Over the past two decades, China’s financial market has achieved great development, and the types of financial instruments continue to increase. However, compared with developed countries and even some emerging market countries, there is still a big gap. Therefore, increasing the innovation intensity of financial instrument is of great significance for realizing the internationalization of China’s financial market. While expanding bond issuance is of great significance to China’s multi-level financial system construction, increasing the proportion of direct financing and reducing the fragility of the financial system.
2. Literature Reviewed
Zhang Sheng  summarized the comparative advantages of corporate bond financing based on the analysis of western corporate financing theory, and conducted empirical research based on the current situation of China’s financing market. The empirical results show that the cost advantage of Chinese listed companies’ corporate bond financing is not significant, but the financing “signal” transmission effect is obvious. Shi Wenchao  constructed a regression model by the relationship between money supply in wild sense, loan balance and the amount of bond custody. It is concluded that the direct debt financing market can effectively reduce the currency derivation effect of the total financing behavior of the society, and the injection of funds into the real economy has an incomparable certainty in the credit market. Ji Chengpeng  used the panel data of 37 countries from 1991 to 2010 to study the relationship between the development of the bond market and the net inflow of bond capital. Finding the size of a country’s bond market is an important factor in attracting net inflows of bond capital. Gu Xian  found that in countries with stronger creditor rights protection, its enterprises tend to issue bond financing, and the resulting proportion of capital investment is higher; Conversely, in countries with weaker creditor rights protection, the lower the proportion of enterprises through bond financing, the lower the level of capital investment.
From the above literature, we can hardly see the use of PVAR model in measuring the financing ability of bonds and stocks and examining the regional effects of bonds and stocks on social financing. Therefore, based on Generalized Methods of Moments, Impulse Response and Variance Decomposition, which can preferably measure the short-term and long-term dynamic relationship among variables, this thesis empirically analyzes the impact of bonds and stocks on social financing to measure difference of the financing ability of bonds and stocks in financial markets, and this paper examines the regional effects of bonds and stocks on social financing in order to measure difference of the financing ability of bonds and stocks under different economic background. Ultimately, proposing relevant countermeasures and suggestions.
3. Data Description and Model Assumptions
3.1. Data Selection
The data sample is the quarterly data of 31 provinces, autonomous regions and municipalities in China from the fourth quarter of 2013 to the first quarter of 2018. The data comes from the statistics database of China Economic and Trade Network. This paper uses STATA14.0 statistical software to analyze the relevant data. In order to study the regional effects of bonds, it refers to the “China Marketization Index” compiled by Fan Gang, etc. in 2010 and the results of Xiong, Qiyue and Zhang Yiru  (2012). The provinces in the sample are divided into three regions according to the degree of economic development: economically developed regions, economically less-developed regions, and economically underdeveloped regions1.
3.2. Theory Model
The Panel Vector Auto Regression  model is used in the study of this thesis to examine the regional effects of China’s bond financing ability.
Therein, . is estimated parameter matrix, represents the cross-sectional effect of each region, represents the time effect. The dependent variable of this thesis is , represents the social financing increment of i province in t quarter, the independent variable is and , represents the bond volume of i province in t quarter, represents the stock volume of i province in t quarter. In order to eliminate the heteroscedasticity, all variables are taken logarithm here.
4. Empirical Analysis
4.1. Establishment and Estimation of PVAR Model
1Economically developed regions: Beijing, Tianjin, Shanghai, Jiangsu, Zhejiang, Shandong, Guangdong; economically less-developed regions: Hebei, Liaoning, Anhui, Fujian, Jiangxi, Henan, Hubei, Hunan, Guangxi, Sichuan, Xinjiang, Shanxi, Chongqing; economically underdeveloped regions: Inner Mongolia, Jilin, Heilongjiang, Hainan, Guizhou, Yunnan, Tibet, Shaanxi, Gansu, Qinghai and Ningxia.
In the practical application of vector auto regressive model, it is usually desirable to have a large enough lag order to more fully reflect the constructed dynamic features. However, the longer the lag order, the more parameters to be estimated in the model, and the less the degree of freedom. Therefore, in order to find a kind of equilibrium between the lag period and the degree of freedom, we use the AIC, BIC and HQIC information criteria to judge the optimal lag order of the model. There are three variables of statistical characteristic description, which are sample numbers, mean values, standard deviation, minimum and maximum in Table 1. It can be seen from Table 2 that the optimal lag order of the three models is finally determined to be 7th order. And on this basis, the model is estimated by GMM (Generalized Methods of Moments) to obtain short-term regression coefficients. The test results are shown in Table 3.
4.2. Impulse Response
Because the regression coefficient of the PVAR model is more, it is difficult to explain the continuous relationship between the various variables in the future phases. Therefore, the impulse response figure is used to directly describe the interaction relationship between the various variables in the next eight phases. It measures the short-term, purely unilateral impact of one variable on another variable by the impact of the unit standard deviation.
Table 1. Three variables of statistical characteristic description.
Table 2. Optimal lag order.
Table 3. GMM estimation results of the model.
The impulse response of social financing subject to one unit standard deviation positive impact of bonds and stocks is shown in Figure 1. It can be seen from the figure that the impact of bonds and stocks on social financing has a lag of one phase (i.e., one quarter). Both bonds and stocks have the strongest promotion effect on the fifth phase of social financing in the future, and then the promotion effect has gradually become smaller. And bonds and stocks still have an impact on social financing in the eighth phase, indicating that bonds and stocks have a longer-lasting impact on social financing. However, as can be seen from the figure, the fluctuation amplitude of the impulse response graph of bonds to social financing is larger than that of stocks to social financing, indicating that the impact of bonds on social financing is greater than the impact of stocks on social financing.
It can be seen from Figure 2 that the impulse response of social financing subject to one unit standard deviation positive impact of bonds is greater than the impulse response of social financing by one unit standard deviation positive impact of stocks. And the shock of bonds and stocks on social financing is positive, but the impact of bonds and stocks on social financing is not significant.
Figure 1. Impulse response figure of developed regions.
Figure 2. Impulse response figure of less-developed regions.
It can be seen from Figure 3 that the impulse response of social financing subject to one unit standard deviation positive impact of bonds is greater than the impulse response of social financing by one unit standard deviation positive impact of stocks. The shock of bonds and stocks on social financing is positive but not significant.
4.3. Variance Decomposition
Using the PVAR model, it is also possible to perform dynamic analysis of the variance decomposition study model. The main idea is to decompose the fluctuations (k prediction mean square error) of each endogenous variable (m in total) in the system into m components associated with the information of each equation according to their formation, so as to understand the relative importance of information to the model endogenous variables.
Table 4 shows the variance decomposition results of social financing on bonds and stocks in the three economic regions. From the perspective of the contribution degree of 1 unit fluctuation in social financing increment, in the three economically developed regions, the contribution rate of bonds to the social financing is greater than the contribution rate of stocks to social financing.
Figure 3. Impulse response figure of underdeveloped regions.
Table 4. Sub-regional variance decomposition results.
For economically developed regions, the contribution rate of bonds and stocks to social financing has slowly increased as time goes by. For economically less-developed regions, the contribution rate of bonds to social financing is the largest in the 12th period, and the contribution rate of stocks to social financing is gradually decreasing as time goes by. For economically underdeveloped regions, the contribution rate of bonds and stocks to social financing is constantly increasing.
5. Conclusion and Suggestion
It can be seen from the above analysis that the impact of bonds on social financing is greater than the impact of stocks on social financing. Compared with economically less-developed areas and underdeveloped areas, the social financing in economically developed regions is the most sensitive to bonds and stocks. And the bonds and stocks in economically developed regions have a greater and far-reaching impact on social financing. Also there is a certain time lag effect for the impact of bonds and stocks on social financing, and bond financing has a period of transmission process to have an impact.
It is not difficult to see that the influence of bonds on social financing should not be underestimated. Therefore, in the macroeconomic regulation and control, more attention should be paid to the development potential of bonds. Because the impact of bonds and stocks on social financing has obvious regional effects and certain time lag effects, in the macroeconomic regulation and control, differentiated macroeconomic countermeasures should be adopted, and regulation and control should be made timely and appropriately to avoid further economic gap growth in economically developed regions, economically less-developed regions and economically underdeveloped regions.
This work is supported by the National Natural Science Foundation of China (No.11561056) and Natural Science Foundation of Qinghai (No. 2016-ZJ-914).
Conflicts of Interest
The authors declare no conflicts of interest regarding the publication of this paper.
 Xiong, Q.Y. and Zhang, Y.R. (2012) Research on the Economic Regional Effect of Monetary Policy Credit Channel—Empirical Evidence Based on 31 Provincial Panel Data in China. Review of Investment Studies, No. 7, 78-89.
 Love, I. and Zicchino, L. (2006) Financial Development and Dynamic Investment Behavior: Evidence from Panel VAR. The Quarterly Review of Economics and Finance, No. 2, 190-210.