1
How do deviation from the target leverage affect the payment of
acquisition?
Zhenni Hu 10824286
Msc Business Economics: Finance Supervisor: Tolga Caskurlu
Master Thesis 14/08/2015
Statement of Originality
This document is written by Zhenni Hu who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Zhenni Hu 14/08/2015
122
Abstract
The paper is to test whether the target capital structure exists in sight of merger & acquisition transaction. Among the main capital structure theories, trade-off theory and pecking order theory have a different assumption of target capital structure. Trade-off theory assumes that firms have their optimal leverage, which is opposite to pecking order theory. In this paper, I assumes that firms have their target leverage. Static trade-off theory assumes the target leverage and dynamic trade-off theory explains firms usually deviate from their target leverage in real life. The paper focus on the characteristics before the acquisition (cash fraction of deal payment and deal premium) to explore the relation with deviation from target leverage. First, I predict each firm’s target leverage following the previous studies. Then, I calculate that the firms’ leverage deficit of each sample year. Next, I investigate how leverage deviation affect the cash used in the deal payment and the deal premium for the target firms. The result show that before the acquisition, the higher deviation from the target leverage lead to the lower cash fraction used in the payment and the lower premium paid for the target firms, which follows my related literature. Therefore, I can get the conclusion that firms have their target capital structure and the deviation from the target leverage will affect firms’ further financial decision. Firms are willing to use less liability or issue less new debt when they are already over the target leverage. The extent of deviation also affects the level for firms to use debt.
Content
1. Introduction ...4
2. Literature Review ...6
2.1 Theoretical Support ...6
2.2 Target Leverage Prediction ...8
3. Hypothesis and Methodology ...9
3.1 Hypothesis ...9
3.2 Methodology ...9
3.3 The determinants of target leverage ...11
4. Data and Descriptive Statistics ...13
4.1 Data collection and Process ...13
5. Empirical Results ...15
5.1 The estimation of target leverage ...15
5.2 The calculation of leverage deficit ...17
5.3 How does leverage deficit affect the cash fraction used in payment? ...18
5.4 How does leverage deficit affect the premium paid by bidders? ...20
6. Robustness check ...23
7. Conclusion ...23
124 1. Introduction
Acquisition becomes a popular way for firms to gain capital, extend emerging market, broaden customer base, save costs for tangible and intangible assets and acquire new technology. During the negotiation of acquisition, the payment method of acquisition and premium for target firm is the important part for both counterparties, especially for the bidders. Since the deal value is usually depended by the size of the target firms, the large amount of payment comes with the completed transaction. As mentioned by Harford, Klasa and Walcott (2009), payment of acquisition can be divided into three ways: cash payment, equity payment and mixed payment. Therefore, the choice of the payment method can affect the liquidity of the bidders and acquriors’ original capital structure. As Bharadwaj and Shivdasani (2003) point out that most cash offers are debt-financed, I focus on the capital structure of the bidder to examine the capital structure before the acquisition and how it affect the financial choice of the transaction. The objective of this paper is to test that firms have their target capital structure and the deviation from their target leverage can affect firms’ financing decision of the further investment activities.
Among the theories of capital structure, the discussion whether firms have optimal capital structure lasts for a long time. Trade-off theory and free cash low theory (Jensen, 1986) explain that firms have their optimal leverage in different aspects. However, pecking order theory (Myers & Majluf, 1984), agency theory (Jensen & Meckling, 1976) and equity market timing theory (Baker & Wargler, 2002) consider there is not target leverage for firms, and the choice of raising debt or issuing equity is due to the asymmetric information or the current market value of firms.
The main conflict of meaningful target capital structure is discussed between the trade-off theory and pecking order theory. Trade-off theory can be divided into static trade-off theory and dynamic trade-off theory. Static trade-off theory focus on considering the bankruptcy cost when firms issuing additional debt. It explains that target capital structure exists when the benefits of interest tax shield equals to the cost of financial distress. And dynamic trade-off theory concerns more about the reality of firms. It considers and tests that firms usually deviate from the target leverage but
they will start to adjust back to the target when the cost of deviation from target capital structure is larger than the cost of adjustment toward the target. However, pecking order theory focuses on the asymmetric information inside and outside the firm so it explains that firms are willing to raise money by internal capital, debt and equity in order. Since managers and outside investors have asymmetric information and equity issuing always be considered as overvaluation of the firms, managers usually issue stock when it is no longer sensible to issue any more debt. Pecking order theory does not consider that firms have their target leverage and the capital structure is just the result of the natural choice in order.
Although many previous studies have already tested that firms have target capital structure and investigated how firms’ further financial activity can be affected by the leverage deficit, few literature stands on the sight of acquisition. In this paper, I want to test whether firms have the target capital structure, which is also considered as optimal leverage, and how target leverage affect the financial choice about payment method and acquisition premium from the acquisition aspect.
First, I predict the future year’s target leverage by using the previous year’s relevant variables. Then, I define the leverage deficit (Uysal, 2011) as the real market leverage minus the predicted target leverage of the same year. Here, I define market leverage as book value of total liability divided by market value of total asset. Next, I explore how the deviation from target leverage of bidders affect the percentage of cash used in the acquisition transaction. Finally, I investigate how the leverage deficit affect the transaction premium that acquirers paid for the target firms. In the empirical test, I find that higher leverage deviation lead the firms to acquire lower fraction of cash for the payment and lower premium for the target firms when the transaction is announced. The findings are identical with the previous studies (eg. Haford, 2009; Kayhan and Titman, 2011).
Haford(2009) and Kayhan(2011) study the relation between leverage deficit and transaction characteristics to test that firms have their target leverage and how the optimal leverage plays an important role on the transaction. They raise the research questions to estimate the effect of deviation from target leverage on transaction in
126
three different time period: before transaction, after the transaction completed, and after the transaction for a few years. In this paper, I focus on the time point before the transaction to test whether the firms have optimal capital structure. I raise two research questions about transaction’s payment and premium in order to test the main idea by using different independent variables in two methods.
Compared to the previous studies for testing the target leverage by investigating acquisition transaction, my contribution is that I use the combined control variables and new control variables to predict the target market leverage of the next year. Also, I add panel data of fixed effects to get the less unbiased correlation coefficient for the prediction and compare the result with the estimator from tobit regression. I give the different definition of transaction premium with Uysal (2011), I define premium as the deal value divided by the target firm’s market value of the last twelve-month.
In the rest part of the paper, I review the related literature in Section 2 and discuss the methodology in Section 3. Section 4 describes the data preparation and fundamental descriptive statistics. Section 5 explains my empirical findings. At last, I conclude the whole paper in Section 6.
2. Literature Review
2.1 Theoretical Support
The most influential theories of capital structure in finance are trade-off theory and pecking order theory. The static trade-off theory, which focuses on the benefits and costs of issuing debt, predicts that an optimal target financial debt ratio exists, which maximizes the value of the firm (Atiyet, 2012). The optimal point can be attained when the marginal value of the benefits associated with debt issues exactly offsets the increase in the present value of the costs associated with financial distress for issuing more debt (Myers, 2001). And dynamic trade-off theory points out that the actual debt ratio can adjust back if the capital structure remove from the target leverage because of seasoned equity offerings or other financing issuance (Kayhan and Titman, 2007).
for capital used to finance their businesses because of adverse selection (Myers and Majluf, 1984) and asymmetric information (agency theory) (Jensen and Meckling (1976). However, Halov and Heider (2004) argue that the pecking order theory is the special case for capital structure because of the character of adverse selection. They explain that the pecking order theory works for the adverse selection at first but adverse selection applies when there is asymmetric information about value or risk. Thus, the firms prefer to raise capital from outside and pecking order theory does not work any more. Myers (2003) points out that agency costs of equity could result in a pecking order. Jensen and Meckling (1976) also identify an agency problem of debt as risk shifting. This idea is that if the firm is operated on behalf of equity, only cash flows in non-bankrupt states matter. Therefore, the firm will tend to accept projects that are too risky but with large payoffs in good states.
Shyam-Sunder and Myers (1999) show that their test results have time series explanatory power for basic pecking order against trade-off hypothesis. However, since Shyam-Sunder and Myers (1999) do not conclude the broadly convincing result by only including 157 firms, Frank and Goyal (2003) extend the sample by studying the extent to which the pecking order theory of capital structure provides a satisfactory account of the financing behavior of publicly traded American firms over the 1971 to 1998 period, and conclude that net equity issues track the financing deficit more closely than do net debt issues, which is contrary to the pecking order theory.
Reviewing the previous studies, most support to static trade-off theory and verify the existence of the target capital structure (see in Graham and Harvey, 2001, DeAngelo and Whited, 2011, etc). Furthermore, Myers (1977), Myers and Majluf (1984) and other studies point out that firms usually deviate from their target debt ratio. Kayhan and Titman (2007) explain that based on the information asymmetry, market inefficiencies and transaction costs, cash flows, investment expenditures, and stock price histories will lead debt ratio deviate from the target one. Therefore, the deviation from the target leverage will affect the ability to issue debt and thereby affect the managers’ financing decision of firms. Thus, we examine how leverage deficit affect the financing decision from acquisition aspect for the payment method
128
and premium to the target firms.
2.2 Target Leverage Prediction
To investigate the effect of optimal debt ratio on the acquisition choice, the calculation of target leverage should be done first. Considering different determinants that affect target debt ratio, previous studies show the mixed and complicated regression result. Hovakimian, Opler and Titman (2001) raise that since firm may face impediments of movements toward their target ratio, the target ratio may change over time as the firm’s profitability and stock price change. They emphasize that firm history plays an important role on capital structure, which is also the part that beyond static trade-off theory. Also, they explain that firms are willing to issue new equity when the stock price is on a growth trend, which implies that firms perform well at that time. Therefore, firms with good performance is likely to decrease their leverage. In the Tobit regression for target leverage, they include R&D expenses/sales (R&D), selling expenses/sales (Exp), firm size (Size), defined as natural log of total assets, and the proportion of tangible assets (Tng) in the first stage regression. In particular, R&D and selling expenses/sales capture, among other things, construct like future growth opportunities and product uniqueness that might otherwise be captured by the market-to-book ratio. They define debt ratio to be book value of debt over the sum of book value of debt and market value of equity.
In the subsequent study of Kayhan and Titman (2007), the determinants are more comprehensive. Standing on a long-term horizon, they consider financial deficit (as dummy variable), timing measures, stock returns, profitability (past 5 years EBITDA/ the first year’s total asset), leverage deficit and change in target. Therefore, they add EBITD, R&D dummy variable and industry dummy variable, compared to Hovakimian, Opler and Titman (2001). Also, the results indicate that after controlling for changes in stock prices and other timing and pecking order effects, changes in debt ratios are still explained by the leverage deficit and by changes in target debt ratios. But there is still endogeneity of financial deficit variable that need to be eliminated.
Harford, Klasa and Walcott (2009) considers firm size, growth opportunities, liquidation value of assets, marginal tax rate lead the debt ratio from their target one, based on the short horizon that not as long as 5 years but just 1 year. Their hypotheses only find support if financing effects are strong enough to have a role in the method of payment choice in spite of other factors. In the Tobit regression, the different part from Hovakimian and others (2001) is that they select to use the future year of the EBIT over total assets to proxy future year’s profitability instead of the past year’s profitability. Also, they estimate separate annual regression to predict leverage instead of estimating a pooled regression model.
Uysal (2011) uses total asset instead of sales to be the denominator for profitability and R&D expenses, compared to the previous studies. Also, Uysal considers that the past target leverage may affect the current target leverage, thereby including the target debt ratio of the past year in the model and test p-values by Newey-West method for the autocorrelation of residuals. And he studies on interdependence of capital structure and investment decisions, rather than one-side effect for the counterparty.
3. Hypothesis and Methodology
3.1 Hypothesis
Following the research questions of the paper, I raise two hypotheses to the questions. The first hypothesis is bidders with higher deviation from their target leverage are less likely to acquire the target firm by cash. The second hypothesis is bidders with higher deviation from the target leverage pay less money for the transaction based on the target firms’ total asset value.
3.2 Methodology
In this paper I need the two-step estimation for the regression. First of all, I predict the target leverage for each firm and calculate the deviation of the real market leverage from the target leverage. The deviation of the leverage is defined to be real market leverage minus target leverage of the same year for each firm. Next, I let
12 10
deviation of the leverage to be independent variable, make the percentage of cash in transaction and the premium for the target firms to be dependent variable respectively, including other control variables in the two model. For the first step, I use tobit regression and panel data of fixed effects to compare the result. For the second step, since all the observations are either time-series data or cross-section data, I regress by OLS.
For the first step, I collect balanced panel data for each firm in 15 years and get 14-years effective regression. Here, I need to explain that the first year’s data is to be used for the second year’s regression independent variables, therefore I can only get 14 groups of Beta from the annual regression. To make sure about the consistency of the result, I use the two-sided Tobit regression to censor the target value from 0 to 1 and winsorize the outliers back to the upper and lower bound (eg. McDonald and Moffitt, 1980; Kayhan and Titman, 2007). The reason for using Tobit regression is to censor the leverage between the upper bound and lower bound since debt ratio usually is between 0 and 1 by definition but few firms have the negative market value of equity, which make debt ratio over 1 and increase the outlier number of the regression. Also, since the observation I get is panel data, I use the panel data with fixed effect of entities to estimate the beta of every control variables. So I want to compare the regression result for the different two methods.
The prediction model is: Market Leverageit=iWit1it1. Here W means the control variables, including (Operating income before depreciation and appreciation/book assets)t-1, ln(Sales)t-1, (Property, plant and equipment/book assets)t-1, (Market-to-book ratio of assets)t-1, (Research and development expense/book assets)t-1, (Stock return)t-1, (Dummy for no research and development expense)t-1, (Cost of goods/sales)t-1. Since all the independent variables are the prior-year data, endogeneity can be reduced.
Then, I calculate leverage deviation by subtracting target leverage from real market leverage. Here, the definition of leverage is same with the definition in previous studies (eg. Harford, 2009; Uysal, 2011). I make total book value of
liabilities divided by the sum of total book value of liabilities and total market value of equity to be the market leverage in my paper.
Next, I need to build the relationship between the fraction of the payment in cash for the acquisition and the deviation pre-acquisition. I use OLS model which is clustered by each firm. The model is showed belowed:
1 7 6 1 5 1 4 1 3 1 2 1 1 1 ) / ( Re ) / ln( Re * it it it it it it it it TA EBITDA turn Stock CPI e Marketvalu Sales ansaction lativeoftr bookassets to Market rage Marketleve viation LeverageDe c ofcash Percentage
Finally, I test how the bidders pay for target firms based on the target firms’ market value, which also called premium for the acquisition. I consider the control variables in three aspects, bidders’ firm characteristics, target firms’ basic characteristics, and deal transaction variables. The model is:
1 8 7 1 6 1 5 1 4 3 1 2 1 1 1 arg ) / ( Target of Book -Re Re -Re ' arg / it it it it it it it it et forT TA EBITDA Market lativesize saction lativetran Sales Bookratio Market turn Stock rage MarketLeve Deviation e marketvalu etfirms T nValue Transactio
Here, market leverage, stock return, make-to-book ratio and sales represent the characteristics of bidders. Relative size of transaction is defined as the market value for each bidder scaled by the transaction value of the acquisition they attended, and the relative size here equals that the target firms’ market value divided by the bidders’ market value. So these two variables represent the character of the acquisition deal. And the last two variables of market-to-book ratio and profitability for target firms are also considered in the regression. Since the data I get is the pooled data because we just select one acquiror with one recent deal transaction, I use OLS regression clustered by each firm (each deal).
3.3 The determinants of target leverage
Considering about predicting target leverage, I use two-sided Tobit regression to connect financial control variables of the prior year with the real market leverage in the later year. Also, I use panel data with fixed effects method to deal with the same variables. Then I use the coefficient times the later year control variables respectively
12 12
to get the target leverage. I think about the control variables in these aspects : the profitability, the growth opportunity, the collateral ability for debt, product uniqueness, and size. Since the regression is for prediction, I let all the independent variables to be the prior-year data, and make the dependent variable to the current-year data.
Profitability: I list (EBITDA/BookAsset)t-1 to represent it. Although Haford (2009) considers the (the future year’s operating income/Total assets) to avoid the unexpected profitability or the lagged profitability, I still use the past year’s operating income before depreciation and appreciation because I think the past year’s margin can provide the reliable rate of the profit. As previous study mentioned, firms with higher profitability are less willing to carry debt or issue new loans because firms has the ability to earn more for the future operation and goods production.
Growth Opportunity: I use firms’ market-to-book ratio and stock return to assess firms’ growth opportunity. The two variables have the same meaning that the stock price is higher than before, which represents that firms are considered to have good growth opportunities and are willing to be owned. The firms with great growth opportunities are not likely to carry more debt because they can gain the investment of the new project from their shareholders.
Collateral ability for debt: As we know, firms with more fixed assets are more easily to issue new debt or bank loans because of the carry value of the fixed assets, which we call it collateral ability. Here, I define it as Net property, plant and equipment/Book Asset, and the variable should be positive to the market debt ratio.
Product Uniqueness: The definition is from Titman and Wessels(1988). I include R&D expense/Book asset and Cost of Goods/Sales to represent product uniqueness. Although Haford(2009) and Uysal(2011) use the selling expense/Sales instead, I consider that cost of goods is more direct to the expense of products. The more cost of goods that firms spend scaled by the annual sales, the less likely the firms will carry the debt. And firms with higher R&D expense scaled by the total book asset are thought to be the firms with great growth opportunities of their new product. Therefore, the product uniqueness should be expected to have the negative direction with the current-year market leverage.
Size: Following previous studies(eg. Kayhan and Titman, 2005), I let the size of firm to be expressed by natural logarithm of annual sales of each firm. The firm with larger size are considered to have the better ability for repaying the debt and the greater access to sell the product and earn more. So size is positive to debt ratio.
4. Data and Descriptive Statistics
In this section, I will introduce the process of collecting the data and data processing in excel and stata, and the show my empirical findings from different regression.
4.1 Data collection and Process
In this paper, I need three database - CRSP, Compustat and Thomson One for my analysis. First, I enter into the Center for Research in Security Prices (CRSP) and find the CRSP/Compustat Merged data for the Annual fundamentals. I search the database by the firms’ cusip with the consolidated level (parent and subsidiary accounts combined), USD currency and active status, excluding the financial services. I set the time range from 1992 to 2006, which is one year prior for my sample of M&A, and I let the entire fiscal period must be within link date range to make sure the new merged database meet my need to save time of merge CRSP and Compustat manually. I include CUSIP and Standard Industry Classification Code (SIC) as identifying information, make fiscal data year to be the company descriptor, and select total asset, book value per share, total common equity(market value of equity), cost of goods sold, EBITDA, total liability, net income, total net value of property, plant and equipment, net sales, research and development expense, annual close price represent the financial items. According to the data that I generate new variables of outstanding shares, market-to-book ratio, size, profitability, collateral ability, R&D ratio, cost of goods ratio, each year’s leverage by using the existing data. Also, I include R&D dummy variable and Fama French 48-industry dummy in the file. For the R&D dummy, I let it to be zero if the firm has zero or missing R&D expense because many firms has the missing variable for it every fiscal year and the reason is they invest little money on research and development. For the Fama French 48-industry dummy, I
12 14
use the file that is presented in Appendix A to match different ranges of SIC to the 48 fama french industry numbers in order to control the industry factor. Following the previous study (eg. Fama and French,2002; Kayhan and Timan,2007) for the target leverage prediction, I exclude the SIC from 4900 to 4999 and from 6000 to 6999 to make sure financial firms and regulated utilities not be contained in the dataset. Also, I drop the observation with net sales below 10 million (Uysal,2011) and the data with total asset value under 10 million (Kayhan and Timan,2007). Following Uysal (2011), I winsorize all the financial variables to keep the regression result from the inconsistency by the outliers. Therefore, I gain 26,341 observation. Therefore, I obtain 25,939 observation in CRSP/Compustat dataset.
Then, I enter into Thomson One database for the M&A transaction of my main samples. In the advanced analysis from M&A search, I restrict the acquiror nation to be US., the announcement date from 1993 to 2006, which is also the samples’ time horizon. I exclude the undisclosed value transaction, and make sure that all the transaction are completed without the option acquiring payment. Thereby, I get the M&A transaction data for 52,681 in the 14 years. However, the only identifying number that Thomson One provide is 6-digit CUSIP, which does not match the 9-digit CUSIP that I gain from the CRSP/Compustat merged database.
Next, I enter in to WRDS for the CUSIP database to download all the first 6-digit issuer number, 2-digit issue number and the last 1-digit CUSIP check digit. I use the stata to merge the CUSIP data with the Thomson One database by the first 6-digit issuer number and export it to the excel file by using the CONCATENATE formula to combine all the 9 digits together and transfer the CUSIP into the 9-digit number with the same format of the CUSIP in CRSP/Compustat merged database.
Last, I merge the final two database into one by stata using the CUSIP number and get 12,469 data at last. Since what I need is panel data and I need to get the annually regression of the target leverage for all the firms, I use command to make it balanced panel data. Although CUSIP is the code that combines number and alphabet, I still get the matched gvkey code for each aquirors. Then I use “tsset” to test all the firms and find that my panel data is unbalanced. So I download “xtbalance” command
in stata, make my time horizon to be from 1992 to 2006 and drop the observations with missing variables in the range and get the balanced panel data finally. At last, all my observations in 15 years are 9,488 and the deal transaction observations are 579.
Table 1
Summary statistics on acquirer and deal characteristics
Acquirors’ firm characteristics, acquriors’ dummy variables description, and deal characteristics.The paper contains 9488 sample from 1991 to 2006. In the first prediction Regression, I create 2 dummy variables for the R&D expense and different industry.Sales, EBITDA, PP&E, R&D, COGS, book asset, Transaction value and Target’s total asset are all in millions. Transaction value is in millions.
Panel A Acquirers’ firm Characteristics
Variable Obs Mean Std. Dev. Min Max
Market leverage 9488 0.380 0.251 -2.969 7.727 Market-to-book ratio 9488 1.916 2.243 -125.2 41.07 In(Sales) 9488 6.583 2.096 -4.075 12.75 EBITDA/Book asset 9488 0.131 0.126 -2.671 0.643 PP&E/Book asset 9488 0.305 0.236 0 0.954 R&D/Book asset 9488 0.029 0.074 0 2.276 COGS/Sales 9488 0.685 1.776 0 121.0 Stock Return 9488 0.071 0.913 -1 46.73
Panel B Acquirers’ dummy variables Description
Dummy R&D 0 1 Total
Frequency 5,064 4,424 9488
Percentage 53.37% 46.63 100%
Dummy industry-fama 1-12 13-24 25-36 37-48 Total
Frequency 1296 2336 3232 2624 9488
Percentage 13.66% 24.62% 34.06% 27.66% 100%
5. Empirical Results
5.1 The estimation of target leverage
First of all, I use both Tobit regression and panel data with fixed effects method to estimate each coefficient of control variables from the prior-year data. The result is showed on Table 2. The upper limit for the tobit regression is 1.
Panel C Deal Characteristics
Variable Obs Mean Std. Dev. Min Max
Cash fraction 579 .2298742 0.37515 0 1 Transaction value 579 530.5333 3657.45 0.007 72671 Acquisition premium 148 .9192453 1.96966 -1.3641 20.355
12 16
Considering the tobit regression, all the correlation effect is the same with the studies and my expectation. The upper limit that I give to this tobit regression is 1 because market leverage is defined to be total liability divided by sum of total liability and market value of equity. And right-censored observations that is over 1 is 13. It means that 13 samples have the leverage larger than 1, then tobit regression winsorize the outliers to 1. The tables shows that market value divided by book value is negative to firm’s next year’s market leverage, which make sense because market to book ratio represents firm’s growth opportunity. Size is positive related to dependent variable as well as collateral ability (PP&E/Sales). But profitability (EBITDA/Book assets), R&D ratio, cost of goods ratio and stock return is positive with the next year’s market leverage. Prior year’s market leverage is highly related to the next year’s market leverage, the coefficient is 0.697. Also, I let R&D ratio to be dummy variable, when the firm records no R&D expense or zero expense, the dummy equals zero, otherwise it equals one. And I include Fama industry 48 dummy variable. The industry dummy is based on the SIC (Standard Industry Classification) code and classify the firms into 48 area industry.
Panel data with fixed effects shows the similar result of each control variable. Prior year’s market leverage has the highest correlation with the current year’s market leverage. And other control variables play the same direction role on the current year’s market leverage. Only cost of goods ratio is not significant in the both two regression. Then I exclude cost of goods ratio to be the determinant of prediction calculation. Other independent variables are significant and the R squared is 0.169 in panel data regression.
The difference of result between Tobit regression and panel data regression is that Tobit winsorize the right-censored observations to the upper limit 1. However, some firms have the negative market value of equity in real life. So the regression should show the real data of the samples.
Thereby, I use the beta got from the panel regression and calculate the target leverage for every firm, then gain the deviation of the market leverage from the target one. The deviation of leverage is defined to be market leverage minus target leverage
and scaled by the market leverage. Since I make the right-hand side variables to be the prior-year variable and the market leverage to be the current data, the prediction could avoid revision causal relationship between the independent variables and dependent variable.
5.2 The calculation of leverage deficit
After the prediction of the target leverage, I calculate it by using every significant coefficient times the control variables of the prior year and get the target leverage of the current year for each firm. Then I minus the target leverage from the market leverage of the same year. The number that I get is the deviation from the target leverage, therefore each firm has 14-year leverage deviation. The result can be classified to three parts: over zero, zero, or under zero. The firms are considered to be over leveraged if the deficit is over zero and under leveraged when the deviation from the leverage is under zero.
Table 2
Target leverage prediction analysis
In table 2, the result of Tobit regression and panel regression is showed below. Size means the natural logarithm of firm’s annual sales, EBITDA/book assets represents profitability, property, plant and equipment/book assets means the collateral ability of firms. Cost of goods ratio is defined to be cost of goods divided by sales and the denominator of R&D ratio is also sales. The number of observations is 9488 and number of firms (by gvkey) is 593. In Tobit regression, 13 samples are censored above the upper limit, and be winsorized to 1. The Pseudo R-Squared of Tobit model is 44.169, and R-squared in panel data regression is 0.169. ***,** and * stand for statistical significance at the 1%, 5%, and 10% level, respectively.
Post leverage Tobit Panel
Market to book ratio -0.004*** -0.003*** (0.001) (0.001) Size 0.003*** 0.033*** (0.001) (0.003) Profitability -0.145*** -0.131*** (0.017) (0.025) Collateral ability 0.0102 0.117*** (0.007) (0.025) R&D ratio -0.203*** -0.179***
12 18
(0.029) (0.043) Cost of goods ratio -0.001 -0.002
(0.001) (0.001) Market leverage 0.697*** 0.448*** (0.008) (0.012) Stock return -0.009*** -0.006*** (0.002) (0.002) R&D dummy -0.018*** -0.024* (0.004) (0.014) Fama 48-Industry dummy Yes Yes
Constant 0.109*** 0.403*** (0.007) (0.0244) Observations 9,488 9,488 Number of gvkey 593 593 R-Squared 0.169 Pseudo R-Squared 44.169 Left-censored observations 0 Uncensored observations 9,475 Right-censored observations 13
5.3 How does leverage deficit affect the cash fraction used in payment?
The second step of the paper is to investigate the relationship between deviation from leverage and the characteristics before acquisition transaction. The first research question is to test how leverage deficit affect the cash component used in the payment for the bidders. We can see the result in table 3. The leverage deviation is negative to percentage of cash in the payment, the beta is -0.612 and it is significant. The relative size represents bidders’ market value over target firms’ market value, and it is positive to cash fraction because the relative size means bidders’ capacity to acquire the target firms. The higher relative size, the bigger capacity of money for the bidders to use after the acquisition, therefore the acquirers will consider less about the cash they used in this acquisition transaction. Market to book ratio and stock return of bidders represent the growth opportunity and further prospect, and the coefficient is -0.007 and -0.0453 respectively. The larger growth opportunity will make firm use less debt. As mentioned before, most cash that firms used in the acquisition transaction is from Table 3
Table 3 shows that how leverage deviation and other control variables affect the acquisition financial choice in terms of the cash component used for the transaction payment. The main relationship that I want to know is between leverage deviation and percentage of cash, and I can get the significant coefficient from table 3. The table also shows that the more leverage deficit will lead to the lower cash that used in the transaction for bidders. Among other control variables, relative size, market leverage, sales/book assets have the positive relation with the dependent variable. Here, since sales is too large to estimate the coefficient, I use sales/book assets to represent the size of the acquirers. Also, market to book ratio, stock return, profitability and CPI adjusted index have the negative relation with cash fraction of payment. Only CPI adjusted index is insignificant at 10% level. The regression includes 579 samples, the R squared is 0.035. I cluster the firm by gvkey code and 562 observations are included. ***,** and * stand for statistical significance at the 1%, 5%, and 10% level, respectively.
issuing debt. Therefore, in this condition, firms are willing to use less cash to pay for the target firms. Sales/book assets is positive to cash fraction, it could be consider as size of bidders. Then the firm with larger size has capacity to raise more debt so the percentage of cash in payment is higher as well. Profitability is the same with the first regression, EBITDA/Book assets, and it is negative to cash component because firms are less willing to issue debt with high profitability and they can earn more retained
Variables Cash fraction Leverage deviation -0.612*** (0.516) Relative size 0.002*** (4.003) Market leverage 0.242* (0.142) Market to book ratio -0.007*
(0.006) Sales/Book assets 0.003*** (0.001) Stock return -0.045** (0.0216) Profitability -0.339*** (0.126) CPI adjusted -0.011 (0.009) Constant 0.109** (0.055) Observations 579 R-squared 0.035 Cluster (gvkey) 562
12 20
earnings by their products. CPI adjusted variable is the natural logarithm of CPI index divided by the market value of acquirers. Since the year of transaction announced is different, this variable is used to adjust the macro difference between years by using CPI index. It is negative with the cash component in deal and it make sense because the cash fraction should be positive to market value and market value is in the denominator of the variable. I cluster the regression by each firm and the R-squared is 0.035. In this case, I can get the economical significant result that one cent per dollar increase in leverage deviation lead to 0.6 percentage cash payment reduce in the whole payment, which is consistent with Harford (2009) and Uysal (2011). The main result support the hypothesis of question one that firms are willing pay less cash when their market leverage is higher than the target one in the acquisition transaction.
5.4 How does leverage deficit affect the premium paid by bidders?
Similar to the first research question, this question is to test how deviation from target leverage affect the premium paid for the target firms. Standing by another side to test whether leverage deviation affect the financial decision by bidders, I define premium to be transaction value divided by target firms’ total market value. Uysal (2011) defines premium for target firms as the aggregate value of cash and stock for the target firms divided by the market value of target 40 days prior to the acquisition. Following the definition of the previous study and the data that I already collected, I make premium to be the total transaction value that bidders paid divided by the market value of target firms 4 weeks prior to the acquisition, which is similar to the definition that Officer (2003) used. I calculate the denominator by adding total liability to market value of equity 4 week prior to the acquisition and the market value.
Since the lack of target firms’ deal information that Thomson One provided restricts my sample size, I collect 148 data to explore the second research question in order to keep all the variables feasible. I regress the data by ordinary least squares method because the data are neither time series nor cross section and all the data is random in the time horizon. Following Uysal, I make bidders’ market to book ratio,
stock return, target firms’ market to book ratio, target firms’ EBITDA/TA, and target firms’ stock return to be the part of control variables. Previous study use sales to be the independent variable, but I use the natural logarithm of sales to stand for the size of bidders in order to get a more appropriate coefficient in the regression.
Also, I include relative transaction and relative size to explore the relationship between bidders and target firms. Relative transaction is calculated by dividing bidders’ total asset by the value of transaction, and I define acquirers’ total book assets divided by targets’ total book asset to be relative size. The reason to include these two variables is that buyers usually consider the size of targets compared to their own size and this may affect the premium that they paid for. If the target’s size is large enough to affect the capital turnover for bidders when they are going to deal the acquisition transaction, the relative size between two firms may affect the premium for targets. Also, when the transaction value occupies part of the total value of buyers and the payment is relatively large enough, bidders may consider the premium for target because of it. These two variables are raised by myself and I list them into the model since it makes sense theoretically. I assume that both of the variables are negative with the dependent variable. The result in table 4 shows that relative transaction has the negative coefficient that is significant at 5% significance level but the relative size holds the positive relationship with acquisition premium with the insignificance. Hence, the relative size should not be considered because of the insignificance and objection of the assumption. And the fraction of transaction value accounted in bidders’ total assets affect the acquisition premium during the transaction.
Among the three variables related to target firms, they are supposed to have negative coefficient as Uysal (2011) mentioned. Market to book ratio stands for the growth opportunity, EBITDA/total assets represents the profitability of a firm, and stock return means the overvaluation of the equity. The three characteristics shows the potential value of the target firms in the future, and the results explains that target firms with higher market to book ratio and higher profitability receives lower premium from buyers, which is consistent with Bargeron, Schlingemann, Stulz, and Zutter (2008). However, the target firms’ stock return lead to a higher acquisition
12 22
premium from table 4, and this result is opposite to the result in Uysal (2011). Among all the independent variables, buyers’ characteristics are first-lagged data and the variables related to target firms are the current data because of the limitation of data source. So the difference relation between target firms’ stock return and acquisition premium may caused by limitation of data and this could be the further question that I need to explore. Overall, the main result of the premium regression is consistent with the previous studies and show that higher leverage deviation from bidders’ target leverage will lead to the lower premium.
Table 4
Acquisition premium regression analysis
Table 4 displays how leverage deviation affect the premium that bidders paid for the target firms. The leverage deviation is negative with transaction premium and it is significant at 1% level with 1.692 standard error. Also, market to book ratio, size, target firms’ market to book ratio, target firms’ stock return and the constant term are significant at 5% level. Among other control variables, stock return, market to book ratio, relative transaction, target firms’ market to book ratio, target firms’ EBITDA/total assets have the negative correlation with the acquisition premium. However, bidders’ market leverage, size, relative size, and target stock return have the positive coefficient with the dependent variable. The sample includes 148 observations and R-squared of the regression is 0.192. In the table, ***,** and * stand for statistical significance at the 1%, 5%, and 10% level, respectively.
Variables Premium Leverage deviation -5.332** (1.692) Market leverage 3.082** (1.319) Stock return -2.863** (2.015) Market to book ratio -4.113***
(1.295) Size 0.264*** (0.034) Relative transaction -0.005** (0.003) Relative size 0.183 (0.190) Target market to book ratio -6.226***
(1.091) Target EBITDA/Total asset -4.109*
(2.445) Target stock return 3.561***
(1.118) Constant 1.334*** (0.221) Observations 148 R-squared 0.192 6. Robustness check
In target leverage prediction, I let the future year’s operating income instead of the past year’s operating income to be divided by the total assets to check the robustness of my prediction regression, following Harford (2009). The coefficient is significant (-0.23) at 1% level and the future year’s profitability is also negative with the current year’s market leverage, which is robust to my regression result.
In percentage of cash in payment regression analysis, I add year dummies to the model as same function of CPI adjusted variable. The main variable (deviation from target market leverage) maintains the same (negative) relationship with cash fraction in acquisition payment but leverage deficit is significant (-1.205) at 10% level. The reason for the reduction of significance could be that limitation of sample size. Although the efficient time horizon is 14 years (from 1993 to 2006), the whole sample size is 579. Thus every year’s data is limited for the robustness check.
In acquisition premium analysis, I use the natural logarithm of sales to stand for size of firms. Then I change it to be sales/total assets and find the similar significant negative (-0.581 at 1% significant level) relationship between leverage deviation and acquisition premium in the new model. Also, sales/total assets is positive (0.527 at 5% significance level) with acquisition premium.
7. Conclusion
In this paper, I want to test whether firms have target capital leverage or not in acquisition aspect. I focus on the characteristics before the deal to explore the answer of my research question. First of all, I use panel data with fixed effects to predict the target leverage. The independent variable is the prior year’s characteristics of a firm
12 24
and the dependent variable is the current year’s market leverage of each firm. After the regression, I get significant coefficient of most control variables (exclude cost of goods ratio) and let each beta times the current year’s control variables to calculate the current year’s target leverage. Then, I minus target leverage from the market leverage of the same year to obtain the deviation from the target leverage (leverage deficit) of each firm. Next, I test how leverage deficit affect the cash component used in payment for the target firms by OLS. The result shows that the higher leverage deficit lead the bidders to use less cash component in payment for the target firms. In addition to strengthen the regression, I focus on acquisition premium to explore the relation between leverage deficit and premium by OLS. The result shows that firms with higher leverage deviation pay less transaction premium for target firms. The two main conclusion is consistent with the previous studies (see in Officer,2003, Harford, 2009, Uysal, 2011). To avoid the endogeneity of the regression, I use all the prior-year independent variables and current-year dependent variable in the first research questions and most first-lagged independent variables in the second research question2.
Since firms with higher deviation from the target leverage are less likely to use debt before the transaction and firms are willing to pay less premium for target firms when they are already overleveraged during acquisition transaction, I can conclude that firms have their target capital structure. The proposition that deviation from target capital structure affects the further financial decision is also supported by the result of two research questions. The negative relationship between leverage deficit and cash fraction in payment (amount of acquisition transaction) shows that firms try not to deviate far away from their target capital structure.
2 The reason for not using lagged variables in deal characteristics is that Thomson One only provide with the
current data for deal and target firms. I try to find target firms’ characteristics prior to the acquisition year in WRDS database by changing targets’ 6-digit cusip into 9-digit cusip. But the correctly matched sample of target firms is less than the data provided by Thomson One. In case of the insignificance of the result caused by sample limitation, I choose to the original current data from Thomson One.
Reference
1. Anu Bharadwaj, Anil Shivdasani, 2003, Valuation effects of bank financing in acquisitions, J. Journal of Financial Economics, 67 (2003) 113–148
2. Armen Hovakimian, Tim Opler and Sheridan Titman, 2001, The Debt-Equity Choice, J. Journal of Financial and Quantitative Analysis, Vol.36, No.1, March 2001 3. Ayla Kayhan, Sheridan Titman, 2007, Firms’ histories and their capital structures, J. Journal of Financial Economics, 83 (2007) 1–32
4. Ben Amor Atiyet, Higher Institute of Management of Gabès, The Pecking Order Theory and the Static Trade Off Theory:Comparison of the Alternative Explanatory Power in French Firms, J. Journal of Business Studies Quarterly, 2012, Vol. 4, No. 1, pp. 1-14
5. Edwin O. Fischer, Robert Heinkel, Josef Zechner, 1989, Dynamic Capital Structure Choice: Theory and Tests, J. The Journal of Finance, Vol.44, No.1 (Mar., 1989), pp. 19-40
6. Eugene F. Fama, Kenneth R. French, 2002, Testing Trade-Off and Pecking Order Predictions About Dividends and Debt, J. The Review of Financial Studies, Vol. 15, No. 1 (Spring, 2002), pp. 1-33
7. Jarrad Harford, Sandy Klasa, Nathan Walcott, 2009, Do firms have leverage targets? Evidence from acquisitions. J. Journal of Financial Economics. 93(2009)1-14
8. John R. Graham, Campbell R. Harvey, 2001,The theory and practice of corporate finance:evidence from the field, J. Journal of Financial Economics, 60(2001) 187-243 9. Lakshmi Shyam-Sunder, Stewart C. Myers, 1999, Testing static tradeoff against pecking order models of capital structure, J. Journal of Financial Economics, 51 (1999) 219-244
10. Leonce L. Bargeron, Frederik P. Schlingemann, Rene M. Stulz, Chad J. Zutter, 2008, Why do private acquirers pay so little compared to public acquirers, J. Journal of Financial Economics, 89 (2008) 375–390
11. Mark J. Flannery, Kasturi P. Rangan, 2006, Partial adjustment toward target capital structures, J. Journal of Financial Economics, 79 (2006) 469–506
12 26
capital structure, J. Journal of Financial Economics, 67 (2003) 217–248
13. Murray Z. Frank and Vidhan K. Goyal, 2009, Capital Structure Decisions: Which Factors Are Reliably Important. J. Financial Management. Spring 2009. pages 1 - 37 14. Raghuram G. Rajan, Luigi Zingales, 1995, What do we know about capital structure? Some Evidence from International Data. J. The Journal of Finance. Vol. 50, No. 5. (Dec., 1995), pp. 1421-1460
15. Sheridan Titman, Roberto Wessels, 1988, The determinants of capital structure choice, J, Journal of Finance, Vol. XL, NO.1, March 1988
16. Stewart C. Myers, 1977, Determinants of Corporate Borrowing, J. Journal of Financial Economics, 5 (1977) 147-175.
17. Stewart C. Myers, Nicholas S. Majluf, 1984, Corporate Financing and Investment Decisions when firms have information that investors do not have, J. Journal of Financial Economics, 13 (1984) 187-221. North-Holland
18. Vahap B. Uysal, 2011, Deviation from the target capital structure and acquisition choices. J. Journal of Financial Economics. 102. 602–620
