A risk measurement study evaluating the impact of COVID-19 on China

Annals of Operations Research https://doi.org/10.1007/s10479-023-05178-9 ORIGINAL RESEARCH A risk measurement study evaluating the impact of COVID-19 on China’s financial market using the QR-SGED-EGARCH model Malin Song1 · Zixu Sui1 · Xin Zhao1 Accepted: 6 January 2023 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2023 Abstract Due to the significant impact of COVID-19, financial markets in various countries have undergone drastic fluctuations. Accurately measuring risk in the financial market and mastering the changing rules of the stock market are of great importance to macro-control and financial market management of the government. This paper focuses on the return rate of the Shanghai Composite Index. Using the SGED-EGARCH(1,1) model as a foundation, a quantile regression is introduced to establish the QR-SGED-EGARCH(1,1) model. Further, the corresponding value at risk (VaR) is calculated for a crisis and stable period within each model. To better compare the models, the Cornish-Fisher expansion model is included for comparison. According to the Kupiec test, VaR values calculated by the QR-SGED-EGARCH(1,1) model are superior to other models at different confidence levels most of the time. In addition, to account for the VaR method’s inability to effectively measure tail extreme risk, the expected shortfall (ES) method is introduced. The constructed model is used to calculate the corresponding ES values during different periods. According to the evaluation index, the ES values calculated by the QR-SGED-EGARCH(1,1) model have a better effect during a crisis period with the model showing higher accuracy and robustness. It is of great significance for China to better measure financial risk under the impact of a sudden crisis. Keywords COVID-19 pandemic · Economic security · Financial market · Risk prediction · QR-SGED-EGARCH model 1 Introduction With the development of non-traditional security theory, environmental, energy, ecological, and financial security have drawn a lot of attention (Nguyen et al., 2020). The continuous spread of COVID-19 around the world has significantly impacted the social and economic development of many nations, including China. In particular, China’s financial markets have B Xin Zhao 1 School of Statistics and Applied Mathematics, Anhui University of Finance and Economics, Bengbu 233030, People’s Republic of China 123 Annals of Operations Research suffered severe shocks, which have seriously endangered its financial security and stability (Le et al., 2020). The SSE Composite Index fell 7.72% on the first trading day after the Spring Festival (February 3, 2020), which was the largest one-day decline in nearly five years. Overseas financial markets were also seriously affected, with the global stock market undergoing a dramatic adjustment. The U.S. stock market experienced sharp declines with the Dow Jones index dropping from nearly 30,000 points to around 20,000 points (Afum et al., 2020). In addition, the spread of COVID-19 has severely damaged the fundamentals of listed companies. Understandably, fear of the pandemic and the need for national prevention and control measures have restricted consumer behaviors. Consumption in catering, entertainment, tourism, transportation, among others was greatly reduced, which directly impacted the operating cash flow and income of relevant enterprises. The temporary rise in the unemployment rate caused by forced unemployment of the labor force and the sharp decline in wage income further decreased demand in those areas (Fang et al., 2020). Due to the COVID-19 pandemic, the path for listed companies to issue bonds for share buybacks was also interrupted, which directly affected stock valuations. The COVID-19 pandemic also revealed the sensitivity of the stock market to unusually volatile "overreactions" in the short term during sudden emergencies (Lasfer et al., 2003). This not only amplifies the vulnerability of the secondary financial market system, but also leads to abrupt changes in the original stable correlation state between stock markets (Guidolin et al., 2019), resulting in risk contagion among stock markets and a significant increase in the risk spillover effect (White et al., 2015). Thus, we can conclude that COVID-19 has led to a sharp rise in market panic and market uncertainty, increased market volatility, enhanced risk resonance and the risk spillover effect among regional financial markets, caused obvious risk transmission among stock markets, and significantly increased the cross-market risk spillover effect. In these types of uncertain times, it is extremely important to predict the risks of the financial market, to grasp the changing rules of the stock market, and for the government to conduct macroeconomic regulation and financial market management (Song et al., 2020). To date, many scholars have analyzed the risk measurement of financial return series. In 1993, the G30 Group published a report named, "Practice and Rules of Derivatives," which introduced the concept of value at risk (VaR) to measure market risk for the first time. Because of its excellent ability to quantify and abstract risks, VaR has been widely recognized by the financial community since its publication. Duffie and Pan (1997) provided a comprehensive description of VaR, including its theory, concept, background, estimation method, and stock market application. Penza et al. (2001) reiterated the effectiveness of VaR for measuring financial risk. Since then, VaR has become the basis of future research on financial market risk measurement. In studying VaR risk measurement methods, the generalized autoregressive conditional heteroscedasticity model (GARCH) model has attracted researchers’ attention. Engle (1982) first proposed the autoregressive conditional heteroscedasticity model (ARCH), and Bollerslev (1986) improved it into the generalized GARCH model. Ricardo used the ARCH family model to model and calculate VaR under different distribution conditions. In the same study, he also tested the accuracy of risk measurements. Giot and Laurent (2004), Angelidis and Degiannakis (2005), Giot (2005), and McMillan and Speight (2007) also modeled and estimated VaR based on the ARCH model, the GARCH model, and the GARCH extension model, respectively. Among them, Angelidis and Degiannakis (2005) found that the estimation of VaR by the GARCH model fitted under a t-distribution was accurate enough to predict five indexes, including the Dow Jones Industrial Average. Ansari et al. (2020) used the GARCH model to measure the risk of Chinese open-ended funds under a t-distribution and 123 Annals of Operations Research GED distribution, respectively, and concluded that the VaR model is better under GED distribution. The essence of VaR is quantile. Engle and Manganelli (2004) proposed to introduce the concept of quantile into the calculation of VaR (CAViaR mod (...truncated)