CHINA JOURNAL OF ACCOUNTING STUDIES 2019, VOL. 7, NO. 1, 1–24 https://doi.org/10.1080/21697213.2019.1635322 ARTICLE Are group control associated with excess leverage? Empirical evidence a b,c d Yulan Wang , Xiaochen Dou and Jinglin Li School of Accountancy, Shanghai University of International Business and Economics, Shanghai, China; b c Financial Department, China Investment Corporation, Beijing, China; Chinese Academy of Fiscal Sciences, Beijing, China; School of Accountancy, Hubei University of Economics, Wuhan, China ABSTRACT KEYWORDS Group control; ownership This study investigates the eﬀect of group control on the excess type; excess leverage; leverage of Chinese listed ﬁrms during 2003–2016. We ﬁnd that internal and external the extent and probability of excess leverage in group-aﬃliated governance ﬁrms are signiﬁcantly higher than stand-alone ﬁrms. After consid- ering the ﬁrm ownership, we ﬁnd that the excess leverage is more pronounced among private group-aﬃliations than the state- owned ones. Moreover, the empirical evidence indicates that bet- ter corporate governance and eﬃcient institutional environment can eﬀectively decrease the degree of excess leverage in group- aﬃliated ﬁrms. This paper provides a new perspective for under- standing the over-leverage problem of Chinese ﬁrms and oﬀers practical implications for the implementation of de-leverage policy. 1. Introduction The slowing down and over-capacity of economy have raised the economic and ﬁnan- cial risks in China. According to the report of Chinese Academy of Social Sciences, the total national leverage of China has reached to 168.48 trillion yuan, and the leverage ratio is about 249%. After considering the structure of debt, the excess leverage problem is more severe in non-ﬁnancial sector, and the leverage ratio is up to 131%. The high level of debt may increase the cost of debt and even increase the risk of bankruptcy at the micro level. Meanwhile, the high level of debt will reduce the overall social con- sumption capacity and the investment level, thus leading to the slow-down of economy and ultimately triggering ﬁnancial crisis. Therefore, the Chinese government has put forward the supply-side policy and one of its core contents is requiring the Chinese ﬁrms to reduce their debt level. The real meaning of ‘de-leverage’ refers to reducing the existing debt ratio to a reasonable level in order to avoid the occurrence of ﬁnancial risks (Wang, 2017). So the execution of de-leverage is not applicable for all ﬁrms, but only for ﬁrms whose leverage are too high. There are many questions to address with regard to excess leverage before the de-leverage mechanism can be eﬀectively implemented. For CONTACT Yulan Wang firstname.lastname@example.org School of Accountancy, Shanghai University of International Business and Economics, Shanghai, China Paper accepted by Kangtao Ye. © 2019 Accounting Society of China 2 Y. WANG ET AL. example, which ﬁrms are over-leveraged? Which factors lead to the excess leverage phenomenon? And how to eﬀectively governance the excess leverage among our Chinese ﬁrms? We address the following issues in this research: (1) Whether group-aﬃliated ﬁrms are more likely to have a higher level of excess leverage than independent ones; (2) What’s the diﬀerence between private group-aﬃliations and the state-owned ones in terms of excess leverage? (3) How corporate governance, institutional factors and audit quality aﬀect the excess leverage of group-aﬃliated ﬁrms. Our results show that in China, group-aﬃliated ﬁrms are more likely to have excess leverage than stand-alone counterparts, both in probability and extent. This phenom- enon is more signiﬁcant among private group-aﬃliations than the state-owned ones. Furthermore, we ﬁnd that the increasing of board size and the percentage of indepen- dent boarders can signiﬁcantly reduce the level of excess leverage of group-aﬃliated ﬁrms. In addition, the development of a sound institutional environment and the improvement of audit quality contributes to the decrease of excess leverage of group- aﬃliated ﬁrms. This study is associated with several strands of the literature. First, we contribute to the excess leverage literature by providing a new perspective for understanding the excess leverage phenomenon in China. We investigate diﬀerent types of organizational forms of controlling shareholders aﬀect the excess leverage problem of ﬁrms. This will help us to better understand the excess leverage issue in China. Second, the results of this study contribute to the business group literature. Prior literature usually uses the level of debt to examine how business group control aﬀects ﬁrms’ debt policy, ignoring the debt capacity heterogeneity of diﬀerent ﬁrms. While the excess leverage ratio has considered this heterogeneity, and it is more eﬀective to judge the appropriateness of a ﬁrms’ leverage. Finally, our results can oﬀer some practical implications for the implementation of de-leverage policy. As we all know, most Chinese ﬁrms’ leverage is too high, which has attracted the attention of the government, and the de-leverage policy has been put forwarded to solve this problem. However, before we implement this policy, we should clear about which types of ﬁrms have excess leverage, and the factors that associate with it. The ﬁndings of our study can be useful for the government to identify the excess leverage ﬁrms and provide some suggestions in relation to developing an eﬀective governance mechanism to reduce the level of excess leverage. The remainder of the paper is organized as follows. In Section 2, we brieﬂy introduce the background of Chinese business groups and the literature review. In Section 3,we develop our hypotheses. In Section 4, we describe the database and discuss our methodologies. In Section 5, we present the empirical results and the results of robust- ness checks. In Section 6, we show the results of further study. We conclude the paper in Section 7. 2. Literature review 2.1. Formation and function of business groups Since the late 1990s scholars in economics, ﬁnance and management have extensively studied business groups. Today in the business world, business groups represent CHINA JOURNAL OF ACCOUNTING STUDIES 3 a dominant organizational form in developed and developing countries alike, with the GDP contributions of business groups accounting for upwards of 35% of GDP in some countries (Colpan, Hikino, & Lincoln, 2010; Granovetter, 1995). Despite their historical economic dominance in many economies, business groups received virtually no atten- tion until 1970s (Berglöf & Perotti, 1994; Flath, 1992; Gertner, Scharfstein, & Stein, 1994; Weinstein & Yafeh, 1995). Three main theoretical perspectives give diﬀerent explanations for the formation of business groups. These theories include the expropriation perspective, the rent-seeking perspective and the institutional voids perspective. The expropriation perspective por- trays business groups as organizations created by controlling shareholders to divert or expropriate minority shareholders’ interests. The rent-seeking perspectives postulates that business groups are structures set up to serve as a mechanism through which a handful of owners receive access to scarce resources for private beneﬁts (Khanna & Palepu, 2000). The institutional voids perspective describes the business groups as a social good that compensates for weak external institutions by creating more eﬃcient markets and saving transaction costs. The formation of Chinese business groups originates with China’s market-oriented reform. In the mid-1980s, the Chinese government started to promote business groups because it believed that such groups can absorb new technology, deliver stable ﬁnancial performance, and achieve international competitiveness (He, Ma, Rui, & Zha, 2013). One of the institutional characteristics of Chinese business groups is that most of these groups are sponsored and controlled by government-related entities and only small numbers of business groups are privately controlled. The state-owned enterprises (SOEs) have multiple and conﬂicting objectives. On one hand, the SOEs have to maintain and improve the eﬃciency of capital resources. On the other hand, the SOEs have the responsibility to sustain the level of employment and take on other social responsibil- ities to fulﬁll their political objectives. These diﬀerent and multi-dimensional objectives make SOEs diﬃcult to identify their operating loss. Meanwhile, China has maintained a state-dominated ﬁnancial system in which the government at various levels controls the allocation of ﬁnancial resources, particularly provided through the banking sector. Government-guided ﬁnancial resource allocation usually favors SOEs, and this leads to the problem of soft budget constraints. The empirical results of business groups research are mixed. On one hand, some researchers argue that business groups serve as an internal ﬁnancial market through which capital can be allocated among aﬃliated ﬁrms, which can support those aﬃliated ﬁrms whose performance is not good, avoiding the occurrence of ﬁnancial distress or bankruptcy. Strachan (1976) suggests that business groups provide insurance for the instability of the outside capital markets. Bena and Ortiz-Molina (2013) ﬁnd that business groups in the form of pyramids provide a ﬁnancing advantage in setting up new ﬁrms when the pledgeability of cash ﬂows from outside ﬁnanciers is limited. Khanna and Yafeh (2005) contend that the risk-sharing and co-insurance function of business groups can diversify risks and enhance the production eﬃciency. On the other hand, some researchers ﬁnd that business groups also have a dark side. The business group structure can be a value destroyer. For example, through tunneling among group members in a pyramid structure, the controlling shareholder may use high excess leverage as a channel to place more resources at their disposal to facilitate tunneling activities. 4 Y. WANG ET AL. Meanwhile, the principal-principal conﬂict also encourages the risk-taking behavior of the managers of aﬃliated ﬁrm. So, a consensus has not yet been reached concerning the net advantages resulted from aﬃliation with a business group. 2.2. Business groups and debt ﬁnance Researches about the relation between business groups and debt ﬁnance have formed two contradictory opinions. On one hand, some studies argue that business groups allow the formation of internal capital markets that can pool funds from diﬀerent aﬃliates and reallocate them to the most proﬁtable uses, and create enough cash ﬂow within the group, reducing the need for external debt ﬁnance (Gopalan, Nanda, & Seru, 2007). Hence, the level of debt ﬁnancing in group-aﬃliated ﬁrms is lower than the independent ones. On the other hand, business groups aﬃliation can create severe agency problems, and thus destroy ﬁrm value. Moreover, debt ﬁnancing enables con- trolling shareholders to control more economic resources without diluting their control over the corporation, so business groups usually have an incentive to use more debt to expand their control of resources in order to commerce tunneling, resulting in higher leverage. 3. Theoretical analysis and hypothesis development 3.1. Business groups and excess leverage Modigliani and Miller’s(1958) irrelevance proposition concerning capital structure choices and ﬁrm value led to a large body of theoretical and empirical studies. Three alternative theories of capital structure emerged, including the trade-oﬀ theory, the pecking order theory and agency cost theory. Despite these theories refer to a stand- alone ﬁrms, they can also apply to the capital structure of business groups. First, the trade-oﬀ theory predicts that a ﬁrm’s target leverage is determined by the trade-oﬀ between tax shields of debt and the cost of ﬁnancial distress. Generally, group-aﬃliated ﬁrms have tendency to be more diversiﬁed, which can reduce the potential risk of default and increase the group’s debt raising capacity. Meanwhile, group-aﬃliated ﬁrms are likely to cross-subsidize other members and cover debt obligations in the event of a default to protect the group’s reputation. In addition, the costs arisen from information asymmetries at debt renegotiation is smaller for group-aﬃliated ﬁrms. Thus, decreased potential ﬁnancial distress costs prompt group-aﬃliated ﬁrms to take on more debt. Second, the pecking order theory states that ﬁrms choose to ﬁnance new investment, ﬁrstly by internal retained earnings and then by debt and equity. The target leverage concept does not exist. This theory is based on the assumption that the insiders have more information than the outsiders. As for group-aﬃliated ﬁrms, they have greater access to internal funds and this would reduce their desire to use external debt, and the problem of information asymmetry between group members is less severe than non-group-aﬃliated ﬁrms. Therefore, the information advantage within groups would allow group-aﬃliated ﬁrms to take less debt. Finally, the agency theory proposes that the optimal capital structure is determined by agency costs, which includes both debt and equity issue. Compared with equity, debt enables the controlling shareholder CHINA JOURNAL OF ACCOUNTING STUDIES 5 to control more resources without diluting their control over the corporation. Thus, the controlling shareholders usually have the incentive to expand their control of resources in order to commence tunneling, and they enjoy all the beneﬁts without bearing the full ﬁnancial costs. Moreover, the severe agency problem within group-aﬃliations make it diﬃcult for the outsiders to monitor their transactions. This expects to lead to the excess leverage of group-aﬃliated ﬁrms. Thus, ex ante, it’s unclear whether group-aﬃliated ﬁrms have a higher debt than stand-alone ﬁrms. If the pecking order theory holds, we would expect the probability and degree of excess leverage within group-aﬃliated ﬁrms are lower than stand-alone ones. If the trade-oﬀ theory and the agency theory hold, the probability and extent of excess leverage of group-aﬃliated ﬁrms expect to be higher than the independent ones. Therefore, we develop the following two competing hypotheses: Hypothesis 1a: Group-aﬃliated ﬁrms are more likely to have a higher level of excess leverage than independent ones. Hypothesis 1b: Group-aﬃliated ﬁrms are more likely to have a lower level of excess leverage than independent ones. 3.2. Business groups, ownership and excess leverage In China, the behavior of debt ﬁnance diﬀers greatly between SOEs and private enterprises. SOEs face less ﬁnancial constraints. As China has maintained a state-dominated ﬁnancial system in which the government at various levels controls the allocation of ﬁnancial resources via China’s banking system. The four state-owned commercial banks dominate the banking market, and government-guided ﬁnancial resource allocation usually favors SOEs that are considered important to the economic development of the country. It is diﬃcult for most private ﬁrms to secure ﬁnancing through the state-controlled ﬁnancial system, and as a result they suﬀer from serious ﬁnancial constraints. In such a context, a business group is likely to serve as an internal capital market to mitigate the ﬁnancial constraints faced by private ﬁrms. Lu, He, and Dou (2015) ﬁnd that the debt ﬁnancing advantage brought by business groups is more prominent in non-SOEs. Through establish business groups, private ﬁrms can enhance their capability of debt guarantee, and debt ﬁnancing. However, this may also brings the problem of excess leverage. Li, Chen, and Huang (2007) propose that in order to solve the problem of ﬁnancing constrains, many private ﬁrms in China get together to form business groups to improve their debt ﬁnancing capability, and this leads to a higher level of excess leverage within private business groups. Overall, the less ﬁnancial constraints of SOEs may lead to a higher target leverage of SOEs, and the probability of excess leverage may be lower than non-SOEs. The severe problem of ﬁnancing constraints and the desire for debt ﬁnancing promote private business groups to issue more debt and ﬁnally lead to the excess leverage problem. As a result, we expect to see that business groups play diﬀerent roles in SOEs and private ﬁrms. Thus, we propose the following hypothesis: Hypothesis 2: Private group-aﬃliated ﬁrms are more likely to have a higher level of excess leverage than SOEs. 6 Y. WANG ET AL. 4. Methodology and measurement of variables 4.1. Data collection We collect our data from the China Stock Market and Accounting Research (CSMAR) database. Our sample include 2341 listed non-ﬁnancial ﬁrms for the year 2003 to 2016 and 19,000 ﬁrm-year observations for which data are available for all the variables we require for the analysis. We examine ﬁnancial leverage from the perspective of the form of controlling shareholders. To this end, our sample period began in 2003 when the information of controlling shareholders was issued by the China Securities Regulation Commission (CSRC). Based on the database from the CSRC and CSMAR, we identify a ﬁrm’s group-aﬃliated in each year if its ultimate controlling shareholder entity has more than one ﬁrm in that year. We also identify each ﬁrm’s group aﬃliation and the aﬃliation and disaﬃliation years. We exclude: (1) Financial ﬁrms (ﬁrms with unique accounting standards and capital structure), (2) ‘ST’ ﬁrms or negative-equity ﬁrms (ﬁnancial distressed ﬁrms), (3) ﬁrms that went public during the research date, (4) ﬁrms whose leverage ratio exceed one or below zero, and (5) ﬁrms whose relevant data are not complete or cannot be acquired. The ﬁnal sample consists of 19,000 ﬁrm-year observations from 2003–2016, and includes all the 12 industries in the Chinese capital market. To mitigate the eﬀect of outliers, all variables are winsorized at the 1% and 99% levels. 4.2. Measuring variables 4.2.1. Excess leverage Following previous studies, we deﬁne the excess leverage as the diﬀerence between a ﬁrm’s actual and predicted book leverage in a given year. To calculate a ﬁrm’s predicted leverage, we use predicted values from the same Tobit regression model used in Harford, Klasa, and Walcott (2009). The variables that are used to predict leverage are proﬁtability (ROA), asset tangibility (FATA), ﬁrm sizes (SIZE), median industry leverage (IND_LEVB), ownership (SOE), growth opportunities (GROWTH), shareholding percentage of the ﬁrst largest shareholder (SHRCR1). By estimating separate annual regressions, we are able to exclude expected inﬂation from the model as this variable is uniform across all ﬁrms for any given year. We establish the following equation to calculate the target leverage: LEVB ¼ α þ α SOE 1 þ α ROA þ α IND LEVB þ α GROWTH þ α FATA t 0 1 t 2 t1 3 t1 4 t1 5 t1 (1) þ α SIZE þ α SHRCR1 þ ε 6 t1 7 t1 The dependent variable is the book leverage (LEVB), which is deﬁned as the total debt to total assets. Our selection of independent variables is motivated by Chang, Chen, and Liao (2014), who ﬁnd that the most reliable factors inﬂuencing leverage among Chinese public traded ﬁrms are: proﬁtability (ROA), which is computed as total pre-tax proﬁtto total assets; asset tangibility (FATA), which equals total tangible assets to total assets; ﬁrm sizes (SIZE) equals to the natural logarithm of total assets; median industry leverage (IND_LEVB) is computed each year for each four digit SIC code; ownership (SOE) is deﬁned by the ultimate controlling shareholder of a ﬁrm. The dummy variable equals 1 if CHINA JOURNAL OF ACCOUNTING STUDIES 7 the controlling shareholder is the central or a local government, or a central or State Asset Supervision and Administration Commission (SASAC), and 0 if others; growth opportunities (GROWTH) is deﬁned by the percentage change of sales revenue. 4.2.2. Group identiﬁcation Our sample includes all ﬁrms listed on either the Shanghai or Shenzhen Stock Exchange. The information on each ﬁrm’s group aﬃliation is from its ﬁnancial statements which contain the information about the ﬁrm’s ownership structure, ultimate controlling share- holder, and other related ﬁrms within the same group. The information covers the period from 2003 to 2016. Based on this database, we identify a ﬁrm’s group-aﬃliation in each year if its ultimate controlling entity has more than one ﬁrm in that year. As for the central government-controlled ﬁrms, their ultimate controlling shareholder are the Central Asset Supervision and Administration Commission (SASAC). However, the SASAC won’t participate in the operation management of these ﬁrms. So following the research of Cai and Hu (2016), if the ﬁrm’s ultimate controlling shareholder is SASAC, we continue to ﬁnd the groups that directly controlled by the SASAC, and deﬁne the group as the ultimate controlling shareholder. We set the GROUP dummy variable to 1 if the ultimate controlling entity has more than one ﬁrm in that year, and 0 otherwise. To investigate the eﬀect of business group aﬃliation on excess leverage, we use the same approach as those of Lu et al. (2015). They regress excess leverage on group aﬃliation dummy variable, and other control variables. We adopt the following basic panel speciﬁcation to examine the group aﬃliation on the probability and degree of excess leverage: LogitðEXLEV dumÞ¼ γ þ γ GROUP þ γ SIZE þ γ SOE i;t i;t i;t 0 1 2 3 þ γ ROA þ γ GROWTH þ γ FATA i;t i;t i;t 4 5 6 þ γ IND LEVB þ γ SHRCR1 þ γ MB i;t i;t i;t 7 8 9 (2) þ γ NDTS þ γ EXP þ γ ETR þ γ VCFOTA i;t i;t i;t i;t 10 11 12 13 þ γ VEBITTA þ γ MANAHOLD i;t i;t 14 15 þ INDUSTRY þ YEAR þ ε EXLEV ¼ β þ β GROUP þ β SIZE þ β SOE þ β ROA i;t i;t i;t i;t i;t 0 1 2 3 4 þ β GROWTH þ β FATA þ β IND LEVB i;t i;t i;t 5 6 7 þ β SHRCR1 þ β9MB þ B NDTS þ β EXP (3) i;t i;t 10 i;t i;t 8 11 þ β ETR þ β VCFOTA þ β VEBITTA i;t i;t i;t 12 13 14 þ β MANAHOLD þ INDUSTRY þ YEAR þ ε i;t Where i = 1,2 . . ., N refers to the i-th ﬁrm in period t = 1,2 . . ., T in our sample. ε refers to errors. Deﬁnitions of these variables are provided in Table 1. We use Logit model to estimate Equation (2), and use panel data ﬁxed eﬀects to estimate Equation (3). The dependent variable in Equation (2) is the dummy variable of group aﬃliation (EXLEV_dum), it equals 1 if the ultimate controlling shareholder is a group, and 0 otherwise. We try to examine the likelihood of excess leverage between group aﬃliated ﬁrms and the independent ﬁrms. We use Equation (3) to examine the relation between group aﬃliation and excess leverage, and the explanatory variable is the degree of 8 Y. WANG ET AL. Table 1. Variable deﬁnition. Variables Deﬁnition LEV Year-end total debt/year-end total assets EXLEV The degree of excess leverage: it equals to the actual leverage for each ﬁrm minus the target leverage in a speciﬁc year EXLEV_dum Excess leverage dummy variable: it equals 1 if the actual leverage exceeds the target leverage, and 0 otherwise GROUP A dummy variable that equals 1 if the ﬁrm is a group aﬃliated, and 0 if the ﬁrm is an independent ﬁrm SIZE The natural logarithm to total assets at the beginning of year t SOE Equals 1 if the ﬁrm is a state-owned enterprise, and 0 otherwise ROA Net income for year t/total assets at the beginning of year t GROWTH Percentage change of sales revenue FATA Total ﬁxed assets to total assets IND_LEVB Industry median leverage SHRCR1 Percentage of shares held by the largest shareholder MB Market value/replacement value NDTS Non-debt tax shield, it equals to the interest expense to total assets EXP Overhead expenses to total assets ETR Eﬀective tax rate ratio, equals to the total tax paid to total assets VCFOTA Standard deviation of cash ﬂow VEBITTA Standard deviation of cash ﬂow EBIT MANAHOLD Ratio of shares held by top management LAWSCORE Regional marketization index BIG4 Equals 1 if the ﬁrm is audited by the BIG4 and 0 otherwise BOARSIZE Equals to the total number of directors DIR Equals to the number of independent directors/total number of directors excess leverage (EXLEV), which is a continuous variable. We also include the following control variables: MB refers to the market value/replacement value; NDTS is the non- debt tax shield, which equals to the interest expense to total assets; EXP refers to the overhead expense, which is calculated by the overhead expenses to total assets; ETR refers to the eﬀective tax rate, which equals to the total tax paid to total assets; VCFOTA refers to the cash ﬂow volatility, and VEBITTA refers to the volatility of EBIT; MANAHOLD refers to the shares held by top managers to total shares outstanding. 4.3. Descriptive statistics Table 2 reports the descriptive statistics for the variables used in our regression models. The average leverage ratio of Chinese listed ﬁrms is 46.2%, which seems not too high. From the perspective of excess leverage, we can ﬁnd that the mean and median excess leverage ratios are respectively −0.1% and 0.5%, which indicates that in this sample the mean and median diﬀerence between the actual leverage and optimal leverage is −0.1% and 0.5% respectively. In addition, we ﬁnd that the max and min value of excess leverage is −52.4% and 59.8%. It indicates that the debt level exhibits a great diﬀerence within diﬀerent ﬁrms, and the key to solve the over-leverage problem of Chinese ﬁrms is to ﬁnd the over-leveraged ﬁrms and encourage their de-leverage behavior. The mean value of Group is 58.5%, indicating that the number of group-aﬃliated ﬁrms is higher than independent ﬁrms. The results are the same as previous studies about Chinese business groups(Ji & Liu, 2014; Pan & Yu, 2010). The mean size of Chinese listed ﬁrms is 21.83. The average of proﬁtability is 3.9%. The mean value of industry median leverage is 45.7%. The average ETR is only 17.9% in our sample, which is much lower than the typical number in other economies, implying that Chinese listed companies bear CHINA JOURNAL OF ACCOUNTING STUDIES 9 Table 2. Summary statistics for Chinese listed ﬁrms. Variables Obs. Mean Median Std. Dev Min Max LEV 19000 0.462 0.468 0.207 0.05 0.999 EXLEV 19000 −0.001 0.005 0.171 −0.524 0.598 GROUP 19000 0.585 1.000 0.493 0.000 1.000 SIZE 19000 21.83 21.70 1.192 15.980 25.520 SOE 19000 0.531 1.000 0.499 0.000 1.000 ROA 19000 0.039 0.036 0.060 −0.192 0.224 GROWTH 19000 0.210 0.105 0.446 −0.425 3.089 FATA 19000 0.250 0.217 0.177 0.002 0.748 IND_LEVB 19000 0.457 0.423 0.100 0.205 0.708 SHRCR1 19000 0.364 0.343 0.153 0.089 0.750 MB 19000 0.965 0.676 0.895 0.084 6.516 NDTS 19000 0.022 0.019 0.016 0.000 0.074 EXP 19000 0.048 0.042 0.030 0.004 0.170 ETR 19000 0.179 0.165 0.175 −0.508 0.824 VCFOTA 19000 0.049 0.037 0.044 0.002 0.254 VEBITTA 19000 0.029 0.016 0.036 0.001 0.232 MANAHOLD 19000 0.061 0.000 0.142 0.000 0.657 a slightly-low real income tax burden. The ratio of shares held by the largest shareholder (Blockshares) has a mean value of 36.4%, suggesting that the ownership structure of Chinese listed companies is highly concentrated. The average ratio of shares held by top management is only 6.1%; the low management holding implies relatively high agency costs. In our sample, approximately 53.1% of listed ﬁrms are state-owned, which indi- cates that the SOEs take up more than a half of the ﬁrms in the sample. Overall, most descriptive statistics are consistent with those described by Lu et al. (2015), who studies the type ﬁrm ownership on excess leverage. 4.4. Correlation matrix Table 3 reports the correlation matrix between our main variables. The correlation coeﬃcient between EXLEV and GROUP is 0.142, which is positive and statistically signiﬁcant, indicating that group aﬃliations may be the main factor that leads to the excess leverage of Chinese listed ﬁrms. Both size and SOEs are positively and signiﬁ- cantly correlated with EXLEV, indicating that the level of leverage of SOEs is higher than non-SOEs; this result is close to the ﬁndings of numerous previous studies. ROA, NDTS, EXP and MANAHOLD exhibit negative and signiﬁcant correlations with EXLEV, indicating that management share-holding can eﬀectively reduce the agency costs and lead to the decrease of excess leverage. These ﬁndings are generally consistent with the results of previous studies. 5. Empirical analysis and discussion of results 5.1. Univariate analysis Table 4 reports univariate analyses for group aﬃliation ﬁrms and independent ﬁrms, respectively. Group aﬃliations have signiﬁcantly higher leverage and excess leverage than independent ﬁrms. In addition, following previous studies (Antoniou, Guney, & Paudyal, 2008; Booth, Aivazian, Demirgüc-Kunt, & Maksimovic, 2001;Rajan &Zingales, 10 Y. WANG ET AL. Table 3. Correlation matrix. (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (1) EXLEV 1 (2) GROUP 0.142*** 1 (3) SIZE 0.168*** 0.238*** 1 (4) SOE 0.137*** 0.458*** 0.254*** 1 (5) ROA −0.284*** −0.049*** 0.085*** −0.087*** 1 (6) GROWTH 0.087*** −0.050*** 0.145*** −0.091*** 0.183*** 1 (7) FATA 0.018** 0.157*** 0.043*** 0.227*** −0.120*** −0.147*** 1 (8) IND_LEVB 0.217*** 0.153*** 0.139*** 0.223*** −0.067*** −0.026*** −0.045*** 1 (9) SHRCR1 0.017** 0.258*** 0.240*** 0.243*** 0.132*** 0.044*** 0.085*** 0.084*** 1 (10) MB 0.408*** 0.216*** 0.498*** 0.277*** −0.239*** −0.028*** 0.140*** 0.325*** 0.126*** 1 (11) NDTS −0.036*** 0.180*** 0.011 0.197*** −0.115*** −0.185*** 0.788*** −0.112*** 0.102*** 0.058*** 1 (12) EXP −0.093*** −0.074*** −0.340*** −0.058*** 0.003 −0.135*** −0.014** −0.273*** −0.104*** −0.289*** 0.122*** 1 (13) ETR 0.048*** 0.059*** 0.086*** 0.069*** 0.145*** 0.022*** −0.023*** 0.150*** 0.035*** 0.112*** −0.039*** −0.096*** 1 (14) VCFOTA 0.155*** −0.024*** −0.124*** −0.044*** −0.028*** 0.064*** −0.178*** 0.140*** 0.020*** −0.012* −0.126*** −0.049*** 0.043*** 1 (15) VEBITTA 0.051*** −0.033*** −0.213*** −0.043*** −0.229*** 0.022*** 0.048*** −0.019*** −0.071*** −0.113*** 0.108*** 0.159*** −0.145*** 0.256*** 1 (16) MANAHOLD −0.230*** −0.444*** −0.185*** −0.441*** 0.111*** 0.068*** −0.170*** −0.284*** −0.100*** −0.215*** −0.160*** 0.077*** −0.072*** −0.052*** −0.045*** 1 CHINA JOURNAL OF ACCOUNTING STUDIES 11 Table 4. Univariate test of independent ﬁrms and group aﬃliated ﬁrms. Non-Group aﬃliations Group aﬃliations Diﬀerence Tests Mean Median Mean Median T value Z value LEV 0.400 0.400 0.500 0.510 −0.100*** −33.284*** EXLEV −0.030 −0.030 0.020 0.030 −0.050*** −19.675*** EXLEV2 −0.030 −0.040 0.040 0.050 −0.080*** −27.760*** EXLEV3 −0.030 −0.030 0.040 0.040 −0.070*** −25.327*** 1995), we also use a ﬁrm’s leverage minus the optimal leverage measured by the industrial mean and median book leverage level of all Chinese listed ﬁrms, to measure the ﬁrm’s excess leverage. We then create two new variables for our analysis, i.e., the excess leverage measured by industrial mean leverage (EXLEV2) and the excess leverage measured by the industrial median leverage (EXLEV3). Applying these two proxies of excess leverage (i.e., EXLEV2 and EXLEV3), we have the same results that the level of excess leverage is signiﬁcantly higher than independent ﬁrms, which is consistent with our prediction. Overall, our univariate analysis as reported in Table 4 are consistent with our hypothesis H1b that group aﬃliations are more likely to have excess leverage than independent ﬁrms. 5.2. Multivariate analysis To test the relationship between group controlling and excess leverage, we run Logit regression of the excess leverage dummy variable on our group controlled variable and a vector of control variables that are typically used in the excess leverage literature. Results presented in Table 5 are consistent with our prediction. Column (1) of Table 5 shows that the coeﬃcient on Group is positive and statistically signiﬁcant at the 5% level, indicating that ﬁrms that are group controlled are more likely to have excess leverage than independent ﬁrms. Column (2) of Table 5 that repeats the same regression but have controlled the problem of heteroscedasticity reveals the results are unchanged. In addition, we also examine the relationship between the degree of excess leverage and group control. We run the OLS regression of the level of excess leverage on the GROUP Control Variable. Column (3) of Table 5 shows that the coeﬃcient on GROUP is more economically and statistically signiﬁcant than column (1), which is signiﬁcant at the 5% level as the former is signiﬁcant at the 1% level. Column (4) replicates column (3) but has controlled the heteroscedasticity problem, however, results are not altered by this addition. These results specify that compared with independent ﬁrms, group- aﬃliations are more likely to have excess leverage and the degree of excess leverage is much higher. Despite many studies have acknowledged the over-leverage problem of Chinese ﬁrms, prior researches have rarely reported which ﬁrms exhibit excess leverage. Thus, an investigation of this issue is important to implement the de-leverage policy that has put forward by the Chinese government. Overall, the results in Table 5 suggest that group control has a positive eﬀect on excess leverage both in the likelihood and extend. Our results are consistent with those of He et al. (2013), who reveals that the internal capital markets in Chinese groups are not eﬀective and the Chinese group aﬃliations rely heavily on loans from banks, resulting in the high level of leverage within the Chinese group aﬃliations. 12 Y. WANG ET AL. Table 5. Multivariate analysis of group control and excess leverage. (1) (2) (3) (4) Logit Cluster OLS Cluster EXLEV_dum EXLEV_dum EXLEV EXLEV GROUP 0.083** 0.088** 0.013*** 0.013** (2.10) (2.00) (4.97) (2.20) SIZE 0.498*** 0.507*** 0.003*** 0.003 (20.64) (21.07) (2.82) (1.38) SOE −0.144*** −0.130*** −0.009*** −0.009 (−3.36) (−3.02) (−3.48) (−1.49) ROA −10.631*** −11.030*** −0.631*** −0.631*** (−27.94) (−28.79) (−31.35) (−17.66) GROWTH 0.339*** 0.345*** 0.053*** 0.053*** (7.99) (8.10) (19.71) (13.97) FATA 1.453*** 1.482*** 0.059*** 0.059*** (7.95) (8.11) (5.78) (2.73) IND_LEVB −3.911*** −6.405*** 0.099*** 0.099*** (−7.66) (−12.40) (8.04) (3.67) SHRCR1 −0.316** −0.266** −0.022*** −0.022 (−2.50) (−2.11) (−2.83) (−1.33) MB 1.685*** 1.605*** 0.061*** 0.061*** (31.82) (31.35) (39.57) (22.71) NDTS −14.800*** −14.123*** −1.106*** −1.106*** (−7.52) (−7.18) (−9.44) (−4.47) EXP 5.451*** 5.633*** 0.457*** 0.457*** (8.02) (8.27) (11.36) (5.31) ETR 0.276** 0.317*** 0.022*** 0.022** (2.56) (2.94) (3.50) (2.28) VCFOTA 5.806*** 5.378*** 0.524*** 0.524*** (13.26) (12.30) (19.75) (11.47) VEBITTA 0.579 0.789 0.013 0.013 (1.06) (1.43) (0.40) (0.22) MANAHOLD −2.095*** −2.048*** −0.170*** −0.170*** (−13.18) (−13.13) (−16.88) (−9.29) CONSTANT −10.426*** −9.545*** −0.194*** −0.194*** (−18.84) (−17.33) (−7.28) (−3.43) YEAR YES YES YES YES INDUSTRY YES YES YES YES Observations 19000 19000 19000 19000 Adjusted R 0.273 0.272 Pseudo R 0.271 0.267 Our results are consistent with the Trade-oﬀ theory and the Agency Cost theory, which predict a higher leverage level of group aﬃliations than independent ﬁrms. However, this ﬁnding is inconsistent with the Pecking order theory predictions that group aﬃliations are less likely to have excess leverage because of its eﬃcient internal capital market. This is consistent with the opinion of Admati, Demarzo, Hellwig, and Pﬂeiderer (2018), who ﬁnds that shareholders pervasively resist leverage reductions no matter how much such reduction may enhance ﬁrm value. Shareholders would instead choose to increase leverage even if the new debt is junior and would reduce ﬁrm value. He calls this phenomenon the ‘leverage ratchet eﬀect’. From this point of view, ﬁrms with excess leverage are more likely to increase their debt level, which will increase the ﬁnancial risks and even lead to the bankruptcy of the ﬁrm. So, it becomes important and necessary to oversee the problem of excess leverage within Chinese groups. In addition, considering diﬀerent ﬁnancing environments faced by SOEs and non- state enterprises, we further investigate the eﬀect of state-ownership on the relation CHINA JOURNAL OF ACCOUNTING STUDIES 13 between group aﬃliations and excess leverage. We look at all group-aﬃliated ﬁrms and then separate them into state-owned and private ones. Results are presented in Table 6. The coeﬃcient of SOE is −0.171, which is negative and statistically signiﬁcant at 1% level, as shown in column 1 of Table 6. The results suggest that the state-owned group- aﬃliates are less likely to have excess leverage than private ones. To control for the heteroskedasticity in the data, robust standard errors are used (White, 1980). Firm- clustered standard errors are also employed to account for time-series dependence (Petersen, 2009). The coeﬃcients of SOEs are still negative and statistically signiﬁcant, which suggests that our results are robust. In order to further investigate the extent of excess leverage, we use the continuous variable of excess leverage as the explanatory variable. Column 3 shows that the extent of excess leverage within state-owned group- aﬃliates is much lower than the private ones, and the coeﬃcient of SOE is −0.015, which is negative and statistically signiﬁcant at 1% level. The results are consistent with those of Lu et al. (2015), who ﬁnd that the equity ﬁnancing advantage of SOEs reduced their need for debt ﬁnancing. As China maintained a state-dominated ﬁnancial system in Table 6. Business groups, ownership type and excess leverage. (1) GROUP = 1 (2) GROUP = 1 (3) GROUP = 1 (4) GROUP = 1 Logit Cluster OLS Cluster EXLEV_dum EXLEV_dum EXLEV EXLEV SOE −0.171*** −0.140** −0.015*** −0.017** (−3.12) (−2.54) (−4.61) (−2.10) SIZE 0.425*** 0.415*** 0.000 0.003 (13.89) (13.64) (0.10) (0.77) ROA −10.622*** −10.869*** −0.686*** −0.688*** (−21.16) (−21.55) (−26.47) (−15.27) GROWTH 0.393*** 0.442*** 0.058*** 0.052*** (6.36) (7.06) (16.20) (10.91) FATA 1.623*** 1.705*** 0.061*** 0.094*** (6.94) (7.29) (4.88) (3.13) IND_LEVB −5.782*** −8.185*** 0.073*** −0.022 (−8.64) (−12.13) (4.64) (−0.44) SHRCR1 −0.314* −0.277* −0.041*** −0.033 (−1.91) (−1.68) (−4.32) (−1.61) MB 1.669*** 1.591*** 0.058*** 0.060*** (25.83) (25.58) (32.23) (16.26) NDTS −17.337*** −16.746*** −1.275*** −1.535*** (−7.24) (−6.98) (−9.21) (−4.89) EXP 4.158*** 3.898*** 0.477*** 0.399*** (4.54) (4.24) (9.06) (3.42) ETR 0.396*** 0.395*** 0.021*** 0.023* (2.82) (2.81) (2.71) (1.95) VCFOTA 5.829*** 5.387*** 0.545*** 0.514*** (9.77) (9.08) (15.80) (8.52) VEBITTA 1.063 1.267* 0.006 0.013 (1.42) (1.69) (0.13) (0.16) MANAHOLD −3.238*** −3.562*** −0.298*** −0.264*** (−4.69) (−5.17) (−6.57) (−3.10) CONSTANT −8.021*** −6.825*** −0.079** −0.126 (−11.33) (−9.70) (−2.39) (−1.41) YEAR YES YES YES YES INDUSTRY YES YES YES YES Observations 9447 9447 9447 9447 Adjusted R 0.258 0.285 Pseudo R 0.269 0.268 14 Y. WANG ET AL. which the government at various levels controls the allocation of ﬁnancial resources in both banking sector and securities market. Government-guided ﬁnancial resource allo- cation usually favors a few large-scale SOEs that are important to the economic devel- opment of the country and the speciﬁc region. SOEs may also face the problem of soft budget constraints. However, it is rather diﬃcult for most non-state owned enterprises to secure ﬁnancing through the government-controlled ﬁnancial system. Consequently, private ﬁrms suﬀer from serious ﬁnancial repression. In such a context, a business group is likely to serve as an internal capital market to mitigate the ﬁnancial constraints faced by private ﬁrms. Nonetheless, some studies have found that the internal capital market of Chinese groups is ineﬃcient (Ji & Liu, 2014; Shao & Liu, 2008). Many groups ﬁnance themselves through tunneling among group members in a pyramid structure. Large and inﬂuential ﬁrms may be propped up at the expense of other ‘weak’ members in the group; this also leads to the excess leverage of group-aﬃliated ﬁrms. 5.3. Robustness tests 5.3.1. Endogeneity and sample-selection issues The OLS regressions, as noted in prior studies are prone to have selection bias and a potential endogenous problem, as the group-aﬃliation sample may be selected based on some unobservable factors and these factors could inﬂuence the variation in excess leverage across ﬁrms. This potential may create a bias in the estimation of the coeﬃcients of group dummy. We use the Heckman (1979) two-stage method to take into account self-selection bias. In the ﬁrststage,weestimateaprobit modelofgroup aﬃliation(dummy) on a set of variables that tend to inﬂuence a ﬁrm’sgroup aﬃliation choice. Then we include the Lambda (inverse Mills’ ratio) based on the probit estimate in the previous regression speciﬁcation to control for potential self-selection bias. We use the following variables as control variables in the above regressions: corporate size, growth and earnings performance(ROA). To capture the potential determination of group aﬃliation choice, we also include cash and cash equivalent, sales revenue as control variables. The results of the ﬁrst-stageestimationare shownincolumn1in Table 7. Then based on the estimation with the use of above control variables, we get Lambda(inverse Mills’ ratio). We include the Lambda in the regression, this does not weaken our previous results, as shown in column 2, although the coeﬃcient of Lambda is signiﬁcantly negative, which partially reﬂects the eﬀect of group aﬃliation choice on excess leverage. 5.3.2. Alternative excess leverage measures The literature provides alternative ways to measure excess leverage, which include a ﬁrm’s leverage minus the optimal leverage to be measured by the industrial median and mean book leverage level of all Chinese listed ﬁrms. We then create two new variables for our regressions: the excess industrial median based leverage (EXLEV2) and industrial mean based leverage (EXLEV3). The results are shown in Table 8. Column 1 and column 3 in Table 8 show the results that group-aﬃliations have a much higher excess leverage than stand-alone ones as the coeﬃcients are signiﬁcantly positive, which is consistent with the result presented previously. This ﬁnding provides further evidence that our results are not determined by the measure errors of excess leverage. CHINA JOURNAL OF ACCOUNTING STUDIES 15 Table 7. Robustness tests for potential self-selection bias. (1) First-stage estimation to (2) Regression controlling forself- getLambda (inverse Mills’ ratio) selection bias (Lambda) GROUP EXLEV GROUP 0.011** (2.41) SOE 1.102*** −0.065*** (46.05) (−4.60) ROA −1.101*** −0.406*** (−5.52) (−12.54) SIZE −0.022 −0.033*** (−1.05) (−10.36) GROWTH −0.043* 0.038*** (−1.78) (13.10) CASHEQ −0.403*** (−4.45) SALES 0.241*** (13.52) FATA −0.007 (−0.34) IND_LEVB −0.138*** (−3.58) SHRCR1 0.016 (1.34) MB 0.027*** (13.82) NDTS −0.822*** (−3.85) EXP −0.128 (−1.60) ETR −0.002 (−0.27) VCFOTA 0.114*** (3.42) VEBITTA 0.204*** (4.27) MANAHOLD −0.074*** (−4.41) IMR (λ) −0.109*** (−4.88) CONSTANT −5.038*** 0.941*** (−8.39) (10.26) YEAR YES YES INDUSTRY YES YES Observations 19000 19000 Pseudo R 0.224 Adjusted R 0.215 To control for the heteroskedasticity in the data, robust standard errors are used. Firm- clustered standard errors are also employed to account for time-series dependence (Petersen, 2009). 5.3.3. Change sample period A characteristic of the Chinese capital market before year 2007 was a split share structure where almost 70 percent of Chinese listed ﬁrms’ outstanding shares were non- tradable shares (NTS) mainly held by block holders, including controlling shareholders. The remaining shares were tradable and mainly held by domestic individuals and institutional investors (Firth, Lin, & Zou, 2010; Li, Xin, & Yu, 2011). The Chinese 16 Y. WANG ET AL. Table 8. Robustness tests using an alternative measure. (1) (2) (3) (4) OLS Cluster OLS Cluster EXLEV2 EXLEV2 EXLEV3 EXLEV3 GROUP 0.015*** 0.015** 0.015*** 0.015** (5.78) (2.54) (5.67) (2.49) SIZE 0.044*** 0.044*** 0.045*** 0.045*** (38.40) (18.55) (39.07) (18.81) SOE 0.001 0.001 0.001 0.001 (0.49) (0.21) (0.23) (0.10) ROA −0.988*** −0.988*** −0.999*** −0.999*** (−49.69) (−27.65) (−50.34) (−28.02) GROWTH 0.027*** 0.027*** 0.026*** 0.026*** (10.75) (7.99) (10.45) (7.80) FATA 0.088*** 0.088*** 0.089*** 0.089*** (8.67) (4.14) (8.83) (4.21) IND_LEVB −0.414*** −0.414*** −0.602*** −0.602*** (−34.08) (−15.57) (−49.71) (−22.58) SHRCR1 −0.049*** −0.049*** −0.050*** −0.050*** (−6.58) (−3.06) (−6.65) (−3.10) MB 0.060*** 0.060*** 0.060*** 0.060*** (39.03) (22.88) (39.17) (22.94) NDTS −0.826*** −0.826*** −0.865*** −0.865*** (−7.12) (−3.40) (−7.46) (−3.56) EXP 0.416*** 0.416*** 0.407*** 0.407*** (10.50) (4.89) (10.31) (4.79) ETR 0.027*** 0.027*** 0.027*** 0.027*** (4.24) (2.76) (4.33) (2.82) VCFOTA 0.528*** 0.528*** 0.519*** 0.519*** (20.30) (11.74) (19.98) (11.55) VEBITTA 0.101*** 0.101* 0.102*** 0.102* (3.06) (1.66) (3.11) (1.69) MANAHOLD −0.195*** −0.195*** −0.191*** −0.191*** (−21.66) (−11.68) (−21.26) (−11.43) CONSTANT −0.826*** −0.826*** −0.752*** −0.752*** (−31.75) (−14.80) (−28.97) (−13.45) YEAR YES YES YES YES INDUSTRY YES YES YES YES Observations 19000 19000 19000 19000 Adjusted R 0.390 0.390 0.401 0.401 government implemented the NTS reform in 2005 to solve the problems associated with the split share structure. The reform was expanded to include all listed ﬁrms in August 2005, and by the end of 2007, the reform represented over 97% of all Chinese A-share listed companies (Li, Xin, & Yu, 2011). The NTS reform expects to aﬀect the debt ﬁnancing behaviors of companies and inﬂuence the relationship between group control and excess leverage. Therefore, we exclude the sample prior to 2008 to avoid the possible eﬀect of such special events. Table 9 shows the results, indicating the coeﬃ- cients of group dummy are all signiﬁcantly positive when we exclude the sample before 2008. This ﬁnding suggests that our results are not determined by the NTS reform. Group-aﬃliations show a relatively higher leverage ratio than unaﬃliated ones. 5.3.4. Considering the direction of excess leverage In order to reduce the eﬀect of the direction of excess leverage on the relationship between group-control and excess leverage, we split our sample into two sub-samples, including the over-leveraged(whose actual leverage is higher than the optimal leverage) CHINA JOURNAL OF ACCOUNTING STUDIES 17 Table 9. Robustness tests with changed sample period (2007–2016). (1) (2) (3) EXLEV EXLEV2 EXLEV3 GROUP 0.012*** 0.013*** 0.013*** (4.10) (4.68) (4.69) SIZE 0.012*** 0.050*** 0.049*** (8.05) (34.60) (34.41) SOE −0.007** 0.006** 0.006** (−2.47) (2.03) (2.03) ROA −0.698*** −1.036*** −1.035*** (−31.18) (−46.83) (−46.80) GROWTH 0.045*** 0.019*** 0.019*** (16.37) (7.55) (7.34) FATA 0.074*** 0.104*** 0.104*** (6.01) (8.53) (8.55) IND_LEVB −0.005 −0.408*** −0.675*** (−0.12) (−10.31) (−17.08) SHRCR1 −0.010 −0.033*** −0.033*** (−1.21) (−4.03) (−4.05) MB 0.060*** 0.064*** 0.065*** (31.07) (33.26) (33.71) NDTS −1.194*** −0.905*** −0.916*** (−8.98) (−6.87) (−6.95) EXP 0.456*** 0.449*** 0.448*** (9.73) (9.75) (9.72) ETR 0.020*** 0.027*** 0.026*** (2.98) (3.83) (3.79) VCFOTA 0.524*** 0.542*** 0.541*** (17.91) (18.91) (18.89) VEBITTA −0.019 0.071* 0.074** (−0.51) (1.95) (2.03) MANAHOLD −0.139*** −0.170*** −0.169*** (−13.34) (−18.28) (−18.27) CONSTANT −0.285*** −0.899*** −0.766*** (−7.58) (−24.39) (−20.76) YEAR YES YES YES INDUSTRY YES YES YES Observations 16000 16000 16000 Adjusted R 0.299 0.428 0.435 sample and the under-leveraged(whose actual leverage is lower than the optimal leverage) sample to do the regression respectively. Table 10 presents the results of this analysis. The relation between group-control and excess leverage is signiﬁcantly positive when the ﬁrms are over-leveraged, but the coeﬃcients of group dummy are not signiﬁcant when the ﬁrms are under-leveraged. It suggests that the positive relation between group control and excess leverage only exhibits in ﬁrms whose actual leverage exceeds the optimal leverage. This ﬁnding also veriﬁes our results are not aﬀected by the direction of excess leverage. 5.3.5. Considering the size eﬀect The existing literatures suggest that ﬁrm size plays an important role in the determina- tion of a ﬁrms’ debt ﬁnancing. Generally speaking, the size of group-aﬃliations are much larger than those independent ones. In order to reduce the impact of ﬁrm size, outlying size at the top and bottom 25% are excluded from the sample. We repeat the regression using the sample whose size are between the 25% to 75%. The results are shown in 18 Y. WANG ET AL. Table 10. Robustness tests considering the direction of excess leverage. (1) (2) (3) (4) (5) (6) EXLEV > 0 EXLEV < 0 EXLEV2 > 0 EXLEV2 < 0 EXLEV3 > 0 EXLEV3 < 0 GROUP 0.013*** 0.003 0.010*** 0.004 0.010*** 0.004 (5.91) (1.33) (4.21) (1.60) (4.29) (0.87) SIZE −0.018*** 0.008*** 0.006*** 0.027*** 0.007*** 0.027*** (−16.65) (6.00) (5.26) (19.34) (5.81) (9.78) SOE −0.004* −0.005* 0.007*** −0.003 0.007*** −0.004 (−1.80) (−1.78) (2.85) (−1.14) (2.81) (−0.77) ROA −0.431*** −0.017 −0.608*** −0.185*** −0.631*** −0.173*** (−22.54) (−0.88) (−29.84) (−9.53) (−30.13) (−5.21) GROWTH 0.033*** 0.023*** 0.014*** 0.009*** 0.013*** 0.009*** (15.53) (8.59) (6.44) (3.63) (5.75) (3.00) FATA −0.012 0.084*** −0.006 0.095*** −0.004 0.097*** (−1.33) (8.15) (−0.60) (8.95) (−0.45) (4.80) IND_LEVB −0.104*** 0.034 −0.429*** −0.197*** −0.563*** −0.304*** (−4.20) (1.11) (−15.91) (−6.82) (−19.88) (−6.65) SHRCR1 0.010 −0.014** −0.006 −0.012* −0.010 −0.013 (1.59) (−1.98) (−0.90) (−1.75) (−1.43) (−0.97) MB 0.026*** 0.060*** 0.035*** 0.076*** 0.035*** 0.077*** (20.92) (19.51) (26.74) (22.40) (26.10) (12.42) NDTS −0.637*** −0.697*** −0.524*** −0.579*** −0.547*** −0.667*** (−6.27) (−6.49) (−5.11) (−5.14) (−5.18) (−3.31) EXP −0.034 0.527*** −0.088** 0.522*** −0.103*** 0.511*** (−0.95) (13.90) (−2.33) (13.80) (−2.67) (6.48) ETR 0.002 0.013* −0.001 0.025*** −0.001 0.024** (0.42) (1.96) (−0.14) (3.61) (−0.17) (2.35) VCFOTA 0.137*** 0.253*** 0.186*** 0.267*** 0.221*** 0.276*** (6.50) (9.42) (8.10) (10.60) (9.26) (6.97) VEBITTA 0.210*** −0.202*** 0.327*** −0.139*** 0.318*** −0.143*** (7.34) (−6.46) (11.10) (−4.44) (10.55) (−2.94) MANAHOLD −0.086*** −0.047*** −0.104*** −0.060*** −0.116*** −0.057*** (−7.54) (−5.79) (−9.52) (−8.21) (−10.66) (−4.71) CONSTANT 0.527*** −0.402*** 0.143*** −0.721*** 0.192*** −0.667*** (19.72) (−12.76) (5.19) (−22.31) (6.70) (−11.28) YEAR YES YES YES YES YES YES INDUSTRY YES YES YES YES YES YES Observations 9447 9104 9447 9104 9447 9104 Adjusted R 0.202 0.139 0.301 0.214 0.323 0.224 Table 11, indicating our results remain unchanged. The relationship between group- control and excess leverage is still signiﬁcantly positive. 6. Additional tests In order to ﬁnd out some eﬀective mechanisms to solve the problem of excess leverage within our business groups, we conduct the following analysis from the perspective of inside and outside governance. 6.1. Business groups, corporate governance and excess leverage Agency theory suggests that debt can act as a self-enforcing mechanism to mitigate the conﬂicts of interest between managers and shareholders (Grossman & Hart, 1982; Jensen, 1986). Emphasizing the use of debt as an eﬀective way to mitigate agency problem, agency theory also highlights the importance of board monitoring in con- straining managerial behavior (Jensen & Meckling, 1976). In previous empirical studies, CHINA JOURNAL OF ACCOUNTING STUDIES 19 Table 11. Robustness tests considering the size eﬀect. (1) (2) (3) EXLEV EXLEV2 EXLEV3 GROUP 0.009** 0.009** 0.009** (2.21) (2.31) (2.39) SIZE 0.020*** 0.057*** 0.057*** (3.93) (11.31) (11.31) SOE 0.001 0.013*** 0.014*** (0.18) (3.26) (3.41) ROA −0.547*** −0.892*** −0.890*** (−16.63) (−27.38) (−27.29) GROWTH 0.055*** 0.028*** 0.027*** (12.17) (6.58) (6.43) FATA 0.116*** 0.139*** 0.141*** (6.75) (8.19) (8.30) IND_LEVB 0.015 −0.445*** −0.666*** (0.29) (−9.08) (−13.56) SHRCR1 0.015 −0.009 −0.010 (1.20) (−0.74) (−0.81) MB 0.106*** 0.112*** 0.112*** (25.82) (27.33) (27.31) NDTS −1.459*** −1.174*** 0.112*** (−8.00) (−6.51) (27.31) EXP 0.559*** 0.534*** 0.538*** (8.81) (8.55) (8.59) ETR 0.025** 0.035*** 0.035*** (2.54) (3.56) (3.52) VCFOTA 0.533*** 0.535*** 0.538*** (12.70) (13.04) (13.11) VEBITTA 0.057 0.126** 0.127** (1.07) (2.41) (2.42) MANAHOLD −0.170***(−16.88) −0.134***(−10.11) −0.132***(−10.01) CONSTANT −0.103*** −1.162*** −1.063*** (−7.02) (−10.64) (−9.72) YEAR YES YES YES INDUSTRY YES YES YES Observations 7728 7728 7728 Adjusted R 0.306 0.354 0.361 board size and board independence are frequently used as proxies for board monitoring and board eﬀectiveness. Following the literature, we also use these two proxies to examine their eﬀects on group-aﬃliations’ excess leverage. We believe that the eﬀec- tiveness of the board of directors plays an important role in the governance of excess leverage in business groups. Column 1 of Table 12 reports the results concerning board governance, showing that the excess leverage of group-aﬃliations decreased signiﬁ- cantly when the ﬁrm has an eﬀective board. The result suggests that a better corporate governance mechanism can exert a great help in reducing excess leverage. 6.2. Business groups, institutional environment and excess leverage The study of Faccio, Lang, and Young (2010) shows that leverage enables controlling shareholders to control more resources without diluting their control over the corpora- tion. However, the institutional environment with a better investor protection can prohibit the tunneling behavior of the controlling shareholders through excess leverage. One of the prominent characteristics of the Chinese economy is the very high level of 20 Y. WANG ET AL. Table 12. Business groups, internal and external governance mechanisms and excess leverage. (1) Corporate governance (2) Institutional environment (3) Big 4 auditing variables EXLEV EXLEV EXLEV GDSIZE −0.004** (−2.52) GDIR −0.105* (−1.93) GROULAW −0.001** (−1.97) GROUBIG4 −0.035*** (−2.75) GDRA 0.009 (1.32) GROUP 0.089*** 0.022*** 0.014*** (2.94) (4.68) (5.24) DIR 0.009 (0.23) DSIZE 0.004*** (2.70) DRA −0.002 (−0.55) LAWSCORE −0.001*** (−3.77) BIG4 −0.010 (−0.87) SIZE 0.048*** 0.046*** 0.049*** (31.35) (39.84) (36.01) SOE 0.010*** −0.002 0.000 (3.02) (−0.81) (0.14) ROA −1.059*** −1.001*** −0.997*** (−42.10) (−50.50) (−49.10) GROWTH 0.021*** 0.026*** 0.021*** (7.12) (10.19) (8.54) FATA 0.067*** 0.079*** 0.103*** (5.01) (7.80) (9.61) IND_LEVB −0.640*** −0.600*** −0.644*** (−44.27) (−49.58) (−21.32) SHRCR1 −0.046*** −0.048*** −0.038*** (−5.03) (−6.48) (−5.13) MB 0.064*** 0.059*** 0.067*** (32.52) (38.25) (38.63) NDTS −0.504*** −0.830*** −1.047*** (−3.33) (−7.17) (−8.98) EXP 0.439*** 0.436*** 0.399*** (8.80) (10.99) (9.76) ETR 0.022*** 0.028*** 0.028*** (2.73) (4.49) (4.53) VCFOTA 0.606*** 0.511*** 0.506*** (18.30) (19.68) (19.51) VEBITTA 0.068 0.087*** 0.128*** (1.58) (2.66) (3.94) MANAHOLD −0.155*** −0.188*** −0.182*** (−15.78) (−20.67) (−20.04) CONSTANT −0.851*** −0.770*** −0.827*** (−21.36) (−29.50) (−25.74) YEAR YES YES YES INDUSTRY YES YES YES Observations 19000 19000 19000 Adjusted R 0.439 0.403 0.424 uneven distribution of economic resources across the country. We believe the diﬀerence in regional market developments has profound eﬀects on the relationship between business groups and excess leverage. To determine whether the eﬀect of business CHINA JOURNAL OF ACCOUNTING STUDIES 21 groups on excess leverage diﬀers under diﬀerent market environments, we use an index of market intermediaries and legal environment. The index covers the development score for each province and major municipality (Fan & Wang, 2017). Complied by China’s National Research Institute (NERI), the index covers a number of catalogue including the percentages of lawyers and registered certiﬁed public accountants to the total popula- tion, market order, legal enforcement eﬃciency, intellectual property rights protection, and consumer rights protection. We use the continuous variable Lawscore as the proxy of investor protection to study the governance eﬀect of external institutional environ- ment on the excess leverage phenomenon within group-aﬃliations. Column 2 of Table 12 shows the results, revealing the coeﬃcient of GROULAW (the interaction of group dummy and Lawscore) is signiﬁcantly positive. This ﬁnding suggests that a better institutional environment can eﬀectively reduce the excess leverage within business groups. The results are consistent with the view of He et al. (2013). 6.3. Business groups, audit quality and excess leverage Financial statements as primary sources of information for capital markets. External audits contribute to the quality of ﬁnancial reporting by providing an independent assessment of the accuracy and fairness of ﬁnancial statements representing the results of operation, ﬁnancial position and cash ﬂow in conformity with generally accepted accounting principles. High auditing quality can reduce the information asymmetry between the shareholders and creditors through providing fairly and timely accounting information. This expects to reduce a company’s excess leverage behavior and a creditor’s willingness to supply more capital. Meanwhile, higher auditing quality can help the shareholders to monitor the risk-taking behaviors of aﬃliations’ managers more eﬀectively. Teoh and Wong (1993) argue that Big 4 auditors actually provide higher quality service than non-Big 4 auditors and their study ﬁnds that the earnings response coeﬃcients (ERCs) of Big 4 auditors’ clients are signiﬁcantly higher than those of non-Big 4 clients. Becker et al. (1998) document that discretionary accruals of Big 4 auditors are smaller than those of non-Big4 auditors. Similarly, Wu, Yang, and Lu (2015) suggest that investors’ perception of ﬁnancial reporting quality is higher when a ﬁrm’s ﬁnancial reports are audited by Big 4 auditors. Similar to Teoh and Wong (1993), we deﬁne a high-quality audit as an audit that improves the credibility of ﬁnancial statement information and allows investors to make a better estimate of the ﬁrm value. We operationalize audit quality by using auditor size, Big4 and non-Big4 to identify a high- quality service, and further investigate the governance eﬀect of audit quality on the excess leverage of group-aﬃliations. The empirical results are shown in Column 3 in Table 12, presenting the coeﬃcient of GROUBIG4 is signiﬁcantly negative, which indi- cates that high audit quality can exert an eﬀective governance eﬀect on group- aﬃliations’ excess leverage behavior and reduce ﬁrm ﬁnancial risks. 7. Conclusion The issue of de-leverage of the non-ﬁnancial sector has attracted much attention since the proposal of the supply-size structural reform years ago. Previous empirical studies have proved that the phenomenon of excess leverage is prominent in non- 22 Y. WANG ET AL. ﬁnancial sector, and reducing the level of debt in non-ﬁnancial sector is the key to solve the excess leverage problem in China. However, few studies have systematically investigated the eﬀect of group control on the excess leverage policy and examined the role of internal and external governance mechanisms on excess leverage. In this study, we address the issues of ‘which ﬁrms have excess leverage’ and ‘how to reduce the excess leverage of these ﬁrms’.We ﬁnd that the likelihood and extent of excess leverage in group-aﬃliations are much higher than independent ones, which indicates that the business groups’ internal capital market is ineﬃcient. Also, we ﬁnd that the excess leverage phenomenon is more pronounced in private group- aﬃliations than the aﬃliated ﬁrms who are state-owned. This is because SOEs in China have advantages in equity and debt ﬁnancing, while the non-SOEs receive low priority from either the banking sector or equity markets in acquiring external ﬁnance for their investment projects. Therefore, if a private ﬁrm is group aﬃliated, it is more likely to show greater marginal eﬀect than a SOE, ceteris paribus. Business groups are likely to serve as capital acquiring channel for group members, mitigating the ﬁnancial constraints faced by private ﬁrms in China through the provision of insurance and cross funding to their members, while it also creates the problem of excess leverage. We also ﬁnd an eﬀective corporate governance mechanism can reduce the excess leverage within the group-aﬃliations. In addition, institutional environment with a better investor protection can prohibit the tunneling behavior of controlling shareholders with the use of excess leverage. Finally, we ﬁnd that employing the Big 4 auditors to audit ﬁnancial reports of a business group can eﬀectively decrease the level of excess leverage of group-aﬃliations. This study presents empirical evidence that can be used to shed new light on the theory of business groups’ capital structure, and provides a better understanding of China’s high debt issue. Disclosure statement No potential conﬂict of interest was reported by the authors. References Admati, A.R., Demarzo, P.M., Hellwig, M.F., & Pﬂeiderer, P. (2018). The leverage ratchet eﬀect. The Journal of Finance, 73(1), 145–198. Antoniou, A., Guney, Y., & Paudyal, K. (2008). The determinants of capital structure: Capital market oriented versus bank oriented institutions. Journal of Financial and Quantitative Analysis, 43(1), 59–92. Becker, C.L., DeFond, M.L., Jiambalvo, J., & Subramanyam, K.R. (1998). The eﬀect of audit quality on earnings management. Contemporary Accounting Research, 15(1), 1–24. Bena, J., & Ortiz-Molina, H. (2013). Pyramidal ownership and the creation of new ﬁrms. Journal of Financial Economics, 108(3), 798–821. Berglöf, E., & Perotti, E. (1994). The governance structure of the Japanese ﬁnancial keiretsu. Journal of Financial Economics, 36(2), 259–284. Booth, L.V., Aivazian, V., Demirgüc-Kunt, A., & Maksimovic, V. (2001). Capital structures in devel- oping countries. Journal of Finance, 56(1), 87–130. Cai, W., & Hu, Z. (2016). Business groups, the nature of ownership, and corporate cash holdings. Management Review, 28(7), 236–251. CHINA JOURNAL OF ACCOUNTING STUDIES 23 Chang, C., Chen, X., & Liao, G. (2014). What are the reliably important determinants of capital structure in China. Paciﬁc-Basin Finance Journal, 30,87–113. Colpan, A., Hikino, T., & Lincoln, J. (2010, August). The oxford handbook of business groups. Oxford, UK: Oxford University Press. Faccio, M., Lang, L.H., & Young, L. (2010). Pyramiding vs leverage in corporate groups: International evidence. Journal of International Business Studies, 41(1), 88–104. Fan, G., & Wang, X. (2017). NERI index of marketization of Chinese Provinces. Beijing: Economic Science Press. (in Chinese). Firth, M., Lin, C., & Zou, H. (2010). Friend or foe? The role of state and mutual fund ownership in the split share structure reform in China. Journal of Financial and Quantitative Analysis, 45(3), 685–706. Flath, D. (1992). Indirect shareholding within Japan’s business groups. Economics Letters, 38(2), 223–227. Gertner, R., Scharfstein, D., & Stein, J. (1994). Internal versus external capital markets. The Quarterly Journal of Economics, 109(4), 1211–1230. Gopalan, R., Nanda, V., & Seru, A. (2007). Aﬃliated ﬁrms and ﬁnancial support: Evidence from Indian business groups. Journal of Financial Economics, 86(3), 759–795. Granovetter, M. (1995). Coase revisited: Business groups in the modern economy, industrial and corporate change. Business History Review, 4(1), 93–130. Grossman, S.J., & Hart, O.D. (1982). Corporate ﬁnancial structure and managerial incentives. In J. McCall (Ed.), The Economics of Information and Uncertainty. Chicago: University of Chicago Press. Harford, J., Klasa, S., & Walcott, N. (2009). Do ﬁrms have leverage targets? Evidence from acquisitions. Journal of Financial Economics, 93(1), 1–14. He, J., Ma, X., Rui, M., & Zha., X. (2013). Business groups in China. Journal of Corporate Finance, 22 (5), 166–192. Heckman, J. (1979). Sample selection bias as a speciﬁcation error. Econometrica, 47(1), 153–161. Jensen, M.C. (1986). Agency costs of free cash ﬂow, corporate ﬁnance and takeovers. American Economic Review, 76(2), 323–329. Jensen, M.C., & Meckling, W. (1976). Theory of the ﬁrm: Managerial behaviors, agency cost and ownership structure. Journal of Financial Economics, 58,3–28. Ji, F., & Liu, X. (2014). Group control, ﬁnancing advantages and investment eﬃciency. Journal of Industrial Engineering and Engineering Management, 28(1), 26–38. Khanna, T., & Palepu, K. (2000). The future of business groups in emerging markets: Long-run evidence from Chile. Academy of Management Journal, 43(3), 268–285. Khanna, T., & Yafeh, Y. (2005). Business groups and risk sharing around the world. Journal of Business, 78(1), 301–340. Li, Y., Chen, C., & Huang, L. (2007). Collectivized operations, ﬁnancing constraint and ﬁnancial risks. Management World, 12, 117–135. Li, Z., Xin, X., & Yu, X. (2011). Financial development, debt ﬁnancing constraints and pyramid structure: Evidence from private enterprise groups. Management World, 1, 123–135. Lu, Z.F., He, J., & Dou, H. (2015). Who is more excess leverage: State-owned or non-state-owned. Economic Research Journal, 12,54–67. Modigliani, F., & Miller, M. (1958). The cost of capital, corporation ﬁnance and the theory of investment. American Economic Review, 49(4), 261–297. Pan, J., & Yu, M. (2010). Business groups, bank loan and capital allocation eﬃciency. Journal of Financial Research, 10,83–102. Petersen, M.A. (2009). Estimating standard errors in ﬁnance panel data sets: Comparing approaches. The Review of Financial Studies, 22(1), 435–480. Rajan, R.G., & Zingales, L. (1995). What do we know about capital structure: Some evidence from international data. Journal of Finance, 50(5), 1421–1460. Shao, J., & Liu, Z.Y. (2008). The impact of the internal capital market of business groups on the ﬁnancing constraints of the aﬃliations: Evidence from Chinese. China Accounting Review, 3, 275–288. 24 Y. WANG ET AL. Strachan, H. (1976). Family and other business groups in economic development: The case of Nicaragua. New York: Praeger Publishers. Teoh, S.H., & Wong, T.J. (1993). Perceived auditor quality and the earnings response coeﬃcient. The Accounting Review, 68(2), 346–366. Wang, G.G. (2017). De-leverage: Category deﬁnition, operational focus and optional strategies. Economic Perspectives, 7,16–25. Weinstein, D., & Yafeh, Y. (1995). Japan’s corporate groups: Collusive or competitive? An empirical investigation of keiretsu behavior. The Journal of Industrial Economics, 43(4), 359–376. White, H. (1980). A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica, 48(4), 817–838. Wu, X., Yang, Y.L., & Lu, Z.F. (2015). Are accounting estimate changes associated with an aggressive earnings eﬀect? Empirical evidence. China Journal of Accounting Studies, 3(1), 1–23.
China Journal of Accounting Studies
– Taylor & Francis
Published: Jan 2, 2019
Keywords: Group control; ownership type; excess leverage; internal and external governance