How Do Firms Price Optimal Capital Structure? Evidence from Mergers and Acquisitions

Transcription

1 How Do Firms Price Optimal Capital Structure? Evidence from Mergers and Acquisitions James S. Ang Bank of America Eminent Scholar and Professor of Finance, Department of Finance, College of Business, Florida State University, Tel: 001-(850) , Fax: 001-(850) Mai M. Daher Assistant Professor of Finance, Olayan School of Business American University of Beirut Tel , Fax: md88@aub.edu.lb Ahmad K. Ismail Associate Professor of Finance, Olayan School of Business American University of Beirut Tel , Fax: ai05@aub.edu.lb and phd98ai@yahoo.com Corresponding author: Ahmad Ismail, Olayan School of Business, American University of Beirut, Bliss Street, P.O. Box: , Beirut, Lebanon. Tel Ext. 3731, Fax: , ai05@aub.edu.lb and phd98ai@yahoo.com 1

2 How Do Firms Price Optimal Capital Structure? Evidence from Mergers and Acquisitions Abstract This study examines how capital structure considerations affect acquisition pricing. Overlevered bidders pay higher premiums in leverage rebalancing deals and create leverage slack, while under-levered acquirers are not equally incentivized to lever up towards target leverage; rather they offer higher premiums using overvalued equity as they consider the market timing opportunity more valuable than leverage rebalancing. This asymmetry is more pronounced for financially constrained firms with non-rated debt. Over-levered acquirers also pay higher premiums for potential debt capacity improvement. Rebalancing and debt capacity improvement are value-enhancing for overlevered acquirers only when the option to borrow is exercised in the long-run. JEL Classification: G34 Keywords: acquisitions; optimal leverage; market timing; premium, debt capacity 2

3 How Do Firms Price Optimal Capital Structure? Evidence from Mergers and Acquisitions I. Introduction We study how a firm s investment decision is affected by its capital structure concerns. While earlier research (e.g. Harford, Klasa, and Walcott, 2009; Uysal, 2011 and Vermaelen and Xu, 2014) documents the impact of target capital structure on acquisition likelihood and financing, limited evidence exists on how capital structure considerations affect acquisition pricing. Because the capital structure of the combined firm depends on the leverage of each party and on the payment form, acquirers may, through their choice of payment method, achieve both the acquisition of more assets (the target firm), and a rearrangement of capital structure. However, acquisition decisions could also be driven by acquirer stock overvaluation (Shleifer and Vishny, 2003; Rhodes-Kropf, Robinson, and Viswanathan, 2005; Dong, Hirshleifer, Richardson and Teoh, 2006; and Ang and Cheng, 2006) and therefore, firms may rationally choose to create value either by rebalancing towards optimal capital structure (as implied by the trade-off theory) or by exploiting market timing (Baker and Wurgler, 2002). Hence, it is imperative to know what price acquirers associate with leverage rebalancing and whether the trade-off between market timing and leverage rebalancing is also reflected in the merger premium. Moving towards optimal capital structure not only increases value, but also creates the option to raise debt when needed; this is achieved by positioning the merged firms capital structure below optimal to create leverage slack 1. However, the benefits of increasing leverage may be smaller than the benefits of decreasing it 2 (Faulkender, Flannery, Hankins and Smith, 2012) which may explain 1 We refer to having a leverage ratio below the target ratio as leverage slack. 2 Faulkender, Flannery, Hankins and Smith (2012) argue that Even if leverage adjustment costs were equal for under- and overlevered firms, the benefits may be asymmetrical. Under-levered firms forego tax benefits of leverage and have little concern with financial distress costs. Yet potential financial distress costs loom quite large for over-levered firms. There is no theoretical reason why the net tax benefit minus 3

4 the faster leverage adjustment for over-levered compared to under-levered firms (Harford, Klasa, and Walcott, 2009 and DeAngelo, DeAngelo, and Whited, 2011). So, over-levered and under-levered firms may perceive the benefits from capital structure rearrangement differently. For instance, creating leverage slack should be more valuable for acquirers that need it most, i.e., over-levered acquirers who need leverage slack. We therefore predict that, in mergers that could enable them to have leverage slack (from deficit), over-levered acquirers would: a) offer a higher premium and b) stand to gain from the joint action of acquiring assets and rearranging capital structure. On the other hand, under-levered acquirers are not necessarily equally incentivized to lever up towards target leverage. These acquirers, therefore; face a tradeoff between rebalancing towards optimal leverage and paying with overvalued stock 3 ; and because the opportunity from stock overvaluation is temporary, while gains from moving toward optimal leverage can be recaptured at a later time, under-levered acquirers may consider the prospect from market timing more valuable than leverage rebalancing, and thus would offer a higher premium in such circumstances using their overvalued stocks. Another proposition of acquisition motives is that the debt capacity of the combined firm can increase due to a reduction in borrowing costs if the merger results in a decrease in the variability of the combined firm s cash flows, and hence in the probability of financial distress (Lewellen, 1971; Levy and Sarnat, 1970), otherwise known as the coinsurance effect (Kim and McConnell, 1977). Empirical evidence shows that firms increase their leverage after merger as a result of, presumably, increased debt capacity (Ghosh and Jain, 2000). However, the value that acquirers associate with this increased debt capacity is still unexplored, thus we examine it thoroughly in this paper. expected financial distress costs should be symmetrical around the firm s optimal leverage ratio, and therefore no reason to maintain that the absolute distance from target leverage fully captures a firm s incentives to adjust. 3 Knowing that underleverage is associated with stock overvaluation (Baker and Wrugler, 2002 and Uysal, 2011) 4

5 Using a sample of 1,810 U.S. M&A deals, completed between 1990 and 2013 we compute Expected Excess Leverage, or Leverage Deficit, as the difference between the target debt ratio and the potential combined leverage to serve as our main variable of concern. We report evidence supporting our predictions. Namely, we observe that the premium offered is negatively related to the combined firm s Expected Excess Leverage 4, an effect that is stronger for over-levered acquirers who offer a much larger premium to rebalance towards target leverage ratio and reduce Excess Leverage. Under-levered acquirers also offer a higher premium to further reduce Excess Leverage. These results suggest that, although over-levered and under-levered acquirers may have the same motive of maximizing firm value in acquisitions, they choose to do so differently. While over-levered acquirers rationally choose to maximize firm value through leverage rebalancing, under-levered acquirers believe that the benefits of exploiting market timing and paying with overvalued stock (Baker and Wurgler, 2002; Dong, Hirshleifer, Richardson, and Teoh, 2006; Ang and Cheng, 2006) outweigh the gain from rebalancing towards optimal leverage. This becomes apparent when looking at the Market-to-Book ratio which turns out to be positively associated with the premium paid in acquisitions for under-levered but not over-levered acquirers. Our findings reiterate the difference in value allocated by over-levered and under-levered acquirers to leverage rebalancing acquisitions: over-levered acquirers choose to pay with shares for the purpose of reducing leverage, irrespective of their stock valuation; in contrast, under-levered acquirers choose to pay with overvalued equity to gain from market timing and reduce Excess Leverage further instead of rebalancing towards target leverage. Our findings also reveal that over-levered acquirers offer a higher premium for deals with potential increase in debt capacity which is not the case for under-levered acquirers who are not 4 Premium is the pre-run-up premium offered relative to target share price 40 days prior to deal announcement. In order to remove outliers, this measure is truncated between 0 and 2 following Officer (2003). 5

6 necessarily in need for debt capacity improvement. Additionally, all our findings are more pronounced for acquirers with non-rated debt that are believed to be more financially constrained. Finally, we examine how markets react to the announcement of acquisition decisions. Our results document significantly lower announcement returns for over-levered acquirers that reduce Excess Leverage, but higher 12 to 24 months post-announcement returns for these same acquirers. These results imply that the market may not fully understand the option value of reduced leverage until 12 to 24 months after announcement when acquirers are eventually observed to put the extra borrowing capacity into good use (i.e. when the option to borrow is exercised to finance good investment opportunities that could not otherwise be revealed at merger announcement). Results also imply that over-levered acquirers do engage in the most value-enhancing acquisitions (Uysal, 2011). Our results also provide evidence that an increase in debt capacity is value-enhancing in the long-run for over-levered, but not for under-levered acquirers. We account for potential endogeneity between the premium and the method of payment and we report robust results. For further robustness tests, we replicate the analysis using book leverage instead of market leverage. We also account for various market conditions that could drive acquirers behavior, namely: tight financial markets, M&A liquidity, financial crisis, and bear stock market. Our findings are all robust. In additional robustness tests, we use industry median leverage as a proxy for optimal leverage and we report qualitatively similar results. Our study is related to the literature examining capital structure decisions through acquisition choices. In particular, Harford, Klasa, and Walcott (2009) show that deviations from target leverage levels affect acquisition finance and capital structure adjustments post-acquisition. Also, Uysal (2011) finds that the likelihood of acquisitions, method of payment, and premium are largely affected by acquirers deviation from target capital structure. Our study goes one step further 6

7 and shows that the gains from achieving target leverage are more valuable for over-levered acquirers that need to create the option to borrow in the future and expand their debt capacities. Additionally, this paper relates to the acquisition finance literature. Vermaelen and Xu (2014) shows that acquirers are able to pay with overvalued stock when such financing decision can be justified by motives relating to the achievement of optimal capital structure, others relate stock payment to the taxation of cash vs. stock offers (Gilson et al., 1988), risk sharing (Hansen, 1987), and different SEC requirements (Martin, 1996). This paper adds to these studies by showing that over-levered acquirers offer a higher premium with a high share fraction in order to rebalance towards optimal leverage, irrespective of their stock s valuation, while under-levered acquirers pay higher premiums to reduce leverage further, when the benefits of paying with overvalued equity exceed the benefits of reaching optimal leverage. The paper is also related to Harford and Uysal (2014) who show that firms with rated debt are more likely to undertake acquisitions than non-rated firms and that they offer a higher premium. Our paper extends these findings by showing that non-rated over-levered acquirers offer a higher premium in order to rebalance towards optimal capital structure and create leverage slack. The rest of the paper proceeds as follows: Section II describes the sample selection procedure and methodology. Section III discusses our main empirical findings of the study and section IV concludes. II. Sample and Methodology We extract our M&A sample from the Securities Data Corporation (SDC) database of Thomson Financial. We include all M&A deals announced and completed in the U.S. market between January 1 st 1990 and Dec. 31 st 2013 where both the acquiring and target firms are publicly 7

8 listed on the U.S. stock markets. For a deal to make it into our sample, it should have resulted in a transfer of control so that the percentage of shares acquired after completion must be at least 50%. We exclude all financial and utility firms (SIC and respectively) as per the M&A and capital structure literatures (Hovakimian, Opler, and Titman, 2001; Fama and French, 2002; Flannery and Rangan, 2006; Uysal, 2011). We collect share price data from the Centre for Research in Security Prices (CRSP) database, and accounting and financial information from COMPUSTAT. To reduce the effect of errors caused by outliers, we winsorize all variables at the 5% and 95% levels. The final sample consists of 1,810 acquisitions. We estimate optimal leverage from a fitted regression following Kayhan and Titman (2007) with industry fixed effects. 5,6,7 We also define L (Cash Fraction) as the expected or potential post-acquisition leverage ratio of the combined firm conditional on the cash fraction in the method of payment after considering that cash paid is obtained through new borrowing (Bhardwaj and Shivdasani, 2003; Faccio and Masulis, 2005; Harford, Klasa, and Walcott, 2009; and Elsas, Flannery and Garfinkel, 2014). We present a detailed procedure explaining the estimation of optimal leverage and combined leverage, L (Cash Fraction), in appendix B. Panel A of Table 1 reports the annual sample statistics for mean Leverage and Excess Leverage for acquirers, targets and the combined entity. This panel shows that both parties are 5 We believe that the notion optimal leverage to an average firm is closer to our regression-based estimate. In the sense that the predicted leverage may not be the elusive true optimal (which is still subject to controversy in the literature), rather what we need in this paper is not the true optimal but what the firm considered as optimal at the time they made the acquisition decision, where an empirical relations from existing firm s leverage decision captures the idea better. 6 Studies applying Kayhan and Titman (2007) for leverage prediction include, among others, (Chang and Dasgupta, 2009; Harford, Klasa, and Walcott, 2009; Uysal; 2011, Vermaelen and Xu, 2014). 7 For robustness, we also rerun our tests with the Fama and French (2002) estimation models, and we use several calculation methods of the dependent variable, that is, we use book leverage, market leverage and also net leverage, defined, as in Vermaelen and Xu (2014), as leverage net of the firm s excess cash. In additional robustness tests, we use industry median leverage as a proxy for optimal leverage. Results are unchanged. 8

9 under-levered in most of the sample years whereas the combined entity has positive Expected Excess Leverage after 2002 mainly due to more debt financing for such deals. Panel B reports a summary of deal, target, and acquirer statistics. The average transaction value in our sample is approximately $839 million, with a median of $215 million. On average the equity (cash) fraction in the consideration offered is 49% (40%). The mean premium is 50.77%. The Excess Leverage statistics are negative for both targets and acquirers, implying that both parties are on average under-levered, although acquirers appear to be less levered than targets. In contrast, the average combined entity becomes over-levered with mean (median) Expected Excess Leverage of 4.98% (2.73%). This is apparently due to the higher mean (median) leverage of the combined firm of 35.34% (32.77%) which is also a function of large deal size and relatively high debt-financed cash fraction. Finally, the relative size of the target to acquirer is quite substantial with a mean of 43% implying that these deals are expected to have a significant impact on the acquirer s future prospects. Insert Table (1) Here III. Results A. Univariate Analysis: 1. Expected Excess Leverage and Merger Premiums: Panel A of Table 2 reports univariate analysis by groups of low and high Expected Excess Leverage. Consistent with our predictions, the table shows that the merger premium is significantly higher for the group of firms with a closer distance to optimal leverage, i.e. firms with lower Expected Excess Leverage. To illustrate, the mean (median) premium is 55.32% (46.15%) for the Low Excess Leverage sub-sample compared to a mean (median) of 47.80% (39.55%) for the High Excess Leverage sub-sample, with a difference in mean (median) significant at the 1% (1%) level. The willingness of acquirers to pay higher premiums for reduced Excess Leverage is also reflected in 9

10 the significantly lower value of (L (Cash Fraction)), the combined firm s post-merger expected leverage, in the Low Excess Leverage sub-sample. To further highlight this result we illustrate the premium by quintiles of Expected Excess Leverage in figure 1 below: Figure 1: Premium vs. Expected Excess Leverage: The figure shows the merger premium for five quintiles of Expected Excess Leverage. The sample consists of 1,810 acquisitions between 1990 and 2013 and is formed by the intersection of the Securities Data Corporation (SDC) database, CRSP, and Compustat. Financial and utility firms (SIC and respectively) are excluded. Expected Excess Leverage represents the deviation from optimal leverage and is calculated as (L (Cash Fraction) - Optimal Leverage). L (Cash Fraction) is the expected market leverage of the combined firm, given the cash fraction offered in the method of payment. Examining stock market reaction, we notice that acquirers and targets generate lower announcement returns in the Low Excess Leverage sub-sample. For acquirers, it seems that rebalancing towards optimal leverage is not well received by the market, an observation that is arguably inconsistent with the notion of value creation from rebalancing. However; this result may be driven by irrationally overestimating the incremental contribution to firm value, implying that such acquirers may have overpaid for targets. Additionally, settling the deal with more acquirer shares, which have a higher M/B ratio, may have resulted in the lower returns for acquiring and 10

11 target firms in the Low Excess Leverage sub-sample. Namely, for this sub-sample, the mean Share fraction is double what it is in the High Excess Leverage sub-sample. Finally, it could be argued that the leverage effect may be caused mainly by higher stock market valuation, that is, a low Excess Leverage is simply due to high market value, or high Q ratio. However, we address this concern using book leverage instead of market leverage in our regressions and we find robust results 8. Insert Table (2) Here Panel B replicates Panel A by groups of over-levered and under-levered acquirers. 9 We are particularly interested in distinguishing between the behavior of under-levered and over-levered acquirers in mergers and acquisitions because, first, their investment opportunities are different (Uysal, 2011), and second, the benefits from leverage adjustment are asymmetrical for these two groups; therefore the speed at which leverage is adjusted differs (Korteweg, 2010; Faulkender, and Flannery, Hankins and Smith, 2012; Harford, Klasa, and Walcott, 2009; Korteweg, 2010; DeAngelo, DeAngelo, and Whited, 2011 and Uysal, 2011). As a result, the value that firms allocate to acquisitions might differ greatly between over-levered and under-levered acquirers, and that is what we aim to examine in what follows. As with Panel A, the subsample of Low Excess Leverage deals is associated with a significantly higher premium than the High Excess Leverage subsample, and this is evident for both groups of under-levered and over-levered acquirers. So, on the one hand, over-levered acquirers associate a higher value to deals that help rebalance towards optimal leverage, which is consistent with our prediction, and on the other hand, under-levered acquirers associate a lower value for rebalancing 8 Tables are available upon request. 9 Due to the small number of observation obtained in some of these groups, especially for over-levered acquirers with Low Expected Excess Leverage, we classify a firm as over-levered (under-levered) if its debt is strictly above (below) its optimal debt ratio. Although this definition sounds strict, in our regressions afterwards, we also report results for an additional definition being Lowest (Highest) quartile for under-levered (over-levered) firms. 11

12 toward optimal as is evident by the higher premium they offer for deals that reduce Expected Excess Leverage. Additionally, we observe that over-levered acquirers who decrease (increase) Excess Leverage offer a higher (lower) share fraction for target firms. The same applies to under-levered acquirers. However, the M/B ratio is higher and significant at the 1% level in the Low Excess Leverage subsample for under-levered but not over-levered acquirers, which supports the literature suggesting that acquirer stock overvaluation is a major incentive to many stock-financed acquisitions (Shleifer and Vishny, 2003; Rhodes-Kropf, Robinson, and Viswanathan, 2005; Dong, Hirshleifer, Richardson, and Teoh, 2006; Ang and Cheng, 2006). Thus, under-levered acquirers may opt for even more underleverage as they perceive the benefits of issuing overvalued equity exceed the benefits of approaching optimal leverage (Faulkender, Flannery, Hankins, and Smith, 2012). Under-levered acquirers, therefore; choose to take advantage of equity market timing (Baker and Wurgler, 2002) rather than reach optimal leverage because stock overvaluation is temporary, while gains from moving toward the optimal can always be recaptured in any later date. 2. Increase in Debt Capacity and Merger Premiums: Increase in debt capacity has been cited as one of the merger motives; as a matter of fact, Ghosh and Jain (2000) show that acquirers increase their leverage after merger and associate this with the increased debt capacity of the merged firm. In this subsection, and for the first time, we examine the value of increased debt capacity to acquirers. Table 3 reports the analysis of acquisition premiums based on acquirer pre-merger Excess Leverage and whether or not the deal results in increased debt capacity. We expect that the increase in debt capacity matters most for acquirers that have already exhausted their own pre-merger debt capacity and are in need of raising financing (Lewellen, 1971; Levy and Sarnat, 1970; Ghosh and Jain, 2000). Consequently, such acquirers are expected to pay 12

13 more for acquisitions with the potential to increase the debt capacity of the combined firm. We define Potential Increase in Debt Capacity as (Combined Firm Optimal Debt Ratio - Acquirer Optimal Debt Ratio) in year Insert Table (3) Here The table reveals that, on average, over-levered acquirers pay a higher premium if the merger leads to potential increase in debt capacity; to illustrate, the mean (median) premium is 53.54% (43.57%) when there is potential increase in debt capacity vs. a mean (median) of 45.80% (38.06%) when there is no potential increase. The difference in means (medians) is statistically significant at the 5% (10%) level, reflecting the value of increased debt capacity for over-levered bidders. However, for under-levered acquirers the difference in premium is not statistically significant. These values are depicted in figure 2 below. Figure 2: Premium vs. Potential Increase in Debt Capacity by Acquirer Excess Leverage: The figure shows the merger premium vs. the increase in debt capacity for over-levered and under-levered firm separately. The sample 10 In this context if the optimal debt ratio of the combined firm is higher than the optimal debt ratio of the acquirer, then this is a simple sign of the increase in debt capacity of the combined firm and vice versa. 13

14 consists of 1,810 acquisitions between 1990 and 2013 and is formed by the intersection of the Securities Data Corporation (SDC) database, CRSP, and Compustat. Financial and utility firms (SIC and respectively) are excluded. Increase in Debt Capacity is defined as Combined Firm Optimal Debt Ratio minus Acquirer Optimal Debt Ratio) in year -1. B. Multivariate Analysis: 1. Do Acquirers offer a higher premium in deals that close the gap with optimal leverage? Table 4 reports results of our OLS regressions. The merger premium is regressed on our main variable of interest, Expected Excess Leverage, and other independent variables which include a set of dummy variables that take the value of 1 if the acquirer and target are within the same industry (Industry-Relatedness), if the acquirer had at least 5% ownership stake in the target before acquisition (TOEHOLD), if the acquisition is hostile as per the SDC definition (Hostile), if there is a competing bidder in the deal as reported in SDC (Competed), if the payment method for the deal is pure equity (Shares), or a mixed offer of cash, equity and other forms (Mixed), zero otherwise. Other independent variables include other characteristics of deals, acquirers, and target firms such as: the relative size of the target s market value of equity to the acquirer s market value of equity (Relative Size), the natural logarithm of the market value of equity of the acquirer (Ln (Eq.MV)), the return on assets (ROA), market to book ratio (M/B), OCF-to-Assets MV, Asset Tangibility, Research and Development (R&D) to Sales, and the Selling, General and Administrative expense to Sales. The table shows that the coefficient on Expected Excess Leverage is negative and strongly significant, in line with univariate observations. In other words, the lower the Excess Leverage of the combined firm becomes, the higher the premium offered by bidders. Our results hold when run on groups of over-levered and under-levered acquirers. However, as previously observed, over-levered acquirers offer a higher premium for the perceived gain from favorable capital structure 14

15 rearrangement, which is derived from the value increase from moving closer to their target debt ratio, and the gain in the option to raise debt by creating leverage slack. Additionally, when comparing the lowest Excess Leverage Quartile (under-levered) to the highest Excess Leverage Quartile (over-levered) it becomes apparent that the premium is not significantly related to Expected Excess Leverage for the former but highly negatively significant for the latter, consistent with the other two regression models of over-levered acquirers; the magnitude of the coefficient on Expected Excess Leverage however is the highest (-1.027) among all models, which implies that acquirers with the greatest need to create leverage slack do in fact offer the highest premium to reach that target. Among other control variables, we notice that overvalued acquirers offer a higher premium and overvalued targets attract a lower premium. These are reflected in the positive (negative) coefficients of the M/B ratio for the acquirer (target) firms. This is particularly observable for under-levered acquirers only where the coefficient of the M/B ratio is highly statistically significant, while it is not for over-levered acquirers, implying that under-levered acquirers have incentive to pay higher premiums when they are paying with overvalued equity, while over-levered acquirers are mainly concerned with rebalancing their capital structure, irrespective of their equity valuation. For robustness, we control for possible endogeneity between the premium offered and the payment form. Although, the premium is affected by the choice of the method of payment, as previous research show (e.g. see Betton, Eckbo and Thorburn, 2008), it could be argued that the choice of the payment form may be affected by how much premium management is willing to pay depending on the relative valuation of target and acquiring firm. In order to account for possible endogeneity, we run a Two-Stage Least Square (2SLS) regression using the size of the acquiring firm as an instrumental variable since the literature indicates that size variables are major determinants of the payment form (Faccio and Masulis, 2005 and Chemmanur, Paeglis and Simonyan, 2009). Results 15

16 are reported in Table C.2 of the appendix; for brevity we only report results of the second stage regression, and we observe that the negative and strongly statistically significant relationship between Expected Excess Leverage and acquisition premium is preserved in all specifications. We also replicate Table 4 using book leverage instead of market leverage. Our findings are not altered by this substitution and are reported in Table C.3 of the appendix. Finally, we account for the fact that various market conditions could drive acquirers, especially over-levered ones, to pay a high price for rebalancing towards optimal leverage. Therefore, we also replicate Table 4 with various market conditions controls, namely: tight financial markets, M&A liquidity, financial crisis and bear stock market. Results are reported in Table C.4 of the appendix. Insert Table (4) Here 2. How do acquirers price the potential increase in debt capacity? In Table 5, we formally test the effect of potential increased debt capacity on acquisition premiums. Our dependent variable is the Increased Debt Capacity defined as (Combined Firm Optimal Debt Ratio - Acquirer Optimal Debt Ratio) in year -1. We use the same independent variables as in Table 4. Results reveal that the coefficient on Increased Debt Capacity is not significant for under-levered acquirers. Under-levered acquires do not necessarily pay a high premium for deals that result in increased debt capacity. In contrast, the evidence for over-levered acquirers reveals that the coefficient on Increased Debt Capacity is positive and highly statistically significant. These results imply that over-levered acquirers, who need debt capacity most, place a 16

17 high value on increased debt capacity, consistent with our hypotheses and observations in the univariate analysis 11. Insert Table (5) Here 3. Does Access to Debt Markets Impact Rebalancing Preference? So far, we have examined how valuable rebalancing towards optimal leverage is for acquirers with different Excess Leverage. However, irrespective of their Excess Leverage, acquirers don t all face the same constraints when accessing markets to raise capital. The literature uses credit ratings to proxy for financial constraints. The existence of a credit rating is found to reduce information asymmetry, therefore easing firms access to capital markets to raise funds whenever their investment needs call for it (Whited, 1992; Gilchrist and Himmelberg, 1995; Almeida et al., 2004; Koziol and Lawrenz, 2010; Campello and Chen, 2010). In addition, because financial constraints keep managers from overinvesting, constrained firms are observed to undertake only the best investments (Harford and Uysal, 2014; Karampatsas, Petmezas, and Travlos, 2014). In Table 6, we explore whether having an S&P credit rating affects the value of reduced Excess Leverage to acquirers. The coefficient on Expected Excess Leverage turns out negative and statistically significant only for acquirers that do not have a credit rating, both under-levered and over-levered. This result implies that acquirers difficulty in accessing capital markets is a leading motivation to reduce Excess Leverage in acquisitions. Hence, non-rated over-levered acquirers pay higher premiums for acquisitions that would reduce combined leverage and relieve their financial constraints, whereas non-rated under-levered acquirers also offer a higher premium to further reduce Excess Leverage to avoid being financially 11 We also conducted another related test using univariate and multivariate analyses in which we test whether acquirers who intend to (actually) use more debt offer a higher premium and we find they indeed do, which supports also the notion and our findings above that the increase in debt capacity of the combined firm is valuable for acquirers and thus commands a higher premium. 17

18 constrained. Additionally, further investigation reveals that non-rated acquirers are relatively smaller, more cash rich, and have lower optimal leverage ratios than rated acquirers 12, which could explain why they may desire to keep low leverage levels. In contrast, the results for rated acquirers simply show that the latter pay higher premiums on average (reflected in the higher regression intercept) which is consistent with Harford and Uysal (2014), but this is not necessarily done in relation to leverage rebalancing as the coefficient on Excess Leverage Deficit is insignificant. Insert Table (6) Here 4. Stock market announcement returns of Expected Excess Leverage and Increase in debt capacity? In our previous subsections, we observed that over-levered acquirers are willing to pay higher premiums for reducing Excess Leverage and increasing debt capacity. In this section, we examine market reactions to acquisition announcements to test whether these acquisitions create value. For this purpose, we use two methodologies: the Stock Market Announcement Stock Returns (CARs) and the Buy-and-Hold Abnormal Returns (BHARs). Results are reported in Tables 7 through 9. a) Announcement Returns Table 7 presents regressions of acquirer stock market announcement returns calculated as the three-day acquirer cumulative abnormal stock return. All specifications include, in addition to our main variable of interest, Expected Excess Leverage, the same set of control variables as in Tables 4 to 6. The CAR is positively associated with Expected Excess Leverage for all groups examined in the table. Abnormal returns are also observed to be higher for the subsample of over-levered acquirers especially in the highest quartile with respect to other quartiles which confirms our 12 We prepared a separate univariate table for rate vs. non-rated firms, however we did not add it to the paper to conserve space. This table is available upon request. 18

19 univariate results in Table 2. Therefore abnormal returns seem to be higher with higher expected Excess Leverage. At first sight, these results appear to be inconsistent with value creation expectations from rebalancing towards target leverage and creating leverage slack by over-levered acquirers. However, the results may imply that the market is commending acquisitions that are accompanied by lower premiums (as high Excess Leverage is associated with lower premium). The market may also consider that over-levered acquirers irrationally overpaid for these deals and overestimated the incremental contribution to firm value of leverage rebalancing. As an alternative explanation, at merger announcement, the market may not fully understand the option value of reduced leverage as it is not obvious yet whether the acquirer will put the extra borrowing capacity into good use via financing good projects. Insert Table (7) Here b) Buy-and-Hold Abnormal Returns Performance The results reported above assume that stock prices correctly reflect information perceived by investors at announcement dates; however, investors may make systematic errors in evaluating acquisitions at announcement and hence inaccurately incorporate information in stock prices in the short run. Consequently, the mis-valuation of acquisitions on the short run is likely to be corrected and hence price reversals may be observed in the long-run (Uysal, 2011). For this purpose, we estimate portfolio-matched buy-and-hold abnormal returns (BHARs) of the acquirer for two post-announcement periods, namely 12 and 24 months. Following the method of Bhojraj, Hribar, Picconi and Mcinnis (2009), we calculate the BHAR for each firm as: T BHARi = t=1 (1 + Rit ) t=1(1 + Rbenchmark,t ), T 19

20 Where R benchmark,t is the return on a matched portfolio at month t, and R it is the stock return of firm i at month t. 13 Results in Table 8 support price reversals in the long run for over-levered acquirers. As a matter of fact, the table shows a negative and significant association between Expected Excess Leverage and BHARs at the 12 and 24 months following the announcement for over-levered acquirers only. Therefore deals resulting in lower Excess Leverage experience better acquirer longrun post-acquisition returns for up to 24 months. These results are consistent with the view that over-levered acquirers pursue the most value-enhancing acquisitions (Uysal, 2011) even though the market only incorporates this value in the long-run and corrects earlier short-run mis-valuation. The table also reflects that better long-run stock returns are more pronounced for larger acquirers as is observed in the positive and significant coefficients on (Ln (Eq.MV)). Returns are also negatively associated with the target Market-to-Book ratio (M/B), namely for under-levered acquirers, consistent with equity overvaluation as per the market timing hypothesis (Baker and Wurgler, 2002), and strongly positively associated with R&D spending (R&D/Sales) for underlevered acquirers, reflecting a higher propensity to spend on Research and Development when debt capacity is readily available. Insert Tables (8) Here Finally, in table 9, we conduct a similar test to that in table 8 on the long-run value gain of acquisitions that lead to potential increase in debt capacity. Results reveal that a potential increase in debt capacity is positively perceived by the market for over-levered acquirers as is mirrored in the positive and significant coefficient on Increased Debt Capacity for the 12 and 24 months following 13 To do so, we match each acquirer in our sample to a matched portfolio constructed following the Fama and French (1993) procedure of 25 equally-weighted Size/Book-to-Market (B/M) portfolios at the beginning of the announcement year using the Size/BM breakpoints from Kenneth French s website. 20

21 announcement. For under-levered acquirers however, the coefficient on Increased Debt Capacity is not significant. These results indicate that a potential increase in debt capacity is highly priced by the market for over-levered acquirers, consistent with our predictions and with the notion that debt capacity improvement is a source of value gain for acquirers who need it most. On the other hand, in unreported analysis we examine whether acquirers that offered a higher premium to increase debt capacity, actually used that additional debt capacity and increased their leverage further after acquisition. Compared to year -1, univariate analysis shows that acquirers with increased debt capacity experienced a larger and significant increase in leverage compared to those without increased debt capacity in years +1, +2 and +3 respectively. 14,15 Moreover, multivariate analysis shows consistent results as the post merger changes in leverage (from year -1 to years +1, +2 and +3) are significantly positively associated with the potential increase in debt capacity Additional Robustness Tests Insert Table (9) Here This section examines alternative measures for optimal leverage. Using empirically estimated optimal leverage levels assumes the existence of a completely specified theoretical capital structure model that is universally agreed upon. In practice, it is more likely that a firm may follow what it perceives as a desired leverage from an incomplete capital structure model. In other words, using an estimated empirical capital structure model in place of what the firms have in mind and rely on, may involve estimation error. Therefore, there may be good reasons to use alternate measures of desired/optimal capital structure that practitioners rely on. For this purpose, we estimate optimal 14 The mean change in leverage are 7.56% vs. 3.35%, 8.71% vs. 3.86% and 9.5% vs. 4.97% in years +1, +2, and +3 respectively for acquirers with vs. those without increased debt capacity, significant at the 1% level. 15 Ghosh and Jain (2000) associate the post-acquisition increase in leverage with possible increased debt capacity, but they do not provide a measure or evidence of debt capacity improvement. Our findings provide a more direct relation between increase in leverage and debt capacity improvement. 16 All these tables are available upon request. 21

22 leverage as the industry median leverage during a given year using the Fama-French 48 industries. We also use market leverage and book leverage, replicate the analysis, and report robust results in the Appendix (Table C.5 and A.6). Additionally, it could be argued that the optimal/desired capital structure could be a range instead of a point, and thus, movement within the range should have no effect. For this purpose we also re-run our regressions of Table 4 with various ranges of the Expected Excess Leverage variable instead of a continuous variable, and find qualitatively similar results. 17 IV. Conclusion Using a sample of 1,810 US M&A deals completed between 1990 and 2013, we study how capital structure considerations affect acquisition pricing. We find evidence that firms associate a high value for reducing Excess Leverage and are willing to pay a price for that reduction. We show that the gains from rebalancing towards target capital structure and reducing excess leverage are more valuable for over-levered acquirers that need leverage slack and create the option to borrow in the future. Namely, we find that over-levered acquirers offer a much larger premium to rebalance towards target leverage ratio and reduce Excess Leverage. The results also hold for under-levered acquirers who allocate a lower value to capital structure rebalancing and instead offer a higher premium to further reduce Excess Leverage. While over-levered acquirers rationally choose to maximize firm value through leverage rebalancing, under-levered acquirers believe that the benefits of exploiting market timing and paying with overvalued stock exceed the gains from rebalancing 17 In another variation, we have redefined the Expected Excess Leverage and split it into two variables as follows: Leverage Slack being equal Expected Excess Leverage if that is negative (under-levered) and zero otherwise, and Leverage Deficit being equal Expected Excess Leverage if that is positive (over-levered) and zero otherwise. We run separate regressions and included Leverage Slack in the over-levered acquirers model while we added Leverage Deficit in the under-levered acquirers model. In both cases we found similar results to those reported in table 4, that is, overlevered acquirers offer a higher premium the closer it is expected to be to target leverage ratio while under-levered acquirers offer a higher premium to get farther away from target leverage, i.e. to reduce leverage further. These tables in addition to all previous robustness tables are available upon request. 22

23 towards optimal leverage. Over-levered acquirers choose to pay with equity, irrespective of their stock valuation, with a main goal of rebalancing leverage toward optimal, whereas, under-levered acquirers face a trade-off between rebalancing towards optimal leverage and paying with overvalued stock. They chose the latter because they perceive the opportunity from stock overvaluation to be temporary, while gains from moving toward optimal leverage can be recaptured at a later time. These results are more pronounced for acquirers with non-rated debt who have no access to public debt markets and are believed to be more financially constrained (Harford and Uysal, 2014). The paper further examines the value of increased debt capacity to bidders. Our results show that debt capacity improvement is more valuable to over-levered than under-levered bidders, as is reflected in the higher premiums paid by the former for deals that exhibit a potential increase in debt capacity. Our findings are corroborated by an observed stock price reversal in the 12 to 24 months following the acquisition announcement. While stock market announcement returns reflect a drop in firm value following leverage-rebalancing acquisitions, mis-valuations are corrected on the long-run as echoed in the stock price reversals observed in the 12 to 24 months following the announcement. These results may indicate that that over-levered acquirers may have irrationally overpaid for these deals and overestimated the gain from leverage rebalancing. Alternatively, they also may indicate that the market may not fully understand the option value of reduced leverage until 12 to 24 months after announcement the acquirer exercises the option to borrow to finance good investment opportunities that could not be revealed at merger announcement. Overall, our results lend support to the view that over-levered acquirers engage in the most value-enhancing acquisitions (Uysal, 2011). 23

24 Appendix A: Variable Definitions Assets BV Assets MV Book Debt Book Equity Optimal Leverage Market Leverage Book Leverage Excess Leverage Tobin s Q OR Market-to-Book ratio of Assets t-1 (R&D Exp./Sales) t-1 R&D Missing Dummy t-1 Leverage t-1 Ln(Sales) t-1 Ln(Assets Bk) t-1 This is book value of Total Assets (Item AT). This is Market Value of Assets and is defined as liabilities(item LT) minus balance sheet deferred taxes and investment tax credit (Item TXDITC) plus Preferred Stock (as defined below) plus Market Equity (Item CSHO*Item PRCC_F). This is Total Assets (Item AT) minus Book Equity This is Total Assets (Item AT) minus liabilities (Item LT) plus balance sheet deferred taxes and investment tax credit (Item TXDITC) minus Preferred Stock. The predicted value of the leverage regression presented in Table C.1. This is Book Debt over Market Value of assets (as defined above). This is Book Debt over Total Assets (Item AT). Market Leverage minus Optimal Leverage (as defined above). Market Value or Assets MV (as defined above) over book value of Total Assets (Item AT). R&D expenses (Item XRD) over Total Sales (Item Sale). Dummy that takes a value of one if COMPUSTAT reports R&D expense (Item XRD) as missing, and of zero otherwise. This is the leverage ratio (book leverage or market leverage depending on the model used in the Leverage prediction regression) in year -1 relative to the merger announcement year. Natural logarithm of Sales in year -1 relative to the merger announcement year. Natural Logarithm of Total Assets in year -1 relative to the merger announcement year. Final Offer Premium relative to day -40 This is Pre run-up premium calculated as [(Final Offer price / P - 40) -1]. CAR (-1,+1) CAR (-1, +1) is the 3-day cumulative abnormal returns estimated using the market model over the (-210,-21) interval using the CRSP value-weighted index returns as the benchmark. The statistical significance of the returns is tested using the Patell (1976) test corrected for time-series and cross-sectional variation 24

25 of abnormal returns. OCF-to-Assets MV Cash-to-Assets BV (M/B) Tax shield (Deprec./Assets BV) EBITDA/Assets BV Debt Issue/Assets MV Net Equity Issue/Assets MV Tangible Assets (PPE/Assets MV) Selling Exp./Sales Deal value Relative size Industry Relatedness Cash Shares Mixed Hostile Operating Cash flow to MV of Assets Ratio and the Operating cash flow is sales minus cost of goods sold, selling and general administrative expenses, and working capital change, items (SALE-COGS-XSGA-WCAPCH). Cash to Book value of Assets ratio item (CHE) over item (AT). Market to Book ratio: Market value of Equity calculated as share price multiplied by number of shares outstanding Divided by Book value of shareholders equity. This is Depreciation over Book value of Assets. Item DP over item AT. This is operating income before depreciation (Item OIBDP) over Total Assets (Item AT). Net debt issues long-term debt issuance minus long-term debt reduction (DLTIS-DLTR) over Assets MV. The ratio of sale of common and preferred stock(item SSTK) minus purchase of common and preferred stock( Item PRSTKC) to Assets MV. This is net property, plant and equipment (Item PPENT) over Assets MV defined above. This is Selling, general and administrative expense (Item XSGA) to sales (Item SALE). Deal Value is the total consideration paid as reported in SDC. Target market value of equity Divided by Acquirer market value of Equity. Dummy equal one if the acquisition is between firms with the same two-digit SIC code. Dummy equal one if the method of payment is Pure Cash. Dummy equal one if the method of payment is Pure Share. Dummy equal one if the method of payment is a mixed offer of cash, equity and other forms. Acquisition is Hostile as in SDC database. TOEHOLD Dummy equal one for deals where the acquirer had at least 5% ownership in the target firm prior to the acquisition. 25

26 Family Control Financial Crisis Bear Market Spread Credit Rating Dummy Industry M&A liquidity Firm is classified in SDC as has family control. We classify a deal announced between July 1 st, 2007 and Dec. 31 st 2008 as falling in the financial crisis period (this classification is in line with Beltratti and Stulz (2012) We classify the period Mar. 13, 2000 to Sept. 30, 2002 as a Bear Market period 18 (Goldfarb, Kirsch and Miller, 2007) It proxies for capital liquidity, following Harford (2005), this is the spread between the average rate on commercial and industrial loans and the Federal Funds rate. It is provided by the Federal Reserve Senior Loan Officer's (SLO) survey. Dummy equal one if the acquirer has a Standard & Poor s credit rating of BBB or higher in year -1 This is calculated as the value of all acquisitions transactions for $1million for each year and two-digit SIC code divided by the total book value of assets of all Compustat firms in the same twodigit SIC code and year (Schlingemann, Stulz, and Walkling, 2002). 18 This classification is meant to differentiate this period from the financial crisis period, in which the stock market also suffered large decline. 26