Systemic risk in Taiwan stock market

Many financial crises have resulted from systemic risk which is caused by

idiosyncratic distress, leading to the crunch of the whole system. The interaction between financial distress and systemic risk due to the effect of idiosyncratic distress has been discussed recently ( Fu, 2009; Campbell et al., 2008; Ang et.

al., 2006). Amongst the idiosyncratic distress, sector-specific risk, which is caused by a group of interconnected institutions, has been given as the main reason for sudden increases in systemic risk, leading to the formalization of financial crises. In this research, taking Taiwan’s stock market as an example and collecting data from 2000 to 2010 which contained the 2001 dot-com bubble and the 2007-09 financial crisis, we adopted the CoVaR model to empirically explore the impact of sector-specific idiosyncratic risk on the systemic risk of the whole financial system and attempt to investigate the links between financial crises, systemic risk, and the idiosyncratic risk of a sector-specific anomaly.

International stock market transmission across different markets has been comprehensively studied. Many recent studies have discussed the volatility spillover effect of stock market returns. These papers have shown some typical characteristics of the volatility spillover effect among different stock markets (Martens and Poon, 2001; Goetzmann , Li and Rouwenhorst , 2001; Worthington and Higgs, 2004; Michelfelder RA, 2005; Scheicher M, 2001; Bekaert and

Harvey, 1997; Gokcan S, 2000; Sheu and Cheng , 2011). In addition, within a stock market, during times of financial crisis, losses tend to spread from a single sector across other sectors, leading to increased system-wide risk and probable

deterioration of the whole stock market system. This financial system instability or potential catastrophe, caused by idiosyncratic events and resulting in risk to the entire financial system, is defined as systemic risk. Many financial crises are initially caused by a “sector-specific” idiosyncratic distress of a country, then spill over across other sectors to increase systemic risk, consequently leading to worldwide crashes. These crises result from systemic risk, caused by

idiosyncratic distress, which cannot be reduced through portfolio diversification.

Several reasons might lead to systemic risk and there are two commonly used assessments for measuring systemic risk, i.e. the “too big to fail” and “too interconnected to fail” test. The “too big to fail” test considers an asset size relative to the marketplace, i.e. market share concentration and the “too

interconnected to fail” measures the likelihood and extent of negative impact to the overall economic system from the failure of a group of correlated institutions.

Traditionally, Value-at-Risk (VaR) is widely used to assess the risk of loss of specific financial assets and provides a measure to manage the market risk of assets. However, VaR focuses on these assets in isolation and does not consider external impacts. Using value-at-risk(VaR) to assess assets, it seems negligible to capture the systemic risk and the true risk is often underestimated when other assets come under stress. For investors to control the risk for underlying assets, the appropriate risk measure could not only assess the risk of a sector’s economic activities itself, but also consider the impact to systemic risk from the idiosyncratic distress. Thus, to supplement the drawback of VaR for estimating market risk, it is necessary to employ more interdependent and comprehensive measures that could consider the interconnected nature of the financial system and gauge the increased systemic risk due to the distress of other financial assets. However, it was not until the financial crisis of 1998 that

some researches begun to discuss systemic risk and develop approaches to measure it. Among these approaches, the “CoVaR” method, proposed by Adrian and Brummermeier (2008), is a more interdependent and comprehensive method and has been successfully employed to capture systemic risk.

Following this CoVaR method, the Taiwan stock market during periods of 2000-2010 was taken as an example to explore systemic risk caused by

“sector-specific” distress for some reason. First, Taiwan rose to second in the world for global IT competitiveness through its strengths in R&D and nurturing technology talent（Business Software Alliance, 2008）. Second, given the economic success of Taiwan in the last several decades, many global investors have taken an active interest and hold an index investment position in this stock market. Finally, the technology industry and financial industry rank as the top two important industries in Taiwan’s stock market. Thus, it is an appropriate objective to measure and backtest the systemic risk during the 2001 dot-com bubble and the 2007-09 financial crisis.

The purpose of this paper is to empirically explore the impact of “sector-specific”

idiosyncratic risk on the systemic risk of the whole financial system. First, this paper examines the magnitude of systemic risk and the marginal risk

contribution caused by sectors to the overall systemic risk on the Taiwan stock market. Further, the differences, between VaR of sectors and sector-specific marginal systemic risks, were also compared. Finally, we endeavored to

investigate the links between marginal systemic risk caused by a sector-specific anomaly and the impact of global financial crises on Taiwan stock market. It is hoped the results of this study will be a useful tool for those stock investors to accurately identify the true systemic risk of Taiwan’s stock market and to

properly allocate their investment portfolios across sectors according to their true

risk contributions.

The structure of this paper is as follows. Section 3.2 discusses risk contagion theory. Section 3.3 describes the methodology used to measure the systemic risk of Taiwan’s stock market and the implementation of the model, which is designed to solve the research problems. Section 3.4 demonstrates the results of the research and other computation analysis. Finally, section 3.5 summarizes some findings and conclusions.

3.2 Risk contagion theory and literature with regarding to sector