1
University of Amsterdam
Amsterdam Business School
Master Thesis
Effects of Government Interventions on Capital
Structure and Investment Behaviors
Finance (Quantitative Finance)
Yanli Xu (11828234)
Thesis Supervisor: Vladimir Vladimirov
2
Statement of Originality
This document is written by Yanli Xu who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3
Acknowledgements
Here I would like to express my very great appreciation to my supervisor Professor Vladimirov for his guidance and help on the thesis.
Also, I would like to extend my gratitude to my dear parents for supporting me and staying with me whatever decisions I made.
I would like to offer my special thanks to my friends from University of Amsterdam, for their company in the past year.
4
Abstract
This paper discusses the effects of government interventions on corporate capital structures and investment behaviors based on analysis of Chinese A-share market. Utilizing the marketization index to proxy for indirect government interventions by provinces, the result shows that indirect government interventions lead to higher leverage and lower firm performance, indicating that government involvement is another type of friction that drive firms from optimal capital structures and investment behaviors.
5
Table of Content
1. Introduction ... 6
2. Literature review and development of hypothesis ... 9
3. Data and descriptive statistics ... 12
3.1 Accounting Data ... 12
3.2 Stock price and equity structure ... 12
3.3 Auxiliary data ... 13 3.4 Marketization index ... 13 3.5 Descriptive statistics ... 14 4. Methodology ... 18 5. Regression results ... 22 5.1 Leverage ... 22 5.1.1 Book leverage ... 22 5.1.2 Market Leverage ... 25 5.1.3 Net leverage ... 27 5.2 Investment ... 27 5.3 Investment Performance ... 29 5.4 Robustness check ... 31 5.4.1 Leverage ... 31 5.4.2 Investment ... 33 5.4.3 Investment Performances ... 34 6. Concluding Remarks ... 35 References ... 38
Appendix I: Definition of variables ... 41
6
1. Introduction
It is known to us that the assumptions of capital structure irrelevance (Modigliani and Miller, 1958) is not valid in real world, and investment decisions are not solely determined by investment opportunities (Myers and Majluf, 1984). There are plenty of frictions existing in the real world that would influence the capital structure of firms, such as information asymmetry (Myers and Majluf, 1984; Fazzari et al., 1988), deadweight cost of bankruptcy (Kraus and Litzenberger, 1973), taxation (Stiglitz, 1973; Kane et al., 1984; Brennan and Schwartz (1984) look at the bankruptcy cost and taxation together), transaction cost (Fischer et al., 1989), etc. In addition to factors mentioned above, this paper discusses another type of friction that may affect the corporate capital structures and investment decisions. That is, the effect of government involvements, which is a prevailing factor that plays a role in the process of corporate decision-making in some emerging markets, based on an empirical study of Chinese listed firms.
Ferrell and Gresham (1985) argue that the social and cultural environment may largely influence ethics decisions. Fisman (2001) argues that political connections are valuable resource for firms. Hunt and Vitell (1986) conclude that the religion, legal environment, and political system may largely affect individual’s ethics decision-making process. Firms operate in countries that are under transition economies may perform differently than firms operate in mature markets. However, it is extremely difficult to compare the effects of cultures and conventions on corporate decisions, as the effects of cultures and conventions are imperceptibly large, but very difficult to quantify and test, especially among different countries. The relationship between governments and enterprises is one of the typical reflections of the effects of cultures and conventions on corporate decision-making process. Thus, an alternative way to examine the effects of cultures and conventions on corporate decisions, is to examine the relationship within one single country.
7
Chinese market provides a good setting for testing the effects empirically. Before 1978, the ownership structure for firms is solely state-owned. The ownership forms for enterprises are diversified since 1978, thanks to the implementation of ownership reform for state-owned enterprises by the central government of China. However, the development of the economy and the legal system is very unbalanced among different provinces. In less developed provinces, the relationship between government and corporates remained “close”, local governments could largely influence the decisions of local enterprises, both directly and indirectly. In more developed provinces, where the legal system and financial systems are highly-developed, the role of local governments are much weaker.
Government interventions may affect corporate capital structures through different channels: by providing subsidies directly, by influencing bank’s lending decisions (La Porta, Lopez-de-Silanes and Shleifer, 2002), or by holding shares of listed enterprises, etc. There are three leading theories dominating this topic (Sapienza, 2004), the social view, the political theory, and the agency theory. In this paper, I mainly examine the effects of last two channels of government involvements and try to contribute to the literature.
The first form of government intervention that I focus on is the direct form of government interventions, by holding significant shares in listed firms. In China, state-owned enterprises dominate the stock market. Among 2678 listed Chinese firms during 2008-2014, 1450 out of 2678 listed firms are state-controlled or has significant state-ownership. In this paper, I will test the effects of direct form by dividing the listed firms into several groups based on percentage of state-owned shares and compare the corresponding capital structures and investment behaviors.
Another form of government interventions on firms’ capital structures is by controlling commercial banks and influencing their lending policies. In mainland China, the 5 biggest banks are all state-owned banks, which lend more than 80% of commercial loans to enterprises in China (Shao, Hernandez, Liu, 2015). The benchmark of lending rates is determined by the Central Bank of China, but the
8
actual rates could be adjusted by individual banks in practice. Commercial banks are in favor of providing loans to state-owned enterprises and large enterprises. With the endorsement of governments, banks are willing to (or be forced to) provide loans to firms, although the “endorsement” normally takes informal forms. In less developed provinces, sometimes the power is so strong that banks could neglect the collaterals while lending.
Most existing papers and theories that examine the effects of government interventions on banks look from a macroeconomic scope, by looking at state-owned banks all over the world and their overall effects on economic development or growth, like Shleifer and Vishny (1994), Shleifer (1998), La Porta, Lopez-de-Silanes and Shleifer (2002). Another typical perspective is looking at the behaviors of state-owned banks and private-owned banks and compare their lending decisions, like Sapienza (2004).
Instead of looking at the macroeconomic influence of government interventions or observing the behavior of state-owned banks, I focus more on the microeconomic scope: the individual behavior of firms which are influenced by government, by answering three basic questions: How does government interventions affect firms’ capital structure? How does government interventions affect firms’ investment decisions? What’s the investment performance of the influenced firms?
My main finding is that, firms that suffer from high level of indirect government interventions maintain higher leverage, and the mitigation effect of government interventions on collateral values does exist. Secondly, the direct interventions boost firm investments, and the effects are significant both statistically and economically in firms with more than 50% of state ownership. Thirdly, firms suffer from higher indirect government interventions perform worse, and the effects are extremely robust when excluding firms from most developed provinces.
The distinctive features of Chinese market may partly explain the results. Chinese market is a market, where the key determinant of market behaviors is not solely
9
supply and demand, but with other factors playing a role, such as the need for the government to “maintain social stability”. From this perspective, the results may appear to be consistent with both the social view and political view of state-ownership (La Porta, Lopez-de-Silanes and Shleifer, 2002). Moreover, standing by Leuz and Oberholzer-Gee (2006), this paper supports the argument that the effects of political connections should be considered when investigating the financial performance of firms that operating in relationship-based markets.
The paper is structured in following ways: Section 1 is the introduction and research question. Section 2 is the literature review. Section 3 is data and construction of variables, as well as the descriptive statistics. Section 4 is the methodology and Section 5 is the regression results. Section 6 is the concluding remarks and Section 7 is the reference list. Additionally, there is an appendix in the end.
2. Literature review and development of hypothesis
Regarding government intervention, one of the most controversial topics is the choice between private and state ownership. Under what circumstances are state-ownership more favorable? What kind of goods or services should be provided by the government (Shleifer, 1998)? There are three leading theories trying to answer these questions (Sapienza, 2004).
The first theory, known as the social view (Atkinson and Stiglitz, 1980), states that SOEs are used to address the market failures when the social benefits exceeds the costs. This view is consistent with the “development” view of Alexander Gerschenkron (1962), which states that government is playing an irreplaceable role in some strategic economic sectors, for example, banking systems, utility industry and educations. In China, the state capital has absolute control over the biggest banks. The government has strong influence over banks’ lending decisions, especially in less developed provinces. Thus, the determinants of capital structures for state-owned firms in China or similar markets may deviated from the
10
determinants for firms in mature market. Myers (1984) asserts that internal financing is preferred than external financing due to transaction cost. However, when we take the political connection into consideration, the pecking-order theory may no longer be true. Johnson and Mitton (2003), Cull and Xu (2003), Faccio et al. (2005) etc., examine from different angles the effects of government interventions on corporate debt level, and arrive at the conclusion that firms with political connections have higher leverage than their non-political connected counterparts. Thus, the first hypothesis is built based on above arguments:
Hypothesis 1:higher the degree of government interventions, heavier firms will rely on debt financing.
The second theory regarding government intervention is the “political” view, stating that governments acquire control of enterprises and banks for political objectives rather than social objectives (Shleifer and Vishny, 1994; Sapienza, 2004). According to political view, politicians are motivated by political achievements, for example, the reduced unemployment rate, the increased GDP, the completion of large social projects and of course, bribes, under the political view of SOEs. As a result, SOEs are forced to make investments on projects that are not solely for profitability, but for employment and public welfare, etc.
The third view, agency view, can be taken as an extension of the social view, that SOEs are built to maximize social welfares, but with cost of corruptions and misallocation as “agency cost” (Banerjee, 1997; Hart et al., 2003). So, the trade-off between the social benefits it brings and the cost it takes determines the efficiency of SOEs (Tirole, 1994). The two theories lead to the second and third hypothesis:
Hypothesis 2: Higher government interventions are related to higher investment. Hypothesis 3: Higher government interventions are related to lower performance.
La Porta, Lopez-de-Silanes and Shleifer (2002) looks at the efficiency of state-owned banks from a macroeconomic perspective. They compare bank ownership of 92 countries and conclude that state-owned-banks are more
11
prevailing in poor countries, where the financial markets are under-developed and financial institutions are not functioning well. My thesis relates closely to the conclusions of La Porta et al. (2002), that higher government ownership is associated with slower subsequent development of the financial system, slower economic growth, and lower productivity. Chinese market is an example of the kind of market summarized by La Porta et al. (2002), which provides a good setting to test the effects of government involvement on firm performance.
Shao, Hernández, and Liu (2015) tested the effects of government interventions on corporate policies based on Chinese market. They examine the direct form of interventions via government ownership and indirect channel, proxied by a dummy indicating whether a firm is in EDA (economic development areas) or not. Shao et al. (2015) conclude that EDAs would ease the direct government interventions on firm’s financial policies, and both forms of interventions will lead to higher investment and lower firm performance. I will test the effects of government interventions from a slightly different perspective than Shao et al. (2015). For the direct form of government interventions, unlike Shao et al. (2015) who apply the single threshold of 30% SOE shares to measure the direct influence from banks, I divide the sample into four groups based on their percentage of SOE shares. The advantage of this more detailed classification is the improved precision. The direct form of government influence is not only reflected in the percentage of government ownerships, but also takes the form of political connected executives. That is, some listed firms have relatively low percentage of state-owned shares, but still suffer from high government interventions, due to the politically connected executives. Simply classify the firms into state-owned and non-state-owned with the threshold of 30% of state-ownership is too rough and imprecise. Moreover, to examine the indirect form of state-ownership, taking Shao et al. (2015) as a starting point, I additionally introduce two interaction variables to the regression, namely the interaction terms between government interventions and the asset tangibility, and the interaction terms between government interventions and profitability. The
12
two interaction terms are used to examine whether the indirect government interventions would influence the order of preference for external financing, as suggested by pecking-order theory.
3. Data and descriptive statistics
The sample used for the empirical analysis is Chinese listed A-share firms between 2008 and 2014. A-shares are shares of the Renminbi currency that are purchased and traded on the Shanghai and Shenzhen stock exchanges. The reason for setting the time range of 2008 t0 2014 is for remaining consistency with the time range of the 7th marketization index developed by Fan et al. (2016). The authors modified some calculation approach and standards to make the index more reliable in this version of index, so it is somewhat not comparable with previous index published before.
3.1 Accounting Data
Most of the financial statement data come from CSMAR (China Stock Market & Accounting Research) database, which provides data on China stock markets and the financial statements of Chinese listed companies. I dropped firms with zero or negative total assets and firms that corporate in financial industries (industry code “J”), as those firms usually exhibit different capital structures and investment behaviors compared to firms corporate in other industries. After dropping above sample, there are 2,762 companies left, summing up to 14,277 company-year observations.
3.2 Stock price and equity structure
To calculate the market value of the equity, I downloaded the stock price of listed firms from CSMAR. The closing price is the year closing price, which is the closing price of the last trading day of each year.
Equity structure, as well as the number of different types of shares, are obtained from the equity structure file database of CSMAR. It is merged to the main data
13
based on stock code and fiscal year.
3.3 Auxiliary data
Other important data includes firms’ location of operations. The location of operation is summarized from the statistical data from the official site of Shenzhen Stock Exchange and Shanghai Stock exchange. Most listed firms are in Eastern China and Southern China, with the top five provinces being Guangdong, Jiangsu, Zhejiang, Shanghai, and Beijing. Observations from above five provinces and cities account for 48.6% of the total observations. There are 31 provinces, which is in line with the marketization index, and the provinces are numbered as 1 to 31 and merged with the main accounting data.
3.4 Marketization index
The effects of indirect form of government intervention is captured by the marketization index developed by Fan, Yu, and Wang in 2016. The authors started the program from 2000 and it is the 7th publication of the program. According to Fan et al. (2016), the average GDP growth rate for China during the period of 1978 to 2016 has reached 9.6%, and the average GDP per capita increased from 200 USD to 8000 USD. The reason behind this astonishing growth is mainly due to the marketization reform during the past decades. However, the marketization process in China is not smooth in many aspects. The development of marketization is very unbalanced in terms of regions and industries. The process of marketization even experienced recessions in some years. They carried out the program to evaluate the marketization process and provide some reference for the government to make administrative decisions, for scholars to carry out academic research, and for investors and corporate managers to make relevant decisions.
The marketization index includes five sub-indices (Fan et al., 2016), they are: 1) the relationship between the government and the market 2) the development of non-state-owned economy 3) the development of product market 4) the development of factor market, and 5) the development of market intermediaries
14
and legal environment of the market. Under each sub-index, there are basic index and sub-basic index to reflect the level of marketization exhaustively. 2008 is regarded as basic period, and the index ranges from 0 to 10 comparing to basic period. The statistical data are obtained from authoritative organizations based on scientific statistical methods, and subjective judgements are avoided to ensure the objectiveness of the index. The overall index is the arithmetic mean of each sub-index, and it is calculated by provinces. In this paper, I apply the main index as a proxy of government interventions among different provinces. The index is merged with main data by the province numbering and fiscal year.
The details of variable explanations and constructions are available in the appendix.
3.5 Descriptive statistics
Based on the percentage of SOE shares, I divided the observations into 4 groups. Group1 includes firms whose state-owned shares are less than 1%, and Group 2 are firms with SOE-shares between 1% and 25%, Group 3 are firms with the percentage of SOE-shares between 25% to 50% and Group 4 includes firms with percentage of SOE-shares more than 50%. There are 4,726 observations in Group 1, 2,584 observations in Group 2, 2,911 and 4,056 observations in Group 3 and Group 4, respectively.
All the financial indicators and accounting variables used in the regressions are winsorized at 5%. The explanations and constructions of variables are available in the appendix.
Table A shows the summary statistics for the evolution of Chinese listed firms’ equity structure (excluding firms operating in financial industries) from 2007 to 2015, as well as the number of observations in each fiscal year. We can notice an obvious trend that the percentages of state-owned shares in listed firms are declining year by year, which is a result of the policy of privatization of state-owned enterprises implemented in mainland China in past decades.
15
Table A:
Summary of the percentage of state-owned shares of listed firms (2007-2015)
Note: The table shows the percentage of state-owned shares within listed firms from 2007 to 2015. Column (1) is the number of firms. Column (2) to column (5) are the corresponding percentiles of SOE percentage. Column (6) is the mean of the sample, and column (7) is the standard deviation of the sample.
Table B shows the descriptive statistics of the sample between 2008 to 2014. The total number of observations is 14,277. The natural logarithm of total assets indicates that the average firm size in each group is very similar, around 21.5. Average return on assets is 3.6%, and the average ROA increase with the increase of SOE-percentage. The ratio of short-term debt and long-term debt is calculated by short and long-term bank borrowings divided by total assets. The ratio of long-term debt is relatively low for both SOEs and Non-SOEs. Again, strikingly, the ratio of short term debt is much lower for state-controlled firms than non-state-controlled firms.
Table B: Descriptive statistics (2008-2014)
Note: This table looks at the descriptive statistics for the full sample from fiscal year 2008 to 2014. Column (1)
SOE percentage N p10 p25 p50 p75 mean sd
(1) (2) (3) (4) (5) (6) (7) 2007 1,386 23.2% 33.7% 46.1% 57.3% 46.1% 0.17 2008 1,504 14.0% 26.9% 41.0% 54.9% 40.9% 0.20 2009 1,600 0% 0% 25.5% 51.6% 29.6% 0.28 2010 1,930 0% 0% 22.3% 64.9% 31.5% 0.31 2011 2,191 0% 0% 21.3% 63.5% 31.0% 0.31 2012 2,353 0% 0% 19.1% 59.0% 29.4% 0.30 2013 2,330 0% 0% 11.1% 44.6% 23.3% 0.26 2014 2,369 0% 0% 11.0% 40.6% 22.3% 0.26 2015 2,549 0% 0% 16.0% 43.3% 24.9% 0.27 2008-2014 14,277 0% 0% 23.4% 53.4% 29.0% 0.28 (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Median Mean Median Mean Median Mean Median Mean Median Mean
Ln(assets) 21.504 21.631 21.784 21.849 21.689 21.765 21.301 21.448 21.223 21.423
ROA 0.036 0.039 0.027 0.033 0.029 0.035 0.032 0.036 0.052 0.053
Short-term debt ratio 0.007 0.102 0.089 0.118 0.087 0.113 0.082 0.108 0.029 0.073
Long-term debt ratio 0.00 0.039 0.004 0.047 0.001 0.041 0.00 0.037 0.00 0.029
Book leverage 0.134 0.161 0.186 0.193 0.163 0.179 0.137 0.161 0.056 0.111 Market leverage 0.057 0.094 0.086 0.116 0.072 0.105 0.059 0.093 0.021 0.063 Investment Ratio 0.463 1.023 0.368 0.918 0.466 1.025 0.467 1.005 0.683 1.207 EBIT/total assets 0.050 0.055 0.043 0.051 0.045 0.051 0.046 0.051 0.062 0.065 Cash/PPE 0.879 3.522 0.662 3.381 0.739 3.072 0.725 2.645 1.704 4.606 N 14,277 4,726 2,584 2,911 4,056
16
and (2) are descriptive statistics for the full sample, and column (3) to (10) are the descriptive statistics for different sample groups. Odd-numbered columns correspond to the median of the corresponding sample group, and even-numbered columns correspond to the mean of the sample. The explanations and constructions of variables are available in the appendix.
The book leverage is calculated by short-term borrowing plus long-term borrowing plus bonds payable, divided by total assets. And the market leverage is calculated by the same numerator divided by the market value of assets. As is shown in the descriptive statistics, the median book leverage is 16.1%, and the median market leverage is 9.4%. Both the book leverage and the market leverage are decreasing with the increase of the percentage of state-owned shares.
Regarding investment ratio, there is an obvious trend of increasing with the increase of state-owned shares within firms. The average percentage of investment ratio for firms with more than 50% of state-owned shares is 31.5% higher than the investment ratio of firms with less than 1% state-owned shares.
Interestingly, the state-owned firms appear to have strong preference towards hoarding cash. The average Cash/PPE ratio for firms in Group 4 is 36.2% higher than firms in Group 1, and the median Cash/PPE ratio is 157% higher, which probably could be explained by agency problems and precautionary preference (Bates et al.,2009) by management of state-owned enterprises.
17
Graph 3.2 shows that the number of listed firms increased dramatically from 2008 to 2014. In more developed provinces, the number of listed firms almost doubled, such as Guangdong Province, Jiangsu Province, Zhejiang Province and Beijing. In under developed provinces (provinces that are listed on the right-hand side of the graph), the number of listed firms also increased, but much more slowly than in developed provinces.
At the same time, the graph shows that more developed provinces suffer from lower government interventions compared to under developed provinces. For example, the number of listed firms in Guangdong province is 184 in 2008 and 361 in 2014, and the marketization index is 7.51 and 9.35 respectively. Contrastively, the number of listed firms in Xizang province is 8 and 9 in 2008 and 2014, and the marketization index is 1.36 and 0.62 respectively, the degree of government intervention increased in Xizang during this period.
Table C: Descriptive statistics for 10 provinces with lowest marketization index
Note: This table looks at the descriptive statistics for firms located in 10 provinces with lowest marketization index (highest government interventions) in 2008 and 2014. The 10 provinces are: Yunnan Province, Guizhou Province, Shanxi Province, Shaanxi Province, Hainan Province, Ningxia Province, Gansu Province, Xinjiang Province, Qinghai Province and Xizang Province. The explanations and constructions of variables are available in appendix.
Table C shows the summary statistics for 10 provinces with highest government interventions in 2008 and 2014. ROA for both years are lower than the ROA of full sample (Table B). Both the short-term debt ratio and the long-term debt ratio are
Median Mean s.d Median Mean s.d Ln(assets) 21.38 21.50 1.09 21.66 21.75 1.06 ROA 2.52% 2.85% 0.05 3.40% 3.96% 0.05 Short-term debt ratio 13.09% 14.63% 0.12 4.61% 8.34% 0.10 Long-term debt ratio 0.99% 5.15% 0.07 0.00% 2.50% 0.05 Book leverage 20.29% 20.91% 0.15 10.18% 13.45% 0.14 Market leverage 13.11% 14.87% 0.12 3.46% 7.04% 0.09 Investment ratio 0.31 0.59 0.89 0.78 1.65 1.82 Cash/PPE 0.48 1.59 3.88 1.02 4.48 7.76 EBIT/TA 4.31% 4.65% 0.06 4.83% 5.57% 0.05 N 2014 2,369 1,504 2008
18
higher than the relevant ratios in Table B, but both display a trend of decreasing as time goes by.
The book leverage and market leverage for firms located in above provinces are much higher than the full-sample average. For example, the average book leverage for firms located in those 10 provinces is 20.91%, which is 29.9% higher than the average book leverage for the full sample, indicating that firms located in provinces with higher government interventions have more debts on average. Additionally, the average book leverage and average market leverage decreased significantly, by 35.66% and 52.64%, from 2008 to 2014.
The investment ratio sees a trend of rocketing from 2008 to 2014 for the 10 provinces, which reflects the progress of the policy of “Stimulating the development of western regions in China”, implemented by the Central Government. The decision of implementing the program was passed in 2000, and the Central Government passed the “Eleventh Five-Year Plan for Western Development” in 2007. The scope of the policy includes 8 provinces out of the 10 provinces mentioned above (excluding Shanxi Province and Hainan Province). A series of preferential policies and subsidies are implemented to the western region of China, together with a series of encouraging policies regarding foreign direct investment. As a result, the investment opportunities rocketed in this period, and the investment ratio of firms located in these provinces therefore increased greatly.
The cash ratio also sees a trend of increase during this period, probably due to the increase of R&D expenses (Bates et al., 2009), the precautionary purpose of firms (Bates et al., 2009), or the increase of firm profitability and government subsidies.
4. Methodology
Based on rationale and backgrounds discussed in above sections, I will focus on the financing and investment behaviors of listed firms in mainland China.
19 on debt financing.
Utilizing the attributes which are proposed by Titman and Wessels’s (1988) that determine capital structures, I build the regression model that combine the determinants of capital structure as well as government interventions as follows.
The basic regression equation is:
𝐿𝐿𝐿𝐿𝐿𝐿𝑖𝑖,𝑡𝑡 = 𝛼𝛼 ∗ 𝑀𝑀𝑀𝑀𝑀𝑀𝑗𝑗,𝑡𝑡+ 𝛽𝛽1∗ 𝑇𝑇𝑇𝑇𝑖𝑖,𝑡𝑡+ 𝛽𝛽2∗ 𝑀𝑀𝑀𝑀𝑀𝑀𝑗𝑗,𝑡𝑡× 𝑇𝑇𝑇𝑇𝑖𝑖,𝑡𝑡+ 𝛾𝛾1∗ 𝑅𝑅𝑅𝑅𝑇𝑇𝑖𝑖,𝑡𝑡+ 𝛾𝛾2
∗ 𝑀𝑀𝑀𝑀𝑀𝑀 × 𝑅𝑅𝑅𝑅𝑇𝑇𝑖𝑖,𝑡𝑡+ 𝛿𝛿 ∗ 𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝑛𝑛+ 𝜋𝜋 ∗ 𝐶𝐶𝐺𝐺𝐶𝐶𝑀𝑀𝐺𝐺𝐺𝐺𝐶𝐶𝑖𝑖,𝑡𝑡+ 𝜀𝜀𝑖𝑖,𝑡𝑡
The dependent variables are book leverage and market leverage. There exist disagreements of whether to use book leverage or market leverage to measure the corporate capital structures. Myers (1977) maintains that corporate debt is better supported by assets that it is by firm’s growth opportunities. Additionally, supporters of book leverage argue that market leverage is not reliable as it may fluctuate fiercely with other factors. However, many scholars prefer market-based leverage than book leverage for various reasons. For example, Welch (2004) argue that book leverage is just a number to balance the financial statement and to keep the left and right-hand side of the balance sheet equal, thus should not be used as a financial indicator to make managerial decisions. In this thesis, I will regress with both market and book leverage and compare the results.
𝑀𝑀𝑀𝑀𝑀𝑀𝑗𝑗,𝑡𝑡 is the main variable of interest, which is an index developed by Fan et al.
(2016) to measure the level of marketization in province j in year t. It is used as a proxy for government interventions. As it is an index to measure the level of marketization of a province, higher the marketization index, higher the level of marketization, which means the province suffers from lower government interventions. If the indirect government interventions, alongside with other factors argued by Titman and Wessels (1988), indeed influence the capital structures of firms, the coefficient would appear significantly negative.
Moreover, I used two interaction terms to further measure the indirect government interventions. The first one is the interaction term between marketization index
20
(Fan et al., 2016) and the asset tangibility, 𝑀𝑀𝑀𝑀𝑀𝑀𝑗𝑗,𝑡𝑡× 𝑇𝑇𝑇𝑇𝑖𝑖,𝑡𝑡, to investigate the effects of the government interventions on the importance of collateral when lending from banks. Normally, tangible assets are used as collateral for bank loans, and higher the assets tangibility, easier it is for firms to borrow money from banks. However, in high government intervened regions, even if firms do not have enough tangible assets to use as collateral, they can still borrow money from banks. Thus, if the hypothesis is true, the coefficient of this term is expected to be negative. Another one is 𝑀𝑀𝑀𝑀𝑀𝑀 × 𝑅𝑅𝑅𝑅𝑇𝑇𝑖𝑖,𝑡𝑡, the interaction between marketization index (Fan et al., 2016) and firm profitability. According to pecking-order theory, profitable firms would prioritize internal financing when they intend to make new investments. Thus, the profitability of a firm is supposed to be negatively correlated with its leverage. Based on this intuition, the interaction term examines whether firms in high government-intervened provinces would borrow a lot of bank loans even if they are profitable enough, as political connection is considered valuable for many firms (Fisman, 2001). If this effect exists, the coefficient should be significantly positive.
To examine the direct government involvement, I adopt the group dummies into the regressions. There are four groups, representing firms with less than 1%, 1% to 25%, 25% to 50% and more than 50%, of state-owned shares, respectively. If the direct political connections would influence firm leverage, the coefficient of the group dummies should be statistically and economically significant.
𝐶𝐶𝐺𝐺𝐶𝐶𝑀𝑀𝐺𝐺𝐺𝐺𝐶𝐶𝑖𝑖,𝑡𝑡 includes main factors that are believed to influence leverage includes
(Titman and Wessels, 1988; Eckbo, 2009): firm size, tangibility of assets, growth, profitability, industry median debt ratios and so on. Here I use the tangibility of assets, profitability, growth opportunity, industry median debt ratio and firm size as control variables.
Hypothesis2: the government interventions boost firms’ investments.
21
(2017), I adopt the model as follows:
𝐼𝐼𝐶𝐶𝐿𝐿𝑖𝑖,𝑡𝑡 = 𝛼𝛼 ∗ 𝐼𝐼𝐶𝐶𝐿𝐿𝑖𝑖,𝑡𝑡−1 + 𝛽𝛽 ∗ 𝑀𝑀𝑀𝑀𝑀𝑀𝑗𝑗,𝑡𝑡+ 𝛿𝛿 ∗ 𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝑛𝑛+ 𝛾𝛾 ∗ 𝐶𝐶𝐺𝐺𝐶𝐶𝑀𝑀𝐺𝐺𝐺𝐺𝐶𝐶𝐶𝐶 + 𝜀𝜀
The dependent variable is investment ratio, following Chaney et al. (2012). The main variable of our interest is again 𝑀𝑀𝑀𝑀𝑀𝑀𝑗𝑗,𝑡𝑡, the marketization index (Fan et al., 2016). If the government involvement has effects on corporate investment decisions, the coefficient of the index should be significantly negative, which means firms suffer from higher government intervention (lower index) tend to invest more.
The group dummies look at the effects of direct government involvements on investment decisions. If firms with higher state-owned percentage maintain significantly higher investment ratio, the second hypothesis cannot be rejected.
𝐶𝐶𝐺𝐺𝐶𝐶𝑀𝑀𝐺𝐺𝐺𝐺𝐶𝐶𝐶𝐶 includes growth opportunity, firm size, ratio of cash to PPE (Chaney et al., 2016) and province-mean investment ratios. I use the province-mean investment ratios as control variables to account for other regional effects. In highly developed provinces, such as Beijing and Shanghai, the regional economic environment is better built than less developed provinces, and the investment opportunities are more accessible. Thus, on average, firms located in more developed regions are expected to invest more. To reduce the omitted variable bias, I include the control variable of province-mean investment ratios to reflect the regional economic development. As is shown in previous sections, the number of listed firms in varies by provinces, in more developed provinces, the number of listed firms can be 10 times more than less developed provinces. In this situation, I think province-mean investment ratio would be more representative than province-median investment ratio.
Hypothesis3: higher local government interventions result in worse performances.
For the investment performance, I implement a similar methodology as in hypothesis 1 and 2, to investigate the effects of government interventions on firm performance.
22
𝑃𝑃𝐿𝐿𝐺𝐺𝑃𝑃𝐺𝐺𝐺𝐺𝑃𝑃𝑃𝑃𝐶𝐶𝑃𝑃𝐿𝐿𝑖𝑖,𝑡𝑡 = β ∗ 𝑀𝑀𝑀𝑀𝑀𝑀𝑗𝑗,𝑡𝑡+ 𝜎𝜎 ∗ 𝑃𝑃𝐿𝐿𝐺𝐺𝑃𝑃𝐺𝐺𝐺𝐺𝑃𝑃𝑃𝑃𝐶𝐶𝑃𝑃𝐿𝐿𝑖𝑖,𝑡𝑡−1 + 𝜋𝜋 ∗ 𝐼𝐼𝐶𝐶𝐿𝐿𝑖𝑖,𝑡𝑡−1
+𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝑛𝑛+ 𝛾𝛾 ∗ 𝐶𝐶𝐺𝐺𝐶𝐶𝑀𝑀𝐺𝐺𝐺𝐺𝐶𝐶𝐶𝐶 + 𝜀𝜀
The dependent variable is firm performance, calculated by EBIT divided by total assets. The independent variable of our main interest is the marketization index (Fan et al., 2016). If the arguments of Shleifer and Vishny (1994) and Sapienza (2004) on the political view is true for Chinese market, that state-owned enterprises sometimes invest in projects for purposes other than profit, the coefficient of the index is expected to be positive, as lower marketization index means higher government interventions.
The group dummies are used to examine the effects of direct government intervention.
The control variables are lagged performance, lag investment and firm size. Of course, group dummies are again included in the regressions.
In all the regressions, I control for three types of fixed effects. The firm fixed effects control for unobserved, time-invariant differences across firms, the year fixed effects control for unobserved, time-varying fixed effects, and the province fixed effects control for unobserved, time-invariant differences across provinces.
5. Regression results
5.1 Leverage5.1.1 Book leverage
Table 5.1.1 shows the regression output for book leverage. Column (1) reports the result of the basic regressions with the marketization index, and column (2) is the regression with the interaction term between the government interventions and assets tangibility. Column (3) is the regression with the interaction term between profitability (ROA) and government interventions. And Column (4) is the regression with all the above independent variables, as
23
well as the group dummies.
Table 5.1.1: Effects of Government Interventions on Book Leverage
The table looks at how government interventions(GI) affects book leverage between 2008 and 2014. Column (1) is the basic regressions including the marketization index as the main independent variable. Column (2) includes the interaction term between marketization index and asset tangibility. Column (3) includes the interaction term between marketization index and ROA. Column (4) is the regression with all the above factors as well as group dummies. The control variables are Asset tangibility, ROA, Growth opportunities, firm size and industry-median debt ratio. For the detailed explanation of the control variables, please refer to the appendix. All regressions include firm fixed effects and year fixed-effects, robust standard errors are in parentheses. The standard errors are clustered at firm-level. Constants were included in the regressions but are not reported. ***, ** and * indicates significance at 1%, 5% and 10%, respectively.
The coefficient for the marketization index is negative for all the four regressions, means there exists negative correlations between the marketization index and firms’ book leverage. However, none of the coefficients for the four regressions are significant.
(1) (2) (3) (4)
VARIABLES GI GI × TA GI × ROA Group dummy Marketization Index -0.0007 -0.0003 -0.0011 -0.0011
(0.003) (0.003) (0.003) (0.003) Asset Tangibility 0.1205*** 0.2021*** 0.1206*** 0.1897***
(0.013) (0.044) (0.013) (0.044) Gov × Asset Tangibility -0.0124** -0.0133**
(0.006) (0.006) ROA -0.4827*** -0.4856*** -0.5581*** -0.5083*** (0.032) (0.032) (0.111) (0.109) Gov × ROA 0.0115 0.0066 (0.016) (0.015) Group 1 dummy 0.0470*** (0.004) Group 2 dummy 0.0294*** (0.004) Group 3 dummy 0.0185*** (0.003) Group 4 dummy -Growth opportynity -0.0030** -0.0031** -0.0031** -0.0024** (0.001) (0.001) (0.001) (0.001) Size 0.0412*** 0.0413*** 0.0413*** 0.0481*** (0.004) (0.004) (0.004) (0.004) Industry median debt ratio 0.3291*** 0.3276*** 0.3288*** 0.3035***
(0.049) (0.049) (0.049) (0.049)
Observations 14,151 14,151 14,151 14,151
R-squared 0.792 0.792 0.792 0.797
Firm FE YES YES YES YES
Year FE YES YES YES YES
24
The coefficient for assets tangibility is significantly positive for the four regressions, all at 1% significance level. The intuition behind is very clear and simple, that firms with higher tangible assets could offer more collaterals while borrowing from banks, thus they normally maintain higher leverage. Based on this intuition, the coefficient for the interaction term between tangible assets and government interventions are also intuitive. If government interventions indeed have effects on banks’ lending decisions, the listed firms may have easy access to bank loans regardless of collateral values. And the coefficient for this interaction term support our hypothesis: the coefficient for column (2) and column (4) are -0.012 and -0.013, both significant at 5%. Intuitively speaking, the effects of government interventions weaken the importance of collateral when firms are seeking bank loans. The role of government between the relationship of enterprises and banks could be viewed as guarantor, though not in the form of paper documents. Listed enterprises understand that the government would rescue them when they are at the risk of default, either through direct government subsidy, or through pushing the banks to offer some exemption clauses or extension terms. From the perspective of banks, they are also “willing to” lend money to listed firms, especially in less developed provinces. The first reason is that the government is a reliable “guarantor”, the banks also understand that the government would rescue the firms while they are at the risk of default. At the same time, the listed firms are normally the “breadwinner” of local economy, thus they are regarded as more competitive borrowers than most private firms and small enterprises. Poorer the province, smaller the number of listed firms, stronger the motivations for local governments to intervene local banks.
The coefficients of profitability (ROA) are significantly negative for the four regressions. This result is believed to be a proof of pecking-order theory (Myers and Majluf, 1984). According to Myers and Majluf (1984), firms will prioritize internal financing before they turn to external financing, thus the
25
coefficient for ROA is negatively correlated with net leverage. However, the order of priority would probability change when the government is involved. The effect of government involvement is reflected in the interaction term. The coefficient of the interaction term for column (3) is positive, showing a reversed effect. However, it is not significant at 10%.
The coefficient for group dummies in column (4) indicates that firms with higher percentage of government ownership appear to maintain a lower leverage, compared to private firms and firms with lower state-owned shares. This result is contrary to the result of Shao, Hernández, and Liu (2014), that government ownership has a significant positive relationship with leverage. The regression output indicates that the effects of indirect government interventions (through marketization index) on leverage are mitigated by the effects of direct government interventions (through state-ownership).
5.1.2 Market Leverage
Table 5.1.2: Effects of Government Interventions on Market Leverage
The table shows how government interventions(GI) affects firms' market leverage between 2008 and 2014. Column (1) is the basic regression, using the marketization index as the main independent variable. Column (2) includes the interaction term between marketization index and asset tangibility. Column (3) includes the interaction term between marketization index and ROA. Column (4) is the regression with all the above factors as well as group dummies. Control variables includes asset tangibility, ROA, growth opportunities, firm size and industry-median debt ratio. For the detailed explanation of the control variables, please refer to the appendix. All regressions include firm fixed effects and year fixed-effects, robust standard errors are in parentheses. The standard errors are clustered at firm-level. Constants were included in the regressions but are not reported. ***, ** and * indicates significance at 1%, 5% and 10%, respectively.
26
Table 5.1.2 is the regression output for market leverage. The coefficient for marketization index for the first two columns are significantly negative, at 10%. One unit decrease of marketization index is correlated with 0.3% increase of market leverage. For firms operating in western China, where the marketization index is generally much lower than firms operating in eastern China, the effects of government interventions on market leverage is significant. For example, the marketization index of Shanghai is 9.77 in 2014, and the marketization index for Qinghai Province is 2.53, the difference is 7.24. If we take the product of the difference of marketization index and the coefficient derived from the regression, the corresponding difference of market leverage caused by indirect government intervention would be 2.17%. As a reference, the actual average market leverage for firms in Qinghai Province is
(1) (2) (3) (4)
VARIABLES GI GIxTA GIxROA Group dummy
Marketization Index -0.0033* -0.0034* -0.0030 -0.0033 (0.002) (0.002) (0.002) (0.002) Asset Tangibility 0.0629*** 0.0565** 0.0629*** 0.0508*
(0.008) (0.029) (0.008) (0.029)
Gov × Asset Tangibility 0.001 0.0001
(0.004) (0.004) ROA -0.3524*** -0.3522*** -0.2883*** -0.2651*** (0.018) (0.018) (0.066) (0.064) Gov × ROA -0.0098 -0.0114 (0.009) (0.009) Group 1 dummy 0.0296*** (0.002) Group 2 dummy 0.0179*** (0.002) Group 3 dummy 0.0121*** (0.002) Group 4 dummy -Growth opportynity -0.0072*** -0.0072*** -0.0071*** -0.0067*** (0.001) (0.001) (0.001) (0.001) Size 0.0399*** 0.0399*** 0.0398*** 0.0440*** (0.003) (0.003) (0.003) (0.003) Industry median debt ratio 0.2508*** 0.2509*** 0.2511*** 0.2362***
(0.031) (0.031) (0.031) (0.031) Observations 14,151 14,151 14,151 14,151
R-squared 0.828 0.828 0.828 0.831
Firm FE YES YES YES YES
Year FE YES YES YES YES
27
10.1%. Obviously, it brings huge effects.
However, neither of the two interaction terms are significant in the regression of market leverage.
5.1.3 Net leverage
In some empirical research on leverage, there is a practice of subtracting cash from debt to calculate the term of net leverage. It reflects the conventional view of “cash is negative debt” (Archarya, Almeida and Campello, 2005). Various scholars carry out research on this topic and hold contradicting views about this topic, including Opler, Pinkowitz, Stulz, and Williamson (1999), Archarya, Almeida and Campello (2005), etc. In Section 3.2, I show that the cash ratio for listed firms increase with the increase of direct government involvement. To test if the cash holding of firms will influence the result, I replace the dependent variable with net leverage.
The regression output is listed in Table A of Appendix II. Basically, the effects of marketization index on net leverage is not significantly different from zero. However, the interaction term between marketization index and asset tangibility shows that governments have significant influence on bank’s lending decisions, at 1%, which mitigate the importance of collateral. However, the interaction between government involvement and firm profitability is not significant from zero.
The effect of the direct form of government interventions is consistent with the result of book leverage, that higher degree of direct government involvement is correlated with higher net leverage. Contrary to the conclusion of Shao et al. (2015) that higher direct government involvement is correlated with lower level of cash holding. One possible explanation for this result is the agency problem in state-owned enterprises.
28
Table 5.2 looks at how government interventions affect investments between fiscal year 2008 and 2014. Column (1) measures the indirect effects from the government, measured by marketization index. The coefficient for the marketization index is negative, but it is not significant at 10%. There are not significant correlations between indirect government interventions and firm’s investment ratio.
Table 5.2: Effects of Government Interventions on Investments
The table shows how government interventions(GI) affects firms' investment between 2008 and 2014. Column (1) is the regression of indirect effects, using the marketization index as the main independent variable. Column (2) is the “direct” effects, by using the group dummies as main independent variables. Control variables includes lag investment ratio, growth opportunities, firm size and province-mean investment ratio. For the detailed explanation of the control variables, please refer to the appendix. All regressions include firm fixed effects and province fixed-effects, robust standard errors are in parentheses. The standard errors are clustered at firm-level. Constants were included in the regressions but are not reported. ***, ** and * indicates significance at 1%, 5% and 10%, respectively. (1) (2) VARIABLES lag investment 0.0649*** 0.0605*** (0.017) (0.017) marketization index -0.0088 (0.040) Growth opportunity 0.0375** 0.0343* (0.019) (0.019) Ln(assets) 0.4189*** 0.3941*** (0.056) (0.055) Cash/PPE 0.0336*** 0.0327*** (0.004) (0.004) Province-mean investment ratio 0.9118*** 0.9006***
(0.085) (0.084) Group 1 dummy 0.0102 (0.041) Group 2 dummy -Group 3 dummy 0.0681 (0.044) Group 4 dummy 0.2452*** (0.052) Observations 11,484 11,484 R-squared 0.505 0.507
Firm FE YES YES
Province FE YES YES
Year FE YES YES
29
Column (2) measures the direct effects from the government, by holding significant shares in the listed firm. The group dummy variables represent different percentage of SOE shares in the firm. Table 5.2 shows that the coefficient for group dummies are basically in a rising trend. However, the coefficients for Group 1 to Group 3 are not significant. Especially, the coefficient for Group 4, which is the coefficient of firms with 50% and more state-owned shares, is significant both economically and statistically. It indicates that the investment ratio for firms with more than 50% SOE shares is on average 0.245 higher than that of other firms, ceteris paribus.
5.3 Investment Performance
Table 5.3 investigates the effects of government intervention on corporate performance. Firm performance is proxied by EBIT over total assets. The regression output indicates that one unit decrease of marketization index is correlated with 0.003 decrease of EBIT/total assets, but the effects are weakened when the firms are non-SOEs or when they have lower percentage of state-owned shares (firms in Group 1, 2 or Group 3). The effects are significant both statistically and economically.
Table 5.3: Effects of Government Interventions on Investment Performance
The table shows how government interventions(GI) affects firms' performance between 2008 and 2014. Column (1) is the basic regression, using the marketization index as the main independent variable. Column (2) is the regression with group dummies. Control variables includes lag performance, lag investment ratio, firm size and province-median performance ratio. For the detailed explanation of the control variables, please refer to the appendix. All regressions include firm fixed effects, year fixed effects and province fixed-effects, robust standard errors are in parentheses. The standard errors are clustered at firm-level. Constants were included in the regressions but are not reported. ***, ** and * indicates significance at 1%, 5% and 10%, respectively.
30
To measure the effects intuitively, we can take an example of Heilongjiang province and Shanghai in fiscal year 2014. The marketization index for Heilongjiang province is 6.22, and the marketization index for Shanghai is 9.77, so the difference of the index is 3.55. Thus, we can derive from the regression output that the average difference of performance (EBIT/total assets) due to the difference of marketization index for these two provinces is 1.2%. More intuitively, the average performance for firms located in Heilongjiang province has 0.012 lower EBIT/total assets than firms located in Shanghai, caused indirect government interventions. As a reference, the actual mean of EBIT/total assets for firms in Heilongjiang province is just 0.045 Obviously, it brings huge effects.
There are a lot other factors that would make the firms from those two provinces perform differently. To account for those factors, I control for three
(1) (2)
VARIABLES Basic regression Group dummies Marketization Index 0.0031* 0.0033**
(0.002) (0.002)
Lag performance 0.0370** 0.0346**
(0.015) (0.015) Lag investment ratio 0.0021*** 0.0019***
(0.000) (0.000) Ln(assets) -0.0071*** -0.0080*** (0.002) (0.002) Industry-median performance 0.8802*** 0.8709*** (0.079) (0.078) Group 1 dummy -0.0086*** (0.002) Group 2 dummy -0.0103*** (0.002) Group 3 dummy -0.0059*** (0.002) Group 4 dummy -Observations 11,542 11,542 R-squared 0.530 0.532
Firm FE YES YES
Province FE YES YES
Year FE YES YES
31
types of fixed effects in the regression. The firm fixed effects control for unobserved, time-invariant differences across firms, the year fixed effects control for unobserved, time-varying fixed effects, and the province fixed effects control for unobserved, time-invariant differences across provinces.
5.4 Robustness check
The potential endogeneity may arise from the unbalanced distribution of listed firms and the unbalanced economic development among different provinces. The marketization index not only reflect the level of government interventions, but also reflect the economy development and other factors that accompanying the degree of economic development in that province, such as the financial environment, the development of bond market, the perfection of financial market and legislations. When the firm performance increases with marketization index, it is hard to distinguish whether it is due to decreased government interventions, or due to the level of regional economic development (normally along with more investment opportunities). Moreover, the unbalanced distribution of listed firms among different provinces further strengthen the effects caused by unbalanced economic development. For example, around half of listed firms are from Guangdong province, Shanghai, Beijing, Zhejiang province and Jiangsu province (out of 31 provinces), while the above five provinces are among the list of top developed provinces in China. Although I add the control variables of province mean/median debt/investment ratio, it is difficult to totally rule out the effects.
To address the potential omitted variable bias mentioned above, I run the regressions based on a smaller sample, that exclude the 5 most developed provinces. In that way, we could to some extent relieve the effects brought by the level of local economy.
5.4.1 Leverage
32
The table looks at how government interventions(GI) affects book leverage for firms located in provinces of mainland China (excluding Guangdong, Jiangsu, Zhejiang, Shanghai, and Beijing) between 2008 and 2014. Column (1) is the basic regressions including the marketization index as the main independent variable. Column (2) includes the interaction term between marketization index and asset tangibility. Column (3) includes the interaction term between marketization index and ROA. Column (4) is the regression with all the above factors as well as group dummies. Control variables includes Asset tangibility, ROA, Growth opportunities, Size and Industry-median debt ratio. For the detailed explanation of the control variables, please refer to the appendix. All regressions include firm fixed effects and year fixed-effects, robust standard errors are in parentheses. The standard errors are clustered at firm-level. Constants were included in the regressions but are not reported. ***, ** and * indicates significance at 1%, 5% and 10%, respectively.
The regression output shows that excluding those five provinces brings significant impact to the results. The coefficient for the marketization index becomes much more significant than before. For example, for column (1), the coefficient of marketization index is -0.011, and it is significant at 5%. However, the regression coefficient is -0.001 for the basic regression in Table
(1) (2) (3) (4)
VARIABLES GI GIxTA GIxROA Group dummy
Marketization Index -0.0107** -0.0095* -0.0111** -0.0097* (0.005) (0.005) (0.005) (0.005)
Asset Tangibility 0.1190*** 0.2041*** 0.1191*** 0.1779*** (0.017) (0.061) (0.017) (0.062)
Gov × Asset Tangibility -0.0145 -0.0129
(0.010) (0.010) ROA -0.4843*** -0.4857*** -0.5508*** -0.4816*** (0.043) (0.043) (0.134) (0.131) Gov × ROA 0.0113 0.0034 (0.021) (0.021) Group 1 dummy 0.0217*** (0.004) Group 2 dummy -Group 3 dummy -0.0109** (0.004) Group 4 dummy -0.0305*** (0.005) Growth opportynity -0.0045*** -0.0045*** -0.0046*** -0.0036** (0.002) (0.002) (0.002) (0.002) Size 0.0333*** 0.0333*** 0.0334*** 0.0420*** (0.006) (0.006) (0.006) (0.006)
Industry median debt ratio 0.3627*** 0.3660*** 0.3630*** 0.3473*** (0.059) (0.059) (0.060) (0.060)
Observations 7,287 7,287 7,287 7,287
R-squared 0.788 0.788 0.788 0.794
Firm FE YES YES YES YES
33
5.1.1. However, the coefficient for the interaction term between the marketization index and asset tangibility becomes insignificant.
It is a reasonable result. The five provinces that are excluded from the regression are provinces that suffer least from government interventions, and the financial environment and legal environment for those provinces are much better than other provinces.
The regression output for market leverage is similar with the result of book leverage. The coefficient for marketization index becomes more significant after excluding sample from the five provinces. Please refer to the Table B of Appendix II for the regression output of market leverage.
5.4.2 Investment
Table 5.4.2 looks at the effects of government interventions on investments, after excluding the samples from the five provinces. For the remaining sample, the investment ratio is still not significantly correlated with the marketization index. Overall speaking, there is not a significant correlation between indirect government interventions and firm’s investment ratio. The intuition behind the result may be partially explained by the “Eleventh Five-Year Plan for Western Development” implemented since 2007. The Central Government implemented a series of economy stimulus policies in western China. Although the overall intensity of indirect government interventions in western China, as is indicated by marketization index, remain strong, the investment ratio of firms operating in western part of China increased significantly because of the policies implemented.
Another interesting finding is that the regression coefficients for Group 1 to Group 3 dummies are significantly negative, which also display the mitigation effects of direct government interventions on indirect government interventions (Shao, Hernández, and Liu, 2015).
34
The table shows how government interventions(GI) affects firms' investment for firms located in provinces of mainland China (excluding Guangdong, Jiangsu, Zhejiang, Shanghai, and Beijing) between 2008 and 2014. Column (1) measures the indirect effects, using the marketization index as the main independent variable. Column (2) measures the direct effects with group dummies. Control variables includes lag investment ratio, growth opportunities, firm size and province-mean investment ratio. For the detailed explanation of the control variables, please refer to the appendix. All regressions include firm fixed effects and province fixed-effects, robust standard errors are in parentheses. The standard errors are clustered at firm-level. Constants were included in the regressions but are not reported. ***, ** and * indicates significance at 1%, 5% and 10%, respectively.
5.4.3 Investment Performances
Table 5.4.3 shows the effects of government interventions on firm performance after dropping samples from the five provinces. The results appear quite robust for the smaller sample group. The coefficient of marketization index for column (1) increased from 0.031 to 0.050, compared with Table 5.4.2, indicating that higher indirect government intervention is correlated with lower firm performance, and the effects are more
(1) (2) VARIABLES lag investment 0.0617*** 0.0570** (0.022) (0.023) marketization index -0.0043 (0.052) Growth opportunity 0.0494** 0.0456** (0.022) (0.022) Ln(assets) 0.4286*** 0.3968*** (0.061) (0.059) Cash/PPE 0.0549*** 0.0534*** (0.009) (0.009) Province-mean investment ratio 0.8621*** 0.8516***
(0.096) (0.094) Group 1 dummy -0.2313*** (0.063) Group 2 dummy -0.2502*** (0.066) Group 3 dummy -0.1616*** (0.059) Group 4 dummy -Observations 6,155 6,155 R-squared 0.502 0.505
Firm FE YES YES
Year FE YES YES
Province FE YES YES
35
overwhelming in less-developed provinces.
Table 5.4.3:
Effects of Government Interventions on Investment Performance
The table shows how government interventions(GI) affects firms' performance for firms located in provinces of mainland China (excluding Guangdong, Jiangsu, Zhejiang, Shanghai, and Beijing) between 2008 and 2014. Column (1) is the basic regression, using the marketization index as the main independent variable. Column (2) is the regression with group dummies. Control variables includes lag performance, lag investment ratio, firm size and province-median performance ratio. For the detailed explanation of the control variables, please refer to the appendix. All regressions include firm fixed effects, year fixed effects and province fixed-effects, robust standard errors are in parentheses. The standard errors are clustered at firm-level. Constants were included in the regressions but are not reported. ***, ** and * indicates significance at 1%, 5% and 10%, respectively.
6. Concluding Remarks
This paper contributes to the literature of how government interventions affect firm’s financial policies, investment decisions and the corresponding performance.
(1) (2)
VARIABLES Basic regression Group dummies Marketization Index 0.0050** 0.0050**
(0.002) (0.002)
Lag performance 0.0189 0.0167
(0.021) (0.021) Lag investment ratio 0.0031*** 0.0029***
(0.001) (0.001) Ln(assets) -0.0025 -0.0038 (0.003) (0.003) Industry-median performance 0.9742*** 0.9595*** (0.093) (0.092) Group 1 dummy 0.0012 (0.002) Group 2 dummy -Group 3 dummy 0.0021 (0.002) Group 4 dummy 0.0124*** (0.003) Observations 7,287 7,287 R-squared 0.530 0.532
Firm FE YES YES
Province FE YES YES
Year FE YES YES
36
By analyzing two different forms of government interventions, several findings are reached: 1) Similar to the findings of Shao, Hernández, and Liu (2014), the degree of indirect government interventions are positively correlated with firm leverage, by affecting banks’ lending decisions. In addition to Shao et al. (2014), I noticed that the indirect government interventions could weaken the importance of collateral value for obtaining bank loans. The effects are specifically strong and robust in sample firms operating in less developed provinces. 2) Firms suffer from higher indirect government interventions have worse performance, and the effects are more significant in sample firms operating in less developed provinces. 3) I did not find significant correlations between the indirect government interventions and firm’s investment decisions, like Shao et al. (2015) did, not even in sample groups that suffer higher indirect government interventions. 4) Sometimes, the direct form of government interventions has some “mitigation” effects on indirect government interventions. The potential endogeneity may arise from the unbalanced distributions of listed firms, as well as the unbalanced regional economic development. Excluding the sample firms from top developed provinces helps to relieve the potential endogeneity problem.
The paper contributes to the literature of corporate capital structure and investment behaviors, and it supports the argument that government intervention is another type of friction that drives the state-owned firms, as well as firms that suffer from indirect government interventions, away from the optimal capital structure and investment behaviors (Chen et al., 2011). Moreover, the paper shed light on the growing literature of political connections and bank financing, by showing that firms suffer from higher indirect government involvement have easier access to bank loans (Sapienza, 2004; Faccio et al., 2006; Leuz and Oberholzer-Gee, 2006; Fan et al., 2007; Chen et al., 2010).
Additionally, some findings of this paper support the “development view” of Alexander Gerschenkron (1962), that government is playing an irreplaceable role in some strategic economic sectors, including banking systems. Although not
37
empirically tested directly, the results of the paper may be partially explained by political view (Shleifer and Vishny, 1994) and agency-cost view (Banerjee, 1997; Hart et al., 1997): Politicians would seek for political achievements, e.g. the reduced unemployment rate, the increased GDP, the completion of large social projects etc. when making political decisions (political view). Local governments of western provinces of China are extremely concerned with those targets, resulting in higher government interventions and stronger motivations to intervene banks’ lending decisions. The government involvement would inevitably lead to some inefficiency and misallocation of resources, but it may probably be the best way out for those provinces at a certain stage, where the private sectors are “not sufficiently developed enough to play the crucial development role” (Laporta, Lopez-de-Silanes and Shleifer, pp.265, 2002).
From a practical perspective, it could be argued that the results may serve as additional supporting to implement marketization reform not only in mainland China, but also in other countries that suffer relatively high degree of government interventions, both directly and indirectly.
38
References
Almeida, H., Acharya, V. and Campello, M., 2006, Is Cash Negative Debt? A Hedging Perspective on Corporate Financial Policies. SSRN Electronic Journal.
Alexander Gerschenkron, 1962, Economic backwardness in historical perspective, a book of essays, Cambridge, Massachusetts: Belknap Press of Harvard University Press.
Atkinson, A.B., Stiglitz, J.E., 1980, Lectures on Public Economics. London, McGraw Hill.
Baker, M. and Wurgler, J., 2001, Market Timing and Capital Structure. SSRN Electronic Journal.
Banerjee, A. (1997). A Theory of Mis-governance. The Quarterly Journal of Economics, 112(4), pp.1289-1332.
Bates, T., Kahle, K. and Stulz, R., 2009, Why Do U.S. Firms Hold So Much More Cash than They Used To? The Journal of Finance, 64(5), 1985-2021.
Brennan, M. and Schwartz, E., 1984, Optimal Financial Policy and Firm Valuation. The Journal of Finance, 39(3), p.593.
Chen, S., Sun, Z., Tang, S. and Wu, D., 2011, Government intervention and investment efficiency: Evidence from China. Journal of Corporate Finance, 17(2), pp.259-271.
China State Council, 2007, The Eleventh Five-Year Plan for Western Development. Cull, R. and Xu, L., 2003, Who gets credit? The behavior of bureaucrats and state banks in allocating credit to Chinese state-owned enterprises. Journal of Development Economics, 71(2), pp.533-559.
Duchin, R., Ozbas, O. and Sensoy, B., 2010, Costly external finance, corporate investment, and the subprime mortgage credit crisis. Journal of Financial Economics, 97(3), pp.418-435.
Deng, L., Jiang, P., Li, S. and Liao, M., 2017, Government intervention and firm investment. Journal of Corporate Finance.
39
North-Holland.
Faccio, M., Masulis, R. and McConnell, J., 2005, Political Connections and Corporate Bailouts. SSRN Electronic Journal.
Fan Gang, Wang Xiaolu, Yu Jingwen, 2016, Marketization Index of China’s Provinces: Neri Report 2016. Beijing: Social Science Academic Press.
Fazzari, S., Hubbard, R., Petersen, B., Blinder, A. and Poterba, J., 1988, Financing Constraints and Corporate Investment. Brookings Papers on Economic Activity, 1988(1), p.141.
Ferrell, O. and Gresham, L., 1985, A Contingency Framework for Understanding Ethical Decision Making in Marketing. Journal of Marketing, 49(3), p.87.
Fischer, E., Heinkel, R. and Zechner, J., 1989, Dynamic Capital Structure Choice: Theory and Tests. The Journal of Finance, 44(1), p.19.
Fisman, R., 2001, Estimating the Value of Political Connections. American Economic Review, 91(4), pp.1095-1102.
Hart, O., 2003, Incomplete Contracts and Public Ownership: Remarks, and an Application to Public-Private Partnerships. SSRN Electronic Journal.
Hunt, S. and Vitell, S., 1986, A General Theory of Marketing Ethics. Journal of Macromarketing, 6(1), pp.5-16.
Hunter, H., Gerschenkron, A., 1963, Economic Backwardness in Historical Perspective. Russian Review, 22(3), 316.
Johnson, S. and Mitton, T., 2001, Cronyism and Capital Controls: Evidence from Malaysia. SSRN Electronic Journal.
Kane, A., Lee, Y. and Marcus, A., 1984, Earnings and Dividend Announcements: Is There a Corroboration Effect? The Journal of Finance, 39(4), p.1091.
Kraus, A. and Litzenberger, R., 1973, A State-Preference Model of Optimal Financial Leverage. The Journal of Finance, 28(4), p.911.
La Porta, R., Lopez-De-Silanes, F. and Shleifer, A., 2000, Government Ownership of Banks. The Journal of Finance, Vol 57, No.1, 265-301.
Leuz, C. and Oberholzer-Gee, F., 2003, Political Relationships, Global Financing and Corporate Transparency. SSRN Electronic Journal.
