MSc Finance
Quantitative Track
Master Thesis
Is corporate innovation efficiency affected by stock markets?
--Natural experiment based on step-by-step expansion of Margin Trading Policy
by
Jingyi Liang
11845430
July 2018
EC 15
Period 2-3
Statement of Originality
This document is written by Student Jingyi Liang 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.
Abstract
Previous research of the opening or restrict of short selling effect focus on stock market volatility and price discovering, and the research of stock share reform on corporate outcomes raises the question that whether short selling mechanism affects specific corporate outcome – innovation efficiency. In 2010, China’s Margin Trading Policy has launched, indicating short selling mechanism started its performance in mainland China. Using R&D/Total assets, R&D/Revenue and natural logarithm of company patent applying as proxies for innovation input and output, I investigate effect of the short selling mechanism implementation on the company's innovation efficiency, and its influencing process by Difference-in-Difference model. I find that implementation of short selling mechanism from Margin Trading Policy does not affect company innovation input significantly, while it does enhance company innovation output with significance level. Therefore, my finding indicates that the Margin Trading Policy has an “innovative incentive effect” and can significantly increase the company’s innovation efficiency.
Content
1. Introduction ... 1
2. Background and Literature Review ... 5
2.1 Margin Trading Policy ... 5
2.2 Literature Review ... 7
2.3 Hypothesis ... 9
2.3.1 Theory of Constraints ... 9
2.3.2 Theory of Pressure... 9
3. Methodology ... 10
3.1 Data and variables ... 10
3.2 Econometric Model & Variables ... 12
4. Data and descriptive statistics ... 13
5. Empirical findings ... 15
5.1 General empirical finding ... 15
6. Research conclusions and inspirations ... 26
1. Introduction
This paper examines the effect of short selling mechanism from Margin Trading Pilot Policy on corporate innovation. I focus on innovation input and output of target companies in Shanghai and Shenzhen Exchange (the biggest 2 exchange in China) during sample period 2007-2016. First, motivated by Jensen and Meckling (1976) that shareholders and management have an agency relationship, I hypothesize that innovation efficiency will be improved by short selling mechanism implementation under the theory of constraints. Second, motivated by Mitchell et al. (2004), He and Tian (2016) that introduction of short-selling mechanism will bring great short-term price pressure to the company's stock price, and short-sellers in the market inherently have very limited tolerance for short-term projects failure, I hypothesize that under the theory of pressure, introduction of the short selling mechanism will significantly reduce the company’s innovation efficiency. I use R&D divided by Revenue and Total assets as innovation input measurements, and use natural logarithm of patent applying as a proxy of innovation output. The implementation of short selling is step by step, which gives natural setting to test these hypotheses using Difference-in Difference model.
Given the background that on March 31, 2010, China officially launched a Margin Trading Pilot Policy, allowing qualified investors to finance from securities companies and buy securities, or to borrow securities and sell them (short selling mechanism), which is the investigated target in my article. From the perspective of theory, unlike developed capital markets around the world, China's margin trading policy has undergone a special process of promotion after the first chosen target company’s trial then following by step-by-step expansion of the underlying stocks. The introduction of the margin trading pilot policy has been provided to explore the short-selling mechanism. With the ideal natural experiment environment, the
gradually step-by-step expansion has provided excellent conditions for my research. From the perspective of practice, after the implementation of China's Margin Trading Pilot Policy, along with the opening of the financial restructuring policy and the improvement of corresponding regulations, the transaction scale of margin trading has significantly improved. From the beginning of the implementation in March 2010, the total market transaction volume was less than 7 million yuan. In Dec 2017, the transaction volume of the two main financial markets (Shenzhen Exchange & Shanghai Exchange) has reached an amount of 1,026.26 billion yuan (Figure1). The rapid development of short selling mechanism in China makes it important to clarify the impact of short-selling on corporate behavior, especially in innovation perspective.
From existing literature, the research of the short selling mechanism mainly focus on the company’s stock pricing efficiency, financial statements’ quality, and short-term effects of investment behavior, such as Li and Zhang (2015) studies the impact of the short-selling mechanism on stock pricing efficiency; Karpoff, Lou (2010), Hirshleifer et al. (2012), Chen et al. (2015) examine the impact of the short-selling mechanism on the quality of financial information. Grullon et al. (2015) examines the impact of the short selling mechanism on corporate investment behavior. Besides, Campello, Ribas and Wang (2014) examines the influence of stock markets on corporate outcomes performance by multiple institutional characters under the Chinese Conversion Program. He and Tian (2016) studies the impact of the implementation of SHO policies on corporate innovation within US market. They use the short selling restriction in United States as an exogenous event, exploring how much the impact of short selling mechanisms on corporate innovation output is. This article breakthroughs the limitations of short-term effects of inspections, going into the perspective of corporate innovation, linking the short selling mechanism with the long-term behavior of corporate innovation together, and systematically studies how the implementation of Margin Trading Pilot Policy affects the objective performance and value of corporate innovation.
As China's Margin Trading Pilot Policy has experienced the promotion procedure, the progress of the step-by-step expansion of the underlying stock has provided us with natural treatment groups (targeted listed stocks) and control groups (non-targeted listed stocks). Based on this, this paper adopts the experiment method of difference-in-difference (DID) based on the event of the implementation of China's Margin Trading Policy, and systematically examines how the short-selling mechanism affects the company's innovation and its value.
From the research, the finding support that with implementation of Margin Trading Policy, more and more listed companies are qualified to be short selling targets and these companies show significantly improvement in innovation efficiency. The research conclusions confirmed that the implementation of the margin trading policy is in line with the “the theory of constraints” rather than the “the theory of pressure”. The Chinese short-selling mechanism (margin trading policy) has a significant “innovative incentive effect”.
Comparing with the long-term history of short selling mechanism in European countries, short selling market in China has been underdeveloped from 2010 to 2018 by only less than 8 years. The corresponding regulations, theory, trading system and risk management are not mature enough, where there should be some improvement to upgrade.
The contributions of this research are, firstly, it expands the research perspective of the short selling mechanism. From the view of corporate innovation, this article analyzes and tests the short selling mechanism's effect on corporate innovation and its value mechanism. It studies the link between financial innovation and long-term corporate behavior, and analyzes changes in decision-making behaviors of management and shareholders under the short selling, providing a new way to test the value of short selling mechanism. Secondly, it enriches and supplemented research on corporate innovation. As for the determinants of corporate innovation, most of the previous literature research from company's internal characteristics such as ownership
structure (Wang and Zhao, 2015), corporate financial factors (Manso, 2011), executive characteristics (Hirshleifer et al., 2012) and corporate governance (Edererand Manso, 2013), but the present research lacks from the perspective of changing external policies to analyze the effect of corporate innovation behavior and its mechanism after policy environment changes. This article uses the “quasi-natural experiment” implemented by China’s Margin Trading Pilot Policy to thoroughly examine the effect and mechanism of short selling on company innovation. Research does not stop at intuitive and general analysis, but it is based on difference-in-difference. The research, after examining the effect of the Margin Trading Pilot Policy on the performance of the company's innovation generally, further analyzes and tests the influencing factors and the corresponding value-raising effect of the relationship between the two. Besides, I also divide the sample period for a better presentation of changing effects. (3) Provides policy enlightenment for the effect of the short selling mechanism. After the introduction of the Margin Trading Pilot Program in March 2010, there was a lot of controversy in both theoretical and practical issues. In particular, the stock market crash in China's capital market in the second half of 2015 pushed problems of short selling mechanism to the desk. Some people think that the short-selling mechanism is the criminal of the stock crisis in China.
The paper constructs as follows. I describe the literature review and policy background in Section 2. In Section 3, it follows the empirical methodology; and data and descriptive statistics are in Section 4. I deliver results to give economic meaning in Section 5 and conclude a final discussion in Section 6.
Figure 1 Margin Trading Scale Evolution in China Data source: CSMAR.
2. Background and Literature Review
2.1 Margin Trading Policy
Margin trading is defined that stock market investors pay a certain margin according to their quality and credit to brokers, who lend money for investors to buy securities or lends securities prior for them to sale later (short selling mechanism). This is integration in the financial markets. From a global perspective, the margin trading program is a basic credit trading system. The Margin Trading Policy is an inevitable arrangement for establishing short selling rules, since the lack of short selling mechanism will enable investors to only profit from stock price rising, which will lead to irrational rise in stock price and follow irrational decline. On March 30, 2010, the Shanghai Stock Exchange and the Shenzhen Stock Exchange issued announcements stating that they will formally open a margin trading system from March 31, 2010,
and officially start the Margin Trading Pilot Program. The introduction of short-selling mechanism ended the unilateral market era in China's securities market for more than 20 years. It is a milestone of the Chinese capital market. Though it is a new policy in China's securities market, Margin Trading Policy has a “high-risk and high-yield” characteristic. With leverage, it further enlarges the profit and loss of securities investment. Therefore, we should recognize risks exist indeed, such as, forced close out, investment loss enlargement, and cancelation of credit line, and we should master relevant risk by prevented methods.
After the program started, the underlying stocks of margin trading have undergone a step-by-step process of expansion. More and more stocks have been included in the scope of the underlying stocks of margin trading. As of December 31, 2016, the underlying stocks of margin financing and securities lending were changed as follows: On November 25, 2011, the Shanghai and Shenzhen stock exchanges announced the expansion of the scope of the underlying stocks for margin trading, the number of which has been adjusted from the original 90 to 278, mainly covering the constituent stocks of the Shanghai 180 Index and the Shenzhen 100 Index; On January 25, 2013, the stocks of the securities and securities lending companies of the two exchanges were expanded to 500; On September 16, 2013, the stocks of the underlying securities of the two exchanges were expanded to 700; On September 22 2014, the stocks of the securities and stock markets of the two exchanges were expanded to 900; On December 22, 2016, the expansion of the company’s stocks under margin trading was up to 950. Specific expansion events were shown in Table 1.
Table 1 Step-by-step expansion of target stocks under Margin Trading Policy
This table shows the expansion of target stocks under the process of Margin Trading Policy implementation. More and more stocks are included in the target stock pool of Margin Trading Policy. As of December 31, 2017, the stocks of the Margin Trading Policy were changed as follows: On November 25, 2011, the Shanghai and Shenzhen Stock Exchanges
announced the expansion of Margin Trading Policy. The scope of the target stock has been adjusted from the original 90 to 278, mainly covering the constituent stocks of the SSE 180 Index and the Shenzhen 100 Index. On January 25, 2013, the target securities of the two exchanges expanded to 500. On September 16, 2013, the target securities of the two exchanges expanded to 700. On September 12, 2014, the target securities of the two exchanges expanded to 900. The latest expansion was on December 12, 2016, the total target stocks reached 950. This table shows number of new-added target stock, number of eliminated target stocks, total target stocks, total A-shares stocks, and the proportion of target stocks.
Step by step expansion of stocks
Number of new-added target stock Number of eliminated target stocks Total target stocks Total A-shares stocks The proportion of target stocks (%) Implementation （2010/3/31） 90 90 1627 5.53 ⅠExpansion （2011/12/5) 189 1 278 1935 14.37 ⅡExpansion （2013/1/31) 222 0 500 2048 24.41 ⅢExpansion （2013/9/16) 206 6 700 2468 28.36 ⅣExpansion （2014/9/22) 205 5 900 2559 35.17 ⅤExpansion （2016/12/12） 77 27 950 3026 31.39
2.2 Literature Review
Most of papers available now research the effect on stock volatility and price efficiency from the perspective of macro. Henry and Mc Kenzie (2006) applies the
model using the short selling data from Hong Kong stock market and find increasing market volatility, and asymmetric impact after completion of short selling mechanism implementation. Zheng, Yan and Liu (2015) investigate the reason of stock market volatility within sample period 2005 to 2014, using GARCH and EGARCH models and dummy variables. They find a non-statistically significant effect on stock volatility though in a positive way. Woolridge and Dickinson (1994) find short selling mechanism are in positive correlation with stock price using data from New York Stock Exchange, Futures Exchange and Over the Counter. They show that short selling play a role of price discovery, encouraging the stock price to converge to its intrinsic value, instead of dropping the stock price. Saffi and Sigurdsson (2010) enlarge a research sample by using 26 countries stock markets and find that short selling mechanism on the one hand can achieve positive yield but on the other hand it cannot decrease extreme losses probability, which indicates an uncertain impact of short selling mechanism on stock market. Chang and Yao (2014) find that short selling mechanism implemented by Margin Trading Policy enhance price discovery efficiency and eliminate over-priced. The results also show Margin Trading Policy reduces stock market volatility by progressive implementation. Cai (2010) finds that short selling mechanism has a positive incentive to stock market volatility in Taiwan stock market, using VAR model. Figlewshki and Webb (1993) find that there exists a positive correlation between stock index and short selling trading volume, which indicates short selling mechanism, can enable a more stable stock market and decrease the stock volatility. He and Tian (2016) finds that short-sellers inherently have very low tolerance for short-term projects failure. Bogen and Krooss (1960) is the first paper proposes the correlation between stock market volatility and implementation of short selling mechanism. The research thinks if the stock price goes up, leading to a better expectation of the certain stock, then there is a better and larger performance in price. On the contrary, there is also an exacerbation when stock price declines by the leverage effect of short selling mechanism.
2.3 Hypothesis
2.3.1 Theory of Constraints
The constraint effect under the short-selling mechanism will improve the internal quality and basic characteristics of the company's innovation. (1) From the perspective of agency theory. Jensen and Meckling (1976) proposed that in the framework of the contract, shareholders and management have an agency relationship. In the real business decisions, the management often reveals moral hazard in the company’s operating and investment decisions. The introduction of the short selling mechanism can inhibit management's agency problems in innovation decisions. Firstly, the introduction of the short selling mechanism has strengthened the long-term incentives of management. Secondly, the introduction of the short-selling mechanism has increased the external supervision mechanism. When short sellers discover that the company’s management has lax mentality or deliberately avoid investing in high-risk innovation projects due to evasion of responsibilities, they will use the advantages of their own information diggers to actively arrange short selling of the company’s stocks. This leads to negative market reactions. At this time, short sellers as an effective external regulatory will force management to actively lay out high-quality, innovative projects aimed at enhancing the long-term value of the company for the purpose of maximizing the interests of the company rather than its own interests.
Hypothesis-A: If the short selling mechanism meets the “Theory of Constraints”, the
implementation of the short selling mechanism will significantly improve the company's innovation efficiency.
2.3.2 Theory of Pressure
The introduction of the short-selling mechanism will change the mode of management decision-making and will have a significant negative impact on the company's
innovation. Because the introduction of short-selling mechanism will bring great short-term price pressure to the company's stock price (Mitchelletal., 2004), this price pressure will make the management investment decision-making mode more emphasis on short-term projects and ignore long-term projects, so short-selling mechanism will strengthen management's short-term behavior and short-sighted issues. He and Tian (2016) finds that short-sellers inherently have very low tolerance for short-term projects failure. In the short term, the extremely high tolerance for failure and the long-term high motivation for success are the two most important factors in the success of a company's innovation (Manso, 2011). Therefore, under the pressure hypothesis, the company's management is forced by the price pressure of the short-selling mechanism, and will pay more attention to short-term stock prices and operational performance, and sacrifice and neglect the company's long-term value investment. Therefore, the following hypothesis is proposed.
Hypothesis-B: If the short selling mechanism meets the “Theory of Pressure”,
implementation of the short selling system will significantly reduce the company’s innovation efficiency.
3. Methodology
3.1 Data and variables
I use panel data and collect all data from CSMAR data base for all listed A-shares stocks both in Shanghai Exchange and Shenzhen Exchange to construct my samples. I collect R&D, Total assets, Revenue then construct R&D/Total assets and R&D/Revenue as measurements of innovation input indicators. Second, I collect patent of applying of every listed stock company during the whole sample period (omitting missing data) and use the natural logarithm of patent as proxy of innovation output. I also collect Total liabilities and Net profit as to make 3 control variables,
which are Leverage ratio, constructed by Total liabilities over Total assets; ROA, constructed by Net profit over Total assets; and natural logarithm of Total assets as proxy of firm size. My sample period is 2007-2016. Starting from 2007 is due to a new accounting standard starting in 2006. When I collect all data including total assets, total liabilities, revenue, R&D, net profits, patent, company stock code and industry code directly from CSMAR database, I construct the variables I need including natural logarithm of total assets as ln(size), R&D/total assets and R&D/revenue as innovation input proxies, natural logarithm of patent as innovation output dependent variables. I use Shanghai and Shenzhen Exchange official website to assess target companies of Margin Trading Policy by year and construct dummy variables LIST and POST. LIST is a dummy variable that indicates the whether the company is a target company implemented Margin Trading Policy, which means implemented the short selling mechanism when it equals to 1. POST is a dummy variable equals to1 after the years (including the year it is targeted) the company is targeted. LIST*POST is also a dummy variable and an interaction of LIST and POST (which is not included in this regression since I include year fixed effects), therefore it represents target companies after (include) the year they are implemented the policy when it equals to 1. I drop all financial companies in my samples since they are not substantial economy.
This article uses the Difference-in-Difference model to examine the impact of short-selling mechanism on company research and development input and innovation output. It uses the non-financial companies of target margin trading companies in the Shanghai and Shenzhen Stock Exchanges as the treatment group, and the non-financial companies that have not entered the margin trading list as the control group. Since the new accounting standards for listed companies in China were implemented in 2006, the starting point for this study was 2007, and the period from 2007 to 2017 was taken as the sample period. For sample companies, some of the observations are excluded based on the following criteria: (1) financial companies; (2) B-share companies; (3) other companies with missing financial and governance
variables. In addition, due to the need to combine datasets for patent and R&D expenditures in the research, missing data was deleted. All the financial data, stock trading data, margin trading data and patent data of this article are based from the CSMAR database.
3.2 Econometric Model & Variables
Based on existing literature, I adopt the following difference-in-difference model to examine the impact of the implementation of margin trading on corporate innovation input and output.
(1) (2) (3)
This article measures company innovation from two points of view: R&D spending and innovation output. It uses R&D expenditures accounted for total assets at the beginning of the period (RD_AT) and R&D expenditures accounted for revenue at the beginning of the year (RD_SALE). In the measurement of innovation output, this article is based from the references of Le Wenyu, and Zheng Manni (2016) and applies a indicator: LNPAT, which represents the logarithm of total number of patents applied plus one. LIST is a dummy variable for margin financing and short selling. When the stock of the company is included in the stocks of the financing and short selling margin during the sample period, it is set to 1; otherwise, it is set to 0. POST is the virtual variable after the company enters into the securities market. After the company enters the target list of margin trading (including the corresponding year), the company takes 1; otherwise, it is 0. This article focuses on the regression
coefficients β2 in model (1), (2) and model (3). If β2 in model (1), (2) are significantly positive, it indicates that the implementation of the margin trading has a significant positive impact on the company's innovation investment. If β2 in model (1), (2) are significantly negative, it indicates that the implementation of the margin trading has a significant negative impact on the company's innovation expenditure. Similarly, β2 in the model (3) is significantly positive, which indicates that the implementation of the margin trading program has a significant positive impact on the company's innovation output. If β2 in model (3) is significantly negative, it indicates that the implementation of the program has a significant negative impact on the company's innovation and output. In choosing control variables CV, this article controls leverage ratio (lev), firm size (lnsize) and return on assets (roa). is time fixed effect, is firm fixed effect and is industry fixed effect. is error term.
4. Data and descriptive statistics
I use panel data and collect all data from CSMAR data base for all listed A-shares stocks both in Shanghai Exchange and Shenzhen Exchange to construct my samples. I collect R&D, Total assets, Revenue then construct R&D/Total assets and R&D/Revenue as measurements of innovation input indicators. Second, I collect patent of applying of every listed stock company during the whole sample period (omitting missing data) and use the natural logarithm of patent as proxy of innovation output. I also collect Total liabilities and Net profit as to make 3 control variables, which are Leverage ratio, constructed by Total liabilities over Total assets; ROA, constructed by Net profit over Total assets; and natural logarithm of Total assets as proxy of firm size. And I use Stata to winsorize data to eliminate the extreme values before running the models. My sample period is 2007-2016. Starting from 2007 is due to a new accounting standard starting in 2006. When I collect all data including total assets, total liabilities, revenue, R&D, net profits, patent, company stock code and
industry code directly from CSMAR database, I construct the variables I need including natural logarithm of total assets as ln(size), R&D/total assets and R&D/revenue as innovation input proxies, natural logarithm of patent as innovation output dependent variables. I use Shanghai and Shenzhen Exchange official website to assess target companies of Margin Trading Policy by year and construct dummy variables LIST and POST. LIST is a dummy variable that indicates the whether the company is a target company implemented Margin Trading Policy, which means implemented the short selling mechanism when it equals to 1. POST is a dummy variable equals to1 after the years (including the year it is targeted) the company is targeted. LIST*POST is also a dummy variable and an interaction of LIST and POST (which is not included in this regression since I include year fixed effects), therefore it represents target companies after (include) the year they are implemented the policy when it equals to 1. I drop all financial companies in my samples since they are not substantial economy. Descriptive statistics table shows below.
Table 2 Descriptive statistics.
This table provides statistics for the sample of listed companies. R&D/Revenue measure defined as R&D divided by Revenue. Ln(Patent) is the measurement of innovation output in this article, in percentage. LEVERAGE is a control variable, defining total liabilities/total assets. Ln(Size) is the second control variable, represents natural logarithm of total assets. ROA is defined as net profit/total assets, the third control variable.
Variable N Mean Median Standard deviation
R&D/Revenue 10621 0.044562 0.035277 0.000408 R&D/Total assets 10613 0.022077 0.018843 0.000167 Ln(Patent) 10765 2.159489 2.197225 0.014237 Ln(Size) 10757 21.77345 21.58146 0.011385 Leverage 10757 0.378817 0.358964 0.001972 ROA 10757 0.043297 0.041643 0.000487
5. Empirical findings
5.1 General empirical finding
Table 3 Short Selling Mechanism and Corporate Innovation Input: Research & Development Expenditure / Total Assets
The table presents the results of regressions of research and development expenditure/total assets on short selling mechanism. Research & Development Expenditure/Total Assets is the first measurement of innovation input in this article, in percentage. LIST is a dummy variable that indicates the whether the company is a target company implemented Margin Trading Policy, which means implemented the short selling mechanism when it equals to 1. POST is a dummy variable equals to1 after the years (including the year it is targeted) the company is targeted. LIST*POST is also a dummy variable and an interaction of LIST and POST (which is not included in this regression since I include year fixed effects), therefore it represents target companies after (include) the year they are implemented the policy when it equals to 1. LEVERAGE is a control variable, defining total liabilities/total assets. LN(SIZE) is the second control variable, represents log(total assets). ROA is defined as net profit/total assets, the third control variable. Regression (1), (2), (3) all include year, industry and firm fixed effects. Regression (4) only includes year fixed effect. Standard errors are in parentheses.*, **, and *** measure significance at 10%, 5% and 1% level, respectively.
Dependent Variables: R&D / Total Assets (%) (1) (2) (3) (4) LIST 0.00344*** (0.000640) LIST*POST 0.000913** 0.000855** -0.00032 -0.000548 (0.000417) (0.000419) (0.000428) (0.000760) LEVERAGE 0.00575*** 0.00320*** -0.00326*** -0.00159 (0.00108) (0.00106) (0.00104) (0.00109) LN(SIZE) -0.00627*** -0.00595*** -0.00325*** (0.000289) (0.000289) (0.000196) ROA 0.0255*** 0.0614*** (0.00256) (0.00369) Observations 10,367 10,367 10,367 10,613 R-squared 0.845 0.843 0.835 0.085
Table 4 Short Selling Mechanism and Corporate Innovation Input: Research & Development Expenditure / Revenue
The table presents the results of regressions of research and development expenditure/total assets on short selling mechanism. Research & Development Expenditure/Total Assets is the first measurement of innovation input in this article, in percentage. LIST is a dummy variable that indicates the whether the company is a target company implemented Margin Trading Policy, which means implemented the short selling mechanism when it equals to 1. POST is a dummy variable equals to1 after the years (including the year it is targeted) the company is targeted. LIST*POST is also a dummy variable and an interaction of LIST and POST (which is not included in this regression since I include year fixed effects), therefore it represents target companies after (include) the year they are implemented the policy when it equals to 1. LEVERAGE is a control variable, defining total liabilities/total assets. LN(SIZE) is the
second control variable, represents log(total assets). ROA is defined as net profit/total assets, the third control variable. Regression (1), (2), (3) all include year, industry and firm fixed effects. Regression (4) only includes year fixed effect. Standard errors are in parentheses.*, **, and *** measure significance at 10%, 5% and 1% level, respectively.
Dependent Variables = R&D / Revenue (%)
(1) (2) (3) (4) LIST 0.00135 (0.00130) LIST*POST 0.000696 0.000901 0.000165 -0.00215 (0.000981) (0.000994) (0.000993) (0.00153) LEVERAGE -0.0233*** -0.0144*** -0.0204*** -0.0555*** (0.00255) (0.00251) (0.00240) (0.00224) LN(SIZE) -0.00438*** -0.00551*** -0.000835** (0.000681) (0.000686) (0.000421) ROA -0.0897*** -0.0630*** (0.00601) (0.00758) Observations 10,367 10,367 10,367 10,613 R-squared 0.855 0.851 0.850 0.391
In Table 3, model (1) provides the result of short selling mechanism on R&D/Total assets including all control variables leverage, ln(size), ROA and fixed effect industry, firm and time. In model (1), a standard deviation increase in LIST*POST causes 0.0913% significantly increase in dependent variable R&D/Total assets, though it is a very small amount of improvement. In Table 4, model (1) presents a result of short selling mechanism on R&D/Revenue including all control variables leverage, ln(size), ROA and fixed effect industry, firm and time. In model (2), a standard deviation increase in LIST*POST causes 0.0696% increase in dependent variable R&D/Revenue, which is not significant. As from the model setting, R&D/Total assets
and R&D/Revenue are 2 measurements of corporate innovation input. Results of model (1) presented in Table 3 and Table 4 indicate that the effect of Margin Trading Policy on innovation input is not significant.
Model (2), (3) and (4) in Table 3 and Table 4 are robustness checks of empirical equations. From (2) and (3), I omit 1 and 2 control variables respectively and omit firm fixed effect. In model (2) and (3), I find that model (2) in Table 3 has a lower amount of coefficient in the most important dependent variable POST*LIST, model (3) in Table 3 even has a negative insignificant coefficient. Model (2) and (3) in Table 4 are still insignificant. As for model (4), both coefficients of POST*LIST are negatively insignificant, and dependent variable LIST of model (4) are positive, which may be explained that firm fixed effect is a indispensable effect when I run this model. Besides, if I look at the R-squared in both Table 3 and Table 4, I find that both model (1) have the biggest R-squared. It can be concluded that, the power of explanation of model (1) is the largest and most convincible among all 4 models both in Table 3 and Table 4.
Table 5 Short Selling Mechanism and Corporate Innovation Output: Ln(Patent)
The table presents the results of regressions of log(number of patent) on short selling mechanism. Ln(Patent) is the measurement of innovation output in this article, in percentage. LIST is a dummy variable that indicates whether the company is a target company implemented Margin Trading Policy, which means implemented the short selling mechanism when it equals to 1. POST is a dummy variable equals to1 after the years (including the year it is targeted) the company is targeted. LIST*POST is also a dummy variable and an interaction of LIST and POST (which is not included in this regression since I include year fixed effects), therefore it represents target companies after (include) the year they are implemented the policy when it equals to 1. LEVERAGE is a control variable, defining total liabilities/total assets. LN(SIZE) is the second control variable, represents log(total assets). ROA is defined as net profit/total assets, the third control variable. Regression (1), (2), (3) all include year,
industry and firm fixed effects. Regression (4) only includes industry and year fixed effects. Standard errors are in parentheses.*, **, and *** measure significance at 10%, 5% and 1% level, respectively.
Dependent Variables = Ln(Patent)
(1) (2) (3) (4) LIST 0.0971* (0.0500) LIST*POST 0.0827* 0.0821* 0.124*** 0.0323 (0.0441) (0.0441) (0.0441) (0.0592) LEVERAGE -0.219* -0.252** 0.0451 -0.207** (0.115) (0.112) (0.107) (0.0872) LN(SIZE) 0.269*** 0.273*** 0.422*** (0.0307) (0.0306) (0.0163) ROA 0.332 3.653*** (0.272) (0.294) Observations 10,513 10,513 10,513 10,756 R-squared 0.757 0.757 0.755 0.245
In Table 5, model (1) provides the result of short selling mechanism on natural logarithm of patent including all control variables leverage, ln(size), ROA and fixed effect industry, firm and time. In model (1), a standard deviation increase in LIST*POST causes 8.27% significantly increase in dependent variable Ln(Patent), which indicates that the implementation of short selling mechanism improve the amount of patent applying of target companies.
Model (2), (3) and (4) in Table 5 are robustness checks of empirical equations. From (2) and (3), I omit 1 and 2 control variables respectively and omit firm fixed effect. In model (2) and (3), I find that model (2) in Table 5 has almost a same amount of coefficient in the most important dependent variable POST*LIST, model (3) in Table
3 has a similar significant coefficient. Model (2) and (3) in Table 5 are both significant. As for model (4), coefficient of POST*LIST is positively insignificant, and dependent variable LIST of model (4) are positive and significant under 10% significance measurement, which may be explained that firm fixed effect is a indispensable effect in this equation. Besides, if I look at the R-squared in Table 3, I find that model (1) and (2) have the biggest R-squared. It can be concluded that, the power of explanation of model (1) and model (2) are the largest and most convincible among all 4 models. For model (4), its R-squared suddenly drop sharply comparing with other 3 models, which strengthens the conclusion that firm fixed effect is also indispensable in this equation.
Table 6 Summary of Short Selling Mechanism and Corporate Innovation Input & Output
The table provides the summary results of regressions of R&D/Total assets, R&D/Revenue and log(number of patent) on short selling mechanism. R&D/Total assets, R&D/Revenue mean to be innovation input and Ln(Patent) is the measurement of innovation output in this article, in percentage. Independent variables are LIST*POST, a dummy interaction term of LIST and POST (which is not included in this regression since I include year fixed effects), therefore it represents target companies after (include) the year they are implemented the policy when it equals to 1. Control variables include LEVERAGE, defining total liabilities/total assets, LN(SIZE), representing log(total assets), and ROA, defining as net profit/total assets. Three regressions all include year, industry and firm fixed effects. Standard errors are in parentheses.*, **, and *** measure significance at 10%, 5% and 1% level, respectively.
Dependent Variables
R&D/Total assets R&D/Revenue Ln(Patent)
LIST*POST 0.000913** 0.000696 0.0827*
(0.000417) (0.000981) (0.0441)
Observations 10,367 10,367 10,513
R-squared 0.845 0.855 0.757
Table 4 is a summary table for all 3 equations with the most appropriate setting in the models. Table 1，2，3 shows the regression results of the margin trading policy and company innovations input and output. The dependent variables of Model (1) and Model (2) are the innovation input indicators measured by R&D expenditure over Revenue or Total assets, and the Model (3) is the innovative output indicators measured by natural logarithm of patent amount. From the results in the table, after controlling for industry, year, and firm fixed effects, the regression coefficient of LIST is insignificant in either the innovation input model or the innovation output model, indicating that before margin trading program implemented, there was no significant difference of innovation input and output between treatment group and control group. This regression focuses on LIST*POST(interaction term), which is significantly positive in both the innovation input model and the innovation output model, indicating that after the inclusion in the target list of margin trading policy, compared with non-target companies, the target company’s innovation investment significantly improves. At the same time the huge improvements is also significant in company’s innovation output. The results show that the implementation of margin trading policy has significantly increased the company's investment in innovation, and significantly increased the company’s innovation output to a bigger extent, thus the implementation of margin trading policy has improved the company's innovation efficiency. The research conclusions confirmed that the implementation of the margin trading policy is in line with the “the theory of constraints” rather than the “the theory
of pressure”. The Chinese short-selling mechanism (margin trading policy) has a significant “innovative incentive effect”.
Under the theory of constraint in my hypothesis, the constraint effect will improve the company’s innovation by internal quality and other basic characteristics. The implementation of short selling mechanism could inhibit management agency problem (Jensen and Mechling, 1976) when innovation decision making happens. Then short selling mechanism can strengthen long-term incentives of management board and increase external supervision to some extent. When management team try to avoid risky innovation projects, short sellers discover this condition and use their information advantages to trade actively the corresponding company’s stock using short selling. In this case, it will occur negative direction of stock price reaction. Furthermore, short sellers are acting as external supervision regulators to enforce management team execute high-quality innovative projects for the benefits of company itself instead of management team’s own interest. Therefore, short selling mechanism improves company’s innovation efficiency by reducing agency problems under theory of constraint.
5.1 Estimations dividing sample periods
Table 7 Under different time period, Short Selling Mechanism and Corporate Innovation Input & Output: Research & Development Expenditure / Total
Assets; Research & Development Expenditure / Revenue; Ln(Patent)
This table provides corporate innovation input and output models estimated results under different manually divided time period. It tests 4 sample periods separately as 2009-2010, 2011-2012, 2013-2014, and 2015-2016. I mainly focus on estimated coefficient of interaction term LIST*POST, and I control LEVERAGE, LN(SIZE) and ROA variables for all sample periods. Regression (1), (2), (3) all include year, industry and firm fixed effects. Standard errors are in parentheses.*, **, and *** measure significance at 10%, 5% and 1% level, respectively.
Dependent variables
(1) (2) (3)
R&D/Total assets R&D/Revenue Ln(Patent)
2009-2010 LIST*POST -0.00288 -0.00160 0.148 (0.00538) (0.0123) (0.448) LEVERAGE 0.00961 -0.00841 1.404* (0.00927) (0.0212) (0.769) LN(SIZE) -0.0177*** -0.0156* -0.0265 (0.00387) (0.00883) (0.322) ROA 0.0504** -0.0579 0.201 (0.0207) (0.0473) (1.723) Observations 498 498 502 R-squared 0.905 0.902 0.895 2011-2012 LIST*POST -0.0961 -0.302 0.262* (77,065) (174,404) (0.147) LEVERAGE -0.00366 -0.0199 -0.519*** (0.00606) (0.0137) (0.195) LN(SIZE) -0.00736*** -0.00613 0.400*** (0.00209) (0.00474) (0.0380) ROA 0.0131 -0.149*** 1.603** (0.0103) (0.0234) (0.717) Observations 1,368 1,368 1390 R-squared 0.935 0.956 0.233 2013-2014 LIST*POST 0.00153 0.00493** 0.184*** (0.000964) (0.00249) (0.0669) LEVERAGE 0.00283 -0.0151** -0.230 (0.00256) (0.00659) (0.157)
LN(SIZE) -0.00873*** -0.00449** 0.441*** (0.000778) (0.00201) (0.0296) ROA -0.00284 -0.123*** 3.896*** (0.00521) (0.0134) (0.533) Observations 3,036 3,036 3,309 R-squared 0.955 0.951 0.271 2015-2016 LIST*POST -0.000706 -0.00142 -0.113* (0.00157) (0.00409) (0.0626) LEVERAGE 0.00299* -0.0176*** -0.154 (0.00175) (0.00457) (0.152) LN(SIZE) -0.00780*** -0.00292** 0.406*** (0.000471) (0.00123) (0.0269) ROA 0.00996*** -0.0616*** 4.242*** (0.00347) (0.00905) (0.480) Observations 3,618 3,618 3,926 R-squared 0.965 0.962 0.252
As a typical short selling mechanism, China's Margin Trading Policy has significant differences with other short selling mechanisms, reflecting by the progress of step-by-step expansion of the underlying stocks. Till 2018, the Margin Trading Policy has first launched in March 2010, the first expansion in December 2011, the second and third expansion in January 2013 and September 2013, the fourth in September 2014, and the fifth in December 2016. In order to further examine the systemic effects of Margin Trading Policy, I conduct a time-period test based on the advancement of Margin Trading Policy. The whole sample time period is divided into 4 time periods, of which there are periods of 2009-2010, 2011-2012, 2013-2014, and 2015-2016. Table 7 shows the results of the phase inspection of the Margin Trading Policy implementation. From the results in the table, it can be seen that in the start-up phase of short selling mechanism from Margin Trading Policy in 2010, the regression
coefficient of LIST*POST is not significant either in the innovation input model or the innovation output model. After the first expansion in 2011, the regression coefficient of LIST*POST in the innovation output model turns to be significant, while coefficients of interaction term in innovation input model are negative. 2011-2012 first expansion period starts to indicate an improvement in innovation efficiency. Regression coefficients LIST*POST on R&D/Revenue and Ln(Patent) in Model 2013-2014 become significant, and they are statistically significant at the 1% level, indicating that after the second expansion, innovative incentive effect on target companies has gradually become significant. Compared with non-target companies, the target companies can increase the patent output by 18.4% after the implementation of short selling mechanism.
When it goes to 2015-2016 period, all LIST*POST coefficients become negative. A standard deviation increase in LIST*POST even causes 11.3% decrease in Ln(Patent). This conflicts with the results and conclusions above. But if I look in to the Chinese stock markets during 2015-2016, there was a abnormal shock influencing all the listed company and might be the reason of negative incentive to corporate innovation. In 2015, China’s substantial economy was weak and it required the stock market to support. Under a loose monetary policy throughout the year, it indirectly stimulated the soaring of the stock market. In 2015, The Shanghai Composite Index continued its momentum after the 52.87% increase in 2014. In the context of loose monetary policy and regulation reform, the Shanghai Composite Index soared to a highest point 5178 in only three months. In June, the bull market quickly turned around and the Shanghai Composite Index fell to 3373 points. Then a two-month shock-down mode was opened. A discovery of huge risk hiding in the allocation of capital and a de-leveraging prevented systemic financial risks. With the completion of the liquidation of the fund, the Chinese stock market started going back on normal track.
6. Research conclusions and inspirations
So far, I investigate a general estimation of short selling mechanism on corporate innovation input and output within the whole sample period. With control variables leverage, ln(size), and ROA and fixed effects firm, industry and year, the finding shows that R&D/Total assets and R&D/Revenue, the proxy of innovation input, do not have a significant influence by short selling mechanism in general. Regarding to innovation output measurement, ln(patent), it does have significant improvement, which is consistent with Hypothesis-A. Under theory of constraint, introduction of short selling mechanism effectively raise the amount of corporate patent, and reducing the agency problems between management board and company at the same time(Jensen and Meckling, 1976). Therefore, I reach a conclusion that short selling mechanism has a significant positive effect on corporate innovation and enhances corporate innovation efficiency.
Early in the trial period of the Margin Trading Policy, the China Securities Regulatory Commission has expected that the introduction of the Margin Trading Policy would exert four basic functions for the construction of China's capital market: price discovery, market stability, liquidity enhancement, and risk management. These functions focus on the short-term impact of the margin trading program on pricing efficiency. But in this paper, a new path is used. I start taking into consideration from long-term impact of the implementation of the Margin Trading Policy on the company's innovation behavior and its value effect. My research find that, firstly, the implementation of the short selling mechanism does not significantly affect the company's innovation input, but it significantly promotes the company's innovation output, indicating that the Margin Trading Policy can improve the company's innovation efficiency, with a significant "innovative incentive effect". Secondly, the "innovative incentive effect" of Margin Trading Policy is gradually emerging as the policy proceeding by years. But it is influenced by the stock markets abnormal volatility.
The conclusions of this paper have important policy implications. First, in addition to the traditional four functions of price discovery, market stability, liquidity enhancement and risk management, my research also finds that short selling mechanism (from Margin Trading Policy) has long-term innovation incentive effect. Under the constraint of short selling mechanism, the management board will pay more attention to the cultivation of internal strength of the enterprise and strengthen the R&D investment and management quality, which has a significant strategic impact on the improvement of corporate value.
Since the second half of 2015, the phenomenon of stock market crash caused by the frequent outflow of thousands of collective stocks in China's capital market has raised investors' high suspicion and negative evaluation of short selling mechanism. Investors believe that short selling mechanism aggravates stock market volatility. But in this research, I find that the implementation of the short selling mechanism can enhance the company's innovation efficiency, and this innovation incentive effect is gradually positively affected by the deepening of the Margin Trading Policy. Therefore, the Margin Trading Policy should not be interpreted as a bad game for investors and listed companies, but should be understood as a exploring corporate creation. It can be seen that after certain market conditions and institutional environment turn into normally growing, the Chinese government should continue to promote the short selling mechanism construction, financial creation and expansion of the short selling mechanism, such as considering the implementation of the comprehensive stocks in the market, and consider reducing margins of short selling. Based on above, it may consider introducing various short-selling mechanisms such as futures and options, promoting the construction of multi-level short-selling mechanism and market system. This action undoubtedly has significant strategic value for the long-term development of China's capital market and listed companies' value added.
In all, the introduction of Margin Trading Policy, as a short-selling mechanism introduction, is of epoch-making significance in Chinese capital market. As a
financial innovation, its introduction is a meaningful reform of trading system that enables dramatically change in financial markets, profit earning models of investor and broker, the product design and risk management. All parties in capital market should work together to play its positive functions, actively exploring the developing rules of margin trading. Regulators should strengthen daily supervision, actively preventing and resolving the risks of margin trading to promote a smooth development and a healthy financial market.
7. Reference
Bogen, J. I., & Krooss, H. E. (1960). Security credit: Its economic role and regulation. Prentice-Hall.
Campello, M., Ribas, R. P., & Wang, A. Y. (2014). Is the stock market just a side show? Evidence from a structural reform. The Review of Corporate Finance
Studies, 3(1-2), 1-38.
Chang, E. C., Luo, Y., & Ren, J. (2014). Short-selling, margin-trading, and price efficiency: Evidence from the Chinese market. Journal of Banking & Finance, 48, 411-424.
Chen, Z., Dong, G. N., & Gu, M. (2016). The causal effects of margin trading and short selling on earnings management: A natural experiment from China.
Ederer, F., & Manso, G. (2013). Is pay for performance detrimental to innovation?. Management Science, 59(7), 1496-1513.
Figlewski, S., & Webb, G. P. (1993). Options, short sales, and market completeness. The Journal of Finance, 48(2), 761-777.
Grullon, G., Michenaud, S., & Weston, J. P. (2015). The real effects of short-selling constraints. The Review of Financial Studies, 28(6), 1737-1767.
He, J., & Tian, X. (2015). SHO time for innovation: The real effects of short sellers. Kelley School of Business Research Paper.
Henry, Ó. T., & McKenzie, M. (2006). The Impact of Short Selling on the Price‐Volume Relationship: Evidence from Hong Kong. The Journal of
Business, 79(2), 671-691.
Henry, Ó. T., & McKenzie, M. (2006). The Impact of Short Selling on the Price‐Volume Relationship: Evidence from Hong Kong. The Journal of
Business, 79(2), 671-691.
Hirshleifer, D., Low, A., & Teoh, S. H. (2012). Are overconfident CEOs better innovators?. The Journal of Finance, 67(4), 1457-1498.
Jensen, M. C., & Meckling, W. H. (1976). Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of financial economics, 3(4), 305-360.
Karpoff, J. M., & Lou, X. (2010). Short sellers and financial misconduct. The Journal
of Finance, 65(5), 1879-1913.
Li, Y., & Zhang, L. (2015). Short selling pressure, stock price behavior, and management forecast precision: Evidence from a natural experiment. Journal of
Accounting Research, 53(1), 79-117.
Manso, G. (2011). Motivating innovation. The Journal of Finance, 66(5), 1823-1860. Mitchell, M., Pulvino, T., & Stafford, E. (2004). Price pressure around mergers. The
Journal of Finance, 59(1), 31-63.
Nie, F., Huang, H., Cai, X., & Ding, C. H. (2010). Efficient and robust feature selection via joint ℓ2, 1-norms minimization. In Advances in neural information
processing systems (pp. 1813-1821).
Saffi, P. A., & Sigurdsson, K. (2010). Price efficiency and short selling. The Review of
Financial Studies, 24(3), 821-852.
Sharif, S., Anderson, H. D., & Marshall, B. R. (2014). Against the tide: The
commencement of short selling and margin trading in mainland China. Accounting &
Finance, 54(4), 1319-1355.
Tian, X., & Wang, T. Y. (2011). Tolerance for failure and corporate innovation. The
Review of Financial Studies, 27(1), 211-255.
Wang, Y., & Zhao, J. (2015). Hedge funds and corporate innovation. Financial
Woolridge, J. R., & Dickinson, A. (1994). Short selling and common stock prices. Financial Analysts Journal, 50(1), 20-28.
Eric C. Chang, Yan Luo, Jinjuan Ren. Short-selling, Margin-trading, and Price Efficiency: Evidence from the Chinese Market [J]. Journal of Banking & Finance, 2014, 48: 411-424
Zheng Xiaoya, Yan Hui, Liu Fei. Long-term Volatility of Margin trading and China Stock Market [J]. Financial Research, 2015(2): 87-93.
