 187-221. Office, Micah S., Annette B. Poulsen, and Mike Stegemoller, 2007, Information asymmetry and acquirer returns, Review of Financial Economy forthcoming. Opler, T., L. Pinkowitz, R. M. Stulz and R. Williamson, 1999, The determinants and implications of corporate cash holdings, Journal of Financial Economics 52, 3-46. Petersen, Mitchell A., 2009, Estimating standard errors in finance panel data sets: Comparing approaches, Review of Financial Studies 22, 435-480. Pinkowitz, Lee, René Stulz, and Rohan Williamson, 2006, Does the contribution of corporate cash holdings and dividends to firm value depend on governance? A cross-country analysis, Journal of Finance 51, 2725-2751. Pinkowitz, Lee and Rohan Williamson, 2007, What is the market value of a dollar of corporate cash? Journal of Applied Corporate Finance 19, 74-81. 29 Rhodes-Kropf, Matthew, and David T. Robinson, and S. Viswanathan, 2005, Valuation waves and merger activity: The empirical evidence, Journal of Financial Economics 77, 561-603. Rhodes-Kropf, Matthew, and S. Viswanathan, 2004, Market valuation and merger waves, Journal of Finance 59, 2685-2718. Robinson, John R. and Philip B. Shane, 1990, Acquisition accounting method and bid premia for target firms, The Accounting Review 65, 25-48. Schlingemann, Frederik P., 2004, Financing decisions and bidder gains, Journal of Corporate Finance 10, 683-701. Shleifer, Andrei, and Robert Vishny, 2003, Stock market driven acquisitions, Journal of Financial Economics 70, 295-311. Stulz, Rene, 1988, Managerial control of voting rights: Financing policies and the market for corporate control, Journal of Financial Economics 20, 25-54. Stulz, Rene, 1990, Managerial discretion and optimal financing policies, Journal of Financial Economics 26, 3-27. Travlos, Nickolaos G., 1987, Corporate takeover bids, method of payment, and bidding firms' stock returns, Journal of Finance 42, 943-963. 30 Table 1 Variable Definitions Used in the Analyses All variables are measured at time t unless otherwise subscripted. Variable Description Calculation and Compustat Data Codes Cash Cash and Marketable securities to assets che / at Excess OPSW OPSW excess liquid assets of the acquiring firm deflated by its asset base OPSW regression residual (see appendix) Excess Harford Harford excess liquid assets of the acquiring firm deflated by its sales Harford regressions residuals (see appendix) Excess industry Excess liquid assets based on median of industry for the year Cash/assets – Median Industry Cash/Assets (see appendix) High (Excess) Cash Top Quartile by year of (Excess) Cash Measure Standardized (Excess) Cash Liquidity measure standardized yearly to have zero mean and unit variance (Cash Measureit - μt) / σt Abret 4 year average of 12 month buy and hold return of the bidder above the 12 month buy and hold return of the appropriate 5x5 FamaFrench size and MB portfolio 12 month buy and hold return – 12 month buy and hold return of Fama-French 25 portfolio Growth 4 year average of Growth rate in sales of the acquiring firm (Sale – Salet-1) / Salet-1 NWC 4 year average of Non-cash net working capital deflated by assets (act – lct – che) / at PE 4 year average of Price Earnings Ratio prcc_f / epspi MB 4 year average of Market to book ratio of the firm’s equity (csho * prcc_f) / (at – lt – mib) Leverage 4 year average of Book value of debt divided by market value of equity (dltt + dlc) / (csho * prcc_f) Size Size of the acquiring firm’s real asset base (year 2000 dollars) ln((at/CPI)*100) 31 Table 2 Distribution of 58,117 firms years through time Firm years are classified as no bid if they do not undertake any acquisition in that year. We classify firm years as a Stock Bid (Cash Bid) if all of the consideration offered for acquisition bids within that year is comprised of equity (cash) components. Bids are classified as Mixed if both equity and cash are offered in acquisitions bids Year No bid All Stock Bids All Cash Bids Mixed Bids Total Bids % Stock 1984 2,124 26 93 19 138 18.84% 1985 2,195 25 120 20 165 15.15% 1986 2,225 24 96 27 147 16.33% 1987 2,127 25 131 27 183 13.66% 1988 2,194 27 120 21 168 16.07% 1989 2,240 25 97 21 143 17.48% 1990 2,127 29 78 24 131 22.14% 1991 2,155 41 78 24 143 28.67% 1992 2,209 50 112 41 203 24.63% 1993 2,352 79 129 53 261 30.27% 1994 2,444 107 164 54 325 32.92% 1995 2,603 146 181 67 394 37.06% 1996 2,749 122 214 86 422 28.91% 1997 2,762 139 253 86 478 29.08% 1998 2,816 136 220 73 429 31.70% 1999 2,820 133 154 55 342 38.89% 2000 2,520 64 149 57 270 23.70% 2001 2,392 37 171 58 266 13.91% 2002 2,299 37 198 53 288 12.85% 2003 2,376 37 260 68 365 10.14% 2004 2,307 22 259 72 353 6.23% 2005 2,113 23 270 61 354 6.50% Total 52,149 1,354 3,547 1,067 5,968 22.69% 32 Table 3 Summary Statistics Cash is cash/assets. Excess Harford [OPSW] is defined as the residual from a first pass regression predicting cash holdings based on Harford (1999) [OPSW (1999)]. Excess Industry is cash/assets less the industry median for that year, where industry is defined using the Fama-French 30 industries. Standardized (excess) cash measures are transformed to have zero mean and unit variance each year. High (excess) cash measures indicate the firm is in the top quartile of cash holdings within a given year. Abret is the 4 year average excess buy and hold return benchmarked against the appropriate 5x5 FamaFrench size and MB portfolio. Growth is the four year average of sales growth. NWC is the four year average of non-cash net working capital deflated by assets. PE is the four year average price earnings ratio. MB is the four year average of the market to book ratio of the firm’s equity. Leverage is the four year average book value of debt deflated by market equity. Size is the natural logarithm of the firm’s real assets deflated to year 2000 dollars using the CPI. Panel A: N Mean 25th Percentile Median 75th Percentile Standard Deviation Cash 58,117 0.140 0.020 0.067 0.195 0.173 Excess Harford 58,117 0.033 -0.121 -0.047 0.025 0.582 Excess OPSW 58,117 0.056 -0.024 0.010 0.125 0.731 Excess Industry 58,117 0.043 -0.036 -0.001 0.095 0.155 Standardized Cash 58,117 0.000 -0.693 -0.422 0.338 1.000 Standardized Harford 58,117 0.000 -0.366 -0.166 -0.007 1.000 Standardized OPSW 58,117 0.000 -0.190 -0.055 0.114 1.000 Standardized Industry 58,117 0.000 -0.550 -0.282 0.342 1.000 High Cash 58,117 0.250 High Harford 58,117 0.250 High OPSW 58,117 0.250 High Industry 58,117 0.250 Abret 58,117 0.056 -0.169 -0.007 0.191 0.416 Growth 58,117 0.249 0.037 0.119 0.267 0.524 NWC 58,117 0.138 0.008 0.130 0.262 0.176 PE 58,117 15.304 3.501 13.412 22.642 28.500 MB 58,117 2.679 1.255 1.963 3.202 2.602 Leverage 58,117 0.532 0.063 0.241 0.621 0.857 Size 58,117 5.373 3.995 5.211 6.576 1.837 33 Table 4 Summary Statistics Split by Bidder Activity and Method of Payment Cash is cash/assets. Excess Harford [OPSW] is defined as the residual from a first pass regression predicting cash holdings based on Harford (1999) [OPSW (1999)]. Excess Industry is cash/assets less the industry median for that year, where industry is defined using the Fama-French 30 industries. Standardized (excess) cash measures are transformed to have zero mean and unit variance each year. High (excess) cash measures indicate the firm is in the top quartile of cash holdings within a given year. Abret is the 4 year average excess buy and hold return benchmarked against the appropriate 5x5 FamaFrench size and MB portfolio. Growth is the four year average of sales growth. NWC is the four year average of non-cash net working capital deflated by assets. PE is the four year average price earnings ratio. MB is the four year average of the market to book ratio of the firm’s equity. Leverage is the four year average book value of debt deflated by market equity. Size is the natural logarithm of the firm’s real assets deflated to year 2000 dollars using the CPI. Bidders (n=5,968) Non – Bidders (n=52,149) Mean Median Mean Median *** *** Cash 0.154 0.073 0.138 0.066 *** * Standardized Cash 0.056 -0.401 -0.006 -0.425 Standardized Harford 0.024** -0.153*** -0.003 -0.168 *** *** Standardized OPSW 0.046 -0.040 -0.005 -0.057 *** *** Standardized Industry 0.035 -0.254 -0.004 -0.287 *** High Cash 0.273 0.247 0.000 *** High Harford 0.283 0.246 0.000 *** High OPSW 0.272 0.247 0.000 ** High Industry 0.263 0.248 0.000 *** *** Abret 0.159 0.067 0.044 -0.016 Growth 0.309*** 0.161*** 0.242 0.116 *** NWC 0.108 0.097 0.141 0.135*** PE 18.01*** 15.64*** 14.99 13.14 *** *** MB 3.10 2.38 2.63 1.92 *** 0.244*** Leverage 0.425 0.211 0.544 Size 6.23*** 6.14*** 5.28 5.09 ***,**, and * indicate significant differences between bidders and non-bidders at the mean or median at the 1, 5, and 10% levels respectively. Panel A: Split by Bidder Activity 34 Table 4 (Continued) Summary Statistics Split by Bidder Activity and Method of Payment Panel B: Just Bidders (n=5,968) Cash Standardized Cash Standardized Harford Standardized OPSW Standardized Industry High Cash High Harford High OPSW High Industry Abret Growth NWC PE MB Leverage Size All Stock Bids (n=1,354) Mean Median *** 0.198 0.121*** 0.365*** -0.071*** 0.181*** -0.130*** -0.024 -0.031*** 0.231*** -0.141*** 0.391*** 0.360*** 0.324*** 0.350*** 0.243*** 0.125*** 0.430*** 0.241*** 0.098 0.092 *** 20.49 17.86*** 4.05*** 3.05*** 0.277 0.120 6.02 5.88 All Cash Bids (n=3,547) Mean Median 0.130 0.058 -0.097 -0.499 -0.062 -0.164 * 0.052 -0.047 -0.063 -0.284 0.215 0.239 0.239 0.221 0.119 0.050 0.240 0.131 *** 0.113 0.100 17.24 15.28 2.72 2.18 ***++ 0.479 0.259***+++ ***+++ 6.46 6.38***+++ Mixed Bids (n=1,067) Mean Median +++ 0.179 0.093+++ 0.171+++ -0.296+++ 0.109+++ -0.138+++ 0.118++ -0.032+++ 0.114+++ -0.232+++ 0.318+++ 0.329+++ 0.315+++ 0.295+++ 0.184+++ 0.074++ 0.384+++ 0.210+++ 0.104 0.089 17.42 15.10 +++ 3.14 2.36+++ 0.432 0.207 5.72 5.60 ***,**, and * indicate significant differences between all stock and all cash bids at the mean or median at the 1, 5, and 10% levels respectively. +++,++, and + indicate significant differences between Mixed and All Cash at the 1, 5, and 10% levels respectively. 35 Table 5 Multinomial Logistic analysis of determinants of bidding Dependent variable set to 0 if there is no bid in the subsequent fiscal year; set to 1 if there is at least 1 bid in the subsequent fiscal year and all are completely financed with stock; set to 2 if there is at least 1 bid in the subsequent fiscal year and all are financed completely with cash; and set to 3 if there is at least one bid in the subsequent fiscal year and the financing contains both cash and equity components. The sample is 58,117 firm-years comprised of 1,354 stock bids, 3,547 cash bids, 1,067 mixed bids, and 52,149 non-bids. Cash is cash/assets. Excess Harford [OPSW] is defined as the residual from a first pass regression predicting cash holdings based on Harford (1999) [OPSW (1999)]. Excess Industry is cash/assets less the industry median for that year, where industry is defined using the Fama-French 30 industries. Standardized (excess) cash measures are transformed to have zero mean and unit variance each year. High (excess) cash measures indicate the firm is in the top quartile of cash holdings within a given year. Abret is the 4 year average excess buy and hold return benchmarked against the appropriate 5x5 Fama-French size and MB portfolio. Growth is the four year average of sales growth. NWC is the four year average of non-cash net working capital deflated by assets. PE is the four year average price earnings ratio. MB is the four year average of the market to book ratio of the firm’s equity. Leverage is the four year average book value of debt deflated by market equity. Size is the natural logarithm of the firm’s real assets deflated to year 2000 dollars using the CPI. All regressions are run including industry and year dummy variables. Industries are defined as the 30 Fama-French industries. Standard errors are clustered by firm. P-values based on the clustered standard errors are reported in brackets. ***,**, and * indicate significant differences between All Stock and All Cash at the 1, 5, and 10% levels respectively. +++ ++ , , and + indicate significant differences between Mixed and All Cash at the 1, 5, and 10% levels respectively. (1) Panel A: High Cash All Stock 0.345*** [0.000] All Cash 0.096 [0.080] (2) Mixed 0.276+ [0.001] Standardized Cash All Stock All Cash Mixed 0.156*** [0.000] 0.036 [0.177] 0.124++ [0.001] Abret 0.578 [0.000] 0.524 [0.000] 0.521 [0.000] 0.571 [0.000] 0.524 [0.000] 0.518 [0.000] Growth 0.200** [0.000] 0.101 [0.012] 0.241+++ [0.000] 0.193* [0.000] 0.100 [0.013] 0.235+++ [0.000] NWC -0.288** [0.245] 0.369 [0.021] -0.044 [0.857] -0.171* [0.510] 0.375 [0.022] 0.023 [0.926] PE 0.001 [0.181] 0.001 [0.212] 0.001 [0.178] 0.001 [0.149] 0.001 [0.205] 0.001 [0.154] MB 0.040*** [0.000] -0.037 [0.000] -0.010++ [0.388] 0.039*** [0.000] -0.037 [0.000] -0.010++ [0.391] Leverage -0.474*** [0.000] -0.126 [0.000] -0.093 [0.039] -0.461*** [0.000] -0.127 [0.000] -0.088 [0.048] Size 0.322 [0.000] 0.346 [0.000] 0.176+++ [0.000] 0.326 [0.000] 0.346 [0.000] 0.178+++ [0.000] Pseudo R2 0.0783 0.0784 36 Table 5 (Continued) Multinomial Logistic analysis of determinants of bidding Panel B: Alternative Liquidity Measures All Stock All Cash Top Quartile by Year 0.119 0.363*** High Harford [0.000] [0.017] 0.224 0.194 High OPSW [0.001] [0.000] 0.311*** 0.089 High Industry [0.000] [0.079] Continuous Measures, Standardized 0.048*** -0.029 Standardized Harford [0.045] [0.211] 0.054 0.122* Standardized OPSW [0.031] [0.000] *** 0.144 0.013 Standardized Industry [0.000] [0.597] 37 Mixed 0.343+++ [0.000] 0.330+ [0.000] 0.196 [0.013] 0.052++ [0.063] 0.180 [0.000] 0.100++ [0.003] Table 6 Multinomial Logistic analysis of determinants of Method of Payment for First Bids Dependent variable set to 1 if the bid is completely financed with stock; set to 2 if financed completely with cash; and set to 3 if the financing contains both cash and equity components. The sample is 5,968 bids comprised of 1,354 stock bids, 3,547 cash bids, and 1,067 mixed bids. The cash alternative is the baseline. Cash is cash/assets. Excess Harford [OPSW] is defined as the residual from a first pass regression predicting cash holdings based on Harford (1999) [OPSW (1999)]. Excess Industry is cash/assets less the industry median for that year, where industry is defined using the Fama-French 30 industries. Standardized (excess) cash measures are transformed to have zero mean and unit variance each year. High (excess) cash measures indicate the firm is in the top quartile of cash holdings within a given year. Abret is the 4 year average excess buy and hold return benchmarked against the appropriate 5x5 Fama-French size and MB portfolio. Growth is the four year average of sales growth. NWC is the four year average of non-cash net working capital deflated by assets. PE is the four year average price earnings ratio. MB is the four year average of the market to book ratio of the firm’s equity. Leverage is the four year average book value of debt deflated by market equity. Size is the natural logarithm of the firm’s real assets deflated to year 2000 dollars using the CPI. Relative value is defined as (deal value / (bidder market capitalization + deal value)), Deal Size is the natural logarithm of deal value deflated to year 2000 dollars using the CPI, Unsolicited is an indicator variable which is set to one if SDC describes the deal attitude as starting off unsolicited, Auction is an indicator variable set to one if there is more than one bidder in the control contest, Defense is set to one if the target uses some sort of takeover defense, Private is set to one if SDC indicates that the target is private, Subsidiary is set to one if SDC indicates that the target is a subsidiary, and Outside Industry is set to one if the bidder and target are from different industries as defined by the Fama-French 30 classification. All regressions are run including industry and year dummy variables. Industries are defined as the 30 Fama-French industries. Standard errors are clustered by firm. P-values based on the clustered standard errors are reported in brackets. 38 (1) Panel A: High Cash Stock vs. Cash 0.258 [0.006] (2) Mixed vs. Cash 0.214 [0.026] Standardized Cash Abret Growth NWC PE MB Leverage Relative Value Deal Size Unsolicited Auction Defense Private Subsidiary Outside Industry Pseudo R2 0.349 [0.000] 0.350 [0.000] -0.248 [0.447] -0.001 [0.639] 0.096 [0.000] -0.451 [0.000] 2.650 [0.000] 0.170 [0.000] -1.341 [0.000] -1.010 [0.000] -1.811 [0.003] -0.711 [0.000] -2.509 [0.000] 0.051 [0.541] 0.229 [0.014] 0.372 [0.000] 0.095 [0.757] 0.000 [0.877] 0.041 [0.018] -0.149 [0.023] 3.102 [0.000] 0.035 [0.264] -0.272 [0.257] 0.079 [0.688] -0.869 [0.039] 0.150 [0.164] -1.064 [0.000] -0.104 [0.204] 0.1935 Stock vs. Cash Mixed vs. Cash 0.142 [0.001] 0.340 [0.000] 0.349 [0.000] -0.147 [0.657] -0.001 [0.682] 0.094 [0.000] -0.434 [0.000] 2.622 [0.000] 0.174 [0.000] -1.333 [0.000] -1.019 [0.000] -1.849 [0.003] -0.713 [0.000] -2.509 [0.000] 0.052 [0.532] 0.112 [0.010] 0.224 [0.017] 0.371 [0.000] 0.160 [0.608] 0.000 [0.845] 0.039 [0.024] -0.141 [0.031] 3.079 [0.000] 0.038 [0.230] -0.268 [0.266] 0.074 [0.707] -0.900 [0.034] 0.149 [0.167] -1.063 [0.000] -0.104 [0.204] 0.1939 39 Panel B: Alternative Liquidity Measures Stock vs. Cash Mixed vs. Cash Top Quartile By Year High Harford 0.276 [0.002] 0.275 [0.001] High OPSW 0.026 [0.773] 0.131 [0.138] High Industry 0.229 [0.011] 0.164 [0.070] Continuous Measures, Standardized Standardized Harford 0.142 [0.000] 0.133 [0.001] Standardized OPSW -0.073 [0.056] 0.022 [0.601] Standardized Industry 0.137 [0.000] 0.101 [0.010] Continuous Measures, Split by Positive versus Negative Standardized Pos. Cash 0.103 [0.062] 0.088 [0.136] Standardized Neg. Cash -0.299 [0.058] -0.190 [0.198] Standardized Pos. Harford 0.092 [0.053] 0.089 [0.069] Standardized Neg. Harford -0.453 [0.017] -0.402 [0.029] Standardized Pos. OPSW 0.009 [0.903] 0.029 [0.675] Standardized Neg. OPSW 0.115 [0.035] -0.018 [0.799] Standardized Pos. Industry 0.086 [0.141] 0.046 [0.454] Standardized Neg. Industry -0.256 [0.027] -0.226 [0.037] 40 Table 7 Regression Analysis of Abnormal Returns Dependent variable is bidder cumulative abnormal returns over the -1 to 1 period using the market model with parameters estimated from days -370 to -253. Cash is cash/assets. Excess Harford [OPSW] is defined as the residual from a first pass regression predicting cash holdings based on Harford (1999) [OPSW (1999)]. Excess Industry is cash/assets less the industry median for that year, where industry is defined using the Fama-French 30 industries. Standardized (excess) cash measures are transformed to have zero mean and unit variance each year. High (Excess) Cash equals 1 if the firm is in the top quartile of (excess) cash holdings in a given year and zero otherwise. Run up is the bidder cumulative abnormal returns over the period -252 to -20, where day zero is the announcement of the bid. Leverage is the bidder’s leverage. Low Q is an indicator variable set to one if the bidder’s market to book value of assets is less than 1. Outside Industry is an indicator variable equaling 1 if the bidder and the target are in different industries as measured by the 30 Fama-French industries. Deal Size is the logarithm of real value of the deal, deflated by the CPI. Relative value is the value of the transaction deflated by the market capitalization of the bidder. Unsolicited is an indicator variable set to one if the bid is indicated as hostile or unsolicited by SDC. Tender Offer is equals 1 if the bid is through a tender offer and zero otherwise. Auction is an indicator variable set to one if there is more than one bidder in the control contest. Defense is set to one if the target uses some sort of takeover defense. Private target equals 1 if the target is listed as private by SDC, and zero otherwise. Subsidiary is set to one if SDC indicates that the target is a subsidiary. Unexpected Bidder is set to one if the acquisition is unexpected based on a first pass logistic regression predicting bidders. Numbers in brackets are p-values based on robust standard errors clustered at the firm level. ***,**, and * indicate significant differences between all stock and all cash bids at the mean or median at the 1, 5, and 10% levels respectively. +++,++, and + indicate significant differences between Mixed and All Cash at the 1, 5, and 10% levels respectively. 41 Panel A: Full Sample All Stock All Cash Mixed High Cash -0.008 [0.006] -0.021 [0.001] -0.001 [0.674] -0.011 [0.143] Run up 0.006 [0.001] 0.008 [0.019] 0.004 [0.127] 0.005 [0.224] Leverage 0.001 [0.852] -0.005 [0.774] -0.002 [0.801] 0.018 [0.329] Low Q -0.006 [0.016] 0.002 [0.825] -0.007 [0.008] -0.009 [0.211] Outside Industry 0.001 [0.635] 0.009* [0.135] -0.003 [0.160] 0.001 [0.846] Deal Size -0.005 [0.000] -0.005 [0.012] -0.003 [0.000] -0.007 [0.006] Relative Value 0.083 [0.000] 0.058 [0.010] 0.085 [0.000] 0.105 [0.000] Unsolicited -0.001 [0.830] 0.000 [0.996] -0.012 [0.095] 0.001 [0.935] Tender Offer 0.019 [0.000] 0.007 [0.657] 0.002 [0.668] -0.013 [0.209] Auction -0.002 [0.679] -0.012 [0.461] -0.005 [0.400] 0.010 [0.316] Defense -0.017 [0.086] -0.047 [0.305] -0.016 [0.169] -0.010 [0.591] Private Target 0.032 [0.000] 0.056*** [0.000] 0.006 [0.145] 0.028++ [0.004] Subsidiary 0.033 [0.000] 0.051*** [0.000] 0.008 [0.061] 0.049+++ [0.000] Unexpected Bidder 0.004 [0.188] 0.004 [0.604] 0.005 [0.083] -0.002 [0.808] Constant -0.005 [0.399] -0.010 [0.487] 0.010 [0.118] 0.003 [0.863] N 5,900 1,334 3,509 1,057 Adjusted R2 0.0611 0.0932 0.0475 0.0667 *** 42 Panel B Full Sample All Stock All Cash Mixed Top Quartile By Year High Harford -0.006 [0.026] -0.020*** [0.001] -0.000 [0.893] -0.005 [0.487] High OPSW -0.004 [0.098] -0.015** [0.026] 0.001 [0.858] -0.009 [0.242] High Industry -0.002 [0.467] -0.013** [0.048] 0.002 [0.470] -0.002 [0.832] Continuous Measures, Standardized Standardized Cash -0.003 [0.021] -0.009** [0.004] -0.001 [0.413] -0.002 [0.513] Standardized Harford -0.003 [0.011] -0.006* [0.001] -0.001 [0.467] -0.002 [0.602] Standardized OPSW 0.001 [0.395] 0.002 [0.398] -0.002 [0.370] 0.004 [0.241] Standardized Industry -0.003 [0.065] -0.007 [0.029] -0.001 [0.362] -0.001 [0.781] 43 Table 8 Overvaluation Decomposition Segmented By Liquidity Overvaluation is calculated using the RKRV (2005) methodology. Firm-specific error, Time-Series Sector Error and Long-Run Value to Book are defined as in RKRV (see Table 5 of their paper, page 579). Total Error is the sum of Firm-Specific Error and Time-Series Sector Error. Cash is cash/assets. Excess Harford [OPSW] is defined as the residual from a first pass regression predicting cash holdings based on Harford (1999) [OPSW (1999)]. Excess Industry is cash/assets less the industry median for that year, where industry is defined using the Fama-French 30 industries. High (Excess) Cash signifies the firm is in the top quartile of (excess) cash holdings in a given year. Numbers in parentheses represent the number of observations for each classification. ***,**, and * indicate the mean is significantly different from 0 at the 1, 5, and 10% levels respectively, using a t-test. P-values within columns are calculated using a t-test correcting for unequal variances. Firm-Specific Error Time-Series Sector Error Total Error Long-Run Value to Book High Cash (14,269) Not High Cash (42,270) -0.0692*** 0.0234*** 0.0280*** 0.0080*** -0.0412*** 0.0313*** 0.9547*** 0.5503*** P-value High vs. Not [0.0000] [0.0000] [0.0000] [0.0000] High Harford (14,221) Not High Harford (42,318) -0.0732*** 0.0246*** 0.0198*** 0.0107*** -0.0535*** 0.0354*** 0.8512*** 0.5856*** P-value High vs. Not [0.0000] [0.0000] [0.0000] [0.0000] High OPSW (14,211) Not High OPSW (42,328) -0.1454*** 0.0488*** 0.0217*** 0.0101*** -0.1237*** 0.0589*** 0.8299*** 0.5928*** P-value High vs. Not [0.0000] [0.0000] [0.0000] [0.0000] High Industry (14,237) Not High Industry (42,302) -0.0760*** 0.0256*** 0.0195*** 0.0109*** -0.0565*** 0.0364*** 0.9199*** 0.5623*** P-value High vs. Not [0.0000] [0.0000] [0.0000] [0.0000] 44 Table 9 Overvaluation Decomposition Segmented By Liquidity and Acquisition Activity Overvaluation is calculated using the RKRV (2005) methodology. Firm-specific error, Time-Series Sector Error and Long-Run Value to Book are defined as in RKRV (see Table 5 of their paper, page 579). Total Error is the sum of Firm-Specific Error and Time-Series Sector Error. Cash is cash/assets. High Cash signifies the firm is in the top quartile of cash holdings in a given year. Firm years are classified as non-bidders if they undertake no acquisition in that year. We classify firm years as a Stock Bid (Cash Bid) if all of the consideration offered for acquisition bids within that year is comprised of equity (cash) components. Numbers in parentheses represent the number of observations for each classification. ***,**, and * indicate the mean is significantly different from 0 at the 1, 5, and 10% levels respectively, using a t-test. P-values within columns and across rows are calculated using a t-test correcting for unequal variances. Not High Cash (N = 42,270) High Cash (N = 14,269) Non – Bidders (50,672) Bidders (5,867) 38,016 4,254 12,656 1,613 Non – Bidders Bidders 0.0155*** 0.0932*** -0.0852*** 0.0567*** P-Value Bid vs. Non [0.0000] [0.0000] Non – Bidders Bidders *** 0.0041 0.0423*** 0.0225*** 0.0718*** P-Value Bid vs. Non [0.0000] [0.0000] Panel A # of Observations Firm-Specific Error Time- Series Sector Error Non – Bidders Bidders 0.0196 0.1355*** -0.0628*** 0.1284*** P-Value Bid vs. Non [0.0000] [0.0000] Non – Bidders Bidders *** 0.5357 0.6811*** 0.9514*** 0.9806*** P-Value Bid vs. Non [0.0000] [0.0145] Not High Cash (N = 3,542) High Cash (N = 1,277) Stock Bid (1,332) Cash Bid (3,487) 809 2,733 523 754 Stock Bid Cash Bid 0.2155*** 0.0601*** 0.2199*** -0.0334* P-Value Stock vs. Cash [0.0000] [0.0000] Stock Bid Cash Bid *** 0.0661 0.0340*** 0.1100*** 0.0410*** P-Value Stock vs. Cash [0.0000] [0.0000] Stock Bid Cash Bid *** 0.2816 0.0941*** 0.3299*** 0.0075 P-Value Bid vs. Non [0.0000] [0.0000] Stock Bid Cash Bid *** 0.7593 0.6654*** 1.1043*** 0.9028*** P-Value Stock vs. Cash [0.0000] [0.0000] Total Error Long-Run Value to Book *** Panel B # of Observations Firm-Specific Error Time- Series Sector Error Total Error Long-Run Value to Book 45 P-value High Cash vs. Not [0.0000] [0.0271] [0.0000] [0.0000] [0.0000] [0.6964] [0.0000] [0.0000] P-value High Cash vs. Not [0.8932] [0.0000] [0.0004] [0.3949] [0.1833] [0.0002] [0.0000] [0.0000] Table 10 Examination of Valuation of Public Targets Overvaluation is measured for the target firms and is calculated using the RKRV (2005) methodology. Firm-specific error, Time-Series Sector Error and Long-Run Value to Book are defined as in RKRV (see Table 5 of their paper, page 579). Total Error is the sum of Firm-Specific Error and Time-Series Sector Error. High Cash signifies the bidding firm is in the top quartile of cash/assets in a given year. We classify firm years as a Stock Bid (Cash Bid) if all of the consideration offered for acquisition bids within that year is comprised of equity (cash) components. Bid premiums are measured using the offer price relative to the target stock price at period t before the bid. Numbers in parentheses represent the number of observations for each classification. ***,**, and * indicate significant differences from 0 at the 1, 5, and 10% levels respectively, using a t-test. P-values across rows and columns are calculated using a t-test correcting for unequal variances. RKV Decomposition for Target High Cash (N = 161) 89 72 0.1526** -0.1281* [0.0073] 0.0671** -0.0333 [0.0056] 0.2197*** -0.1614** [0.0005] 0.8855*** 0.7080*** [0.0346] P-value high cash vs. not Stock Bid (331) Cash Bid (341) Stock Bid Cash Bid P-Value Stock vs. Cash Stock Bid Cash Bid P-Value Stock vs. Cash Stock Bid Cash Bid P-Value Stock vs. Cash Stock Bid Cash Bid P-Value Stock vs. Cash Not High Cash (N = 511) 242 269 0.0347 0.0037 [0.5415] 0.0472*** 0.0233** [0.1287] 0.0820** 0.0270 [0.3031] 0.6866*** 0.5329*** [0.0009] Bid Premium of offer price to stock price at period t before bid Stock Bid (504) # of Observations Cash Bid (469) Stock Bid Cash Bid 1 Day P-Value Stock vs. Cash Stock Bid Cash Bid 1 Week P-Value Stock vs. Cash Stock Bid Cash Bid 4 Weeks P-Value Stock vs. Cash Not High Cash (N = 704) 343 361 30.26 47.52 [0.0001] 33.65 52.46 [0.0005] 40.05 59.82 [0.0004] High Cash (N = 269) 161 108 35.61 44.99 [0.0678] 41.44 52.88 [0.0260] 47.11 67.32 [0.0853] P-value high cash vs. not # of Observations Firm-Specific Error Time- Series Sector Error Total Error Long-Run Value to Book 46 [0.1093] [0.0951] [0.4178] [0.0201] [0.1148] [0.0240] [0.0020] [0.0133] [0.1519] [0.6883] [0.0443] [0.9479] [0.1117] [0.5373] Appendix – Excess Cash Measures We construct three measures of abnormal cash holdings which we refer to as Excess Cash: Harford, OPSW, or Industry. A. Excess Cash - Harford The Harford measure follows equation (1) of Harford (1999) where we run the following regressions: Cash Salesi,t = a i + b1 NetCFO + b5 NetCFO Salesi,t Salesi,t+2 + b 2 ΔRisk Premium t+1 + b3 Recession t + b 4 NetCFO + b6 M Bi,t-1 Salesi,t+1 + b 7 CFOVari + b8Sizei,t-1 + ε i,t where: NetCFO is operating cash flow net of investments23, Risk premium is the difference between rates of AAA and BAA bonds, Recession is a dummy variable set to one for years within recessions as defined by the National Bureau of Economic Research, M/B is the firm’s market to book of assets ratio24, CFOVar is the coefficient of variation using the standard deviation and mean of the firm’s cash flow for up to the prior 10 years, and Size is the natural log of market value of the equity deflated to 2000 dollars using the CPI. As in Harford (1999), the model is separately for each of the Fama-French industries and is estimated with firm-specific fixed effects. The Excess Cash – Harford measure is constructed as the deviation from the industry average prediction (see Harford (1999) page 1977). B. Excess Cash – OPSW We also construct a measure of excess cash based on the model in Opler, et al (1999), where we run the following regression: ln ( Cash ) = α + β1MB + β 2Size + β3Cash Flow + β 4 NWC + β5 RD + β6 Indsigma + β7 Leverage + β8Capex + β9 Dividends + ε where cash is defined as (cash and marketable securities / net assets). Net assets are defined as book value of assets less cash. MB is (market value of equity – book value of equity + book value of assets) / (net assets), Size is the log of assets deflated to 2000 dollars, Cash Flow is (operating income before depreciation – interest – taxes – common dividends) / net assets, NWC is non-cash net working capital deflated by net assets, RD is research and development expenses deflated by sales, Indsigma is the median of the industry’s cash flow volatility for up to the prior 10 years, Leverage is defined as (short term debt + long term debt) / net assets, Capex is capital expenditures deflated by net assets, and Dividends is a dummy variable which equals one if the firm paid common dividends within the year. Following Opler, et al (1999), when R&D is missing, it is set to zero. The model is estimated each year to allow the coefficients to vary across time. The Excess Cash – OPSW measure is the deviation from the exponential of the prediction of the model. C. Excess Cash – Industry Finally, we construct a simple measure of excess cash holdings using the deviation from the annual industry median. Industries are defined using the Fama-French 30 industry classifications. 23 Operating cash flow is operating income before depreciation – interest – taxes – change in noncash working capital (ebitda – xint – txt – Δ(act – lct – che)). Investments are defined as capital expenditures (capx) 24 Constructed as (market value of equity + short term debt + long term debt)/ book value of assets. 47