Interdependence between Yu Ebao and Shibor:An Empirical Study Based on VAR-DCC-GARCH Model

Transcription

1 Interdependence between Yu Ebao and Shibor:An Empirical Study Based on VAR-DCC-GARCH Model Xiangyou Wu 1, Chiang Ku Fan 2 1. Newhuadu Business School of Minjiang University. Institute of Internet Innovation of Minjiang University, Fuzhou, Fujian, China 2. College of Management, Shih Chien University, Taipei, Taiwan China Abstract: As an alternative money market fund arising from the rapid development of internet finance, Yu ebao is inevitably highly related to Shibor with its assets mainly invested in the interbank market. 1) With its disproportionate impact on Shibor,a natural question is whether Yu ebao is a disrupter or builder of the interbank market. 2) Different answers to this question will influence not only the development of Yu ebao, but also the liberalization of China s interest rates, even the growth of China s financial market as a whole. 3) This paper finds that the effect of Yu ebao s yield on Shibor is more significant than that of Shibor on Yu ebao, and that Yu ebao s yield is less volatile than Shibor, which indicates high performance of risk management of Yu ebao. 4) This paper is different from previous literature by discovering a system of error correction in Yu ebao s yield, which is helpful to smooth the fluctuation of Shibor. 5) By introducing an alternative empirical method VAR-CCC-MGARCH, this paper justifies its application of VAR-CCC-MGARCH model to capture the spillover effects of Yu ebao on Shibor.. Keywords: yu ebao; shibor; interdependence; var-dcc-garch model 1. Introduction Yu ebao developed by Alibaba for the users of the third-party payment platform of Alipay in June 2013 is a balance appreciation service. In terms of strict theories, Yu ebao is not a financial innovation, but a new payment method of the purchase of money market funds. Alipay users can transfer some or all of the balance in the Alipay accounts to the Yu ebao account affiliated with the Apiapy accounts to purchase a money market fund named Zenlibao supplied by a key player in China s money market, the Tianhong Fund. By doing so, Alipay users can not only enjoy return from the investment in the money market fund, but also keep the right to make real-time payments with the same fund when doing online shopping on Taobao, Tmall and other E-commerce platforms. Thanks to Yu ebao, Alipay account simultaneously plays the roles of payment media and investment tool. Yu ebao gets separately idle money of enormous Alipay users together with the help of Internet to invest in Certificates of deposits (CDs), agreement deposits, and so on. Agreement deposits enjoy interest rates as high as Shibor (the Shanghai Inter-bank offered rate), which is the real market interest rate with the least risk premium in China. After the global financial crisis, especially after the 2009 fourtrillion stimulus plan, China s central bank began to implement prudent or moderately stringent monetary policies, spare no efforts to carry out oriented fine and precise minor adjustments rather than economy-wide quantity easing policies, which resulted in the liquidity risk of financial industries in the early part of 2013 and leaded to a scarcity of money (Qianhuang, in Chinese).Yu ebao was born in the context of industry-wide money shortage faced by banks, which drove Shibor to very high levels and helped Yu ebao to enjoy a golden period of rapid growth characterized by high yield and scale expansion with most of its fund invested in money market funds. So the seven-day annualized yield of Yu ebao has been much higher than that of one-year term deposit. The interdependence of financial assets is one of the key determinants of asset pricing and risk management. Sometimes, the correlation itself will be a risk nest. Hence, the correlation between Yu ebao s yield and Shibor has been taken as one of the focuses of financial researches. With the background of the liberalization of China s interest rates, these researches have special theoretical and practical values. The first difficulty this paper facing is how to explain the disproportionate influence of Yu ebao on Shibor. Theoretically, as one of money market s newcomers, Yu ebao s yield should be significantly affected by Shibor. Correspondingly, the yields of Yu ebao could only indicate or slightly influence Shibor. In contrast to theories, Yu ebao has a magnified influence on market interest rate with its rapid risingup, excessively high yields and expansive asset size. The second difficulty is to answer whether Yu ebao is destructor or constructor of China s money market. Financial industrialists and academia are distinctly opposed to each other on this question. More and more scholars hold the viewpoint that money market funds based on internet such as Yu ebao accelerates the liberalization of China s interest rates. Money market funds like Yu eabo are absolutely innovative builders of China s money market. While financial industrialists mainly thought Yu ebao as a disrupter or trouble-maker of the development of financial markets because Yu ebao has been making the financial market less predictable and stranger to them. Furthermore, MGARCH models are generally applied to study the spillover effect between two markets (i.e. Shanghai composite index and Hensheng stock index) rather than analyze the interdependence between two financial assets yield (Shibor and Yu ebao in this paper).fortunately, existing literature have some evidences to support the employment of MGARCH model in the research of the correlation between Shibor and Yu ebao. This paper makes some contributions to existing literature by finding that both Yu ebao and Shibor are highly influenced by both of themselves lagged in one period, and that the impact from Yu ebao on Shibor is larger than that of Shibor on Yu ebao, and that the volatility of Shibor is bigger than that of Yu ebao, and that there exists an error correction mechanism in the volatility of Yu ebao while Shibor is more volatile. 2. Empirical strategy ARCH model developed by Engle (1982) could capture the time series characteristics of the yields of financial assets [1]. Bollerslev (1986)extended ARCH model to a more universal GARCH model to including one period lagged variance of financial assets yields [2]. Bollerslev and Wooldridge(1988)further extended GARCH model to MGARCH model to including more variables [3]. MGARCH can capture the volatile characteristics of multiple variables and multiple markets while it cannot reflect the time varying property of the correlation of financial assets and it embodied too many parameters to be estimated. Engle (2002) developed DCC-GARCH model, which solves the complexity of the computation of large scale conditional variancecovariance matrixes, simplifies the estimation of the correlation of multiple variables, can estimate the dynamic and time varying correlation of multiple variables and depict the spillover effect of volatility and the process of information transmission [4]. Hence, DCC-GARCH is a very suitable model to capture the interdependence of Shibor and Yu ebao. With regards to the co-movement of Shibor and Yu ebao, the mean equation VAR (1) can be specified as follows. (1) Journal of Residuals Science & Technology, Vol. 13, No. 6,

2 Here, is the vector of the yields of Yu ebao and Shibor,,,, and, is the information set at period (t-1). The one-period lagged explained variable is introduced into the model to eliminate autocorrelation and analyse the spillover effect of both of Shibor and Yu ebao. Furthermore, the DCC-GARCH model can be expressed as follows. (2) Here, is a 2 2 diagonal matrix of time varying standard deviation. is a 2 2 unconditional correlation matrix. is a 2 2 positive definite matrix. is a 2 2 covariance matrix. and are the parameters to be estimated. According to Engle (2002), the parameters of DCC model can be estimated by maximizing the log likelihood function as follows. (3) 3. Data 3.1. Data sources This paper takes the annualized seven-day yield of Yu ebao and three-month Shibor as key variables to analyze the correlation between Yu ebao and Shibor because both of them belong to middle-term variable and have higher typicality. The data on the annualized seven-day yield of Yu ebao was taken from the website of financing gaining, The data on three-month Shibor comes from Shanghai interbank offered rate's official website, Both of them are daily data. Yu ebao has been releasing its yield data every day since its launch, while Shibor official website issues its data only on workdays. In view of the most important variable which influences Yu ebao s yield is time rather than price, the vacancy data about Shibor can be refilled by the data before holidays to make the two time series data matched. The sample period of the data ranges from June to April ,710 observations in all. 3.2 Descriptive statistics Table 1 the descriptive statistics of Shibor and Yu ebao observations mean Std. Min. Max. skewness Kurtosis Shibor Yu ebao As shown in table 1, the means of Shibor and Yu ebao s yield are not obviously different from each other.yueabo has a larger range and a bigger standard deviation than Shibor indicates that the central bank s control over market liquidity is more stable than the Tianhong Fund Company s management of market risks. From the perspective of normality, Shibor and Yu ebao are both different from normal distribution. Comparatively, Yu ebao s yield is more close to normal distribution than Shibor. The most interesting difference comes from the skewness that Shibor is left skewed while Yu ebao s yield is right skewed reflecting the fact that the central bank enjoys more discretionary power in implementing expansionary monetary policy and that Tianhong Fund has been sparing no efforts to raise Yu ebao s yield even in the context of left skewed Shibor. 3.3 The stationarity of Shibor and the yield of Yu ebao Table 2 unit root test of Shibor and Yu ebao s yield Test mode(c,t,k) Test statistics Critical values results Shibor (c,t,4) stationary Yu ebao (c,t,2) stationary Guided by time series graphs of both variables and AIC and BIC information criteria, ADF test with appropriate test mode in table 2 shows that both of Shibor and Yu eabo are stationary time series variables. This finding laid a sound basis for the empirical strategy to use MGARCH model to capture the interdependence between Shibor and Yu ebao. 3.4 Linear correlation of Shibor and Yu ebao Table 3 linear correlation coefficients of Shibor and Yu ebao Shibor Yu ebao Journal of Residuals Science & Technology, Vol. 13, No. 6,

3 Shibor Yu ebao able 3 shows that there exists strong linear correlation between Shibor and Yu ebao indicating there are co-movements or interdependence between Shibor and Yu ebao. Whether there is nonlinear correlation between Shibor and Yu ebao is to be tested by DCC- GARCH model. 4. Empirical Results 4.1 Results from VAR-DCC-GARCH(1,1)model Literature shows that GARCH(1,1) model can depict the yield volatility of financial assets well. Hence, this paper adopts the generally used GARCH (1,1) model to estimate the volatility of Shibor and Yu ebao s yield respectively. Similarly this paper uses DCC(1,1) to capture the dynamic correlation between Shibor and Yu ebao. The results of the mean equations of DCC(1,1) is shown in table 4. Table 4 results of mean equations of DCC (1,1) intercept * (0.093) ** (0.039) *** *** * (0.076) Loglikelihood e *** Table 4 shows that Yu ebao s yield is significantly influenced by itself lagged one period and less significantly affected by one period lagged Shibor. This finding is slightly different from [5] that of Liu Dongqing (2015).On the contrary, Shibor is influenced significantly by both of one-period lagged Shibor and Yu ebao s yield. This finding is close to [6] that of Zhang Yu and Zhang Jiannan (2015) and concerts with [7] that of Chai Yongdong and Cao Jianfei (2014). This finding is not only interesting but also instructive. Firstly, Yu ebao is only a newcomer in the money market. Secondly, it is merely one of many participants of money market. Finally, Yu ebao is only an indirect participant of interbank lending market. The tact that Yu ebao has disproportionate influence on Shibor shows its feature of so called disturber. On the other side, Yu ebao s asymmetric influence is helpful to accelerate the growth of Shibor, speed up the liberalization of interest rates and improve the price discovering function of money market. In this sense, Yu ebao is not a simple disturber, but an active builder. Table 5 results of variance equations of DCC (1,1) intercept *** *** ** (0.031) 2.86e-06*** *** *** Loglikelihood e+07 From the perspective of variance equation, the variance of Yu ebao s yield is significantly influenced by itself lagged in one period and the squared error term lagged in one period. What is noteworthy is that, at 5 percent significance level, one-period lagged variance of Yu ebao s yield has a negative effect on its present variance. This fact shows that Tianhong Fund can adjust its portfolio quickly in response to the change of market conditions and has a high ability of risk control [8]. Meanwhile, this finding indicates that Yu ebao is closely linked to but not bound with Shibor. Yu ebao mainly invests in interbank lending market, but not simply replicate Shibor. Yu ebao succeeds both in taking advantage of the beneficial changes and avoiding the harmful changes in Shibor [9]. This is consistent with the finding in table 1 that Yu ebao s yield is right-skewed. On the contrary, the variance equation of Shibor indicates that Shibor s volatility is clustering. The variance of Shibor has a long memory that a large variance at time t will lead to a larger variance at time (t-1). 4.2 Results from VAR-CCC-GARCH(1,1)model As mentioned above, most of Yu ebao s asset is invested in the interbank lending market. Agreement deposit is one of its main investment objects. The interest rates of agreement deposit are closely linked to Shibor. Therefore, Yu ebao s yield had been driven high for a long period of time by the continually increasing Shibor resulted from the so called money shortage in 2013.In the second half of the sample period, Yu ebao s yield has kept going down accompanied by a continually shrinking asset size with the continually decreasing Shibor. These evidences show the stability of the correlation between Shibor and Yu ebao. Questions naturally arisen are whether the correlation between Shibor and Yu ebao is constant or dynamic [10], and whether Yu ebao is pegged at or bound with Shibor passively. Journal of Residuals Science & Technology, Vol. 13, No. 6,

4 In order to test whether the findings via VAR-DCC-GARCH is reliable or not, this paper try to employ another model, the VAR-CCC- GARCH model to re-estimate the yield and volatility spillover effect between Shibor and Yu ebao, and compare the results from both models. The results of the mean equations of CCC (1,1) is shown in table 6. Table 6 results of mean equations of CCC (1,1) intercept * (0.432) ** (0.018) *** *** *** (0.003) 0.997*** Loglikelihood e+07 In comparison to the results in table 4, the results of mean equation from CCC (1,1) shown in table 6 do not have substantial differences. The only difference is that the coefficients from CCC (1,1) are more significant than those from DCC(1,1). Table 7 results of variance equations of CCC (1,1) intercept *** ** (0.035) 2.76e-06*** ** (0.048) ** (0.014) *** (0.007) Loglikelihood e+07 Similarly, compared with the results of variance equation from DCC (1,1) in table 5, there is no obvious difference of all the coefficients in table 7. The only slight change comes from the significance of the coefficients. One-period lagged variance of Yu ebao still has a minus coefficient just like it does in table 5. Combined the results in table 6 and table 7 from CCC (1,1), there is no significant difference between the results from VAR-CCC- GARCH (1,1) and VAR-DCC-GARCH (1,1). From a cross verification perspective, the results from VAR-CCC-GARCH (1,1) justify the application of VAR-DCC-GARCH (1,1) to estimate the interdependence between Shibor and Yu ebao. Comprehensively, the log likelihood value of VAR-CCC-GARCH (1,1) is significantly smaller than that of VAR-DCC-GARCH (1,1). The former is only , while the latter is as big as Meanwhile, the Chi-square value from the former gets obviously smaller though the accompanying P value is still zero. Hence, VAR-DCC-GARCH (1,1) is the better model to estimate the correlation between Shibor and Yu ebao. 5. Conclusion Based on VAR-DCC-GARCH model, this paper analyzes the interdependence between Yu ebao s yield and Shibor, finds that there exist bi-directional spillover effect of yield and volatility. With respect to yield spillover effect, both of them are influenced by one-period lagged values of the other. The most noteworthy finding is that Yu ebao s yield has a more significant influence on Shibor than that of Shibor on Yu ebao even though the former is only one of many participants in interbank lending market. This finding indicates that Yu ebao is an accelerator of the liberalization of China s interest rates rather than a disrupter of the development of money market. Similarly, with regards to volatility spillover effect, both of the variances of Yu ebao and Shibor are affected by one-period lagged values of the other. The volatility of Yu ebao s yield has a self-rehabilitation ability reflecting that Yu ebao is only closely linked to rather than bound with Shibor or passively duplicate Shibor. The fund managers of Yu ebao have succeeded in benefiting from increases in Shibor and avoiding negative impact from Shibor s harmful volatility to maintain a relatively high and stable yield. At present, there exist larger divergences of viewpoints about the development of Yu ebao-like internet-based money market funds between the academia and the industrialists. Academic circles as a whole advocate free development, while business circles mainly emphasize normal and regulated development. Comprehensively, the biggest worries that Yu ebao-like internet-based money market funds faced are policy risk, especially the central bank s monetary policy, and the inter-bank boycott. Therefore, Yu ebao-like internet-based money market funds should accelerate product innovation, and should expand investment channels and investment tools to spread risks through diversification. Investors of Yu ebao should raise risk awareness and take a rational view of the volatility of Yu ebao s yield. In essence, Money market funds are a financial product with varying rates of return and without guaranty of principal. Yu ebao is a less risky but not a risk-free financing tool. It will make loss to its investors under some financial shocks. Furthermore, Yu ebao s relatively high yield is achieved at the cost of the sacrifice of some safety.it still has some degree of liquidity risk and interest rate risk. Hence, investors should not have blind trust in Yu ebao and should try best to diversify investment. The government should vigorously support the development of Yu ebao. As a so called market disrupter, Yu ebao will inevitably intensify the competition by diverting some funds and encounter resistance from traditional financial sectors. Policy supports from government will offset some extrusion from hostile traditional financial sectors and cultivate the growth of Yu ebao. Meanwhile, government should play the role of coordinator between Yu ebao-like innovative financial business and traditional financial business, and promote constructive competition and cooperation to accelerate the unfinished financial reform. Journal of Residuals Science & Technology, Vol. 13, No. 6,

5 Acknowledgements This research work was supported by Project of Fujian Provincial Social Science Foundation (FJ2015B209) and project of the Education Department of Fujian Province (JAS150475). This research work was also supported by Project of Fujian Provincial Social Science Foundation (2014B097). Project of Fuzhou Science and Technology Bureau (2015-G-60). The supports are gratefully acknowledged. References [1] Engle Robert F. Autoregressive Conditional Heteroscedasticity with Estimates of the Variance of United Kingdom Inflation. Econometrica, Vol. 50, 1982, No. 4, [2] Bollerslev T. Generalized autoregressive conditional heteroskedasticity. Journal of Econometrics, Vol.31, 1986, No.3, [3] Bollerslev, T., Engle, R., & Wooldridge, J. A Capital Asset Pricing Model with Time-Varying covariance. Journal of Political Economy, Vol.96, 1988, No 1, [4] Engle Robert. Dynamic Conditional Correlation: A Simple Class of Multivariate Generalized Autoregressive Conditional Heteroskedasticity Models. Journal of Business & Economic Statistics, Vol. 20, 2002, No 3, [5] Liu Dongqing.an analysis of the value of Yu ebao based on liner regression model. Journal of Haerbin business university, 2014,No 5, [6] Zhang Yu, Zhang Jiannan.Can Yu ebao influence Shibor-an empirical analysis. Financial Perspective, 2015, No 5, [7] Chai Yongdong, Cao Jianfei. A study of the relationship between the yields of internet-based money market funds and bank financing products and Shibor. Academic forum, 2014, No 10, [8] Lun Jinyun, Xue Zhuozhi, Zhou Qi. An empirical study of the influence of Shibor on Yu ebao yield. Credit Report, 2015, No 11, [9] He Jianmin, Bai Jie.a study of the determinants of Yu ebao yield based on EEMD-VAR model. Journal of Tianjin economics and Finance University, 2015, No 8, [10] Yu Yanjia,Pan Xiuhua. the impact of Yu ebao on the operation of commercial banks and the liberalization of interest rates.ecommerce,2014,no 11,3-5. Journal of Residuals Science & Technology, Vol. 13, No. 6,