Do firms have leverage targets? New evidence from mergers and acquisitions in China

North American Journal of Economics and Finance 40 (2017) 41–54

Contents lists available at ScienceDirect

North American Journal of Economics and Finance j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / e c o fi n

Do firms have leverage targets? New evidence from mergers and acquisitions in China Qizhi Tao a, Wenjia Sun a, Yingjun Zhu b,⇑, Ting Zhang c a

School of Finance, Southwestern University of Finance and Economics, Chengdu 611130, China Shanghai Lixin University of Accounting and Finance, Shanghai 201620, China c School of Business Administration, University of Dayton, Dayton, OH 45469, USA b

a r t i c l e

i n f o

Article history: Received 10 November 2016 Received in revised form 3 January 2017 Accepted 17 January 2017 Available online 28 January 2017 JEL classification: G32 G34 Keywords: M&A Capital structure Trade-off theory Adjustment speed

a b s t r a c t We provide new and consistent evidence supporting the trade-off theory using China’s mergers and acquisitions (M&A) deals between 2000 and 2015 as a sample. We show that acquirers do have leverage targets and they adjust their leverage ratios toward an optimal level at which the cost and benefit of the debt are equal. In examining the leverage adjustment speed during the post-acquisition period, we find that acquirers partially adjust their leverage ratios to the optimal levels; and the adjustment speed is affected by the adjustment cost proxied by the bankruptcy risk. Finally, we are able to successfully replicate the US evidence which is consistent with the trade-off theory as well using our improved methodology. Ó 2017 Elsevier Inc. All rights reserved.

1. Introduction Do firms have leverage targets? A large number of studies have provided mixed answers to this important question regarding a firm’s capital structure. Theoretically, the trade-off theory (Bradley, Jarrell, & Kim, 1984; Myers, 1984; Fischer, Heinkel, & Zechner, 1989) hypothesizes that firms have target leverage ratios and the optimal capital structure level is obtained when firms trade off tax benefits of debt financing against costs of financial distress. Some empirical studies (i.e., Titman & Wessels, 1988; Rajan & Zingales, 1995; Graham, 1996; Hovakimian, Hovakimian, & Tehranian, 2004) find that certain firm characteristics, such as size, growth opportunities, liquidation value of assets, and marginal tax rates are important determinants of leverage ratios. These findings generally support the hypothesis that firms strive to maintain a capital structure target. In contrast, another strand of studies report a negative relation between firm current leverage ratios and past profitability (i.e., Fischer et al., 1989; Shyam-Sunder & Myers, 1999; Strebulaev, 2007), evidence that does not support the trade-off theory. Furthermore, several studies (i.e., Baker & Wurgler, 2002; Welch, 2004) show that changes of leverage ratios are mostly the result of stock performance or firms’ attempt to time the stock market.1 ⇑ Corresponding author at: 2800 Wenxiang Road, Shanghai 201620, China. E-mail address: [email protected] (Y. Zhu). Other theories to explain firm capital structure include control hypothesis (Jensen, 1986), pecking order hypothesis (Myers, 1984) and market timing theory (Baker & Wurgler, 2002). 1

http://dx.doi.org/10.1016/j.najef.2017.01.004 1062-9408/Ó 2017 Elsevier Inc. All rights reserved.

42

Q. Tao et al. / North American Journal of Economics and Finance 40 (2017) 41–54

As the previous evidence is concentrated on the US market, we turn to another important market – the Chinese market to seek for new insight on this important issue. China has achieved remarkable economic growth since the late 1970s, and it has become the second largest economy in the world. Following Harford, Klasa, and Walcott (2009), we use an important type of corporate event in the Chinese market – mergers and acquisitions (M&A) as a unique setting to investigate whether firms have leverage targets. First, M&A activities in China have grown rapidly driven by the country’s economic transformation and development since 2000. The number of domestic deals increases 14-fold from 142 in 2000 to 2226 in 2015 and the total value of domestic deals increases 193 fold from CNY6.98 billion (USD843 million) in 2000 to CNY1358.3 billion (USD210.6 billion) in 2015.2 Examining firm capital structure decisions around M&A events can help us obtain important insights. Second, M&A normally involves great changes of a firm’s (acquirer) capital structure, which enables us to assess the role of leverage targets in financing decisions (Harford et al., 2009). Third, the post-acquisition capital structure changes create an ideal setting to analyze whether acquirers adjust their leverage to a certain optimal level, and if they do, the adjustment speed. Fourth, previous studies on China’s market have documented several interesting findings that seem to be inconsistent with the traditional capital structure theories. For example, Liu, Wei, Zhan, and Tai (2004) test the pecking order by the sample of Chinese listed firms in Shanghai and Shenzhen Stock Exchanges and report that Chinese firms first choose equity when raising capital, not debt, which is not consistent with the pecking order theory. Therefore, studying the changes of leverage ratios around M&A events in China can shed new light on firms’ capital structure decisions. According to the trade-off theory, firms have leverage targets and they will actively adjust the actual leverage ratios to trend the optimal level. Although there is no direct evidence showing that the main motivation of M&A is to adjust the leverage ratios, previous studies have found that leverage ratio is an important factor when firms make M&A. The null hypothesis of this study is that the acquiring firms take the optimal capital structure into consideration when making M&A decisions. The alternative hypothesis is that firms have no optimal capital structure. If the null hypothesis is true, we expect to observe that when acquiring firms’ capital structures deviate from the optimal level before the deals, acquiring firms will use M&A to make an adjustment. When the adjustment is under- or overly corrected subsequent to the M&A, acquirers will continue to change their capital structure during the post-acquisitions until the leverage ratios converge to the optimal level in the longrun. If this hypothesis holds, the results support the trade-off theory. Therefore, the acquiring firms’ leverage ratio deviations i.e. the extent to which the actual leverage ratios is different from the optimal leverage ratios, become a key measure to test the trade-off theory in this study. We measure a firm’s leverage ratio deviation as the difference between the predicted leverage and its actual leverage, whereas the predicted leverage is estimated using a tobit regression model as in Kayhan and Titman (2007) and Harford et al. (2009). However, an important improvement of our estimation method is that we separate the ‘‘estimation window” from the ‘‘event window” when conducting the cross-sectional regression model to predict firm optimal level of leverage. M&A is considered an important corporate event, thus the features of capital structures after M&A are expected to be quite different from those before M&A. As discussed by MacKinlay (1997), the event period itself should not be included in the estimation period to prevent the event from contaminating ‘‘the parameter of the normal return model.”3 Previous studies (i.e.Baker & Wurgler, 2002; Harford et al., 2009; Morellec & Zhdanov, 2008) estimate the predicated leverage ratios using insample models, which could cause an issue by mixing the ‘‘estimation window” and the ‘‘event window”. In contrast, we use an out-of-sample regression to estimate the optimal leverage ratios. We consider the improvement of the estimation method one of our important contributions to the literature. Such an improvement in the methodology makes our empirical results strong and consistent with the trade-off theory, as we will further discuss below. We obtain 257 M&A deals from 2000 to 2015 in the Chinese market as our main sample to investigate the effect of M&A on firm capital structure decisions. When constructing our sample, we exclude firms engaging in two or more successive acquisitions in the estimate window for a couple of reasons. First, for these ‘‘repeat” acquirers, the effect of an acquisition on their capital structure could be offset by another one. This becomes particularly an important issue when different payment methods are used. Second, as the purpose of our study is to examine the effect of M&A on firm capital structure, not to test how M&A affects firm growth or financial performance, the exclusion of ‘‘repeat” acquirers should not bias our sample. In contrast, previous studies (i.e., Ghosh & Jain, 2000; Morellec & Zhdanov, 2008) do not adopt this filter when constructing the sample, which might cause issues in their results. We first conduct a cross-sectional regression on the determinants of firm leverage ratios using [
4,
1] (half year for each period, with a total of 2 years) as the estimation window where leverage ratios are defined as total liabilities over total assets, both measured at book value. The regression results are then applied to the event window [0,+4] (half year for each period) to estimate the predicted optimal leverage ratios. The leverage ratio deviation is defined as a firm’s actual leverage ratio minus its predicted leverage level. Following Harford et al. (2009), we estimate the M&A-induced leverage change as the leverage deviation change from period
1 to 0 around the acquisition event. Based on whether a firm’s M&A-induced

2 Counted in number, the top 3 M&A industries in 2015 are information technology, Internet, and finance; while counted in deal value, the top 3 M&A industries become Internet, real estate, and finance. China Securities Regulatory Commission (CSRC), equivalent to the U.S. Securities and Exchange Commission (SEC) has issued several policies and administration procedures on the M&A activities of listed companies in Shanghai Stock Exchange and Shenzhen Stock Exchange. While some of the large M&As will have to undergo a long and in-depth filing and approval process, the CSRC has been making efforts to simplify the process. 3 ‘‘Including the event window in the estimation of the normal model parameters could lead to the event returns having a large influence on the normal return measure” (MacKinlay, 1997, p20).

Q. Tao et al. / North American Journal of Economics and Finance 40 (2017) 41–54

43

leverage change increases or decreases between period
1 and 0, we classify the sample into two groups: the INCREASE and DECREASE groups. The average M&A-induced leverage changes are 0.054 (from
0.016 in period
1 to 0.038 in period 0) for the INCREASE Group and
0.049 (from 0.035 in period
1 to
0.014 in period 0) for the DECREASE Group, respectively. The result implies that acquirers’ capital structures deviate from the optimal level before the deals, and acquirers make use of M&A to minimize the deviations. Furthermore, in examining the changes of leverage ratios subsequent to the M&A, we report two pieces of important evidence supporting the trade-off hypothesis. First, for the INCREASE Group, we observe a continuous and steady decreasing trend of leverage ratio deviations from the year 0 to year +5. The median leverage ratio deviation becomes close to 0% in year +5, indicating a convergence of observed leverage ratios and predicted optimal level. Second, for the DECREASE Group, the median leverage ratio deviation gradually increases from
0.017 in year 0 to approximately 0% in year +5, again implying a convergence of observed leverage ratios and predicted level. These findings are consistent with the trade-off theory, suggesting that firms adjust their leverage ratios gradually after the M&A until interest tax shield is equal to the costs of financial distress, and at this time firms reach an optimal level of their capital structure. We further enrich our analysis of firm capital structure decisions by examining how the acquirers adjust their leverage ratios in the years subsequent to the acquisition. Following Flannery and Rangan (2006), we use the standard partial adjustment model to quantify the adjustment speed at which firms move their leverage ratios toward the optimal levels. Again, we conduct this analysis after classifying our samples into the INCREASE and DECREASE groups. The results show that for the INCREASE Group, the average adjustment speed is approximately 12.9%
13.5% for the period [+1,+4]. For the DECREASE Group, the average adjustment speed is 15.9%
8.1% for the period [+1,+4]. The magnitude of our estimations is consistent with an adjustment speed reported by Flannery and Rangan (2006) for the US firms. The relatively low adjustment speed further verifies the partial adjustment model (Flannery & Rangan, 2006). According to the trade-off theory, it is the adjustment costs that largely prevent firms from converging their leverage ratios to the optimal levels immediately following acquisitions. As in Leary and Roberts (2005) and Harford et al. (2009), we use firm bankruptcy risk as a measure for adjustment costs, and find that bankruptcy risk is significantly and negatively associated with the post-acquisition changes in the leverage deviation. This result indicates that acquirers already facing higher adjustment costs are more likely to slow in reducing their leverage after the acquisition, consistent with the trade-off theory that firms do have leverage targets which are determined by the trade-off between the benefits and the costs of debt. So far we have provided fresh and strong evidence consistent with the trade-off theory based on the findings from the effect of M&A on firm capital structures in the Chinese market. Our study is closely related to and largely in line with two recent studies that also test capital structure theories using M&A events in the US: Harford et al. (2009) and Morellec and Zhdanov (2008). However, both studies report several interesting findings that call for further investigation. For example, after separating the acquiring firms by the payment method, Harford et al. (2009) show that all of the acquiring firms’ debt levels are low relative to the optimal level before the large acquisitions (as evidenced by negative leverage deviations), and they start to increase after the acquisitions. This finding appears to be different from the anecdotal evidence that the M&A may increase or decrease acquiring firms’ leverage ratios. It is also contradictory with a prediction made by the authors that acquiring firms with low (high) liabilities level may choose cash (equity) payment and increase (decrease) the liabilities level. Morellec and Zhdanov (2008) report that the increase of leverage at the M&A announcement is unable to push firms well above their optimal leverage ratios, i.e., acquiring firms are nearly under-levered throughout their life, which appears to be contradictory with the reality that M&A may make some acquiring firms overly-levered. Our sample is based on the M&A deals in China while Harford et al. (2009) and Morellec and Zhdanov (2008) both use M&A events in the US. So the different findings could be caused by the different samples, particularly in the eye of the significant differences in the institutional background and firm characteristics between these two large markets. It could also be due to the refinery of the empirical methodology used in our study. In particular, we have made three important improvements in the methodology used by previous studies. First, we exclude firms with successive deals when constructing our sample. Second, we draw a clear line between the estimation window and the event window when estimating firm optimal level of leverage. Third, we partition all sample into the INCREASE and DECREASE groups based on the M&A-induced leverage deviations, rather than on the acquisition payment method (i.e., cash, equity and mixed). To further explore the differences in the findings reported by our study and Harford et al. (2009) and Morellec and Zhdanov (2008), we construct a US sample including 659 acquiring firms from 1962 to 2001. We then apply our improved methodology to replicate the findings reported by Harford et al. (2009) and Morellec and Zhdanov (2008). We find that the US evidence becomes consistent with the China evidence. In particular, the actual leverage ratios of the INCREASE Group (416 firms) is lower than the optimal level before the announcement year and the deviations start to reverse at the announcement year (from
4.5% to 7.8%). The actual leverage ratios then gradually converge to the optimal level in a long term subsequent to the acquisitions. Consistently, the DECREASE Group’s leverage ratio deviations show an opposite pattern of the INCREASE Group. The US evidence thus strongly supports the trade-off theory as well. This study makes several important contributions to the literature. First, it provides new evidence to support the static trade-off theory using Chinese firms’ M&A events. We show that Chinese firms do have leverage targets and they adjust their leverage ratios toward an optimal level at which the cost and benefit of the debt become equal. Second, in examining the adjustment speed of the leverage ratios during the post-acquisition period, we find that firms partially adjust their leverage ratios to the optimal levels following acquisitions, and the adjustment speed is reversely associated with the adjustment cost proxied by the bankruptcy risk. This finding provides further consistent evidence to support the trade-off theory. Third, different from previous studies (i.e., Harford et al., 2009; Morellec & Zhdanov, 2008), we improve the empirical methodology in

44

Q. Tao et al. / North American Journal of Economics and Finance 40 (2017) 41–54

this study in the following three important aspects: (1) we exclude firms with successive deals when constructing our sample; (2) we differentiate between the estimation window and the event window when estimating firm optimal level of leverage; and (3) we classify all sample firms into the INCREASE and DECREASE groups based on the M&A-induced leverage deviations. More importantly, using our improved methodology, we are able to successfully replicate the US evidence reported by Harford et al. (2009) and Morellec and Zhdanov (2008). For the first time in the literature, we have reported the US evidence that is fully consistent with the trade-off theory. Our improved methodology plays an important role in helping us truly understand the effect of M&A on firms’ capital structure decisions. The rest of the paper is organized as follows. In Section 2, we discuss our data, variables and sample construction. In Section 3, we report empirical results. We conclude in Section 4. 2. Data, variables, and sample 2.1. Data and sample The M&A deals data in the Chinese market are obtained from Thomson One (TO) database. Acquiring firms’ financial data and equity prices are obtained from WIND and CSMAR, respectively. We first search TO database of acquiring firms listed on the Shanghai and Shenzhen stock markets. We remove firms which take on successive acquisitions within the [
4,+4] window.4 Note that in the study conducted by Ghosh and Jain (2000) and Morellec and Zhdanov (2008), the acquiring firms with two or more acquisitions are not excluded. As we have previously discussed, the effect of an acquisition on the capital structure for ‘‘repeat” acquirers could be offset by another one, which is particularly a potential issue when different payment methods (debt vs. equity) are used. We then delete financial firms and regulated utilities firms and firms with missing financial data. Finally, we restrict firms with the market-to-book ratio between the 0.5th and 99.5th percentiles to avoid the influence of outliers. We obtain a sample of 257 acquiring firms (deals). Based on the China Securities Regulatory Commission Industry Classification, 118 (or 45.9%) firms in the sample belong to manufacturing industry, 35 (13.6%) real estate industry, and 19 (7.4%) wholesale and retail trade industry. Agriculture/forestry industry has the least deals (3, or 1.2%). 2.2. Key variables and model An important variable in this study is a firm’s leverage ratio target. We use a tobit regression model to estimate the target following Hovakimian, Opler, and Titman (2001), Hovakimian et al. (2004) and Kayhan and Titman (2007).5 We first regress the actual leverage ratios in period t on a number of lagged variables in period t
1 that have been found by previous studies to be determinants of firm capital structure. The regression model is specified as follows:

Actual Lev erage Ratiot ¼ a þ b1 ðMarket
to
BookÞt
1 þ b2 Asset Tangibilityt
1 þ b3 Profitabilityt
1 þ b4 R&D Expenset
1 þ b5 R&D Dummyt
1 þ b6 Selling Expensest
1 þ b7 Firm Sizet
1 þ

17 X bj It
1 þ et

ð1Þ

j¼8

The dependent variable, a firm’s actual leverage ratio is defined as total liabilities over total assets, both measured at book value. There are arguments on whether book or market leverage ratio is a better measure of capital structure. Myers (1984) argues that since book value of equity refers to assets already in place while a certain part of market value of equity is counted by assets not yet in place (or the assets counted by the present value of future growth opportunities, and the amount of debt supported by growth opportunities will be less than is supported by assets already in place), book leverage ratio is more practical than market leverage ratio. In studies related with M&As, book leverage ratio is considered as a better measure because it is unaffected by the dramatic stock price changes of the acquiring firms around the M&A announcement period (Rajan & Zingales, 1995; Fama & French, 2002).6 The independent variables include Market-to-Book ratio (M/B), Asset Tangibility, Profitability, R&D Expense, R&D Dummy, Selling Expense and Firm Size. The definitions of these variables are provided in Appendix I. We also control for the industry

4

The period is measured by a half year; so [
4, +4] refers to two years before and after M&A event. Three models have been used in the regression method to estimate firm optimal leverage ratio: OLS regression, tobit regression, and Fama-MacBeth regression. A firm’s leverage ratio can be above one if its value of debt is negative, if the value of its equity is positive, or if the absolute value of debt is larger than its value of equity, which is typical of cases of financial distress or large contingent claims (Bruner, 1988). A firm’s leverage ratio can be below zero if its value of debt is negative, if the value of equity is positive, or if the absolute value of debt is smaller than the value of equity, which is untypical of operating firms with abnormal high reservation of cash or unused debt capacity (Bruner, 1988). Some authors believe that it is most common for the leverage ratio to be between 0 and 1. Therefore, the tobit model is adopted in our study to censor leverage outliers, consistent with Hovakimian, Opler, and Titman (2001), Hovakimian et al. (2004) and Kayhan and Titman (2007). 6 The ratio of total liabilities over total assets is a broader definition of leverage ratio, which is a proxy for what is left for shareholders in case of liquidation. As this measure includes non-debt liabilities such as trade credits and pension liabilities, it may exaggerate the real amount of a firm’s leverage and default risk (Rajan & Zingales, 1995). Alternatively, leverage ratio can be measured as the ratio of total debt (short-term debt + long-term debt) over total assets. However, Chinese firms generally face financial constraints to borrow long-term debt as a means to raise capital when making acquisitions. Therefore, we use total liabilities over total assets as a measure for the leverage ratio in this study. 5

45

Q. Tao et al. / North American Journal of Economics and Finance 40 (2017) 41–54

fixed effect to capture the industry-specific characteristics of leverage ratios that are not captured by other explanatory variables. 2.3. Descriptive statistics Table 1 reports the descriptive statistics of variables used in Eq. (1). In Panel A, we classify the sample (257 acquiring firms or deals) into two groups based on whether the actual book leverage ratios increase or decrease between period
1 and period 0 (half year for each period). Group 1 consists of 153 firms with an increase of actual leverage ratios in period 0 from period
1 whereas Group 2 consists of 104 firms with a decrease of actual leverage ratios in period 0 from period
1. There

Table 1 Descriptive statistics. This table describes the statistics of leverage ratios (Panel A) and other variables (Panel B) during a period of [
4, +4] (half year for each period; so [
4, +4] refers to two years before and after M&A event). The announcement semi-annual is set to be 0, the previous semi-annual of the announcement date is set to be
1, and the next semi-annual of the announcement date is set to be +1. Leverage ratio is defined as a firm’s total liabilities over its total assets, both measured as book value at the semi-annual period. M&A deals data in the Chinese market are obtained from Thomson One (TO) database. Acquiring firms’ financial data and equity prices are obtained from WIND and CSMAR, respectively. We first search TO database of acquiring firms listed on the Shanghai and Shenzhen stock markets, and remove firms which take on successive acquisitions within the [
4, +4] window. The sample consists of 257 acquiring firms (deals). We classify the sample (257 acquiring firms or deals) into two groups based on whether the actual book leverage ratios increase or decrease between period
1 and period 0. Group 1 consists of 153 firms with an increase of actual leverage ratios in period 0 from period
1 whereas Group 2 consists of 104 firms with a decrease of actual leverage ratios in period 0 from period
1. The definitions for variables in Panel B are provided in Appendix I. Panel A: Descriptive Statistics of leverage ratios
4


3


2


1

+2

+3

+4

Panel a (Group 1: firms with an increase of actual leverage ratios in period 0 from period
1) N = 153 Min 0.072 0.074 0.056 0.027 0.042 0.046 Max 1.010 0.956 0.946 1.026 1.201 1.240 Med 0.525 0.525 0.518 0.508 0.557 0.570 Mean 0.508 0.503 0.504 0.499 0.554 0.566 Std. Dev 0.176 0.177 0.181 0.184 0.182 0.188

0.043 1.086 0.584 0.568 0.186

0.017 1.160 0.593 0.565 0.189

0.036 1.357 0.582 0.571 0.195

Panel b (Group 2: firms with a decrease of actual leverage ratios in period 0 from period
1) N = 104 Min 0.173 0.192 0.095 0.136 0.119 0.110 Max 1.208 1.655 1.478 1.469 1.347 1.496 Med 0.547 0.559 0.554 0.564 0.505 0.522 Mean 0.537 0.551 0.543 0.571 0.508 0.534 Std. Dev 0.181 0.221 0.236 0.232 0.213 0.200

0.069 0.888 0.507 0.518 0.172

0.067 0.943 0.498 0.496 0.184

0.061 0.919 0.497 0.507 0.177

Variable

Min

Pane B: The descriptive statistics of independent variables Market-to-Book 0.591 Asset Tangibility 0.351 Profitability
0.969 R&D Expense 0.000 R&D Dummy 0.000 Selling Expense 0.000 Firm Size 8.085

0

+1

Max

Med

Mean

Std. Dev

14.812 1.000 0.949 0.289 1.000 0.420 11.209

1.736 0.967 0.022 0.000 1.000 0.036 9.526

2.159 0.943 0.031 0.002 0.766 0.051 9.562

1.414 0.077 0.099 0.012 0.009 0.055 0.539

Fig. 1. Changes of acquiring firms’ median leverage ratios during [
4, +4] period window. This figure plots the changes of acquiring firms’ median leverage ratios before and after the takeovers during [
4, +4] period window (half year for each period). Leverage ratio is defined as a firm’s total liabilities over its total assets, both measured as book value at the semi-annual period. We classify the sample into two groups based on whether the actual book leverage ratios increase or decrease between period
1 and period 0. Group 1 consists of 153 firms with an increase of actual leverage ratios in period 0 from period
1 whereas Group 2 consists of 104 firms with a decrease of actual leverage ratios in period 0 from period
1.

46

Q. Tao et al. / North American Journal of Economics and Finance 40 (2017) 41–54

are more firms in Group 1, consistent with the convention that M&A generally increases a firm’s debt. The median leverage ratio of the Group 1 increases from 0.508 in period
1 to 0.557 in period 0. The median leverage ratio of Group 2 decreases from 0.564 in period
1 to 0.505 in period 0. Fig. 1 plots the changes of acquiring firms’ median leverage ratios before and after the acquisitions during [
4,+4] period window (half year for each period). It is evident that M&A has significantly changed acquirers’ leverage ratios. For Group 1, acquires’ leverage ratio first gradually increases after the event; and then it starts to decline at period +3. Similarly, for Group 2, acquires’ leverage ratio first starts to decrease after the event; but it turns to increase at period +3. This pattern appears to suggest that about two years after the takeovers, firms then start to make adjustment of the leverage ratios. Panel B of Table 1 shows the descriptive statistics of the independent variables used in Eq. (1). The average (median) of Market-to-Book ratio is 2.159 (1.736), with a standard deviation of 1.414. Asset Tangibility has a mean (median) of 0.943 (0.967), with a standard deviation of 0.077. The average (median) Profitability, measured by EBITDA over total assets, is 0.031 (0.022), ranging from
0.969 to 0.949. The average R&D Expense, Selling Expense and Firm Size is 0.002, 0.051, and 9.562, respectively. 3. Empirical results 3.1. Determinants of capital structure In this section, we first examine the determinants of a firm’s capital structure, based on which we estimate the predicted optimal level of the capital structure. Table 2 reports the estimated coefficients for Eq. (1) on the period [
4,
1] (half year for each period) with a sample of 1028 firm-year observations. The results are generally consistent with those reported by Rajan and Zingales (1995), Baker and Wurgler (2002), Flannery and Rangan (2006), Alti (2006), and Kayhan and Titman (2007). The coefficient for Market-to-Book ratio is
0.017 (t =
6.89), indicating a negative and significant relation between a firm’s market-to-book ratio and its leverage ratio. A number of empirical studies have reported a negative relation between leverage ratio and market-to-book ratio. The trade-off theory, the pecking order hypothesis and the market timing theory have provided different interpretations on this negative relation. In particular, the trade-off theory considers market-tobook ratio as a proxy of a firm’s growth/investment opportunities. As high market performance is usually associated with the presence of good growth/investment opportunities (Hovakimian et al., 2001), it is generally believed that growth opportunities play an important role in determining a firm’s financial decision – firms with good growth opportunities are expected to have low debt (Goyal, Lehn, & Racic, 2002). Fama and French (2002) proxy market-to-book ratio as firms’ expected investment opportunities, and they argue that in a complex pecking order world, firms balance current and future financing costs. Thus firms with larger investment opportunities are likely to maintain low-risk debt capacity in order to finance future investments, resulting in a negative relation between book leverage ratio and investment opportunities. According to the market timing theory, there is a negative relation between debt and market-to-book ratio because firms are likely to decrease (increase) their leverage ratio by issuing (repurchasing) equity when their stocks are overvalued (undervalued) (Rajan & Zingales, 1995; Baker & Wurgler, 2002). Asset Tangibility has a positive and significant coefficient (b = 0.256, t = 5.16), consistent with the trade-off theory. Firms with safe, tangible assets and abundant taxable income to shield tend to prefer for debt financing (Brealey & Myers, 2003).

Table 2 Regression analysis of determinants of firm capital structure. This table reports the results of Eq. (1) using a tobit regression model on the period [
4,
1] (half year for each period) with the sample of 1028 firm-year observations. The dependent variable is leverage ratio, defined as a firm’s total liabilities over its total assets, both measured as book value at the semi-annual period. The independent variables include Market-to-Book ratio, Asset Tangibility, Profitability, R&D Expense, R&D Dummy, Selling Expense, and Firm Size. The definitions of these independent variables are provided in Appendix I. Variable Market-to-Book Asset Tangibility Profitability R&D Expenses R&D Dummy Selling Expenses Firm Size Constant Industry Fixed Effect Pseudo R2: F-test N ***

denotes the significance level at 1%, 5%, and 10%, respectively.

Coefficient ***


0.017 0.256***
0.253***
0.034***
0.004
0.004*** 0.057***
0.067*** Control
0.387 284.37*** 1028

t-statistic (
6.89) (5.16) (
7.06) (
2.90) (
0.51) (
2.89) (7.49) (
2.76)

47

Q. Tao et al. / North American Journal of Economics and Finance 40 (2017) 41–54

Rajan and Zingales (1995) find that when a large proportion of a firm’s assets are tangible, these assets are considered as collateral which reduces the risk level from a lender’s perspective. Moreover, tangible assets retain high value in liquidation; and lenders are thus more willing to provide debts to these firms with more tangible assets. Regarding the effect of firm profitability on the capital structure, the trade-off theory expects a positive relation: profitable firms tend to have higher tax shield of debt and lower probability of financial distress, and they are in favor of debt financing. The pecking order theory anticipates a negative relation between leverage ratio and profitability: profitable firms are likely to use a substantial amount of the internal fund to repay the debt rather than to repurchase equity, and they are likely to experience share price increase. We find a negative and significant relation between firm profitability and leverage ratio, consistent with the pecking order (b =
0.253, t =
7.06). R&D Expense and selling expense are usually considered as the indicators of a firm’s uniqueness, and their relation with leverage ratio can be explained by the trade-off theory. Titman and Wessels (1988) show that the more a firm spends on R&D, the more difficult for its competitors to duplicate its innovations and products. Titman (1983), Titman and Wessels (1988), Grinblatt and Titman (2002), and Hovakimian et al. (2004) expect a negative relation between R&D and leverage ratio, as firms with high R&D and selling expense may have little taxable earnings and they are unable to use debt tax shields. DeAngelo and Masulis (1980) and Fama and French (2002) regard R&D expense as non-debt tax shields. They argue that since higher non-debt tax shields imply no taxable income, a lower expected corporate tax rate means a lower expected payoff from interest tax shields. Since the missing data of R&D Expense do not necessarily mean that firms do not have R&D spending, R&D Dummy is used in the regression to distinguish firms that do not report R&D spending from the firms that report very low spending. R&D Dummy is set to 1 if a firm does not report R&D expense, and zero if otherwise. In our analysis, both R&D Expense and Selling Expense have a negative and significant coefficient, consistent with the trade-off theory. Regarding the size effect on a firm’s capital structure, large firms may have high optimal leverage ratios because they have greater access to capital markets (Fischer et al., 1989; Baker and Wurgler, 2002; Hovakimian et al., 2004; Flannery and Rangan, 2006; Kayhan and Titman, 2007). However, Rajan and Zingales (1995) expect the effect of firm size on leverage ratio to be ambiguous. On the one hand, firm size can be considered an inverse proxy for the probability of bankruptcy as large firms tend to be more diversified and are not easy to fail. Therefore, larger firms tend to have higher leverage ratio (Fama and French, 2002). On the other hand, Titman and Wessels (1988) and Grinblatt and Titman (2002) show that small firms pay more commissions than large firms to issue new equity, and they may be more levered than large firms. We report a positive and significant coefficient for Firm Size (b = 0.057, t = 7.49), consistent with the trade-off theory and the pecking order hypothesis. 3.2. Predicted optimal leverage ratio and leverage ratio deviations We now estimate a firm’s optimal leverage ratio based on the regression results reported in the previous section using the regression Eq. (2):

^1 ðM=BÞ ^ ^ ^þb Predicted Lev erage Ratioi;t ¼ a i;t
1 þ b2 Asset Tangibilityi;t
1 þ b3 Profitabilityi;t
1 ^ ^ ^6 Selling Expenses þ b4 R&D Expense þ b5 R&D Dummy þb i;t
1

i;t
1

17 X ^j Ii;t
1 ^7 Firm Sizei;t
1 þ b þb

i;t
1

ð2Þ

j¼8

After predicting the optimal leverage ratio, we estimate a firm’s leverage deviation (DLeverage Ratio) by subtracting the predicted optimal level of leverage ratio from its actual leverage ratio:

DLev erage Ratioi;t ¼ Actual Lev erage Ratioi;t
Predicted Lev erage Ratioi;t

ð3Þ

We use DLeverage Ratio to indicate the change of each acquiring firm’s capital structure during the period [
4,+4] (half year for each period). If the trade-off theory holds, we expect acquiring firms’ capital structures to deviate from the optimal level before the deals. Acquiring firms will then make use of M&A to reduce the deviations, and their leverage ratios will start to converge to the optimal leverage ratios in the long-run period subsequent to the acquisitions, resulting in the zero leverage deviations. The descriptive statistics of leverage ratio deviations are provided in Table 3. In reality, M&A may increase or decrease the debt of acquiring firms. To avoid the leverage deviations between firms that lever up and firms that lever down may cancel out with each other, we divide the sample into two groups based on the M&A-induced leverage change. Following Harford et al. (2009), we estimate the M&A-induced leverage change as the leverage deviation change from period
1 to 0 around the acquisition event. Based on whether a firm’s M&A-induced leverage change increases or decreases between period
1 and 0, we classify the sample into two groups again: the INCREASE and DECREASE groups. The INCREASE Group consists of 148 firms (59.9%) and the DECREASE Group consists of 99 firms (40.1%). The negative (positive) leverage ratio deviation indicates that the actual leverage ratio is low (high) relative to the optimal levels. We note from Table 3 that the average M&A-induced leverage change is 0.054 (from
0.016 in period
1 to 0.038 in period 0) for the INCREASE Group and
0.049 (from 0.035 in period
1 to
0.014 in period 0) for the DECREASE Group, respectively, suggesting that the M&A events have significantly

48

Q. Tao et al. / North American Journal of Economics and Finance 40 (2017) 41–54

Table 3 Leverage ratio deviations during [
4, +4] period. This table describes the statistics of sample firms’ leverage ratio deviations during a window of [
4, +4] period (half year for each period). We estimate a firm’s leverage ratio deviation by subtracting a firm’s predicted optimal level of leverage ratio from the actual leverage ratio (Eq. (3)), whereas a firm’s optimal leverage ratio is estimated based on Eq. (2). We classify the sample into two groups based on whether a firm’s M&A-induced leverage change increases or decreases between period
1 and period 0.
2


1

0

+1

+2

+3

+4

Panel A (INCREASE Group) N = 148 Min
0.600
0.588 Max 0.331 0.319 Med 0.004 0.012 Mean
0.005
0.008 Std. Dev 0.159 0.162


4


0.618 0.319 0.016
0.008 0.167


0.645 0.391 0.002
0.016 0.169


0.632 0.527 0.050 0.038 0.167


0.616 0.627 0.061 0.050 0.178


0.621 0.209 0.051 0.041 0.172


0.612 0.299 0.051 0.034 0.172


0.612 0.585 0.050 0.046 0.188

Panel B (DECREASE Group) N = 99 Min
0.466
0.475 Max 0.249 0.253 Med 0.031 0.022 Mean 0.026 0.028 Std. Dev 0.155 0.155


0.476 0.314 0.029 0.027 0.165


0.529 0.302 0.022 0.035 0.168


0.549 0.179
0.017
0.014 0.165


0.556 0.133
0.002 0.008 0.158


0.624 0.201 0.007 0.000 0.159


0.559 0.133
0.012
0.011 0.163


0.510 0.196
0.009
0.008 0.166

Panel C (Overall Min Max Med Mean Std. Dev


0.618 0.319 0.023 0.006 0.167


0.645 0.391 0.007 0.004 0.170


0.632 0.527 0.022 0.018 0.168


0.616 0.627 0.046 0.034 0.171


0.624 0.209 0.033 0.025 0.168


0.612 0.299 0.023 0.016 0.170


0.612 0.585 0.022 0.025 0.181

Group) N = 247
0.600 0.331 0.020 0.007 0.157


3


0.588 0.319 0.013 0.006 0.160

Fig. 2. Changes of acquiring firms’ median leverage ratio deviations during the 11 year window. This figure plots the changes of acquiring firms’ median leverage ratios before and after the takeovers during [
5, +5] period window (one year for each period). Leverage ratio is defined as a firm’s total liabilities over its total assets, both measured as book value at the semi-annual period. We classify the sample into two groups based on whether a firm’s M&Ainduced leverage change increases or decreases between period
1 and period 0. We estimate a firm’s leverage ratio deviation by subtracting the predicted optimal level of leverage ratio from its actual leverage ratio (Eq. (3)), whereas a firm’s optimal leverage ratio is estimated based on the regression Eq. (2).

changed the debt levels of the acquiring firms. The results in Table 3 also imply that acquirers’ capital structures deviate from the optimal level before the deals, and acquirers make use of M&A to minimize the deviations. According to the trade-off theory, firms will adjust their leverage ratio until the marginal cost of debt is equal to the marginal revenue of the debt and at that time the capital structures are the optimal value. But some random events may cause the capital structures to deviate from the optimal level. The firms will then actively adjust their debt level to remove the deviations. Although there is no evidence that the main motivation of M&A is to adjust the leverage ratio, managers usually consider leverage ratios as an important factor when they conduct M&A deals. We also note from Table 3 that at the end of the period +4 or 2 years after the acquisition event, the average (median) leverage deviation is 0.046 (0.050),
0.008 (
0.009), and 0.025 (0.022) for the INCREASE, DECREASE groups, and the overall sample, respectively. These numbers suggest that the actual leverage ratios are quite close to the optimal level 2 years subsequent to the acquisitions. Table 3 provides preliminary evidence supporting the trade-off theory. That is, acquirers do have leverage targets and they adjust their leverage ratios toward an optimal level subsequent to the acquisitions. Note that the time window for Table 3 is [
4,+4], or two years around the acquisition event. We obtain even stronger evidence when expanding the time window to five years around the acquisition. As shown in Fig. 2, it is evident that for the INCREASE Group, there is a continuous and steady decreasing pattern of leverage ratio deviations from the year 0 to year +5. The median leverage ratio deviation becomes close to 0% in year +5, indicating a convergence of observed leverage ratios and predicted optimal level.

Q. Tao et al. / North American Journal of Economics and Finance 40 (2017) 41–54

49

Consistently, for the DECREASE Group, the median leverage ratio deviation gradually increases to approximately 0% in year +5.7 These findings provide strong evidence supporting the trade-off theory, suggesting that firms adjust their leverage ratios gradually after the M&A until the marginal cost of debt is equal to the marginal value of the debt, and at this time, firms reach an optimal level of their capital structure. 3.3. Replicating US evidence So far we have reported strong and consistent evidence in the Chinese market that supports the trade-off theory. Two recent studies by Morellec and Zhdanov (2008) and Harford et al. (2009) examine the same issue using M&A events in the US market. They both estimate the leverage deviations surrounding the M&A event dates and plot them in figures, which are reproduced in Fig. 3 for comparison.8 Both Panel (a) and (b) show that acquiring firms are under-levered before the M&A announcement. Harford et al. (2009) separate the acquiring firms by the method of payment. As shown in Panel A, all acquiring firms’ leverage ratio is lower than the optimal level before the acquisition, and it starts to increase after the acquisition. This appears to be different from the anecdotal evidence that the M&A may increase or decrease the leverage ratio of acquiring firms. It is also not consistent with a prediction made by the study that acquiring firms with low (high) liabilities level may choose cash (equity) payment and increase (decrease) the liabilities level. Panel B plots the results from Morellec and Zhdanov (2008) and it reveal that the increase of leverage at the announcement is unable to push firms well above their optimal leverage ratios, i.e., acquiring firms are nearly under-levered throughout their life, which appears to be contradictory with the reality that M&A make some acquiring firms over-levered. We conduct further tests to investigate whether the different results between our study and Morellec and Zhdanov (2008) and Harford et al. (2009) are due to different samples (US vs. China M&A events) or the improvement in the empirical methodology used in this study. The improvement can be summarized as the following three areas: (1) we exclude firms with successive deals when constructing our sample; (2) we draw a clear line between the estimation window and the event window when estimating firm optimal level of leverage and leverage deviations; and (3) we classify all samples into the INCREASE and DECREASE groups based on the M&A-induced leverage deviations, rather than on the acquisition payment method (i.e., cash, equity and mixed). We then proceed to replicate the US findings using our refined methodology. We estimate the leverage deviations in 11 years around the announcement year [
5,+5] based on a sample of 659 US acquiring firms between 1962 and 2001. As shown in Fig. 4, it is evident that the median actual leverage ratios of the INCREASE Group (416 firms) are lower than the optimal levels before the announcement year, and the deviations then start to reverse at the announcement year (from
4.5% in year 0 to 7.8% in year +1). Notably, the actual leverage ratios eventually converge to the optimal level in a long term after the acquisitions. Consistently, the DECREASE Group’s leverage ratio deviations show an opposite pattern of the INCREASE Group, and the actual leverage ratios also converge to the optimal level subsequent to the acquisitions. The US evidence thus strongly supports the trade-off theory and is consistent with the China evidence shown in Fig. 2. 3.4. Leverage ratio adjustment speed of acquiring firms after M&A So far we have shown that, consistent with the trade-off theory, firms have leverage targets and they actively adjust their leverage ratios to toward the optimal level. Naturally, an important question is to estimate the speed at which firms adjust their actual leverage ratios after the M&A. We use the following standard partial adjustment model to estimate the adjustment speed following Flannery and Rangan (2006):

Actual Lev eraget
Actual Lev eraget
1 ¼ kðPredicted Lev eraget
Actual Lev eraget
1 Þ þ et

ð4Þ

where k is the adjustment speed ð0 < k < 1Þ. Firms tend to reduce a proportion (k) of the gap between its actual leverage ratio and the predicted optimal level each year, and eventually the gap becomes zero; that is, the actual and predicted optimal leverage ratios become the same. The predicted leverage ratio takes the form of:

Predicted Lev eraget ¼ bX t
1

ð5Þ

where Xt
1 is a vector of firm characteristics that have been found to affect a firm’s capital structure (as defined in Eq. (1)), including Market-to-Book ratio, Asset Tangibility, Profitability, R&D Expense, R&D Dummy, Selling Expense and Firm Size. Substituting Eqs. (5) into (4), we have:

Actual Lev eraget ¼ ðkbÞX t
1 þ ð1
kÞActual Lev eraget
1 þ et

ð6Þ

7 To be specific, the median (average) leverage ratio deviation is 0.010 (0.027) and
0.006 (
0.016) in year 5 for the INCREASE and DECREASE groups, respectively. For the overall group, the median (average) leverage deviation is 0.008 (0.007). 8 Morellec and Zhdanov (2008) have a sample of 1926 acquiring firms that announced M&A between January 1980 and December 2005. For Harford et al. (2009), the sample for ‘‘All Firms” consists of 1188 acquiring firms that announced M&A between the beginning of 1981 and the end of 2000. The ‘‘Firms with Only Large Acquisitions” consists of about 618 acquiring firms (52% of the 1188 firms) that make just one acquisition within an 11-year period [
5, +5].

50

Q. Tao et al. / North American Journal of Economics and Finance 40 (2017) 41–54

Fig. 3. A reproduction of leverage ratio deviations by Harford et al. (2009) and Morellec and Zhdanov (2008) (one year for a period). (a) shows the leverage ratio deviations by Harford et al. (2009) (Table 5, page 10, Panel B). The sample for ‘‘All” consists of acquiring firms that announced M&A between the beginning of 1981 and the end of 2000. (b) shows the 6-year period leverage ratio deviations by Morellec and Zhdanov (2008) (Fig. 7, page 573). A firm’s leverage ratio deviation is estimated by subtracting the predicted optimal level of leverage ratio from its actual leverage ratio (Eq. (3)), whereas a firm’s optimal leverage ratio is estimated based on the regression Eq. (2). The sample consists of 1926 acquiring firms that announced M&A between January 1980 and December 2005.

Fig. 4. A replication of leverage ratio deviation in the U.S. This figure replicates the leverage ratio deviations in 11 years window [
5, +5] around the M&A announcement year based on a sample of 659 US acquiring firms between 1962 and 2001. We exclude firms with successive deals when constructing our sample; and draw a clear line between the estimation window and the event window when estimating firm optimal level of leverage and leverage deviations. We estimate a firm’s leverage ratio deviation by subtracting the predicted optimal level of leverage ratio from its actual leverage ratio (Eq. (3)), whereas a firm’s optimal leverage ratio is estimated based on the regression Eq. (2). We also classify all samples into the INCREASE and DECREASE groups based on the M&A-induced leverage deviations, rather than on the acquisition payment method (i.e., cash, equity and mixed).

51

Q. Tao et al. / North American Journal of Economics and Finance 40 (2017) 41–54

The reason that firms make only a partial adjustment toward the optimal level is due to the associated adjustment cost. If k ¼ 1, the adjustment cost is zero and the actual leverage ratio is always at its optimal level. If k ¼ 0, the adjustment is infinitely slow or there is no adjustment at all. In this case, the actual leverage follows a random walk and it does not support the trade-off theory. Hence, if 0 < k < 1 and the t-test for (1
k) is significant, the partial adjustment holds true and supports the trade-off theory. 3.4.1. Adjustment speed The standard partial adjustment model (Eq. (6)) is tested by OLS and Fama-MacBeth regression model. Table 4 reports the empirical results after we classify all sample firms into INCREASE and DECREASE groups based on whether their M&A-induced leverage change increases or decreases between period
1 and 0. The adjusted R-squares are between 0.741 and 0.904 for the INCREASE Group and between 0.763 and 0.833 for the DECREASE Group, similar to those reported by Flannery and Rangan (2006) and Fama and French (2002) for the US evidence.

Table 4 Regression results of post-acquisition leverage adjustment speed using a partial adjustment model. This table reports the results of the standard partial adjustment model by OLS and Fama-MacBeth regression model based on Eq. (6). We classify the sample into two groups based on whether a firm’s M&Ainduced leverage change increases or decreases between period
1 and period 0. The dependent variable is leverage ratio, defined as a firm’s total liabilities over its total assets, both measured as book value at the semi-annual period. The independent variables include Market-to-Book ratio, Asset Tangibility, Profitability, R&D Expense, R&D Dummy, Selling Expense, Firm Size, and lagged actual leverage ratio. The definitions of these independent variables are provided in Appendix I. Period +1 OLS Actual Leveraget Panel A (INCREASE Group) Constant
0.400*** (
2.76) Market-to-Book 0.003 (0.86) Asset Tangibility 0.093 (1.37) Profitability
0.129 (
1.09) R&D Expenses
0.793 (
1.54) R&D Dummy 0.012 (0.84) Selling Expenses 0.093 (0.91) Firm Size 0.026* (1.90) Actual Leveraget
1 0.967*** (26.92) Adjusted R2 0.904 F-test 69.45*** N 153 Panel B (DECREASE Group) Constant M/B Asset Tangibility Profitability R&D Expenses R&D Dummy Selling Expenses Firm Size Actual Leveraget
1 Adjusted R2 F-test N


0.225 (
0.94)
0.003 (
0.35) 0.074 (1.17) 0.592** (2.16) 0.432 (0.24)
0.013 (
0.57)
0.636 (
0.30) 0.028 (1.31) 0.837*** (17.71) 0.807 24.91*** 104

Period +2 OLS Actual Leveraget

Period +3 OLS Actual Leveraget

Period +4 OLS Actual Leveraget

Period [+1 to +4] OLS Actual Leveraget

Period [+1 to +4 F–M Actual Leveraget


0.679** (
2.21) 0.010* (1.76) 0.214 (1.18)
0.262 (
0.97) 1.115 (0.85) 0.007 (0.21)
0.048 (
0.23) 0.053** (2.03) 0.832*** (11.84) 0.741 25.12*** 153


0.418** (
2.08) 0.007 (1.03)
0.022 (
0.19)
0.266** (
2.13)
0.287 (
0.51)
0.024 (
1.05)
0.120 (
0.91) 0.049*** (2.80) 0.829*** (18.32) 0.806 35.77*** 153


0.063 (
0.29) 0.027*** (4.09)
0.159 (
1.33)
0.456** (
2.00)
0.340 (
1.12)
0.035 (
1.46)
0.104 (
0.76) 0.026 (1.36) 0.855*** (17.20) 0.790 32.80*** 153


0.351*** (
3.17) 0.009*** (
3.37) 0.054 (0.89)
0.235*** (
2.68)
0.146 (
0.57)
0.006 (
0.53)
0.042 (
0.57) 0.033*** (3.37) 0.865*** (33.50) 0.767 166.32*** 612


0.390* (
2.93) 0.012 (2.28) 0.032 (0.46)
0.278** (
4.10)
0.076 (
0.60)
0.010 (
0.89)
0.044 (
0.90) 0.039** (5.15) 0.871*** (28.91) 0.810 9.92** 612

0.237 (0.96)
0.022** (
2.32) 0.098 (0.75)
0.572** (
2.26) 3.918 (1.36)
0.039 (
1.60) 0.355 (1.57) 0.006 (0.27) 0.748*** (15.07) 0.769 20.08*** 104

***, **, * denote the significance level at 1%, 5%, and 10%, respectively.


0.091 (
0.36)
0.019* (
1.93)
0.118 (
0.84) 0.251 (1.12) 4.376 (1.03)
0.016 (
0.57)
0.075 (
0.33) 0.030 (1.32) 0.888*** (13.32) 0.776 16.38*** 104


0.080 (
0.44) 0.017** (2.56)
0.052 (
0.54)
0.476** (
2.19)
0.053*** (
2.66)
0.206 (
1.42)
0.024 (
0.23) 0.031* (1.71) 0.803*** (16.10) 0.833 29.43*** 104


0.080 (
0.67)
0.005 (
1.27) 0.050 (0.78)
0.044 (
0.36)
0.464 (
0.55)
0.031*** (
2.64) 0.028 (0.20) 0.024** (2.01) 0.841*** (29.48) 0.763 195.62*** 416


0.040 (
2.79)
0.007 (
0.76) 0.000 (0.01)
0.051 (
0.18) 2.168 (2.22)
0.069* (
3.13) 0.083 (0.02) 0.021* (2.32) 0.819*** (27.94) 0.796 1.68 416

52

Q. Tao et al. / North American Journal of Economics and Finance 40 (2017) 41–54

Table 5 The effect of adjustment cost on leverage adjustment speed subsequent to acquisitions. This table reports the regression result on whether the adjustment costs influence the adjustment speed by adding bankruptcy risk as an independent variable. Following Flannery and Rangan (2006), we estimate a firm’s bankruptcy risk as: Bankruptcy Risk = (total book assets)/(3.3 times earnings before interest and taxes + sales + 1.4 times retained earnings + 1.2 times working capital). We classify the sample into two groups based on whether a firm’s M&A-induced leverage change increases or decreases between period
1 and period 0. The dependent variable is leverage ratio, defined as a firm’s total liabilities over its total assets, both measured as book value at the semi-annual period. The independent variables include Market-to-Book ratio, Asset Tangibility, Profitability, R&D Expense, R&D Dummy, Selling Expense, Firm Size, and lagged actual leverage ratio. The definitions of these independent variables are provided in Appendix I. Period +1 OLS Actual Leveraget Panel A (INCREASE Group) Constant 0.130 (1.13) Market-to-Book 0.008 (0.89) Asset Tangibility
0.046 (
0.46) Profitability 0.027 (0.19) R&D Expenses
0.139 (
0.24) R&D Dummy
0.030 (
1.65) Selling Expenses
0.076 (
0.65) Firm Size
0.000 (
0.01) Actual Leveraget
1 0.940*** (17.40) Bankruptcy Risk
0.000* (
1.90) Adjusted R2 0.863 F-test 50.25*** N 150 Panel B (DECREASE Group) Constant M/B Asset Tangibility Profitability R&D Expenses R&D Dummy Selling Expenses Firm Size Actual Leveraget
1 Bankruptcy risk Adjusted R2 F-test N Panel C: Full sample Constant M/B Asset Tangibility Profitability R&D Expenses


0.211 (
1.58)
0.004 (
0.44) 0.074 (0.60) 0.612 (1.54) 0.427 (0.76)
0.012 (
0.69)
0.050 (
0.36) 0.028** (2.04) 0.840*** (8.36) 0.000 (0.34) 0.815 24.93*** 102

0.073 (0.69)
0.004 (
0.86)
0.028 (
0.47)
0.150 (
0.93) 0.527 (0.97)

Period +2 OLS Actual Leveraget

Period +3 OLS Actual Leveraget

Period +4 OLS Actual Leveraget

Period [+1to +4] OLS Actual Leveraget

Period [+1to +4] F–M Actual Leveraget


0.385** (
2.60) 0.003 (0.45) 0.070 (0.67)
0.125 (
1.34)
0.778*** (
3.69) 0.012 (0.78) 0.098 (0.95) 0.027** (2.12) 0.967*** (27.69) 0.000 (1.00) 0.888 62.61*** 150


0.650* (
1.91) 0.010* (1.86) 0.229 (1.21)
0.275 (
1.22) 1.002 (1.27) 0.010 (0.37)
0.031 (
0.21) 0.049 (1.56) 0.825*** (5.37)
0.001* (
1.80) 0.641 12.19*** 150


0.070 (
0.31) 0.027*** (4.00)
0.159 (
1.30)
0.436* (
1.81)
0.338 (
1.08)
0.037 (
1.08)
0.104 (
0.74) 0.027 (1.37) 0.850*** (16.15)
0.000 (
0.50) 0.776 28.86*** 150


0.303*** (
2.73) 0.010*** (2.80) 0.027 (0.39)
0.259*** (
2.66)
0.144 (
1.25)
0.005 (
0.44)
0.053 (
0.93) 0.030*** (2.80) 0.862*** (19.01)
0.001*** (
5.42) 0.800 126.58*** 600


0.245 (
1.64) 0.012*** (2.65) 0.024 (0.33)
0.202*** (
2.35)
0.063 (
0.19)
0.011 (
1.00)
0.028 (
0.73) 0.026*** (2.96) 0.896*** (30.13)
0.001 (
0.90) 0.792 16.36** 600

0.250 (1.10)
0.021** (
1.66) 0.108 (0.85)
0.618 (
1.45) 4.085*** (3.53)
0.036 (
1.47) 0.322 (0.96) 0.007 (0.29) 0.737*** (6.28) -0.001 (
1.44) 0.771 19.25*** 102


0.175** (
2.60)
0.006 (
0.88) 0.068 (0.77)
0.390 (
1.61)
0.681** (
2.07)


0.149 (
0.63) 0.011*** (4.97) 0.088 (0.68)
0.162 (
0.91) 1.714 (1.054)
0.082** (
2.50)
0.024 (
0.09) 0.045 (1.57) 0.585*** (3.20)
0.000** (
2.17) 0.605 9.14*** 102


0.380** (
2.11) 0.011*** (3.84) 0.046 (0.47)
0.274* (
1.80) 0.828 (1.21)


0.041 (
0.23)
0.018 (
1.45)
0.070 (
0.99) 0.503 (
0.99) 0.590 (0.47)
0.053*** (
3.30)
0.232 (
1.31) 0.029* (1.68) 0.800*** (9.64)
0.001 (
1.00) 0.826 26.18*** 102


0.147 (
1.40) 0.007 (1.51) 0.070 (1.11)
0.106 (
0.64) 0.056 (0.08)
0.041*** (
3.23)
0.014 (
0.11) 0.033** (2.45) 0.716*** (9.01)
0.000 (
0.82) 0.725 57.46*** 408


0.038 (

0.43)
0.008 (

1.26) 0.005 (1.42) 0.084 (0.33) 1.704** (2.33)
0.046*** (
3.59) 0.004 (0.042) 0.027*** (4.04) 0.741*** (15.27)
0.000* (
1.74) 0.754 2.03 416


0.025 (
0.18) 0.023*** (5.11)
0.108 (
1.47)
0.471*** (
3.06)
0.225 (
0.82)


0.116 (
1.55) 0.006** (2.15) 0.007 (0.18)
0.244** (
2.40)
0.216 (
1.62)


0.127 (
1.48) 0.006 (1.02)
0.006 (
0.16)
0.321*** (
5.30) 0.112 (0.38)

53

Q. Tao et al. / North American Journal of Economics and Finance 40 (2017) 41–54 R&D Dummy Selling Expenses Firm Size Actual Leveraget
1 Bankruptcy risk Adjusted R2 F-test N


0.019 (
1.38)
0.096 (
0.84) 0.007 (0.68) 0.892*** (22.59)
0.000 (
0.45) 0.836 156.71*** 252


0.012 (
0.81) 0.117 (0.74) 0.024 (1.25) 0.824*** (8.88) 0.000 (0.09) 0.890 110.63*** 252


0.024 (
1.29)
0.018 (
0.15) 0.054** (2.54) 0.717*** (6.46)
0.000** (
2.35) 0.641 12.19*** 252


0.036** (
2.28)
0.116 (
1.17) 0.023* (1.74) 0.851*** (23.92)
0.000 (
0.92) 0.801 54.02*** 252


0.021*** (
2.71)
0.004 (
0.06) 0.028*** (2.89) 0.792*** (16.74)
0.000 (
0.37) 0.753 128.26*** 1008


0.023*** (
5.20)
0.028 (
0.62) 0.027*** (3.18) 0.896*** (30.13)
0.000*** (
3.17) 0.792 17.08** 1008

***, **, * denote the significance level at 1%, 5%, and 10%, respectively.

Panel A reports the results for the INCREASE Group. The coefficients for the lagged actual leverage ratios, (1
k), are all statistically significant at 1% level across the board. The results estimated by OLS regression model show that for the INCREASE Group, the adjustment speed k is 3.3% ((1–0.967)  100% = 3.3%) for period +1, 16.8% for period +2, 17.1% for period +3, and 14.5% for period +4, with an average adjustment speed of 13.5% for the period [+1,+4]. For the DECREASE Group, Panel B shows that the adjustment speed k is 16.3% for period +1, 25.2% for period +2, 11.2% for period +3, and 19.7% for period +4, with an average adjustment speed of 15.9% for the period [+1,+4]. The results estimated by Fama-MacBeth regression model show that the adjustment speed is 12.9% for the period [+1,+4] for the INCREASE Group while it is 18.1% for the period [+1,+4] for the DECREASE Group. The adjustment speed we report here is consistent with Fama and French (2002) and Flannery and Rangan (2006). Fama and French (2002) report an adjustment speed at 15%–18% for dividend nonpayers based on the Fama-MacBeth regression model (one year for each period). Flannery and Rangan (2006) report a speed of 13.3% by the Fama-MacBeth regression and 13.6% by the OLS regression model. 3.4.2. Adjustment cost According to the trade-off theory, the leverage ratio adjustment speed of acquiring firms subsequent to the M&A is conditional on the cost of adjustment. In this section, we use acquirer’s bankruptcy risk as a measure of adjustment cost. Following Flannery and Rangan (2006), we estimate a firm’s bankruptcy risk as follows:

Bankruptcy Risk ¼ ðtotal book assetsÞ=ð3:3 times earnings before interest and taxes þ sales þ 1:4 times retained earnings þ 1:2 times working capitalÞ

ð7Þ

We then add Bankruptcy Risk into the regression model (Eq. (6)) to test its effect on the leverage ratio adjustment speed. As reported in Table 5, the coefficients for Bankruptcy Risk are generally negative across the INCREASE, DECREASE, and the whole sample. For example, in Panel A where the sample includes the firms that have increased their leverage ratios after the M&A, the coefficients for Bankruptcy Risk are negative and significant for three out of six specifications. In particular, from the period +1 to +4 subsequent to the M&A, the coefficient for Bankruptcy Risk is
0.001 (t =
5.42) using OLS regression while it becomes negative but insignificant when using Fama-MacBeth regression. The result indicates that the higher the adjustment cost, the slower the leverage ratio adjustment speed subsequent to the acquisition, consistent with the tradeoff theory. 4. Conclusions There are several theories related with a firm’s capital structure. According to the trade-off theory (Bradley et al., 1984; Myers, 1984; Fischer et al., 1989), firms have target leverage ratios and the optimal level is reached when firms trade off tax benefits of debt financing against costs of financial distress. But the empirical evidence is not conclusive. In this study, we use an important type of corporate event in the Chinese market –M&A as a unique setting to investigate whether firms have leverage targets. We first predict acquiring firms’ optimal leverage ratios using a tobit regression model and then estimate leverage ratio deviations, which tracks the extent to which the actual leverage ratios deviate from the optimal leverage ratios. Our empirical results show that acquirers’ capital structures deviate from the optimal level before the deals, and acquirers make use of M&A to reduce the deviations. Their actual leverage ratios converge to the optimal level in the long run subsequent to the acquisitions. Furthermore, we examine how the acquirers adjust their leverage ratios in the years subsequent to the acquisition using a standard partial adjustment model. The average adjustment speed is approximately 12.9%
13.5% for the INCREASE Group for the period [+1,+4] and is 15.9%–18.1% for the DECREASE Group, and such an adjustment speed is conditional on acquirers’ adjustment cost. Overall, our findings are consistent with the trade-off theory, and suggest that firms indeed have a leverage target and they tend to adjust their leverage ratios toward an optimal level, at which the cost and benefit of debts become equal. We also successfully replicate the US evidence using our improved research methodology. The improvement includes excluding firms with successive deals when constructing our sample, differentiating between the estimation window and

54

Q. Tao et al. / North American Journal of Economics and Finance 40 (2017) 41–54

the event window when estimating firm optimal level of leverage and leverage deviations, and classifying all samples into the INCREASE and DECREASE groups based on the M&A induced leverage deviations. Our improved methodology is important to help us truly understand the effect of M&A on firms’ capital structure decisions. Acknowledgement Tao acknowledges the financial support from the Fundamental Research Funds for the Central Universities in China (Grant Number: JBK160921). Appendix I. Variable definitions and predicted signs based on the trade-off theory. This table describes the definitions of the independent variables used in Equation (1) and the predicted signs of the independent variables based on the trade-off theory. All accounting items are measured at the end of the semi-annual period.

Market-to-Book (M/B) Asset Tangibility Profitability R&D Expense R&D Dummy Selling Expense Firm Size

Definition

Trade-off Theory

Market value of equity/Book value of equity Net property, plant and equipment/Total assets EBITDA/Total Assets R&D expense/Net sales R&D Dummy is set to 1 if a firm does not report R&D expense, and zero if otherwise Selling expense/Net sales Logarithm of total asset


+ +
+
+

References Alti, A. (2006). How persistent is the impact of market timing on capital structure? Journal of Finance, 61(4), 1681–1710. Baker, M., & Wurgler, J. (2002). Market timing and capital structure. Journal of Finance, 57(1), 1–32. Bradley, M., Jarrell, G. A., & Kim, E. H. (1984). On the existence of an optimal capital structure. Journal of Finance, 39(3), 857–878. Brealey, R., & Myers, S. (2003). Principles of corporate finance (7th ed.). McGraw-Hill/Irwin Press. Bruner, R. (1988). The use of excess cash and debt capacity as a motive for merger. The Journal of Financial and Quantitative Analysis, 23(2), 199–217. DeAngelo, H., & Masulis, R. (1980). Optimal capital structure under corporate and personal taxation. Journal of Financial Economics, 8(1), 3–29. Fama, E., & French, K. (2002). Testing trade-off and pecking order predictions about dividends and debt. Review of Financial Studies, 15(1), 1–33. Fischer, E., Heinkel, R., & Zechner, J. (1989). Dynamic capital structure choice: Theory and tests. Journal of Finance, 44(1), 19–40. Flannery, M., & Rangan, K. (2006). Partial adjustment toward target capital structures. Journal of Financial Economics, 79(3), 469–506. Ghosh, A., & Jain, P. (2000). Financial leverage changes associated with corporate mergers. Journal of Corporate Finance, 6(4), 377–402. Goyal, V., Lehn, K., & Racic, S. (2002). Growth opportunities and corporate debt policy: The case of the U.S. defense industry. Journal of Financial Economics, 64 (1), 35–59. Graham, J. R. (1996). Proxies for the corporate marginal tax rate. Journal of Finance Economics, 42(2), 187–221. Grinblatt, M., & Titman, S. (2002). Financial markets and corporate strategy. McGraw-Hill Press. Harford, J., Klasa, S., & Walcott, N. (2009). Do firms have leverage targets? Evidence from acquisitions. Journal of Financial Economics, 93(1), 1–14. Hovakimian, A., Hovakimian, G., & Tehranian, H. (2004). Determinants of target capital structure: The case of dual debt and equity issues. Journal of Financial Economics, 71(3), 517–540. Hovakimian, A., Opler, T., & Titman, S. (2001). The debt-equity choice. The Journal of Financial and Quantitative Analysis, 36(1), 1–24. Jensen, M. (1986). Agency costs of free cash flow, corporate finance and takeovers. The American Economic Review, 76(2), 323–329. Kayhan, A., & Titman, S. (2007). Firms’ histories and their capital structures. Journal of Financial Economics, 83(1), 1–32. Leary, M., & Roberts, M. (2005). Do firms rebalance their capital structures? The Journal of Finance, 60(6), 2575–2619. Liu, X., Wei, F., Zhan, Y., & Tai, Benjamin Y. (2004). Empirical tests on the pecking order of listed companies in China. Accounting Research, 6, 66–72 (in Chinese). MacKinlay, C. (1997). Event studies in economics and finance. Journal of Economic Literature, 35(1), 13–39. Morellec, E., & Zhdanov, A. (2008). Financing and takeovers. Journal of Financial Economics, 87(3), 556–581. Myers, S. (1984). The capital structure puzzle. Journal of Finance, 39(3), 575–592. Rajan, R., & Zingales, L. (1995). What do we know about capital structure? some evidence from international data. Journal of Finance, 50(5), 1421–1460. Shyam-Sunder, L., & Myers, S. C. (1999). Testing static tradeoff against pecking order models of capital structure. Journal of Financial Economics, 51, 219–244. Strebulaev, L. A. (2007). Do tests of capital structure theory mean what they say? Journal of Finance, 62(4), 1747–1787. Titman, S. (1983). The effect of capital structure on a firm’s liquidation decisions. Journal of Financial Economics, 13(1), 137–1514. Titman, S., & Wessels, R. (1988). The determinants of capital structure choice. Journal of Finance, 43(1), 1–19. Welch, I. (2004). Capital structure and stock returns. Journal of Political Economy, 112(1), 106–131.