Massive Equity and Debt Issues: What Can we learn from Extreme Capital Structure Changes? ψ

Transcription

1 Massive Equity and Debt Issues: What Can we learn from Extreme Capital Structure Changes? ψ R. David McLean (Alberta) and Berardino Palazzo (Boston University) September 2013 Abstract We document the extent to which firms make massive equity and debt issues, and try to explain why they do so. A massive issue is a firm-quarter in which the total proceeds from either debt or equity issues exceeds 50% of the firm s total assets. We document 3,428 massive equity issues and 3,184 massive debt issues during the period % of massive equity issue proceeds are saved as cash, and this cash is not spent over the subsequent 8 quarters. In contrast, massive debt issues are primarily associated with debt repurchases and some investment. Massive equity issuers are R&D-intensive firms, dependent on equity finance, and have surges in stock prices just prior to issuing. Hence, massive equity issues can be explained by a framework that accounts for precautionary cash needs, equity-dependence, and market timing. ψ s: rdmclean@ualberta.ca and bpalazzo@bu.edu. McLean is grateful to the Dianne and Irving Kipnes Chair for financial support.

2 In this paper we document the extent to which firms make massive equity and debt issues, and try to explain why they do so. We define a massive issue as a firm-quarter in which the total proceeds from either debt or equity issues exceed 50% of the firm s assets. Massive issues represent the largest and most important capital structure decisions that firms make.. Moreover, the majority of debt and especially equity issues are miniscule, and do not reflect important capital structure decisions. Hence, understanding the extent to which firms make massive issues, and their reasons for doing so, can help us to differentiate between various theories of capital structure. During our sample period ( ) there are 200,241 firm-quarter observations with cash flows from equity issues. 1 Most of these are not economically meaningful, as there are only 10,005 observations with equity issuance proceeds that exceed 10% of assets. The majority of quarterly equity issues are less than 1% of assets, and likely represents option exercises (see Fama and French (2008)). Of the 10,005 equity issues that exceed 10% of assets, 3,428 (34%) are massive equity issues, exceeding 50% of assets. Hence, when firms do decide to issue equity, they tend to issue a lot of equity. The distribution of debt issues is very different. In our sample, there are 105,575 observations with cash flows from long-term debt issues. Of these, 23,866 are greater than 10% of assets, a far greater number and percentage of total debt issues as compared to equity issues. Of the debt issues that are greater than 10% of assets, 3,184 (13%) are massive debt issues, which is significantly less than the 34% of the >10% equity issues that are massive. To get an understanding for why firms make massive issues, we study how massive issue proceeds are spent. We find that there is a large contrast between equity and debt. With respect to equity issues, the bulk of the proceeds simply end up as cash. The average massive equity 1 We focus on equity issues that generate cash flow, and therefore do not study stock-financed mergers. 1

3 issue is 115% of assets, and in the same quarter cash increases by 86% of assets. These cash savings are persistent; looking out over the 8 quarters after a massive equity issue, we find that the dollar amount of cash actually increases. In contrast, the most common use of the proceeds from massive debt issues are debt repurchases. The average massive debt issue is 83% of assets. In the same quarter, debt repurchases are 41% of assets, or about half of the issuance proceeds. This likely reflects the rolling over of debt, or refinancing at a lower rate. Investment is about 21% of assets during the massive debt issue quarter, as compared to 3% in the previous quarter. Investment is then elevated for the next three quarters, after which it returns at its pre-issue quarterly average of about 2.8%. Hence, massive debt issues primarily reflect refinancing and investment. On the surface massive equity issues are puzzling, because they are primarily saved as cash rather than spent or invested. We therefore ask if massive equity issues are motivated by precautionary cash needs. Opler, Pinkowitz, Stulz, and Williamson (1999), Passov (2003) and Bates, Kahle, and Stulz (2009), point out that R&D-intensive firms with uncertain cash flow should accumulate precautionary cash savings. Passov (2003) (the former Treasurer of Pfizer) argues that cash reserves are critical for innovative companies because strategically they cannot afford to cut R&D spending, even during downturns when the cost of external finance is exorbitantly high. Consistent with precautionary motives, we find that the majority of massive equity issuers spend on R&D, and have negative internal cash flow. We extend our analyses by reading the quarterly and annual reports of 100 massive equity issuers in our sample. Item 303 of Regulation S-K requires firms to include discussions of financial liquidity and capital resources. Of the 100 reports that we read, 76% report relying on external finance to fund their operations. 2

4 So massive equity issuers are dependent on equity markets, yet need to maintain their R&D spending. 2 Massive debt issuers in contrast, tend to not invest in R&D, and generally do not have negative cash flow. Also consistent with massive equity issuers having high precautionary motives, the cash savings levels of massive equity issuers are very high, even before the massive equity issue. Prior to issuing, the ratio of cash to non-cash assets for massive equity issuers is around This ratio spikes to 4.69 during the massive issue quarter, and then works its way back down to its pre-issue level over the next eight quarters. The post-issuance decline in this ratio is not due to the cash proceeds being spent, but is instead due to non-cash assets growing at a faster rate than cash, which also increases during this period. With massive debt issuers, the cash to non-cash assets ratio is about 0.50, and changes very little around the massive issue quarter. Saving equity issue proceeds as cash makes sense in a world where external finance has fixed costs; firms should make large but infrequent issues. Saving issues as cash also make sense in a world with dynamic issuance costs. In this case, firms should market time, and issue and save when costs are low, to avoid issuing when costs are high. Consistent with this idea, we observe large increases in market values before massive equity issues, but not before massive debt issues. In the four quarters before a massive equity issue, market values increase by 11% (quarter t-4); 14%; 19%; and 27% (quarter t-1). In the four quarters after a massive equity issue, market values increase by about 2%. In contrast, market values increase by about 7% per quarter prior to a massive debt issue, and by 4% per quarter after the issue. Our massive equity issue findings are consistent with Loughran and Ritter (1995 and 1997), Spiess and Affleck-Graves (1995), Baker and Wurgler (2002), and Harvey and Graham 2 Stiglitz (1985), Berger and Udell (1990), Brown, Fazzari, and Petersen (2009), and Hall (2009) all note that debt is unsuitable for R&D-intensive firms, which have few tangible assets and uncertain cash flow. 3

5 (2001), all of which propose market timing as a motivation for share issuance. These papers show that equity issues are associated with high market values, but then low subsequent returns. Deangelo, Deangelo, and Stulz (2010) point out that most firms with these traits do not issue equity, and market timing is therefore not a sufficient explanation for equity issues. Our findings support both arguments. Consistent with market timing, massive issues are precipitated by an increase in stock prices. However, massive issuers are almost universally high precautionary motive firms that depend on equity finance; most firms with large stock price increase do not meet these other two criteria. Deangelo, Deangelo, and Stulz (2010) argue that firms issue equity in public offerings primarily to fund a short-term liquidity squeeze. Most massive equity issuers have negative free cash flow and invest large amounts, so these firms do depend on equity in order to fund their operations and investment. However, the majority (75%) of massive issue proceeds are saved as cash, and remain as cash over the subsequent eight quarters, so the majority of the proceeds are not used to meet any near term liquidity squeezes. Massive issuers finance investment during the subsequent quarters by issuing more equity, although not in massive amounts. Deangelo, Deangelo, and Stulz (2010) also find that cash levels revert to normal levels soon after the issue, suggesting that the issue proceeds were quickly spent. For massive issuers, we also find that the ratio of cash to non-cash assets falls back to the pre-issuance level in the quarters that follow the massive issue. However, this decline is not due to the cash being depleted, but rather non-cash assets growing faster than cash, which also increases. Our massive equity issue findings are not consistent with papers that argue that equity issues reflect growth options exercises (e.g., Carlson, Fisher, and Giammarino (2006) and Lyandres, Sun, and Zhang (2008)). These papers argue that the increase in stock prices prior to 4

6 issuance reflects an increase in growth options, and the low subsequent returns reflect the options being exercised with the investment of the equity issue proceeds. We find that investment barely increases during and around massive equity issue quarters, and that most massive issue proceeds are saved as cash, so it is unlikely that massive equity issues result in abnormal growth options exercises. The growth options story is however a good fit for massive debt issues, which do result in elevated investment, and are associated with low subsequent stock returns (see Spiess and Affleck-Graves (1999)). Prior empirical evidence of share issues going to cash savings is provided in Kim and Weisbach (2008) and McLean (2011). Kim and Weisbach study public offerings, whereas McLean studies aggregate share issues. Both papers estimate average savings rates around $0.45 per $1 of issuance proceeds, finding greater roles for investment and spending than we do. We estimate a savings rate that is closer to $0.75 per dollar, suggesting that the cash savings effects in those papers are driven by massive issues, rather than smaller issues. Kim and Weisbach (2008) contend that their results are consistent with market timing, but do not assign a role to precautionary motives. McLean (2011) emphasizes precautionary motives, and argues market timing is less important, stressing liquidity and economic conditions as the timing variables. 3 Our findings with massive issues partly support both papers, as we find that both market timing and precautionary motives play important roles in explaining massive equity issues. 1. Data and Sample We obtain data from the Compustat quarterly database. Our sample period is from 1985 through We exclude non-u.s. firms, firms that are not listed on one of the three major 3 Bolton, Chen, and Wang (2012), Eisfeldt and Muir (2013), Hugonnier, Malamud, and Morellec (2012), and Warusawitharan and Whited (2013) show theoretically that it can be optimal for firms to time markets to build precautionary cash savings. 5

7 stock exchanges, financial firms (SIC ), and utilities (SIC ). We exclude observations that are missing either total assets or stock price data from the previous or current quarter. The missing stock price restriction ensures that we are not capturing IPO proceeds. We also exclude observations with missing data on quarterly investment expenditure and acquisitions. This process results in 325,078 firm-quarter observations with reported values of total assets. Our main variables of interest are cash flow from equity issues and debt issues. In the Compustat quarterly database, both of these variables are reported as year-to-date proceeds. As an example, equity issues (sstky) in quarter 2 is the sum of the equity issues from quarters 1 and 2. The quarter 3 equity issues is the sum of equity issues from quarters 1, 2, and 3. To get the actual equity issues in quarter 3, we subtract the quarter 2 equity issues from the quarter 3 equity issues. We follow the same procedure for debt issues (dltisy), equity and debt repurchases (prstkcy and dltry), capital expenditures (capxy), and acquisitions (aqcy). More generally, any Compustat quarterly variable that ends in y reflects a year to date sum, whereas variables that end in q reflect proceeds for that particular quarter. We report summary statistics for the primary variables used in this study in Table 1. All of our cash flow variables are scaled by beginning of the quarter assets. ΔCash is the quarterly change in cash and marketable securities (cheq). Equity and debt issues are defined above. Cash flow is net income (niq) plus depreciation and amortization (dpq). R&D spending (xrdq) is set to zero if the value is missing. 4 Investment is the sum of quarterly capital expenditures and acquisitions. Log MB is the log of the market value of equity (prccq*cshoq) scaled by the book value of equity (ceqq). If the book value of equity is negative, we set the market-to-book value to 4 R&D spending at a quarterly frequency is available in Compustat beginning in

8 missing. Our balance sheet items are all scaled by the previous quarter s total assets. All of the variables are winsorized at the top and bottom 0.1%. Table 1 shows that the equity issues, debt issues, and ΔCash are all right skewed. With each of the series, the mean values are greater than the 75 th percentile, and the median values are all zero. With respect to equity and debt issues, this shows that when firms issue, they often issue very large amounts. The right skew of ΔCash suggests that cash savings are often built with equity and/or debt issues. Consistent with this intuition, the cash flow variable is not right skewed; its median (0.019) is greater than its mean (0.003). These results therefore suggest that large issues of securities drive large increases in cash. We explore this idea more in the tables that follow. 2. Main Results 2.1 A First Look at Massive Issues: The Distribution of Equity and Debt Issues In this section we take a first look at massive issues, by documenting their frequency. Table 2 and Figure 1 show the frequency of quarterly debt and equity issues, partitioned by issue size. With respect to equity issues, Table 1 shows that there are 202,241 firm-quarter observations that issue equity. Of these, 190,236 (95%) are less than 10% of assets, and 167,103 (83%) are less than 1% of assets. Hence, most equity issues are not economically meaningful, and are likely just option exercises, rather than managerial share issuance decisions. Among the 10,005 equity issues that are greater than 10% of assets, 3,428 (34%) are massive issues, with proceeds that exceed 50% of assets. So when firms do make meaningful equity issues, a significant percentage of them make massive equity issues. 7

9 Table 2 and Figure 1 reveal a very different pattern for debt issues. There are 105,575 firm-quarter observations that issue long-term debt. Of these, 30,825 (29%) are less than 1% of assets, a far smaller percentage than with equity issues. 23,866 (23%) of debt issues are greater than 10% of assets, a much larger number and portion than with equity issues. There are 3,184 massive debt issues, which represents 13% of the >10% debt issues; much less than the percentage of >10% equity issues that are massive (34%). Figure 2 displays the number of massive equity and debt issues over our sample period. One noticeable feature of the graph is that massive equity issues are more lumpy than massive debt issues. There is a large spike in the number of massive equity issues on around the first quarter of 2000, which was the height of Nasdaq-internet bubble. During that quarter we have 136 massive equity issues, whereas in the typical quarter there are on average 29.. We do not observe any quarters with more than 100 massive debt issues. The Nasdaq-internet bubble spike with massive equity issues suggest that market timing could be one motive for massive equity issues, and we explore this idea more in the analyses that follow The Use of Massive Issue Proceeds In this next section we aim for a better understanding of why firms make massive issues. To do so we look at the evolution of several variables on and around the massive issue quarter. We measure the average issue amount, change in cash, repurchase amount, and investment in the massive issue quarter, and in each of the eight quarters before and after the massive issue quarter. We display the results for equity issues in Panel A of Table 3, and for debt issues in Panel B. The results are also displayed in Figure 3. 8

10 Table 3 and Figure 3 make it clear that massive equity issues are primarily saved as cash, and that the cash savings are persistent. Massive equity issues average 115% of assets; during the quarter of the massive issue cash increases by 85% of assets. Repurchases are virtually zero, and investment is 7% of assets, as compared to 3% in the previous quarter. Table 3 also shows that the massive equity issue cash savings are persistent, and are not spent over the next eight quarters. In fact, changes in cash are positive in all of the eight quarters after the massive issue quarter, ranging from 1% to 5% of assets. Share issuance is equal to about 15% of assets during the quarters before the massive issue, and about 15% of assets during the quarters that come after the massive issue. We note that the amount of share issuance in the quarters around the massive issue quarter is almost exactly equal to the sum of cash used in operations (cash flow is negative) and cash spent on investment, suggesting that massive equity issuers are entirely equity dependent. We examine this idea more closely in the next section. Unlike massive equity issues, massive debt issues primarily reflect repurchases and investment. Massive debt issues average 83% of assets, while debt repurchases in the massive issue quarter average 41% of assets. Hence, about half of massive debt issues reflect the refinancing or rolling over of existing debt. The next largest use of massive debt proceeds is investment, which is 21% of assets in the issuance quarter, significantly higher as compared to the quarters before the massive issue. Finally, change in cash is about 14% of assets in the massive issue quarter, so cash savings are a third-order effect for massive debt issues. Taken in their entirety, Table 3 and Figure 2 show that massive equity and debt issues have very different motivations. The majority of massive equity issue proceeds are saved as cash. These cash savings are persistent, and are added to, not depleted, during the eight quarters that 9

11 follow the massive equity issue. In contrast, massive debt issues are in large part just refinancings and rollovers. Investment also spikes in the massive debt issue quarter, and cash increases too, but by a much smaller amount as compared to with equity issues. Hence, massive debt issues primarily go to repurchases and investment, while massive equity issues primarily go to cash. In the next section we try to better understand the reasons for this difference Massive Issues and Precautionary Motives Thus far we have seen that massive equity issues are primarily saved as cash, whereas massive debt issues are largely refinancings and rollovers coupled with some investment. We now try to explain why firms make massive equity issues, as on the surface it appears that they do not need the cash for an immediate purpose. We therefore turn to the literature on precautionary cash savings. Opler, Pinkowitz, Stulz, and Williamson (1999), Passov (2003), and Bates, Kahle, and Stulz (2009) point out that R&D-intensive firms with uncertain cash flow should accumulate precautionary cash savings. Passov (2003) (the former Treasurer of Pfizer) argues that cash reserves are critical for innovative companies because strategically they cannot afford to cut R&D spending, even during downturns when the cost of external finance is exorbitantly high. To get a better idea of whether massive equity issues are related to precautionary cash needs, we first examine the role of R&D spending. Table 4 shows of the 174,773 observations with equity issue values that are not massive during the period 1989q1-2012q4, 86,389, or 50%, have positive values of R&D spending. On the other hand, of the 3,191 massive equity issuers, 2,481 have positive values of R&D spending: A number three and a half larger than firms with 10

12 zero R&D spending... Hence, massive equity issuers are about 50% more likely to spend on R&D as compared to non-massive equity issuers. With respect to the 2,858 massive debt issues in our sample, only 828, or 29%, spend on R&D. This is the same percentage of the quarter-firm observations that are not massive debt issues and have positive values of R&D spending. Taken together, the results show that massive equity issuers are more than two times as likely to have positive values of R&D spending as compared to massive debt issuers. These effects are further illustrated in Figure 4. Panel A plots the number of massive equity issues separately for firms that spend on R&D and firms that do not spend on R&D. The graph shows that over time it is almost always the case that massive equity issues are more frequent for firms that spend on R&D. Panel B plots the same effects for massive debt issues. Panel B shows that it is almost always the case that massive debt issues are more frequent for firms that do not spend on R&D. Hence, R&D spending is associated with massive equity issues, but not massive debt issues. This could help explain why such a large portion of massive equity issues are saved as cash Massive Issues and Internal Cash Flow The cash savings literature traditionally assumes that firms save from internal cash flow. The reasoning is that external finance is costly, so it makes sense to save from internal funds in order to avoid issuing. Hence, if massive equity issuers are firms that for precautionary reasons need to hold large amounts of cash, then why don t they just save from internal cash flow, why do they need to issue equity? 11

13 As we mention above, Table 3 reports the internal cash flow of massive equity issuers and massive debt issuers around the massive issuance quarter. The results show that on average, massive equity issuers have negative cash flow for all eight quarters before the massive issue; during the massive issue quarter; and for all eight quarters after the massive issue. Internal cash flow averages around -16% of assets before the massive equity issue, and about -11% during the quarters after the issue. The improvement after the massive issue comes from a growth in assets due to the issuance, rather than an improvement in operating performance. Hence, firms that make massive equity issues do not generate cash flow internally, and are dependent on external finance. The only way that these firms can build cash reserves is by issuing equity. Massive debt issuers have internal cash flow that average about 0% of assets both before and after the massive issue, so massive debt issuers are less dependent on external finance as compared to massive equity issuers Financial Statements To further validate the idea that massive equity issuers are dependent of equity finance, we read the quarterly and annual reports of 100 massive equity issuers in our sample. In 1977 the SEC strengthened the reporting requirements for public companies by adopting Regulation S-K. Item 303 of Regulation S-K requires firms to include discussions of financial liquidity and capital resources. As a result, quarterly and annual reports tend to have a section titled liquidity and capital resources. In this section firms typically explain how they fund their operations and investment. As an example, the following is from the Q quarterly report for Aurora Bioscience: 12

14 We have funded our operations through March 31, 2000 primarily through the issuance of equity securities with aggregate net proceeds of $132 million, receipts from corporate collaborations of $107 million, capital equipment lease financing of $13 million and interest income of $7 million. In the statement above the firm indicates that it relies on equity issues to fund its operations, so we define this firm as equity-dependent during the quarter in which it made its massive equity issue. We also define firms as equity-dependent if they mention that the majority of their capital resources are the result of issuing equity. We find that of the 100 firms that we read reports for, 76 meet our criteria for equity-dependence. We pasted the relevant statements from these companies in the Appendix at the end of this paper. The annual and quarterly reports also tell us how the equity was issued. We find that 44% of the massive issues were made via private placements (often to multiple parties); 41% were public offerings; 7% were warrants exercises; and 7% were a combination of warrant exercises and private placements. The warrants were attached to previous private placements or debentures, and in some cases are callable by the firm. If a warrant is exercised by the warrant-holder, then the issuance decision was not the firm s. However, how the warrant proceeds are used is the firm s decision, as the firm could repurchase the exercised shares if it did not want save the issuance proceeds. Hence, the saving of warrant issue proceeds as cash is consistent with precautionary motives. 13

15 2.6. Massive Issues and Cash Holdings As we mention previously, Opler, Pinkowitz, Stulz, and Williamson (1999), Passov (2003) and Bates, Kahle, and Stulz (2009) argue that firms hold cash for precautionary reasons. Opler et al. (1999) show that cash savings are persistent firms appear to have cash targets that they actively try to maintain, and these targets are higher for firms with high R&D spending and more uncertain cash flow. If massive equity issuers are saving cash for precautionary reasons, then it should be the case that these firms have relatively high cash holdings, even before the massive issue. Table 3 and Figure 5 report the cash to non-cash assets ratios for massive equity and massive debt issuers. We see that for massive equity issuers, cash to non-cash assets ratios are very high; the average is over 400% during the quarters preceding the massive equity issue, and there is a slight dip in the quarter just before the issue, to 343%. In the quarter of the massive issue the ratio spikes to 655%, and it then slowly works its way back down to its pre-massive issue values. The reason that the cash to non-cash assets ratio declines during this period is because non-cash assets are growing at a faster rate than is cash, as cash continues to grow after the massive issue as well (see Figure 1 and Table 2). This effect is consistent with Deangelo, Deangelo, and Stulz (2010), who show that cash levels revert to normal levels after public offerings. Deangelo et al. attribute this effect to cash being spent, but with massive issuers the effect is caused by non-cash assets increasing faster than cash. In contrast to massive equity issuers, we see that massive debt issuers hold much less cash, and that for these firms there is not an increase in the cash to non-cash assets ratio around the massive debt issue. This is consistent with Table 3, which shows that a greater portion of massive debt issues are invested than saved as cash. The cash-to-non-cash assets ratio averages 14

16 about 50% for massive debt issuers, and remains around this level throughout the eight quarters before and after the massive debt issue Massive Issues and Market Timing The results so far show that massive equity issues are primarily used to build cash reserves. Yet why don t firms simply wait, and issue equity when they need the proceeds? Why do firms issue so much at one time? Massive and infrequent issues make sense in a world with fixed issuance costs. However, massive issuers also issue in smaller but meaningful amounts (around 15% of assets) before and after the massive issue quarter (see Figure 3). Massive and issues also make sense in a world with dynamic issuance costs. If stock prices vary over time because of mispricing, then it makes sense to make massive issues when prices are inflated, and avoid issuing when prices are low. We test for such effects in this section. Table 6 and Figure 6 report the results for our market timing tests. With respect to equity issues, we see that market values change by about 7% per quarter in quarters t-8 through t-5, and then surge and change by 11%, 14%, 18%, and 26% in quarters t-4 through t-1 respectively, and by 19% in the massive equity issue quarter. Prices then stabilize, and increase by about 2% per quarter during the next 8 quarters after the massive issue. In contrast, we do not observe such a surge in stock prices prior to massive debt issues. Market values change by an average of about 7% up until the massive debt issue, and then by about 4.5% per month thereafter. We observe similar patterns with market-to-book ratios. Market-to-book surges just before the massive equity issue, and then reverts to pre-issuance levels afterwards. In contrast, there is no market-to-book spike for massive debt issues. 15

17 Stock prices should reflect growth opportunities, so we might expect increases in stock prices before all massive issues, if firms were raising capital in anticipation of funding valuable growth opportunities. The relatively greater stock price changes prior to massive equity issues, coupled with the fact that massive issues are primarily saved as cash, points towards market timing as a motivation for massive equity issues. We stress that these results suggest that market timing is only one motivation for massive issues. As Deangelo, Deangelo, and Stulz (2010) point out, most firms that have large increases in stock prices do not issue shares. What our results show is that massive issuers tend to have three traits: high precautionary cash needs; dependence on equity; and increases in stock prices. So a surge in stock prices may be a necessary condition for massive issues, but it is not a sufficient condition The Determinants of Massive Issues: Probit Regressions In this section we more formally explore the determinants of massive issues, by estimating ordered logit regressions. We estimate these regressions for equity (Table 8) and debt issues (Table 9) separately. The dependent variable is equal to 2 for massive issues; 1 for nonmassive issues >10%; and 0 for no issue or issues <10%. The regression equation therefore assumes that firms can either choose a massive issue; a large issue that is not massive; or not to issue. We assume that issues <10% reflect non-issues (in the case of equity issues these are mostly option exercises). If we instead split issues <10% into two groups, zero and greater than zero, our results do not change. The coefficients in these regressions are log-odds ratios. An odds ratio is the likelihood of an event occurring in one group relative to the likelihood of an event occurring in another group. 16

18 So the coefficients reflect how increases in the independent variables affect the likelihood of moving to a higher ordered group. Some of the regressions include industry and time fixed effects, which are denoted in the bottom rows of tables 8 and 9. The standard errors are double clustered on firm and quarter, per Petersen (2009). In the massive equity issue regressions in Table 8, the R&D spending coefficient is always positive and highly significant, while the cash flow coefficient is always negative and highly significant,,showing that, higher R&D spending and lower internal cash flow increase the likelihood of a massive equity issue, consistent with what we find in the previous tables. Consistent with market timing, both change in stock price and market-to-book are positive and highly significant in all of the regressions. Hence, both increases in market values and high levels of market-to-book both increase the likelihood of a massive equity issue. The results further show that massive equity issuers are small growth firms, as investment increases the likelihood of a massive equity issue, whereas size (total assets) decreases the likelihood. GDP growth and the term spread do not have significant effects, probably because these macro variables are subsumed by the change in stock price and market-to-book. As in the previous tables, the effects in Table 9 show that the determinants of massive debt issues are much different than the determinants of massive equity issues. R&D spending decreases the likelihood of a massive debt issue, and cash flow does not have a significant effect. Higher market-to-book ratios decrease the likelihood of a massive debt issue, the opposite effect than that found with massive equity issues. Stock price increases increase the chance of a massive debt issue, however the coefficient is about ¼ the size of the change in price coefficient in the massive equity issue regression. Like the massive equity issue regression, the size coefficients are negative and the investment coefficients are positive, although the coefficients 17

19 are smaller (in absolute value) as compared to the massive equity issue regression. Finally, unlike with equity issues, the GDP growth coefficient is positive and significant, and the termspread coefficient is negative and significant. These effects suggest that firms make massive debt issues when interest rates for corporate lending are lower. This would be consistent with opportunistic refinancings explaining some of the large repurchases that are associated with massive debt issues. 3. Conclusions In this paper we document the extent to which firms make massive equity and debt issues, and try to explain why they do so. Massive issues are not uncommon, and represent the largest and most important capital structure decisions that firms make. Understanding why firms make massive issues sheds light on why firms more generally issue shares and debt, and can help us to differentiate between various theories of capital structure. To get an understanding for why firms make massive issues, we study how massive issue proceeds are spent. The average massive equity issue is 115% of assets, and in the same quarter cash increases by 86% of assets. These cash savings are persistent; looking out over the 8 quarters after a massive equity issue, we find that the dollar amount of cash increases. In contrast, the most common use of the proceeds from a massive debt issue is a debt repurchase, followed by investment. With respect to massive equity issues, we find that the majority of massive equity issuers are equity dependent and have high precautionary cash needs. Massive equity issuers are R&Dintensive and have negative internal cash flow, making equity their only source of funds. We 18

20 further corroborate this conclusion by reading the quarterly and annual reports of 100 massive equity issuers in our sample. Of the 100 reports that we read, 76% report relying on equity to fund their operations. Also consistent with massive equity issuers having high precautionary motives, the cash savings levels of massive equity issuers are much higher than those of massive debt issuers, both before and after the massive issues. Saving issues as cash make sense in a world where external finance has fixed costs; firms should make large but infrequent issues. Saving issues as cash also make sense in a world with dynamic costs. In this case, firms should market time, and issue and save when costs are low, to avoid issuing when costs are high. Consistent with dynamic costs, we observe large increases in market value before massive equity issues, but not before massive debt issues. Our findings are therefore consistent with massive equity issues being motivated by high precautionary cash needs; dependence on equity finance; and market timing. 19

21 References Baker, Malcolm, and Jeffrey Wurgler, 2000, The equity share in new issues and aggregate stock returns, Journal of Finance 55, Baker, Malcolm, and Jeffrey Wurgler, 2002, Market timing and capital structure, Journal of Finance 57, Bates, T., Kahle, K., Stulz, R., Why do U.S. firms hold so much more cash than they used to? Journal of Finance 64, Berger, A. N., Udell, G. F., Collateral, loan quality, and bank risk. Journal of Monetary Economics 25, Blanchard, O., Rhee, C., Summers, L., The stock market, profit, and investment. Quarterly Journal of Economics 108, Bolton, P., H. Chen, and N. Wang, 2013, Market Timing, Investment, and Risk Management, Journal of Financial Economics, Forthcoming. Brown, J. R., Fazzari, S. M., Petersen, B. C., Financing innovation and growth: cash flow, external share, and the 1990s R&D boom, Journal of Finance 64, DeAngelo, H., DeAngelo, L., Stulz, R., Seasoned equity offerings, market timing, and the corporate lifecycle. Journal of Financial Economics 56, Eisfeldt, A. and T. Muir, 2012, Aggregate Issuance and Savings Waves, Working Paper. Fama Eugene F., and Kenneth R. French, 2005, Financing decisions: Who issues stock? Journal of Financial Economics 73, Kim, W., Weisbach, M., Motivations for public share offers: An international perspective. Journal of Financial Economics 87, Loughran, Tim, and Jay R. Ritter, The Operating Performance of Firms Conducting Seasoned Equity Offerings 1997, Journal of Finance 52, McLean, R.D., 2011, Share issuance and cash savings, Journal of Financial Economics 99, Opler, T., Pinkowitz, L., Stulz, R., Williamson, R., The determinants and implications of cash holdings. Journal of Financial Economics 52, Petersen, M., Estimating standard errors in finance panel data sets: comparing approaches. Review of Financial Studies 22,

22 Pinkowitz, L., Stulz, R., Williamson, R., 2006., Does the Contribution of Corporate Cash Holdings and Dividends to Firm Value Depend on Governance? A Cross Country Analysis, Journal of Finance 61, Spiess, D. K., Affleck Graves, J., Underperformance in long run stock returns following seasoned equity offerings. Journal of Financial Economics 38, Warusawitharana, Masa and Toni Whited, 2012, Equity Market Misvaluation, Financing, and Investment, Working Paper. 21

23 Table 1: Summary Statistics This table reports the mean, standard deviation, 25% percentile, 50% percentile, 75% percentile, 90% percentile, 95% percentile, 99% percentile, minimum value, maximum value, and number of available observations for the variables used in the empirical analysis. We exclude non-u.s. firms, firms that are not listed on one of the three major stock exchanges (AMEX, NYSE, and NASDAQ), financial firms (SIC ), and utilities (SIC ). We exclude observations that are missing either total assets or stock price data from the previous and current quarter. We also exclude observations with missing data on quarterly investment expenditure and acquisitions. Compustat reports some variables as year-to-date proceeds. As an example, equity issues (item sstky) in quarter 2 is the sum of the equity issues from quarters 1 and 2. The quarter 3 equity issues is the sum of equity issues from quarters 1, 2, and 3. To get the actual equity issues in quarter 3, we subtract the quarter 2 equity issues from the quarter 3 equity issues. We follow the same procedure for debt issues (item dltisy), equity and debt repurchases (items prstkcy and dltry), capital expenditures (item capxy), and acquisitions (item aqcy). Leverage equals the quarter t total liabilities (item ltq) divided by the quarter t total assets (item atq). Investment is the sum of quarterly capital expenditures and acquisitions. R&D is the quarterly R&D expenditure (item xrdq). This value is set to zero if the value is missing. Cash flows is net income (item niq) plus depreciation and amortization (item dpq). Cash is the quarterly change in cash and marketable securities (item cheq). Cash Holdings is the quarterly value in cash and marketable securities (item cheq) scaled by beginning of the quarter total assets net of cash holdings (item atq minus item cheq). Log MB is the log of the market value of equity (item prccq times item cshoq) scaled by the book value of equity (item ceqq). If the book value of equity is negative, we set the market-to-book value to missing. Price is the quarterly change of the end of the quarter closing price of common equity (item prccq). Size is the log value of total book assets (item atq). Otherwise noted, all of the variables are scaled by beginning of the quarter total assets. The data are at a quarterly frequency over the period 1984q1 2012q4, except for the R&D variable which is only observed over the period 1989q1 2012q4. All the variables are winsorized at the top and bottom 0.1%. mean sd p25 p50 p75 p90 p95 p99 min max count Equity Issuance Equity Repurchase Debt Issuance Debt Repurchase Leverage Investment R&D Cash Flows Cash Cash Holdings Log MB Price Size

24 Table 2: Equity and Debt Frequency Distributions This table reports the distribution of the equity (Panel A) and debt (Panel B) issues variables. Equity issue at the quarterly frequency is calculated using the annualized value in Compustat (item sstky). Similarly, debt issue at the quarterly frequency is calculated using the annualized value of long term debt issue in Compustat (item dltisy). We report the mean, the median, and the number of available observations for six di erent categories of issues. The column > 0 reports the statistics for all the firms with positive issues. The column (0, 0.01] reports the statistics for all the firms with positive issues up to The column (0, 0.10] reports the statistics for all the firms with positive issues up to The column (0.10, 0.5)] reports the statistics for all the firms with an issue value greater than 0.10 and smaller than The column > 0.10 reports the number of observations for all the firms with issues greater than The column > 0.50 reports the number of observations for all the firms with issues greater than The last row reports the fractions of observations in each category relative to the number of firms with a positive issue value. Panel A: Equity Issuance > 0 (0, 0.01] (0, 0.10] (0.10, 0.50) > 0.10 > 0.50 Mean Median Obs Fraction Panel B: Debt Issuance > 0 (0, 0.01] (0, 0.10] (0.10, 0.50) > 0.10 > 0.50 Mean Median Obs Fraction

25 Table 3: Dynamic Evolution of Investment and Financing around Massive Issues This table reports the evolution of equity and debt issue, equity and debt repurchase, investment, change in cash, and cash flows eight quarters before and eight quarters after a massive equity (Panel A) and a massive debt (Panel B) issue episode over the the period 1984q1 2012q4. For each quarter we report the simple average of the observations available in that quarter. Equity issue and repurchase at the quarterly frequency are calculated using the annualized value in Compustat (items sstky and prstkcy). Debt issue and repurchase at the quarterly frequency are calculated using the annualized value of long term debt issue and repurchase in Compustat (items dltisy and dltry). Investment is the sum of quarterly capital expenditures and acquisitions (calculated using items capxy and aqcy). Cash is the quarterly change in cash and marketable securities (item cheq). Cash flows is net income (item niq) plus depreciation and amortization (item dpq). Panel A: Massive Equity Issue q= Equity Issues Equity Rep Investment Cash Cash Flows Panel B: Massive Debt Issue q= Debt Issues Debt Rep Investment Cash Cash Flows

26 Table 4: Equity and Debt Frequency Distributions: R&D vs No R&D This table reports the distribution of the equity (Panel A) and debt (Panel B) issues variables for firms with zero R&D and firms with a positive R&D value over the the period 1989q1 2012q4. We the number of available observations for six di erent categories of issues. The column > 0 reports the number of observations for all the firms with positive issues. The column (0, 0.01] reports the number of observations for all the firms with positive issues up to The column (0, 0.10] reports the number of observations for all the firms with positive issues up to The column (0.10, 0.5)] reports the number of observations for all the firms with an issue value greater than 0.10 and smaller than The column < 0.10 reports the number of observations for all the firms with smaller than The column > 0.50 reports the number of observations for all the firms with issues greater than For each category of issue, the last row reports the ratio of observations in the positive R&D category over observations in the zero R&D category. Panel A: Equity Issuance > 0 (0, 0.01] (0, 0.10] (0.10, 0.50) < 0.50 > 0.50 Obs. R&D= Obs. R&D> Ratio Panel B: Debt Issuance > 0 (0, 0.01] (0, 0.10] (0.10, 0.50) < 0.50 > 0.50 Obs. R&D= Obs. R&D> Ratio

27 Table 5: Dynamic Evolution of Cash Holdings Around Massive Issues This table reports the evolution of cash holdings eight quarters before and eight quarters after a massive equity (Panel A) and a massive debt (Panel B) issue episode over the the period 1984q1 2012q4. For each quarter we report the simple average of the observations available in that quarter. Cash Holdings is the quarterly value in cash and marketable securities (item cheq) scaled by beginning of the quarter total assets net of cash holdings (item atq minus item cheq). Panel A: Massive Equity Issue q= Cash Holdings Panel B: Massive Debt Issue q= Cash Holdings

28 Table 6: Dynamic Evolution of Stock Price and Market to Book around Massive Issues This table reports the evolution of the quarterly change of the end of the quarter closing price and the log market to book eight quarters before and eight quarters after a massive equity (Panel A) and a massive debt (Panel B) issue episode over the the period 1984q1 2012q4. For each quarter we report the simple average of the observations available in that quarter. Log MB is the log of the market value of equity (item prccq times item cshoq) scaled by the book value of equity (item ceqq). If the book value of equity is negative, we set the market-to-book value to missing. Price is the quarterly change of the end of the quarter closing price of common equity (item prccq). Panel A: Massive Equity Issue q= Price Log MB Panel B: Massive Debt Issue q= Price Log MB