China's stock market crash in 2015 aroused scholarly attention on financial systemic risk in China. Using data on China's stock market from January 5, 2007 to September 28, 2018, this study calculated three different measures of systemic risk, i.e., temporal fluctuation of financial institutions' contribution to systemic risk (CoVaR), sensitivity of financial institutions to systemic risk (MES) and long-run expected capital shortfall (SRISK), to assess the formation, occurrence, and consequences of systemic risk in China. The results show that Δ CoVaR and MES exhibited an abnormal rise during the outbreak of the 2008 global financial crisis and the 2015 domestic stock market crash, whereas SRISK showed a steady increase throughout the observation window, indicating China's ever-growing SRISK despite the CoVaR and MES. More importantly, regardless of temporal fluctuation, (
Keywords: Systemic risk; China; financial institutions; CoVaR; MES; SRISK
A financial crisis occurred in the United States in 2007 and spread to global financial markets and other economic sectors. After the outbreak of the crisis, relevant international organizations, national supervisory authorities and scholars conducted in-depth research on the formation and occurrence of the financial crisis, and the external characteristics of financial institutions' risks were widely recognized to harm the stability and security of financial systems. This situation will lead to a series of financial regulatory issues (Basel Committee on Banking Supervision, [
The externality of financial institution risk means that a single financial institution increases its profits by expanding its balance-sheet and off-balance-sheet business scale and leverages and controls its own risks through financial innovation; however, the risks within the entire financial system do not simply disappear but are instead transferred and redistributed. Thus, the health of a single financial institution does not necessarily mean that the entire financial system is safe (Borio, [
Due to the reform and opening-up that began in 1978, China's financial industry achieved rapid development. By the end of 2017, the total assets of China's banking industry, insurance industry, securities industry and trust industry reached 245 trillion yuan, 16.64 trillion yuan, 6.14 trillion yuan, and 25 trillion yuan, respectively. Moreover, in recent years, with changes in the international financial environment and the deepening of China's domestic financial reforms, the trend of China's financial industry shifting from a separate business model to a mixed business model has become more apparent and resulted in the establishment of a number of comprehensive financial holding groups represented by CITIC, Ping An, and China Merchants, whose business scopes cover the major financial sectors: banking, insurance, securities and trusts. In addition, the continuous innovation of financial products and the gradual relaxation of supervision and financing channels have made China's financial market more active and created a large number of systemic risks. In short, the expansion of China's financial industry, the continuous innovation of financial products, and the hybridization of financial formats have reduced the threat of risk contagion among financial institutions and increased the impact of potential systemic risk. As such, it is important to assess the systemic risk of China's financial industry from a holistic perspective (Ding and Tay, [
Derbali ([
Graph
The study finds that
Graph
Graph
Graph
As such, we argue that policy makers should take a holistic approach when monitoring and managing systemic risk of China's financial market, and in particular should establish proactive regulatory measures to address the possibility of the systemic risk accumulating and spilling over from the primary source (i.e., securities and insurance firms) to the ballast of the system (i.e., banks). The establishment of Financial Stability and Development Committee under the State Council of China (国务院金融稳定发展委员会) in 2017 demonstrated the Chinese government's determination to deal with financial mixed operations by enhancing the sharing of information and coordination across the previously divided financial supervisory agencies. A real-time overarching monitoring framework, including the three systemic risk indicators used in this study, will prove to be vital for monitoring and warning against systemic financial risks.
The remainder of this paper is structured as follows. The second section provides a literature review, the third section introduces the theoretical method used in this paper, the fourth section explains the dataset, the fifth section conducts an empirical analysis and discusses the systemic risk rankings of Chinese financial institutions, and the final section concludes the paper.
The research on systemic risk focuses primarily on two types of content: the first is the construction of early warning indicators for systemic risk and the other is the problem of systemic risk transfer and size measurement of financial institutions.
Conducting the first type of systemic risk warning research, Illing and Liu ([
After the US financial crisis, measurements of systemic risk and financial institutions' contribution to systemic risk have become important components of global financial regulatory reform. In view of the shortcomings of traditional measurement methods, after the crisis broke out, regulators and scholars proposed a series of methods to measure systemic risk and marginal risk contribution of individual financial institutions. These methods can be divided mainly into the following two categories according to data types.
The first category is the network analysis approach, which measures the associated data based on actual assets and liabilities between financial institutions. This approach uses data on inter-bank bilateral mutual assets and liabilities to measure the risk of infection in the system and the degree of systemic risk contribution and relative systemic importance of individual banks to the banking sector (IMF, BIS, & FSB, 2009) by simulating bank bankruptcy, bankruptcy loss and the difficulty of inducing a systemic crisis caused by single or multiple banks in the banking network system (Degryse & Nguyen, [
The second type of method is collectively referred to as the reduced-form approach. This approach assumes that financial markets are efficient so that financial market data (including financial institution stock prices, credit default swaps, CDS spreads, etc.) can be used to measure the systemic risk of financial institutions. Such methods have the following advantages. First because the changes in financial institutions' asset prices reflect the market's expectations of their future performance, the adoption of market data is more forward-looking (Duffie, Eckner, Horel, & Saita, [
The simple method can be divided into two types: The first measures the systemic risk of the financial sector by examining the correlation change in the financial institution's stock return rate as performed by Huang et al. ([
The "top-down" analysis first derives systemic risk and then assigns this systemic risk to a single financial institution through a distribution method. For example, Acharya et al. ([
The "bottom-up" analysis estimates the systemic risk of the entire financial system when a single financial institution goes bankrupt. The main representative of such methods is the Conditional Value at Risk (CoVaR), which was proposed by Adrian and Brunnermeier ([
The "top-down" and "bottom-up" approaches measure the degree of loss of a single financial institution in the event of a systemic risk and the likelihood of a systemic risk occurring at the risk of a single institution, respectively. The former assesses the "fire retardance" of financial institutions, while the latter assesses the "combustibility" of financial institutions. As such, we posit that the combination of the two can provide an integrative description of systemic risk and help reveal the formation, occurrence, and consequences of financial system systemic risk. In this study, we analyze the systemic risk of China's financial system (2007–2018) by comparing ΔCoVaR, MES and SRISK across China's banks, insurance firms and securities firms.
In 2009, the first draft of Adrian and Brunnermeier ([
Graph
Graph
To emphasize the systemic nature of the risk measure, Adrian and Brunnermeier ([
We focus on ΔCoVaR as a measure of systemic risks, and it presents several advantages. First, ΔCoVaR focuses on the contribution of each institution to overall system risk, while traditional risk measures focus on the risk of individual institutions. Moreover, regulation based on the risk of institutions in isolation can lead to excessive risk-taking along systemic risk dimensions. Another merit is that ΔCoVaR is generally sufficient to study the risk spillovers from institution to institution across the whole financial network.
There are several possible methods of measuring CoVaR. Here, we primarily use quantile regressions because of their simplicity and efficient use of data. We also provide a time-varying estimation of CoVaR with macro state variables. The specific methodologies of these estimations are presented in the appendices, and the detailed descriptions of macro state variables are described in the next section.
MES measures a firm's equity loss conditional on the event of a market decline (Brownlees & Engle, [
According to Brownlees and Engle, the MES can then defined as the partial derivative of the system's ES with respect to the weight of a firm i in the economy. Therefore, to estimate this measure, we need to first estimate ES. The detailed estimation via the DCC-GARCH model is presented in the appendices.
The SRISK index of an individual firm measures the expected capital shortage a financial firm would suffer in case of a substantial market decline over a given time period, which is subject to the firm's degree of leverage, its size and its equity loss conditional on a market decline (i.e., MES) (Brownlees & Engle, [
SRISK is simply given by the capital shortfall, which tells us how much capital the firm needs to add if other crises were to happen. In this perspective, the firms with the largest capital shortfall are assumed to be the greatest contributors to the crisis and the most systemically risky. Because the numerical definition of SRISK is associated with LRMES, which is closely related to MES, we also apply a DCC-GARCH model to the estimation of SRISK. These definitions and estimations are also available in the appendices.
Here, we notice again the differences between CoVaR (or ΔCoVaR) and SRISK.
Graph
This paper aims to assess the systemic risk of the Chinese financial market from January 5, 2007 to September 28, 2018, with CoVaR, MES and SRISK representing the three indices of systemic risk. The study sample includes all 43 financial institutions that were publicly listed on either the Shanghai Stock Exchange or the Shenzhen Stock Exchange before January 1, 2007, including 16 banks, 4 insurance companies and 23 security companies (see Table 1 for a complete list of the financial institutions included in the study). The primary dataset is the daily return of the 43 Chinese financial institutions as well as the daily return of the Chinese stock market, which was extracted from the China Stock Market and Accounting Research Database (CSMAR). In addition, we obtained the financial statements of the 43 financial institutions from the RESSET database as well as other data on the financial market (e.g., the three-month treasury bill rate, credit spreads, etc.) from the Wind database.
Table 1. Summary statistics of Chinese financial institutions.
Returns Market Capitalization (millions) Mean Std. Dev Max Min Mean Std. Dev Max Min Ping An Bank 0.00076 0.02582 0.19617 −0.10020 107.15 53.28 259.27 25.41 Bank of Ningbo 0.00099 0.03640 1.40544 −0.10016 44.34 21.25 109.61 15.00 Pudong Development Bank 0.00066 0.02415 0.10031 −0.10026 221.56 86.15 400.66 62.33 Hua Xia Bank 0.00067 0.02507 0.10070 −0.10048 80.06 27.46 158.06 28.81 China Minsheng Bank 0.00050 0.02229 0.10101 −0.10000 185.32 60.65 317.09 73.22 China Merchants Bank 0.00073 0.02300 0.10026 −0.10007 308.63 124.72 719.74 134.30 Bank of Nanjing 0.00074 0.02668 0.72182 −0.10013 37.72 17.66 87.11 13.90 Industrial Bank 0.00081 0.02553 0.38799 −0.10020 219.82 83.13 403.23 58.60 Bank of Beijing 0.00049 0.02680 0.81440 −0.10013 97.91 31.09 164.92 41.60 Agricultural Bank of China 0.00049 0.01417 0.10122 −0.09899 892.16 147.58 1396.76 673.39 Bank of Communications 0.00037 0.02466 0.71392 −0.10058 211.23 57.78 435.88 108.13 Industrial and Commercial Bank 0.00032 0.01751 0.10053 −0.10043 1233.80 248.55 2218.51 854.64 China Everbright Bank 0.00045 0.01867 0.18065 −0.09917 144.07 29.76 253.99 93.16 China Construction Bank 0.00043 0.01931 0.32248 −0.10094 51.55 12.49 101.88 33.12 Bank of China 0.00020 0.01751 0.10164 −0.10040 719.93 166.59 1330.09 479.57 China CITIC Bank 0.00052 0.02961 0.96035 −0.10025 182.93 47.70 334.76 106.91 Ping An Insurance 0.00089 0.02585 0.38432 −0.10004 320.05 164.76 853.18 99.46 New China Life Insurance 0.00088 0.02729 0.13720 −0.10008 82.60 26.80 149.57 40.25 China Pacific Insurance 0.00051 0.02770 0.60567 −0.10007 160.25 47.14 387.39 80.70 China Life Insurance 0.00056 0.03192 1.06197 −0.10007 527.23 205.86 1563.43 268.83 Shenwan Hongyuan 0.00239 0.10397 3.04321 −0.10037 141.22 48.68 295.65 95.33 Northeast Securities 0.00248 0.11888 5.96083 −0.10025 19.24 9.18 52.24 1.63 Guoyuan Securities 0.00135 0.07066 3.26916 −0.10016 29.57 13.69 78.25 3.61 Sealand Securities 0.00216 0.06890 2.57540 −0.10024 18.08 12.55 53.35 0.78 GF Securities 0.00011 0.02697 0.10040 −0.10029 94.37 24.16 182.79 56.29 Changjiang Securities 0.00157 0.07662 3.59313 −0.10042 34.11 16.96 91.29 9.40 Shanxi Securities 0.00044 0.03350 0.70128 −0.10032 26.22 10.16 69.67 12.60 Western Securities 0.00115 0.03624 0.66667 −0.10015 40.40 25.07 114.20 12.50 Guosen Securities 0.00099 0.03271 0.44082 −0.10008 130.38 42.28 284.62 65.27 CITIC Securities 0.00072 0.02895 0.10043 −0.10016 163.42 54.81 386.29 81.84 Sinolink Securities 0.00115 0.04154 1.29621 −0.10032 25.71 14.47 100.50 5.58 Southwest Securities 0.00079 0.02765 0.10039 −0.10027 28.40 14.27 76.94 0.72 Haitong Securities 0.00094 0.03214 0.10050 −0.10019 109.28 44.75 244.54 2.29 Orient Securities 0.00044 0.03266 0.43968 −0.10022 92.41 30.73 216.39 47.85 China Merchants Securities 0.00011 0.02477 0.10039 −0.10013 84.62 33.51 222.45 38.17 The Pacific Securities −0.00003 0.02959 0.10095 −0.10030 22.94 11.36 69.35 7.64 Dongxing Securities 0.00078 0.03584 0.44009 −0.10015 53.10 13.94 106.27 26.75 Guotai Junan Securities 0.00014 0.02839 0.43988 −0.10017 145.99 23.24 265.50 105.53 Industrial Securities 0.00046 0.02956 0.48600 −0.10048 40.84 16.79 98.23 17.51 Soochow Securities 0.00046 0.02903 0.13077 −0.10027 29.36 14.55 78.81 12.44 Huatai Securities 0.00030 0.02674 0.10038 −0.10022 81.62 29.52 186.42 40.71 Everbright Securities 0.00009 0.02730 0.29981 −0.10005 56.11 20.12 128.72 25.81 Founder Securities 0.00063 0.02925 0.43590 −0.10050 54.80 24.06 135.58 22.33 Total Liabilities (billions) Total Assets (billions) Mean Std. Dev Max Min Mean Std. Dev Max Min Ping An Bank 1508.66 962.22 3128.04 267.33 1605.31 1036.81 3359.73 274.12 Bank of Ningbo 441.41 294.18 997.30 64.65 468.83 310.97 1062.57 70.17 Pudong Development Bank 3054.32 1616.66 5611.09 671.54 3246.53 1748.27 6032.51 696.74 Hua Xia Bank 1361.69 593.32 2350.42 422.78 1440.28 645.12 2527.18 434.58 China Minsheng Bank 2807.37 1573.43 5449.41 676.63 2989.72 1687.74 5833.69 696.48 China Merchants Bank 3219.23 1504.94 5650.55 903.16 3448.61 1636.24 6140.41 959.40 Bank of Nanjing 476.29 353.33 1127.12 68.49 507.60 373.06 1200.51 78.03 Industrial Bank 3058.45 1815.68 5957.74 639.91 3241.94 1943.23 6387.64 664.57 Bank of Beijing 1119.71 621.51 2290.45 304.09 1202.64 664.38 2474.53 330.49 Agricultural Bank of China 10673.81 7087.84 20640.52 421.41 11398.56 7573.01 22237.32 485.05 Bank of Communications 5024.01 2002.91 8355.18 1867.59 5387.33 2183.64 9007.13 1975.50 Industrial and Commercial Bank 15379.11 4888.26 23958.65 7214.27 16565.23 5417.59 26098.19 7694.96 China Everbright Bank 4174.45 2453.00 9730.78 1406.79 4452.99 2608.55 10324.43 1487.50 China Construction Bank 12802.49 4672.47 20519.11 5040.88 13788.53 5118.29 22366.20 5464.46 Bank of China 10752.02 3471.36 16606.74 5040.88 11591.09 3795.72 17999.80 5464.46 China CITIC Bank 2871.60 1556.43 5161.79 0.60 3078.35 1666.05 5534.50 0.86 Ping An Insurance 641.97 754.98 3555.88 6.78 794.81 760.57 3830.36 133.49 New China Life Insurance 3376.44 4321.82 12150.57 12.50 3597.68 4584.52 12876.08 21.25 China Pacific Insurance 262.69 225.01 739.57 0.74 351.23 232.88 909.92 87.22 China Life Insurance 1595.66 628.35 2714.41 609.15 1825.71 691.12 3040.68 722.83 Shenwan Hongyuan 4.14 8.75 49.66 0.02 17.25 24.72 78.04 0.80 Northeast Securities 23.88 19.92 64.85 0.06 30.82 24.21 75.10 0.54 Guoyuan Securities 18.68 14.12 52.61 0.20 32.59 19.33 71.01 0.61 Sealand Securities 16.48 18.62 53.04 0.18 21.76 23.78 67.42 0.39 GF Securities 111.87 108.63 358.36 0.08 146.26 134.34 427.67 0.35 Changjiang Securities 36.97 28.89 101.52 3.37 49.51 35.01 117.39 2.74 Shanxi Securities 439.88 630.18 1977.81 5.59 462.41 652.47 2080.10 11.68 Western Securities 686.34 837.13 2581.94 0.01 723.05 876.72 2734.81 1.11 Guosen Securities 722.43 810.80 2581.94 0.01 768.90 844.09 2734.81 1.11 CITIC Securities 186.48 115.37 418.94 52.16 267.02 137.88 527.58 84.06 Sinolink Securities 13.35 10.74 43.77 1.72 21.27 16.63 59.50 2.38 Southwest Securities 20.12 18.44 64.01 0.53 30.87 24.50 81.67 0.55 Haitong Securities 120.71 90.98 362.17 0.28 182.96 117.03 459.95 1.19 Orient Securities 118.37 123.25 729.62 43.55 145.65 126.10 763.49 68.14 China Merchants Securities 133.49 129.22 729.62 43.55 167.06 136.34 763.49 68.14 The Pacific Securities 73.26 206.54 816.53 1.96 80.71 214.68 854.48 3.40 Dongxing Securities 5226.89 6069.68 15239.81 34.78 5558.42 6421.87 16315.36 53.37 Guotai Junan Securities 5622.33 6170.07 15928.77 195.79 6000.26 6516.74 17048.73 297.08 Industrial Securities 3081.24 4516.33 12150.57 12.09 3284.74 4790.70 12876.08 20.77 Soochow Securities 194.18 211.09 739.57 6.96 236.96 250.97 909.92 14.48 Huatai Securities 1465.59 2622.12 8374.26 31.06 1599.27 2786.47 8872.13 47.93 Everbright Securities 1415.82 2646.56 8374.26 24.13 1535.22 2817.85 8872.13 46.40 Founder Securities 3283.72 4095.72 10880.73 10.50 3516.43 4353.81 11628.82 24.61
The assets and liabilities in Table 1 show that the bank sector accounts for the largest proportion of the financial industry while security companies account for the smallest proportion. However, few Chinese insurance firms have been listed, which increases the difficulty of drawing certain absolute conclusions. As for the returns, security companies generally have higher returns than the other two sectors, although some of them also suffer relatively low or even negative returns. The volatility of the returns of the security sector is higher than others due to the characteristic of their businesses.
To calculate CoVaR, the macro state variables should be carefully selected. Following Adrian and Brunnermeier ([
Graph
- Real Estate Index (RESI). The real estate index is used to measure changes in real estate prices. With March 1, 2013 as the cutoff point, the previous stock exchange index code is 399200 and the subsequent index code is 399241. We select the index daily logarithmic rate of return as the data sample.
- Volatility (VIX). Generally, the implied volatility is calculated from options that expired at approximately 30 days, thus reflecting the investor's volatility (risk) expectations for the next 30 days. Since the volatilities we extracted from RESSET are calculated from historical data, we choose the daily volatility of logarithmic returns based on the GARCH model of the CSI 300.
- Three-month Treasury Bill Rate (TBR3M). We use the rate of return of the three-month treasury bill announced by Chinabond (www.chinabond.com.cn/).
- Liquidity Spread (LIQSPR). Adrian and Brunnermeier ([
5 ]) measure LIQSPR by the yield spread between the ten-year treasury rate and the three-month bill rate. We choose to use the spread between China's 3-month treasury bill yield and SHIBOR 7-day repo rate as the liquidity spread. - Credit Spread (CRESPR). Here, we use the spread between the 1-year AAA corporate bond yield and the 1-year government bond yield.
- Yield Spread (YIESPR). We indicate this factor by the spread between 10-year and 3-month government bill yields.
Table 2 provides the summary statistics of the above six macro state variables. This table shows that RESI has the highest volatility with a standard variance of 0.0228. The average LIQSPR is negative, which represents that higher liquidity usually brings higher returns.
Table 2. Summary statistics of macro state variables.
Mean Std. Dev Max Min 0.0005 0.0228 0.0997 −0.0961 0.0168 0.0077 0.0410 0.0060 0.0258 0.0079 0.0511 0.0080 0.0119 0.0039 0.0269 0.0043 −0.0034 0.0078 0.0164 −0.0652 0.0112 0.0065 0.0263 −0.0148
When estimating SRISK, we need to know the liability and market capitalization of the individual institutions. We obtain the liabilities from firms' seasonal financial reports and assume that the liabilities of any day in a season are the same and equal to the average of the liabilities at the end of the last season and the end of this season. Regarding market capitalization, we consider the daily data on the total stock market value of each institution.
The study aims to use the three systemic risk measures, i.e., CoVaR, MES, and SRISK, to analyze the Chinese financial market and its systemic risks over time, the differences across financial industries, and the divergent implications of the three measures. Figures 1–4 show the changes in China's financial systemic risk over time as indicated by the arithmetic average of the systemic risks of 43 financial institutions in the three measures. We focus on two special time periods: the 2007–2009 financial crisis and the 2014–2016 China stock market crash. The results showed that ΔCoVaR and MES were both high during these two periods, which is in line with our expectations: when the entire financial system suffers a shock, the risks faced by financial institutions increase. However, SRISK presents no obvious volatility during these periods, although it shows continuous growth, which may be due to the close relationship between SRISK and the financial institutions' liabilities and market capitalization, and the continuous expansion of financial institutions also increases their risk capital.
Graph: Figure 1. CoVaR of the Chinese financial institutions (1/5/2007 – 9/30/2018).
Graph: Figure 2. ΔCoVaR of the Chinese financial institutions (1/5/2007 – 9/30/2018).
Graph: Figure 3. MES of the Chinese financial institutions (1/5/2007 – 9/30/2018).
Graph: Figure 4. SRISK (¥billions) of the Chinese financial institutions (1/5/2007 – 9/30/2018).
Next, we split the sample time into four time periods representing the financial crisis period, the stationary period, the Chinese stock market crash period and the recent period: 2007 to 2009, 2010 to 2013, 2014 to 2016, and 2017 to 2018, respectively. For each time period, we examine the average of the three systemic risk indicators of the banking sector, the insurance sector and the security sector in this interval.
In Figures 5–7, we report the fluctuation of Chinese financial institutions measured by ΔCoVaR, MES and SRISK, respectively, during the period from 2007 to 2009, namely the financial crisis period. The results show that both ΔCoVaR and MES showed an increase trend in 2008, which was a proof of the increasing systemic risks during the financial crisis. For a comparison among the three financial sectors, the banking sector suffered most of the systemic risks and accounted for most of them as well. However, when we focus on SRISK, we cannot find obvious volatility and the banking sector showed the highest SRISK, which was quite different from the other two measures. This difference may result from the sharp increase of the magnitude of Chinese banks during that period.
Graph: Figure 5. ΔCoVaR of the Chinese financial institutions (1/5/2007 – 12/31/2009).
Graph: Figure 6. MES of the Chinese financial institutions (1/5/2007 – 12/31/2009).
Graph: Figure 7. SRISK of the Chinese financial institutions (1/5/2007 – 12/31/2009).
In Figures 8–10, we report the fluctuation of Chinese financial institutions measured by ΔCoVaR, MES and SRISK respectively, during the stationary period from 2010 to 2013. In this time interval, ΔCoVaR and MES were relatively steady. ΔCoVaR indicates that the banking sector played an important role in the contribution of systemic risks, while MES shows that the security sector was the most systemically risky. At the same time, SRISK grew steadily and the banking sector still presented the highest SRISK, followed by the insurance sector.
Graph: Figure 8. ΔCoVaR of the Chinese financial institutions (1/1/2010 – 12/31/2013).
Graph: Figure 9. MES of the Chinese financial institutions (1/1/2010 – 12/31/2013).
Graph: Figure 10. SRISK of the Chinese financial institutions (1/1/2010 – 12/31/2013).
In Figures 11–13, we show the fluctuation of Chinese financial institutions measured by ΔCoVaR, MES and SRISK, respectively, during the Chinese stock market crash period from 2014 to 2016. It is obvious that ΔCoVaR and MES significantly increased in 2015, when the Chinese stock market suffered greatly. Similar to the stationary period, the banking sector showed the highest ΔCoVaR while the MES of the security sector varied most dramatically. Both ΔCoVaR and MES indicated that these three sectors had similar fluctuating trends. However, the SRISK of the banking sector continued to increase, the insurance sector varied in 2014 and then remained stable, and the security sector showed a significant increase in 2015. Therefore, the security companies in China may suffer the most systemic risks when there is negative turbulence in the stock market.
Graph: Figure 11. ΔCoVaR of the Chinese financial institutions (1/1/2014– 12/31/2016).
Graph: Figure 12. MES of the Chinese financial institutions (1/1/2014– 12/31/2016).
Graph: Figure 13. SRISK of the Chinese financial institutions (1/1/2014– 12/31/2016).
In Figures 14–16, we report the recent fluctuations of the Chinese financial institutions measured by ΔCoVaR, MES and SRISK respectively, from 2017 to 2018. The results show that all three measures went through steady fluctuations because a severe crisis was not observed over the past two years. We focus on the sector differences in this interval. The ΔCoVaR results showed that the insurance sector seemed to bear the same systemic risk as the banking sector, which is quite different from the previous periods. As for the MES, the security sector had more obvious variations and was considered the most systemically risky. The changes in SRISK remained the same, where banks still suffered the most systemic risks because of their huge liabilities and market capitalizations.
Graph: Figure 14. ΔCoVaR of the Chinese financial institutions (1/1/2017– 9/28/2018).
Graph: Figure 15. MES of the Chinese financial institutions (1/1/2017– 9/28/2018).
Graph: Figure 16. SRISK of the Chinese financial institutions (1/1/2017– 9/28/2018).
All of these variations are visually represented in Table 3 and Figures 17–19, which shows that 2015 witnessed a distinct increase in all three measures of systemic risk, thus indicating that the domestic financial crush could impact greatly on financial institutions. In contrast, although ΔCoVaR and MES reacted sensitively to the global financial crisis in 2009, the variation was relatively slight compared to that in 2015.
Graph: Figure 17. ΔCoVaR of the Chinese financial institutions (1/5/2007– 9/28/2018).
Graph: Figure 18. MES of the Chinese financial institutions (1/5/2007– 9/28/2018).
Graph: Figure 19. SRISK of the Chinese financial institutions (1/5/2007– 9/28/2018).
Table 3. Systemic risk measures of Chinese financial institutions (2007–2018).
Financial Banking Insurance Security Financial Banking Insurance Security Financial Banking Insurance Security 2007 0.0172 0.0195 0.0191 0.0152 0.0206 0.0262 0.0288 0.0153 133.80 79.06 12.124 0.328 2008 0.0153 0.0194 0.0178 0.0120 0.0089 0.0256 0.0222 −0.0050 143.32 143.25 30.110 0.865 2009 0.0173 0.0227 0.0218 0.0128 0.0312 0.0376 0.0362 0.0258 173.81 174.72 36.184 0.568 2010 0.0141 0.0171 0.0150 0.0119 0.0238 0.0298 0.0274 0.0190 173.72 209.76 44.653 1.407 2011 0.0140 0.0155 0.0135 0.0130 0.0287 0.0251 0.0243 0.0319 182.95 299.91 54.733 1.590 2012 0.0140 0.0134 0.0178 0.0137 0.0285 0.0217 0.0258 0.0337 191.39 349.55 89.942 1.358 2013 0.0166 0.0181 0.0180 0.0154 0.0398 0.0311 0.0344 0.0468 202.54 399.07 54.887 1.372 2014 0.0144 0.0162 0.0148 0.0131 0.0285 0.0282 0.0295 0.0285 221.32 439.51 47.055 2.255 2015 0.0278 0.0302 0.0273 0.0261 0.0596 0.0425 0.0508 0.0729 249.85 480.91 68.380 6.031 2016 0.0177 0.0174 0.0189 0.0176 0.0482 0.0244 0.0376 0.0667 209.91 536.95 55.455 9.263 2017 0.0116 0.0122 0.0128 0.0109 0.0218 0.0092 0.0146 0.0318 235.51 606.12 61.183 8.015 2018 0.0127 0.0136 0.0166 0.0114 0.0177 0.0157 0.0211 0.0185 247.54 636.91 64.627 8.486
This paper aims to use the data on China's financial market from January 5, 2007 to September 28, 2018 to describe the different influences on the formation, occurrence, and consequences of the financial system of Chinese financial institutions from three aspects: ΔCoVaR, MES and SRISK. Among these, ΔCoVaR reflects the impact of the loss of a single financial institution on the entire financial system, which is a "bottom-up" analysis; and MES and SRISK reflect the effects of risks of the system to a single financial institution when the entire financial system is in crisis, which is a "top-down" analysis. These three indicators have different meanings, although they can measure systemic risk to a certain extent and are widely used in the study of systemic risk.
First, we observe the changes in the systemic risk of Chinese financial institutions over the past 11 years. Our research focuses on two special time periods within the sample interval: before and after the 2008 global financial crisis and during the 2015 China stock market crash. For the Chinese financial market, the former is affected by the external financial system risk while the latter is affected by the domestic financial crisis risk. Through the study of these two time periods, we can also explore whether the systemic risks faced by Chinese financial institutions are related to the internal and external nature of the financial crisis. The results show that ΔCoVaR and MES have obvious fluctuations during the financial crisis, which can reflect that financial institutions are facing greater risk during the crisis, while
Graph
Second, for each time period, we consider the differences among the various Chinese financial sectors in accordance with the systemic risk during that time period. For each interval, we show the simple average of the financial risk indicators of the financial institutions in each sector. According to these figures, we obtained the following results. (
The findings suggest that banks are the primary source of systemic risk in China and will be burdened with the greatest capital shortfall in the event of a severe financial crisis in the future. The securities companies are more sensitive to fluctuations of the stock market, and the insurance companies may play a more important role in systemic risks as time passes.
The results as found in this study point to the similarities and differences between China and the USA in terms of the systemic risks of their financial systems. As Brownlees and Engle ([
In terms of the source of the systemic risk, the banking sector in the USA exhibited a higher ΔCoVaR than both the insurance and security sectors prior to the financial crisis (Bernal, Gnabo, & Guilmin, [
Comments and suggestions from Chinese Economy editor Professor Ligang Zhong and reviewers are gratefully acknowledged.
Suppose that a system consists of m different financial firms and that the aggregate result of the system is modeled as follows:
(
Graph
where
Graph
Recall that
Graph
Graph
(
Graph
Definition:
Graph
Graph
Graph
Graph
(
Graph
We denote institution i's contribution to j as follows:
(
Graph
In this paper, we focus on the conditioning event of
Graph
Graph
Graph
Graph
Additionally,
Graph
Graph
(
Graph
This indicator denotes the difference between the
Graph
Graph
To estimate the generalized
Graph
Assume that dependent variable y is a linear function of explanatory variables x:
(
Graph
We can obtain
Graph
Graph
(
Graph
where
Graph
Graph
Graph
Graph
Thus, after minimizing
Graph
(
Graph
Graph
The quantile regressions incorporate estimates of the conditional mean and the conditional volatility to produce conditional quantiles without the distributional assumptions that would be needed for an estimation via OLS. This characteristic makes the quantile regression method useful for broad application.
Consider the predicted value of the financial sector
Graph
(
Graph
Using a particular predicted value of
Graph
Graph
Graph
(
Graph
Graph
(
Graph
In addition to the previous method that focuses on the estimation of a constant
Graph
Graph
(
Graph
(
Graph
Then, we can obtain the following:
(
Graph
(
Graph
With the previous effort, we compute
Graph
(
Graph
Graph
Graph
Graph
Graph
Graph
(
Graph
where
Graph
Graph
Graph
Graph
Graph
Graph
Graph
Therefore, we can obtain the
Graph
(
Graph
Graph
Graph
Here, we estimate
Graph
First, we build a bivariate process of the firm and market returns:
(
Graph
where
Graph
Graph
Graph
Graph
Let
Graph
Graph
(
Graph
(
Graph
Second, we estimate the tail expectation in (
(
Graph
where
Graph
Graph
Graph
Graph
Graph
Third, we estimate
Graph
Graph
(
Graph
where
Graph
Finally, we estimate
Graph
Graph
(
Graph
Based on the previous estimation, we can derive the
Graph
Because
Graph
Graph
Graph
(
Graph
where
Graph
Graph
(
Graph
Here, we notice again the differences between
Graph
Graph
Graph
Graph
Graph
The key to estimating
Graph
Graph
Graph
(
Graph
In the last step, we compute the
Graph
(
Graph
Substituting
Graph
Graph
To maintain consistency for these three indices, let
Graph
Graph
Graph
By Hua Zhou; Wenjin Liu and Liang Wang
Reported by Author; Author; Author