Can the Source of Cash Accumulation Alter the Agency Problem of Excess Cash Holdings? Evidence from Mergers and Acquisitions ABSTRACT

Transcription

1 Can the Source of Cash Accumulation Alter the Agency Problem of Excess Cash Holdings? Evidence from Mergers and Acquisitions ABSTRACT This study argues that the source of cash accumulation can distinguish the competing motives of corporate cash holdings. The cash-rich bidders with more cash saving from equity issuances (internal operations) tend to have positive (negative) announcement returns. Furthermore, while cash holdings associated with internal saving show a negative effect on post-merger operating performance, the effect of saving cash from equity issuance is positive. These findings indicate that the precautionary motive of saving cash from equity issuance provides the investors with higher confidence on acquisition decisions than the concern of the potential agency problem of excess cash holdings. Keyword: Cash Holdings, Mergers & Acquisitions, Agency Theory, Trade-off Theory 1

2 I. INTRODUCTION Corporate cash holdings are well explained by two theories: the trade-off theory (Keynes, 1936; Myers and Majluf, 1984; Opler, Pinkowitz, Stulz, Williamson, 1999) and the agency theory (Jensen, 1986; Harford, 1999). The former predicts the value of cash holdings would be higher but the latter predicts the opposite. Both predictions are empirically supported (for trade-off theory: Faulkender and Wang, 2006; Pinkowitz and Williamson, 2007; Denis and Sibilkov, 2009; for agency theory: Harford, 1999; Dittmar and Mahrt-Smith, 2007; Orler, 2008). However, it remains unknown when one theory would override the other in explaining corporate cash holdings. In this study, we demonstrate that the sources of cash accumulation can explain these competing predictions. When firms accumulate cash from internal operations, managers enjoy a higher degree of discretion in allocating cash for investment projects with limited external monitoring. Hence, it is easier for managers to pursue their self-interests (Jensen, 1986). In contrast, if firms accumulate cash from equity issuances, the potential usages of these cash holdings (including both of investment demands and managerial personal interests) would be taken into account in valuation by investors. With higher transaction costs of saving cash from equity issuance than from internal operations and with stronger monitoring from financial markets than from internal governance mechanism, it is reasonable to predict that the trade-off theory overrides the agency theory when the cash is accumulated from equity issuances rather than from internal operations. Given that external financing is more expensive than internal financing and that cash holdings create zero or lower return than capital investments would, it is therefore necessary to examine whether and how the sources of cash accumulation alter the motive of cash holdings to be more precautionary and less agency motive or vice versa. 2

3 Prior studies verify the agency problem of excess cash holdings by showing that cash-rich firms are more likely to engage in value-decreasing mergers and acquisitions (M&As) and that poorly governed cash-rich firms are more likely to over invest (Harford, 1999; Dittmar and Mahrt-Smith, 2007; Oler, 2008; Harford, Mansi, and Maxwell, 2008). To clarify our objective of distinguishing the agency motive from the precautionary motive, we incorporate our argument (the sources of cash accumulation) to analyze the impact of excess cash holdings on firms' M&As. The reason to examine the M&As is because M&As are major corporate investments requiring large scale of capital, and the prior studies in examining the agency problem of excess cash holdings also use M&As as the samples of analysis (Harford, 1999; Oler, 2008). To better compare with prior studies, we adopt the similar empirical setting as they did. Accordingly, the purpose of this study is to reconcile two competing views by examining the performance of M&As in cash-rich firms with different sources of cash accumulation. Because the value of cash holdings is higher in firms with higher transaction and precautionary motives (Faulkender and Wang, 2006; Pinkowitz and Williamson, 2007; Denis and Sibilkov, 2009), we predict fewer value-decreasing M&As in firms with more cash accumulation from equity issuance and more value-decreasing M&As in firms with more cash accumulation from internal operations. A limited number of studies examine the relation between corporate cash holdings and acquisition performance. On the one hand, Harford (1999) investigates the Jensen s agency theory of cash holdings through the use of a basic cash regression to classify whether a firm is cash-rich and then through the use of a probit model to identify whether a firm is an expected bidder. He finds that the market reacts negatively to unexpected bidders with excess cash, and that cash-rich firms are associated with negative abnormal operating performance after M&As. His findings support Jensen s agency theory. In addition, Oler (2008) investigates whether the level of corporate cash holdings before M&As can be an indicator of 3

4 post-merger performance. Consistent with the prediction based on the agency theory, he finds that firms with higher level of pre-merger cash holdings are associated with poorer post-merger performance. On the other hand, Gao (2013) argues that corporate cash has the adverse selection effect and finds that investors would react negatively to the stock-financed acquisitions of the firms with excess cash, especially when information asymmetry is high. Although Gao (2013) finds that cash-rich firms experience negative stock returns to their M&A announcements, the higher post-merger operating performance in firms with more cash reserves rules out the possible explanation of agency problem to the negative announcement return. Furthermore, Gao and Mohamed (2016) argue that cash reserves do not always cast negative impact on firms, and find a positive effect of excess cash holdings on both announcement return and post-merger operating performance when a bidder demonstrates a stronger precautionary motive, measured by financial constraints, and better shareholder right, measured by E-index. These inconsistent findings between the bidding performance and excess cash holdings from these studies indicate that both precautionary and agency motives are observed in the corporate cash policy. Given that their impacts on firm value are opposite, it is crucial to distinguish the precautionary motive from the agency motive in the corporate cash holdings. In this study, we adopt the model of McLean (2011) to estimate the cash saving rate of equity issuance and then the models of Harford (1999) and Oler (2008) to examine both the announcement effect and post-merger operating performance of unexpected bidders. By examining the U.S. firms from 1978 to 2010, we find that cash-rich bidders with more cash accumulation from equity issuances tend to have a positive announcement effect, but cash-rich bidders with more cash accumulation from internal operations tend to have a negative announcement effect, particularly for unexpected bids. Meanwhile, while cash holdings associated with internal saving show a negative effect on post-merger performance 4

5 (Adjusted ROA), the effect of saving cash from equity issuances is positive. The findings not only identify the potential agency problem of cash holdings associated with cash accumulation from internal operations, but also verify that the behavior of accumulating cash from equity issuances, the expensive source, is not for personal interests but for the firm value enhancement. The high transaction costs and market monitoring from equity issuances make managers accumulate the expensive cash only if they believe the future investment opportunities can lead to a higher return than the cost of accumulating cash from equity issuances. As the cash saving from equity issuances is more expensive than that from internal operations, firms would accumulate cash from internal source before from external sources. That is, the cash holdings within firms would be from both internal and external sources. Therefore, we have to further distinguish the effect of cash sources when both of internal and external sources are considered in empirical analyses. We thus examine the announcement effect and post-merger performance for firms with high level of cash accumulation from equity issuances and with high level of cash accumulation from internal operations versus firms with high level of cash accumulation from equity issuances yet with low level of cash accumulation from internal operations. We find that the positive effect from cash savings from equity issuances is higher when the cash saving from operations is lower. This finding verifies that the sources of cash accumulation can identify the motive of corporate cash holdings and that the precautionary motive generally dominates the agency motive of cash holdings when cash is accumulated from the equity issuances. This study contributes to the literature in three dimensions. First, we provide empirical evidence showing that the source of cash accumulation is the key to distinguish the agency perspective and the trade-off perspective of excessive cash holdings. The prior studies find evidence supporting the two opposite perspectives on cash-rich firms (e.g. Harford, 1999 vs. 5

6 Gao and Mohamed, 2016); however, they focus on the level of excess cash holdings of firms. Our study complement the previous studies by proving that the source of cash accumulation has a more important implication for investors than do the level of cash holdings. Specifically, we provide a relatively intuitive method to measure the effect of the precautionary motive on cash savings. We further contrast the impact of cash holdings consecutively accumulated from external equity issuances versus the impact of cash holdings accumulated from internal operations on acquisition performance. Such distinction allows us to understand the relative impact of the trade-off theory and the agency theory in explaining the corporate cash holdings policy. Second, this study adds to the cash-holding literature. Previous studies have investigated the value of cash holdings in different aspects. For example, the value of cash holdings are found higher when firms have more investment opportunities (Lang, Stulz, and Walkling, 1991); are financially constrained (Faulkender and Wang, 2006; Denis and Sibilkov, 2009); have good corporate governance (Ditmmar and Mahrt-Smith, 2007; Harford, Mansi, and Maxwell, 2008); are in countries with better investor protection (Pinkowitz, Stulz, and Williamson, 2006); have persistently excess cash holdings (Mikkelson and Partch, 2003). Our findings fill in an unexplored gap (i.e. the source of cash accumulation) in this stream of literature. Examining the performance of mergers and acquisitions in cash-rich firms, we verify that the internal source of cash accumulation is more associated with decisions of value-decreasing investment in cash-rich firms than the external source of cash accumulation. Last but not least, this study extends the literature of mergers and acquisitions. Prior studies have focused on the impact of payment method and financing choices on the acquisition performance. For instance, Schlingemann (2004) finds that the bidders valuation is positively associated with the equity financing for acquisitions. Similarly, Martynova and Renneboog (2009) show that acquisitions financed with debt outperform acquisitions 6

7 financed with internal cash. These two studies emphasize the monitoring mechanism and high costs resulted from external financing for explaining acquisition performance. However, they are not able to explain the performance of cash-rich firms, who probably have been accumulating cash from either internal or external source way before the acquisitions, by only looking into payment methods and financing choices at the moment of acquisitions. Different from their studies, the impact of cash accumulation sources emphasized in this study then allows us to explain the value of acquisition in a more comprehensive way. For the organization of this paper, we review the related literature and develop our hypotheses in Session II. Sessions III describes our data and sample selection. The empirical analyses are then discussed in Session IV. Session V concludes the paper. II. LITERATURE AND HYPOTHESES There is a general assumption in prior studies that firms accumulate cash mainly from internal operations (Opler et al., 1999; Almeida, Campello, and Weisbach, 2004; Han and Qiu, 2007). However, some recent studies show that a significant portion of the proceeds from equity issuance is saved as cash reserves (Kim and Weisbach, 2008; Hertzel and Li, 2010). McLean (2011) further finds an increasing proportion of equity issuance saved as corporate cash reserves, while such increasing trend is not significant from debt or internal cash flow. The behavior of saving cash from equity issuances could be motivated by two reasons: market timing and financing for investment (McLean, 2011). The share overvaluation caused by the information asymmetry between insiders and outsiders allows firms to time the market. Prior studies find that on average these issuing firms appear to have high growth and high profitability before they issue; hence the market-to-book ratio and stock returns are high accordingly. However, both of their stock returns and operating performance after issuances 7

8 are apparently poorer than before (Loughran and Ritter, 1995; 1997). According to Loughran and Ritter (1997), both managers and investors are overly optimistic about firms profitability and growing ability. When investors overestimate a firm s value based on its past performance, the firm may choose to issue equity for a relatively low cost, even though it may not really have good investment opportunities. Consequently, it may invest in value-decreasing projects. Similarly, Kim and Weisbach (2008) find that equity-issuing firms with high market-to-book ratio are more likely to hoard the proceeds in cash or invest in R&D, while equity-issuing firms with low market-to-book ratio are more likely to invest in inventory, capital expenditures, or acquisitions. Their finding is in line with the cash holding policy of financially constrained firms (Almeida, Campello, and Weisbach, 2004), suggesting that firms with low market-to-book ratio face a relatively high cost to issue equity. The fact that firms facing high cost of equity capital would still want to issue equity indeed shows their precautionary motive for future investment opportunities. Equity issuance is costly. The low market-to-book firms with investment opportunities face higher transaction cost at issuance. The high market-to-book firms and market-timing firms may enjoy lower cost of equity at issuance, but poor investment decisions could discourage investors and therefore increase future cost of capital. It is the external monitoring that imposes potential costs on equity-issuing firms. Accordingly, the literature further digs into the situation that firms choose equity financing over other less expensive financing sources, e.g. debt. Dittmar and Thakor (2007) find that firms tend to issue equity to fund the investments only if the agreement with investors upon the value of investment is reached. That is, managers are willing to issue equity only when they believe investors are very likely to endorse their investment decisions, i.e. good decisions. Hertzel and Li (2010) decompose market-to-book ratio to identify high-growth-option firms and mispricing firms, and find that high-growth-option firms tend to invest more and experience higher stock return after equity 8

9 issuances than mispricing firms. Their result implies that investors are likely to reach an agreement on the value of high-growth-option firms investments. However, when the agreement is not reached (e.g. mispricing firms), issuing equity will be punished by future stock price drops. Given the high cost of equity issuance, it is then reasonable to hypothesize that firms with higher precautionary motive of cash holdings would be more likely to accumulate cash from equity issuances because they can better convince the investors on the purpose of accumulating cash for future investments than firms with low/none precautionary motive. Furthermore, Mclean (2011) attributes the behavior of accumulating cash from equity issuances to the precautionary motive rather than the market-timing motive. We therefore predict that the investors would react in favor of acquisition announcements for cash-rich firms that accumulate cash from equity issuances rather than other cash-rich firms. When investors do not agree with firms motive of issuing equity, the valuation discount makes the equity issuances very costly. In addition to such market monitoring power via valuation, issuing equity also requires a very high level of flotation cost that is however zero if firms generate cash from internal operation. The cash from internal operation is a lack of market monitoring because the usage of internal cash is within managers discretionary. Especially, if firms can maintain the dividend payouts to shareholders, building large cash from internal operation would not necessarily lead to value discount for such behavior. That is, accumulating cash would cost firms much more from equity issuances than from internal operation. Thus, firms would issue equity and save large portion of net proceeds only if the expected return from investment opportunities is higher than the cost associated equity issuances. In addition to the reason that firms expect to have higher return from future investments than the cost associated with accumulating cash from equity issuance, the other issue is that 9

10 these firms normally have higher cash flow volatility than those firms that can save cash from internal operations. When firms can keep saving cash from internal operations, they have low incentive to build cash because they can easily access to sources either internally or externally whenever they have financing need for their investments. If firms keep building cash from their internal operations, then the market would be concerned with the uses of their large cash holdings. Therefore, we predict that investors would not react in favor of acquisition announcements for cash-rich firms accumulating cash from internal operations. Hypothesis 1a: The investors react favorably to the acquisition announcements for cash-rich acquirers that accumulate cash from equity issuances. Hypothesis 1b: The investors do not react favorably to the acquisition announcements for cash-rich acquirers that accumulate cash from internal operations. The post-merger operating performance should reflect abnormal announcement returns, which are supposed to be based on investors rational expectation on future performance (Healy, Palepu, and Ruback, 1992). Moreover, Oler (2008) suggests that market may not fully understand the implications of an acquirer s behavior immediately, that is, announcement return could only capture part of the early reaction. When Harford (1999) and Oler (2008) both show that cash position has negative relation with post-merger operating performance, the source of cash accumulation is not considered. However, when cash is accumulated through the equity issuance, the post-merger performance would not be still unfavorable especially when firms issue equity for future investments. The theoretical model of Hertzel and Li (2010) suggests that firms saving the proceeds of equity issuance for investments have favorable investment performance. In addition, as suggested by Dittmar and Thakor (2007), equity issuing is more likely to happen when the interests between managers and shareholders are aligned, suggesting that managers consider the long-term value of firms 10

11 in equity issuing. Accordingly, the investors would anticipate better synergies for the acquisition decisions made by the cash-rich firms saving cash from equity issuance than other cash-rich firms. Therefore, we expect that cash-rich firms with cash from equity issuances undertake valuable acquisitions and are supposed to have better post-merger operating performance. Hypothesis 2: Cash-rich firms that accumulate cash from equity issuance experience better long-term post-merger performance than other cash-rich firms. III. DATA AND SAMPLE The data used in this study consists of the acquisition data from Thomson Reuters Securities Data Corporation (SDC) database, firms financial variables from S&P s Compustat database, and stock return data from CRSP database. We identify all acquisitions deals in the United States from 1978 to Following the literature, we exclude the deals in financial (SIC code 6000~6999) and utility (SIC code 4900~4999) industries because their investment decisions are highly influenced by the regulation. Following Harford (1999), we exclude deals in which private target firms are involved. Because the objective of this paper is to investigate the role of firms cash saving behavior in acquisition decisions, we include only acquisitions with full or partial cash payment and require the value of deals to be no smaller than 1% of acquirer s market capitalization. Given that cash deals to be found outperforming stock-exchange deals and often experiencing positive returns in the literature (Ghosh, 2001; Linn and Switzer, 2001; Bouwman, Fuller, and Nain, 2003; Martynova and Renneboog, 2009), our sample selection would be in turn more conservative than the sample 1 Following previous studies, we include deals that are categorized in acquisition of assets, merger, or acquisition of interests and exclude deals that are categorized in buyback, exchange offer, or recapitalization. In SDC, there are some deal are recorded as several deals in one single announcement day. To address this 11

12 consisting of all payment methods. In other words, the result showing consistently negative impact of excessive cash holdings and positive impact of cash saving from equity issuance in cash deals, would be more robust than it would be in deals consisting of all types of payments. In the following tests, the sample firms are classified by the criteria of whether they are cash-rich or non-cash-rich firms. 2 We collect financial variables from the Compustat database from 1971 to After deleting observations with missing value of the variables and observations with negative equity, a total of 83,863 firm-year observations is used to perform the cash management model of Harford (1999). Panel A of Table 1 reports the summary statistics of key variables in the cash model. The mean and median values of key variables for cash-rich and non-cash-rich are reported in the Panel B. On average, cash-rich firms had higher market-to-book ratio of assets, but lower leverage ratio, suggesting that cash-rich firms may have more growth opportunities that normally involve firms with higher risk and thus less capability to raise fund from debt financing. As the acquisition data are available since 1978, the number of firm-year observations used to estimate the bidding probability becomes 62,613. The next dataset is to estimate the market reaction to acquisition announcements, in which both of the complete and incomplete deals are included. For the bidder who announces more than one deal in a single year, we keep the bidder s first completed deal of that year. After deleting missing value, there are 1,260 deals; 173 of which are cash-rich firms and 1,087 of which are non-cash-rich firms. Finally, to test post-merger performance, we require the deal to be completed and then 2 We first group sample firms into 12 industry groups based on the industry classification of Fama and French. Then, following Harford (1999), we perform the cash management regression on the panel data for each industry groups, controlling for the firm-specific fixed effect. For each industry-year, we obtain the median of the Cash/Sales predicted by the cash regression model. One firm is classified as cash-rich if its actual value of Cash/Sales is greater than the sum of the industry median predicted value of Cash/Sales and one standard deviation of its Cash/Sales. 12

13 require non-missing data for the three-year period after the effective date of each deal. After deleting missing data, we end up with 900 observations to perform the model of testing post-merger performance. IV. EMPIRICAL DESIGN AND ANALYSIS To examine whether the source of cash accumulation would affect the relation between cash holdings and acquisition performance, we estimate how much cash the company has been saving from the proceeds of equity issuances and from internal operating cash flows, respectively and we compare the market reaction and post-merger operating performance between firms with higher cash accumulation from equity issuances and firms with high cash accumulation from operating cash flows. 4.1 Measuring Cash Saving from Equity Issuances and from Internal Operations To estimate cash saving from equity issuances and from internal operations, we employ the cash saving model of McLean (2011) to obtain the estimator of the saving rate for each source. 3 Regressions are estimated on cross-sectional data every year for each industry based on the Fama-French 12 industry classification. Since it is not likely to track the exact portion expensed and the exact portion saved out of each source, we assume that firms in the same industry share the same saving rate of each source. This assumption is based on the findings of prior researches that firms investment and financing strategies are subject to industry characteristics. For example, in a competitive industry, firms forced to cut investment due to high leverage would suffer from future growth decline if their industry rivals hold relatively high financial capability (Chevalier and Scharfstein, 1996; Campello, 2003). Furthermore, the 3 McLean (2011) models the change of cash as the function of cash inflows from equity issuance, cash inflows from debt, operating cash flows, other cash inflows, and firm size. The coefficient on each cash source can be viewed as the saving rate of each financing source, according to McLean s interpretation. 13

14 negative impact of high leverage on future growth is significant especially for firms in low Tobin s q industries (Lang, Ofek, and Stulz, 1996). These findings indicate that a firm s financing decision is easily affected by its industry rivals financing structure and investment opportunities within the industry. It is therefore reasonable to use industry saving rate as the firm s saving rate; in particular, the precautionary motive for accumulating cash from equity issuances would also possibly be determined by the threat of industry rivals. Each firm s cash from equity issuances (Issue-saving) in each year would thus be calculated as the firm s new issues multiplied by the industry saving rate of equity issuances in the year. Similarly, we measure the firm s cash saved from internally generated cash flows (OCF-saving) by multiplying the firm s operating cash flows with the industry saving rate of operating cash flows obtained from the model of McLean (2011) Expected vs. Unexpected Bidders Following Harford (1999), we classify M&A bids into the expected bid and unexpected bids. Because investors would anticipate the poor merger performance of cash-rich firms, they would reflect their belief on stock prices at the time of M&A announcements. The negative announcement returns would be more prominent for the unexpected bidders because the investors have reflected the expectation of poor investment decisions before the M&A announcements to those expected bidders but not to unexpected bidders. The probit model of Harford (1999) is adopted to estimate the likelihood of carrying out acquisition decisions. 5 We modify the model by adding one more independent variable, Issue-saving (or OCF-saving) as investors may also modify their anticipation after considering the firm s source of cash saving. The model is specified as follows: 4 Operating cash flows is defined as net income plus depreciation and amortization (McLean, 2011) 5 Following Harford (1999), here we include all the acquisition deals including complete and incomplete deals. 14

15 P(bidder =1) = a i + b 1 CashDev i,t 1 + b 2 Issue-saving i,(t 1,t 4) +b 3 Avg. AR i,(t 1,t 4) + b 4 Sales growth i,(t 1,t 4) + b 5 Noncash WCP i,(t 1,t 4) +b 6 Leverage i,(t 1,t 4) + b 7 MB i,(t 1,t 4) + b 8 PE i,(t 1,t 4) + b 9 Size i,t 1 +Wave Yearly Dummies +ε it, (1) The dependent variable is set to one if the firm announces an acquisition and zero otherwise. CashDev is the measure of excessive cash holdings. The predicted level of cash holdings is first estimated by the cash management model of Harford (1999), which is performed by the panel regression with firm fixed effect for the period from 1971 to We then find the median value of the estimated cash holdings for each industry-year. CashDev is the deviation of a firm s actual cash holdings from the industry median cash holding, measured at the end of year t 1. According to Harford (1999), CashDev is positively related to the probability of bidding. Issue-saving is the firm s cumulative saving amount from equity issuance over the period from t 4 through t 1. The control variables are defined in the same way as in Harford (1999). Avg. AR is the average abnormal stock returns over year t 4 through year t 1. The abnormal stock returns are the residual from the daily market model, which regresses excess stock returns on CRSP value-weighted index returns for the 252-day period before the day that abnormal returns are estimated. Sales growth is calculated as current year sales less prior year sales, divided by prior year sales. Both of the past abnormal return and the sales growth are positively related to the probability of bidding (Roll, 1986; Harford, 1999). Noncash WCP, calculated as net working capital minus cash and cash equivalents, normalized by total assets, is expected positively related to the probability of bidding. Leverage is measured as the ratio of book value of debt to market value of equity. MB is the ratio of the market value to the book value of equity. PE is the ratio of stock price to earnings per share. Size is the natural logarithm of book value of total assets, and expected to be positively related to the probability of bidding. Except that Size is measured at the end 15

16 of year t 1, Sales growth, Noncash WCP, Leverage, MB and PE are averaged over the period from year t 4 to year t 1. Instead of using yearly dummy for all years, we include the variables of Wave Yearly Dummies, those years that are identified as the merger-wave year or the merger-drought year. 6 Table 2 reports the statistics of variables of equation (1). The sample used in this equation covers all firms from 1978 to 2010, with 62,613 observations. On average, there is 2.16% of firm-year observations involved in bidding announcements. This ratio is similar with that of Harford (1999). 7 The mean value of cash deviation in the sample is about The mean values of Issue-saving and OCF-saving are around 0.06 and 0.07, respectively. Table 3 reports the estimations of the probit model of bidding decisions. Model (1) is the same specification as in Harford (1999), and it consistently shows a positive coefficient of Cash Deviation, indicating that firms with excessive cash holdings is more likely to conduct mergers and acquisitions. All other control variables also demonstrate consistent effects on the bidding decisions as expected. The estimations including Issue-saving, OCF-saving, and both together are shown in models (2), (3), and (4), respectively. The coefficients on CashDev are all positive and significant across all models. When controlling for the degree of excessive cash holdings, more OCF-saving is associated with higher likelihood of carrying out an acquisition, while the effect of Issue-saving was insignificant. The positive effect of OCF-saving suggests that, other things being equal, bidding firms are more likely to be the firms who can save more cash from their operating income in the past four years. This is 6 To identify merger-wave year and merger-drought year, we follow the method of Harford (2005), which first identifies the highest 24-month concentration of merger bids involving firms in each decade. Then, a simulated distribution of the highest concentration for each decade is obtained. If the actual highest concentration exceeds the 95th percentile of the simulated highest concentration, that 24-month period is determined as a wave. Similarly, we compare the lowest concentration of merger bids with simulated lowest concentration of merger bids to determine the drought year. Accordingly, 1988, 1989, 1997, 1998, 1999, 2000, and 2001 are identified as merger-wave years. 1980, 1981, 1990, 1991, 2008, and 2009 are identified as merger-drought years. We then create dummy variables for above all years. 7 For the sample period 1977 to 1993, Harford (1999) estimates probit regression with 23,686 firm-year observations, out of which there are 487 bids. 16

17 consistent with the finding of Dasgupta, Noe, and Wang (2011) that the positive operating cash flow shock increases the firm s liquidity by replacing costly external financing in the current year and thus increases the firm s ability to refinance and invest in the later years. The insignificant effect of Issue-saving could be due to the following two reasons. First, after equity issuance, there will be an increasing scrutiny of shareholders that curbs over-investment. Second, it is also possible that the effect of Issue-saving is captured by CashDev as we will show later that cash-rich firms issue and save more than non-cash-rich firms. The result also shows that firms with higher growth and bigger size were more likely to make acquisition bids. From this probit regression, we identify the cutoff points to distinguish expected bidders from unexpected bidders. We first derive the predicted probabilities for each firm-year, and then create the distribution of the probability for bidder and non-bidder, respectively, as in Figure 1. We depict the PDFs from the probit regression with the same design of Harford (1999) in Graph (1) and the PDFs from that probit regression with additional variables, Issue-saving (OCF-saving) in Graph (2) (in Graph (3)). Finally, we incorporate all additional variables in the probit regression and depict its PDFs in Graph (4). The cutoff point is the point on the graph where these two probability distributions cross. As defined in Harford (1999), unexpected bidders are those firms whose predicted probability of bidding is lower than this cutoff point. We then discuss the effect of excessive cash holdings and sources of cash savings on the market reaction to the acquisition announcements and on the post-mergers performance in the following two subsections. 4.3 Market Reaction to the Acquisition Decisions Instead of examining how market reacts to cash-rich firms acquisition decisions by 17

18 examining the bidder s abnormal returns surrounding the date the bid is announced, this study further considers the market reaction to firms with cash accumulated from equity issuances versus from internal operations. Thus, we modify the model of Harford (1999) as follows: CAR(-5,+1) i = a i + b 1 CAR(-252,-20) i + b 2 CashDev i,t 1 + b 3 Issue-saving i,(t 1,t 4) +b 4 CashDev i,t 1 Issue-saving i,(t 1,t 4) + b 5 Leverage i,(t 1,t 4) +b 6 All Cash Offer i + Wave Yearly Dummies +ε it, (2) The dependent variable, CAR (-5, +1), is the acquiring firm s cumulative abnormal returns of the announcement period from five day before to one day after the announcement date. CAR (-252, -20) is the acquiring firm s cumulative abnormal returns from 252 to 20 days before the announcement date. The abnormal return is the residual from the daily market model that regresses excess stock returns on CRSP value-weighted index returns. The estimation window is -253 day through -370 days relative to the day of acquisition announcement. CashDev and Issue-saving is as defined in equation (1). Issue-saving i,(t-1,t-4) is the sum of the cash saving from equity issuance from year t-4 to year t-1. For comparison purpose, we replace Issue-saving with OCF-saving in equation (2). CashDev, Issue-saving, and OCF-saving are interacted with expected bid and unexpected bid indicators, which are distinguished by the cut-off point obtained from Fiqure 1. The Unexpected Bid dummy is also added to the regression. Leverage i,(t-1,t-4) is the average leverage ratio from year t-4 to year t-1. All Cash Offer is the dummy that equals one if the deal is paid wholly in cash, and zero otherwise. Wave Yearly Dummies are set to be one for the years identified as the merger wave years or the merger drought years. The coefficient of CashDev can be negative if it indicates more agency problems, and it can be positive if it indicates higher precautionary motive. A higher level of leverage implies a higher disciplinary effect imposed on managers, so the coefficient of Leverage is expected to be positive. The summary statistics of key variables in equation (2) for all acquirers, and cash-rich 18

19 acquirers versus non-cash-rich acquirers are reported in Panels A and B of Table 4, respectively. The average value of announcement return, CAR (-5, +1), is positive but close to zero. The positive announcement effect is mainly due to that the sample in this study covers the acquisition deals paid by cash either in full or in partial. Acquirers involved in cash acquisitions often experience higher announcement return than that using only stock acquisition because stock acquisition would imply that a bidding firm is overvalued and realization of synergies is uncertain (Travlos, 1987; Wansley, Lane, and Yang, 2005; Martynova and Renneboog, 2009). This explains why our result shows positive cumulative abnormal returns on average. Panel B shows insignificant difference in the cumulative abnormal returns between cash-rich and non-cash-rich firms. This result might be inconsistent with the agency problem with free cash flows; however, it might be due to the possible correlation with other factors in the univariate analysis. Thus, we could not conclude from this two-sample analysis. Another thing to note in Panel B is that Issue-saving is higher and OCF-saving is lower in cash-rich firms than in non-cash-rich firms. This suggests that relative to operating income, equity issuance is a more important source of cash accumulation for cash-rich firms. Consistent to the literature, cash-rich firms tend to have lower leverage than non-cash-rich firms. Higher cash saving from equity and lower leverage in cash-rich firms suggest that cash-rich firms are more likely to be firms with lower borrowing capability but higher growth opportunities. The effect of cash accumulation from equity issuances on the market reaction to acquisition announcements is reported in Table 5. The model (1) shows the estimation of the model similar to the model of Harford (1999). The coefficient on CashDev(Unexpected) is insignificant in our result, while the coefficient is significantly negative in Harford (1999). In an untabulated test, we replicate Harford s model by excluding Wave Yearly Dummies and 19

20 find the similar result with his. We however argue that it is necessary to include Wave Yearly Dummies in this study given that our sample period covers two macroeconomic shocks: the internet bubble during the period from 2000 to 2001 and the financial tsunami during the period from 2008 to 2009, which are found to have significant negative effect on the announcement returns. Before adding Wave Yearly Dummies to the regression, the significantly negative effect of CashDev could be simply driven by the shocks. The result in model (1) suggests that without considering the source of cash saving, we may not be able to fully understand investors reaction to the cash holdings of cash-rich firms. In model (2), Issue-saving is added to serve our purpose of examining the external source of cash saving. The interaction of CashDev with Issue-saving is positive and significant. This finding shows that although the excessive cash holdings would raise the market s concern of potential over-investments, the market is more confident on all M&A decisions made by firms that have accumulated cash from equity issuances. In addition, this positive effect is even stronger for unexpected bids than expected bids, indicating that the decision of accumulating cash from equity issuances demonstrates the managers confidence on future investment opportunities, and the investors reach a common agreement on managers decisions of financing and investment. Additional control variables are included for a more robust result, as shown in models (3) and (4). We further control for growth options, cross-industry mergers and acquisitions, relative size of target firms and acquirer firms, and hostile takeover, which are related to merger performance. The results are similar to models (1) and (2). From this finding, we can conclude that the precautionary motive of cash holdings (measured by Issue-saving) dominates the agency motive of cash holdings (measured by CashDev) in explaining the relation between cash holdings and M&A decisions when firms use the equity issuances to accumulate cash holdings for future acquisitions. To verify our argument that the source of cash could distinguish the agency motive from 20

21 the precautionary motive, we further examine the effect of OCF-saving on the announcement returns, as reported in Table 6. In contrast to Issue-saving, OCF-saving shows an incremental negative and significant effect on the relation between CashDev and announcement returns. This result suggests that the market reacts more negatively to the M&A announcements of firms that hold more excessive cash holdings, especially from their operating income. While the agency concern of cash holdings is well attested by the previous literature, we provide evidence showing that the agency concern is demonstrated if the firm s cash accumulation comes more from internal operations. The findings from Tables 5 and 6 (the negative coefficient on CashDev, the positive coefficient on the interaction between Issuing-saving and CashDev, and the negative coefficient on the interaction between OCF-saving and CashDev) indicate that both theories of cash holdings are co-existed and that firms cash saving behavior can help to distinguish the relative impact between the two theories in explaining corporate cash holdings. 4.4 The Post-Merger Performance An essential question to ask about the role of cash sources in explaining future M&A decisions is whether the acquisitions conducted by firms with cash saving from equity issuance can really create value to shareholders. In addition to the market reaction to the acquisition announcements, we also examine the long-term post-merger performance for sample firms. Following the literature, we investigate the acquiring firm s post-merger operating performance by looking into the average adjusted return on asset (Adj. ROA) for the three-year period after M&As. To measure the Adj. ROA, for each year, the firm s three-year 21

22 average ROA 8 is deducted by the median value of three-year average ROA of all cash-rich firms in its industry. Next, we examine the post-merger stock return in a multivariate analysis. The model is specified as follows: Adj.ROA i,(t+1,t+3) = a i + b 1 CashDev i,t 1 + b 2 Issue-saving i,(t 4,t 1) +b 3 CashDev i,t 1 Issue-saving i,(t 4,t 1) + b 4 Adj. ROA i,(t 3,t 1) +b 5 Diff-industry it + b 6 Hostile it + b 7 NOA it + b 8 Accruals it +b 9 Sales growth it + b 10 Small acquirer it + b 11 Relative size it +Year Fixed Effect +ε it, (3) where Adj.ROA, CashDev, and Issue-saving are as defined above. Similarly, we replace Issue-saving with OCF-saving in equation (3) to test whether cash-rich firms who have been saving higher cash from their earnings would have better or worse long-term performance after M&As. We control for the acquirer s past operating performance with its mean Adj.ROA over the past three years. We follow Oler (2008) for the rest of the control variables. Diff-industry is a dummy variable set to one if the acquirer and the target are in different industries. Prior studies indicate that diversification is more likely to destroy firms value (Healy, Palepu, and Ruback, 1992; Moeller, Schlingemann, and Stulz, 2004). Hostile is a dummy variable that equals one if the deal is denoted as hostile in the SDC database. NOA is net operating assets 9 scaled by the lagged total assets. Accruals are calculated as change of total assets minus change of cash minus change of total liabilities plus change of short-term investment minus change of preferred stock. Sales growth is computed as the change in sales from year t-1 to year t, divided by sales in year t-1. Small acquirer is a dummy variable that equals one if the market capitalization of the acquirer is below the 25th percentile of all 8 Following Harford, Humphery-Jenner, and Powell (2012), ROA is calculated as the operating income before depreciation scaled by total assets. 9 We use the simplified definition following Oler (2008) to calculate net operating assets as book value of debt and preferred stock minus cash and short-term investment minus other investments plus book value of equity and minority interest. 22

23 NYSE firms. Relative size is the target s market capitalization divided by the acquirer s market capitalization. Finally, the year fixed effect is added to the regression. The industry fixed effect is not included because our dependent variable and independent variables have already been adjusted by the industry effect. By so doing, we are able to control for most of the unobserved factors related to industry characteristics. The summary statistics of key variables in equation (3) can be found in Table 4. The mean Adj. ROA was for the whole sample, while the mean Adj. ROA was for cash-rich firms and for non-cash-rich firms, suggesting that non-cash-rich firms have better post-merger operating performance than cash-rich firms. Panel A of Table 7 exhibits the results of testing whether cash-rich acquirers still underperform in the post-merger period if they have saved large proceeds of from equity issuance within the 4-year period before the acquisition. Model (1) shows that the post-merger stock returns significantly decrease with CashDev, consistent with the agency problem of over-investment related to excessive cash holdings. The previous test of announcement returns suggests that investors may not fully understand the implication of acquirers excessive cash holding; however, the post-merger performance of these cash-rich acquirers shows that their acquisitions could reduce their firm values. In model (2), Issue-saving and its interaction with CashDev are added to estimate the effect of saving cash from equity issuances on post-merger performance. While the coefficient on CashDev remains negative and significant, the coefficient on the interaction term is insignificant. In contrast, the result of model (3) shows that the coefficient on CashDev become insignificant, but its interaction with OCF-saving still shows significant and negative effect on post-merger performance. This verifies that most of the agency concern is more likely to concentrate on firms that have saved more cash from operating income before acquisitions. We then include both Issue-saving and OCF-saving in model (4) and find the similar result. 23

24 In Panel A, we are unable to find a significant effect of Issue-saving on post-merger performance. However, we are concerned with the possibilities of positive correlation between Issue-saving and OCF-saving. Specifically, firms that save more cash from operating income (i.e. profitable firms) are able to raise a substantial amount of cash from external financing. We therefore further partition firms with high Issue-saving into firms simultaneously with high OCF-saving and low OCF-saving, as shown in Panel B of Table 7. The result of subsample test shows that only in the case that firms simultaneously have high Issue-saving and high OCF-saving, the excessive cash holdings is significantly negatively related to post-merger performance. This ascertains our argument that cash saving accumulated more from equity issuance relative to internal source is associated with more precautionary motive and less agency problem, and vice versa. Our results regarding post-merger operating performance overall confirm the market reaction around acquisition announcements. As investors expect that firms have better investment performance if they accumulate cash from share issuance to fund their financing needs, the post-merger performance prove that their expectations are right. Our results also support McLean (2011), who suggests that the increase in saving rate of equity issuance is resulted from increasing precautionary motive. 4.5 Robustness for Endogeneity Issue The concern of endogeneity in our study is that whether firms can issue equity depends on some omitted variables that simultaneously affect the market reaction or firm performance. For example, a firm with good reputation or proficiency of seeking good investment targets can attract more investors than other firms. To address the potential endogeneity problem, we adopt the Heckman s (1979) two-stage estimation method. In the first stage, we perform a probit regression to model the likelihood 24

25 of Issue-saving. Following McLean (2011), the model is specified as follows: P(Issue - saving i,( t 4, t 1) = 1) = a + b Leverage 3 + b PREC 6 i + b Cash Flow 1 i,( t 4, t 1) i,( t 4, t 1) + ε, i,( t 4, t 1) + b PPE it 4 + b MB 2 i,( t 4, t 1) i,( t 4, t 1) + b Asset 5 i,( t 4, t 1) (4) where dependent variable is a dummy, set to one if the value of Issue-saving is positive, zero otherwise. Cash Flow is the operating cash flows, scaled by lagged total assets. MB is the market-to-book ratio. Leverage is the ratio of book value of debt to market value of equity. PPE is the property, plant, and equipment, scaled by lagged total assets. Asset is the log of total assets. PREC is the proxy for precautionary motive, which is the first principal component of R&D, cash flows volatility, and cash dividends paid. All independent variables are average measures for the period t-4 through t-1. As McLean (2011) shows that precautionary motive instead of market timing explains the tendency to issue equity, we include the precautionary motive proxy in the probit regression. The result of the first-stage regression is reported in Panel A of Table 8. Model (1) is for the test of announcement returns, and model (2) is for the test of post-merger performance. Consistent with McLean (2011), we find that likelihood of Issue-saving is positively related to market-to-book ratio and precautionary motive but negatively related to firm size. Next, in the second stage, we include the Inverse Mill ratio calculated from the first-stage model into the model of announcement returns and post-merger performance, as shown in Panel B and Panel C, respectively. The result of both tests was very similar to our previous result, showing that excessive cash holdings of firms saving more from equity issuance is associated with less agency problem due to higher transactions costs and scrutiny of equity financing. V. SUMMARY AND CONCLUSION 25