VIETNAM NATIONAL UNIVERSITY HANOI UNIVERSITY OF ECONOMICS & BUSINESSFACULTY OF FINANCE AND BANKING EXAMINING THE TAIL RISK CONNECTEDNESS AMONG VIETNAMESE STOCK SECTORS Instructor : PhD..
Tail Risk
Tail risk refers to rare events with a low probability of occurrence, typically less than 5% or 1%, that can lead to severe and unpredictable consequences In probability theory, these risks are associated with the extreme ends of a probability distribution, highlighting their potential for significant impact when they do occur.
A tail risk event, as defined by Temelkov (2021), is when an investment's value deviates more than three standard deviations from its average, resulting in a low probability but potentially severe negative impacts on portfolios and financial markets These events are thought to contribute to heightened market volatility, with losses from tail risk events significantly exceeding the anticipated drop in asset prices.
Tail risk, as defined by Gordon (2022), is a type of portfolio risk that arises when an investment's value fluctuates beyond three standard deviations from its average price This risk assesses the variability of investment returns in relation to their mean, with occurrences being relatively rare and manifesting at both extremes of the normal distribution curve.
Tail risk, as defined by Hayes (2022), refers to the potential for significant portfolio losses when the likelihood of an investment's movement exceeds three standard deviations from the mean, surpassing what a normal distribution would predict This concept highlights the risk of losses stemming from rare events, which, although unlikely, can have substantial impacts Tail risks encompass events that possess a low probability of occurrence and can manifest at both extremes of a normal distribution curve.
2.1.2.2 The cause of tail risk
The financial and economic sectors would bear the brunt of the fallout if tail risks materialize Here are a few potential reasons why tail risk might occur:
Natural disasters, including earthquakes, tsunamis, floods, and droughts, can significantly impact local economies and hinder regional development, particularly in the agriculture sector Additionally, unpredictable events like plane crashes and industrial accidents contribute to tail risks, further threatening economic stability and safety.
Massive market volatility: Significant market swings, such as large stock market price declines, economic downturns, bank failures, etc., can have a significant influence on communities, regions, and industries.
Political Events: Events like wars, revolutions, political crises, etc can have a major impact on economies.
2.1.2.3 The measure of tail risk
Tail risk, as highlighted by Yieldstreet (2022), is a crucial concept in investment considerations, measuring the likelihood of an asset's performance deviating significantly from its average This concept predicts rare events, indicating the probability of unlikely occurrences that may happen more frequently than many investors realize, particularly in a global context.
Jiang (2014) identifies two primary methods for assessing tail risk dynamics in stock returns: one utilizing option price data and the other relying on high-frequency return data Notable examples of the option-based approach are presented by Bakshi et al.
Research on risk-neutral skewness and kurtosis by 2003, along with Bollerslev et al (2009) investigating the relationship between variance risk premium and equity premium, and Backus, Chernov, and Martin (2012) analyzing disaster risk premia from index options, highlights the significance of tail estimation Bollerslev and Todorov (2012) provide examples of tail estimation using high-frequency data However, these methodologies face data limitations, such as a maximum 20-year sample horizon for returns and inapplicability to low-frequency cash flow data In contrast, our tail risk series, derived from returns and sales growth data since 1963, can be applied in various contexts with a substantial cross-section of data.
Calculating tail risk varies based on the asset class or portfolio in question, but fundamentally, it can be assessed using standard deviation and normal distribution methods.
The standard deviation is determined by taking the square root of the variance of a dataset To calculate variance, sum the squares of the differences between each data point and the mean, then divide by the number of data points minus one.
The normal distribution, characterized by its symmetric bell curve, is a continuous probability distribution To determine the likelihood of a value falling outside the mean plus standard deviation range, one can utilize the normal distribution table or the cumulative normal distribution function available in statistical software.
However, it should be noted that this method is only one way of assessing tail risk and cannot cover all the different risk factors of assets or portfolios.
To effectively assess tail risk, it is essential to combine various methods, as each approach has its own drawbacks and limitations By evaluating risk from multiple perspectives, a more comprehensive understanding can be achieved Here are additional methods for calculating tail risk.
Value at Risk (VaR) is a widely utilized risk assessment tool that quantifies the maximum potential loss over a specified time frame, given a certain probability level The calculation of VaR typically relies on standard deviation and the principles of normal distribution.
Conditional Value at Risk (CVaR), often referred to as Expected Shortfall, is a key metric for evaluating tail risk by measuring the average loss that exceeds the Value at Risk (VaR) threshold This calculation relies on the probability distribution of an asset or portfolio, providing a comprehensive view of potential extreme losses.
Monte Carlo Simulation is a risk assessment technique that generates various market scenarios through stochastic modeling By simulating hypothetical price movements, this method enables the calculation of potential financial outcomes based on these scenarios.
RESEARCH METHODOLOGY 3.1 Research ii
How to measure tail risk oe essessessessecsseeseeseesseeesecsecssecsecssecseecseecseecseesseesseesseesseeeneeses 17 3.1.2 @u ¡0/4/10
In contrast to traditional methods that estimate Value-at-Risk (VaR) by first determining return distributions and then indirectly deriving quantiles, we adopt the asymmetric slope CAViaR approach introduced by Engle and Manganelli (2004), which directly estimates VaR This method is particularly flexible compared to the symmetric absolute value and indirect GARCH(1,1) approaches, as it accommodates asymmetric effects Additionally, the asymmetric slope CAViaR model posits that VaR at a specific quantile, set at 5% in our analysis, follows an autoregressive process, mathematically represented as fz„() = Bo + Brfae-1(B) + Boxiis + Baxi, where Bo is the model constant, and the weights of the lagged VaRs reflect the influence of both positive and negative returns on the VaR.
To investigate the dynamic connectedness over time, the study utilizes the Time-Varying Parameter Vector Autoregression (TVP-VAR) method developed by Antonakakis and Gabauer (2017) This methodology integrates the connectedness framework established by Diebold and Yilmaz, allowing for a comprehensive analysis of interconnectedness in financial markets.
(2009, 2012, 2014) and Koop and Korobilis (2014) This framework allows the variances to vary over time via a Kalman Filter estimation with forgetting factors The TVP-VAR(p) model can be expressed as:
The equations presented describe a model where Ve is influenced by the previous state BrZe-1 and an error term €y, with Fr-1 following a normal distribution The vector representations include y; and Z¿_¡, which are WMx1 and Npx1 dimensional vectors, respectively The model incorporates a time-varying coefficient matrix 6; of dimensions N x Np, alongside an error disturbance vector €; with a time-varying variance-covariance matrix S; Additionally, the vector transformations ứec(,) and 0ec(,_Ă) are defined as N*p x1 dimensional, while R; is an Np x N2p dimensional matrix, indicating the complexity and interdependencies within the system.
To compute the generalized impulse response functions (GIRF) and generalized forecast error variance decomposition (GFEVD), it is essential to convert the time-varying parameter vector autoregression (TVP-VAR) into a time-varying parameter vector moving average (TVP-VMA) model, as outlined by the Wold representation theorem.
= 37=o Air€t—j (5) where L= [Iy, we) Op’ is an NpxXN_ dimensional matrix, W=
The matrix [Pes In@—1) °n(p-1)xw] is an Np X Np dimensional matrix that captures the Generalized Impulse Response Functions (GIRFs) of all variables in response to a shock in variable i To analyze these effects, we calculate the differences between a J-step-ahead forecast when variable i is shocked and when it is not This difference directly reflects the impact of the shock in variable i.
GIRF,(J, Oj,ằ Fr-1) = E (Yes let = Ojtằ Fr-1) — EŒ.+;|F¿—1) (6) g _ AJtStEjt Sit _
The equation Gj, I= = 5S 2 Ay, tSt€j, t (8) defines the GIRFs of variable j, where J indicates the forecast horizon, and the selection vector ổ;; assigns a value of one to the j-th position and zero elsewhere The information set F,_, encompasses data available up to time t - 1 Subsequently, we calculate the GFEVD, which represents the variance share that one variable contributes to others, using the formula y1 29 t=1 Pijt.
$2.0) = with Yj $3.0) = 1and }ÿ;~¡ 6,4) = N Based on the GFEVD, we can build the total connectedness index (TCI) as follows:
>i=1¡+j Pye) x 100 = Ab je Lie] Pity? wyyariej Oe)
The connected approach enables the analysis of how a shock in one variable influences other variables Specifically, it defines the total directional connectedness from variable i to all other variables j.
Second, the shock that variable i receives from all other variables j, ie the total directional connectedness FROM others can be defined as:
Finally, the net total directional connectedness can be given by subtracting the total directional connectedness TO others from the total directional connectedness
Net total directional connectedness measures the influence of variable i on the analyzed network A positive net total directional connectedness indicates that variable i acts as a shock transmitter, exerting more influence on the network than it receives Conversely, a negative net total directional connectedness signifies that variable i is a shock receiver, being driven by the dynamics of the network.
To gain deeper insights into the bilateral transmission process between variables i and j, we aim to calculate the net pairwise directional connectedness (NPDC) Unlike the net total directional connectedness, which can obscure critical underlying dynamics, NPDC provides a more nuanced understanding of the interactions between specific variables.
A positive (negative) value of NPDC;;(J) indicates that variable i is driving
This research analyzes secondary data on historical stock prices and daily trading volumes of companies across 10 sectors listed on Vietnam's stock market from February to October 2022 The data, sourced from HOSE and HNX, categorizes companies into industries based on their business activities, as detailed by Investing.com, a reputable stock market update platform in Vietnam The sectors examined include Insurance, Banking, Technology and Information, Accommodation Services, Oil and Gas, Securities, Healthcare, Real Estate, Retail, and Food & Drinks.
From the above historical stock price data, the sector- average stock price is computed by the following formula :
Pe-1 where P, : The closing price today
EMPIRICAL RESULTS ô-ô-5sessesesesesrsrtstrrrsrersrrrsrnrrrarsrsrsrsrsrar 21 4.1 DesCrIDtiVe StAtẽSECS ô HH HH HH HH HH HH HH HH HH TH HH HH TH HH HH HH HH HH HH HH HH 21 4.1.1 SUMMATLY risn .,ÔỎ 21 4.1.2 Correlation Matrix taÌ@ ô+ sk+xE E111 1111111011111
The development and volatility of stocks in different seCtOFS 26 4.2 The Dynamic spillovers between tail risk of 10 StoCK Se@CtOFS -c-ô-ccô¿ 35
Ox b x, b “Ca b Sp, „5 b Or, Cr, Oy b
Between 2020 and 2021, the stock market experienced a dual impact from rising temperatures and unexpected profits, leading to a significant increase in tail risk for various securities stocks This volatility affected the business cycle, causing securities companies' stocks to respond sensitively, often with high margins, regardless of whether they were large firms with a stable market share at the top.
In 2022, the emergence of securities violations significantly hindered the operations of securities companies, contributing to a persistently high tail risk within the industry Despite the implementation of numerous directives and decisive actions by regulatory authorities aimed at rectifying the market across both business sectors and related areas, challenges remain prevalent.
(Source: Author's own synthesis) Figure 4.1.2 Healthcare
Healthcare stocks are known for their stability and unique market position, yet the COVID-19 pandemic significantly impacted the sector Following the outbreak, social isolation measures led to a notable decline in demand for medical services, adversely affecting pharmaceutical businesses On February 10, 2020, healthcare stocks dropped by 5% due to unmet expectations in healthcare services Although there was a temporary surge in pharmaceutical stock prices during the pandemic, these gains could not be sustained, and sales ultimately declined as prolonged isolation continued to affect the industry.
In 2022, the health sector faced increased tail risks, leading to heightened volatility and uncertainty The industry's recovery remains sluggish, with few positive indicators for business activities, suggesting that challenges will persist in the coming years.
(Source: Author's own synthesis) Figure 4.1.3 Insurance
In early 2022, the insurance market experienced stable growth, with premium revenue increasing by 13% compared to the previous year, driven by a 15% rise in life insurance and a 9% increase in non-life insurance However, by mid-2022, various events, including the Russia-Ukraine conflict that raised energy prices, supply chain disruptions, and rising inflation in the US, began to impact the industry These factors prompted the Federal Reserve to take action, leading to a forecasted series of interest rate hikes in the remaining six meetings of the Federal Open Market Committee (FOMC).
In 2022, the Bank of England raised interest rates for the third consecutive time, following the Federal Reserve's initial hike, bringing the target interest rate to 1.9% by year-end This trend indicates that central banks worldwide are focusing on combating inflation rather than fearing economic downturns, particularly due to the ongoing Russia-Ukraine conflict As deposit interest rates are expected to rise, insurance companies are likely to benefit, enhancing their optimism, as investment plays a crucial role alongside their primary income from insurance operations.
(Source: Author's own synthesis) Figure 4.1.4 Oil
The COVID-19 pandemic significantly impacted the oil industry, causing fluctuations in business, import, and export activities by March 2020 Despite expectations for a quick recovery, the industry's performance remained weak through 2020 and mid-2021, as many customers reduced production and consumption, leading to a forecasted 40-50% decrease in gas output Additionally, rising operational costs and a surge in COVID-19 cases negatively affected oil prices, with a notable drop of 8.3% in January 2021, marking the lowest point for the industry's stock in a decade However, by late 2021 and early 2022, crude oil prices began to rise sharply as post-pandemic recovery efforts increased demand, depleting oil reserves and inventories The easing of social distancing measures further boosted consumption, culminating in a significant price increase of 6.8% on November 18, 2021.
29 increase in oil prices also makes the tail risk of this oil industry gradually stabilized and within a manageable range.
(Source: Author's own synthesis) Figure 4.1.5 Banking
In early March 2020, the COVID-19 pandemic triggered a significant decline in the stock market and banking sector, with the banking industry experiencing its largest drop in five years, plummeting by 6.25% on January 28, 2021 This marked the steepest decline in the history of the Vietnamese stock market, with the VN-Index hitting record lows The surge in COVID-19 infections, reaching nearly 15,000 cases in a single day, was a primary factor behind this downturn However, it is notable that the banking sector quickly adapted, showing positive changes and a continuous upward trend even amidst the ongoing pandemic.
During the recovery period, tail risk surged due to the Tan Hoang Minh event and the Van Thinh Phat situation linked to SCB Bank, leading to the lowest industry valuation in a decade Banking stock prices have dropped significantly, ranging from 10% to 60% ACB Securities Company reports that this sharp market decline has positioned bank tail risk in an appealing zone Despite these challenges, the banking sector is anticipated to maintain growth in 2023, driven by robust operations and increased profitability from recovering retail credit demand.
(Source: Author's own synthesis) Figure 4.1.6 Technology
In 2022, particularly in May and September, technology tail risks have significantly increased While the technology and telecommunications sector has thrived during the COVID-19 pandemic by capitalizing on opportunities and benefiting from ongoing digital transformation and government investments in smart transportation, it remains susceptible to fluctuations in both domestic and international markets.
(Source: Author's own synthesis) Figure 4.1.7 Industrial Goods and Services
In early 2022, the Industrial Goods and Services sector faces heightened tail risks as it navigates the post-pandemic recovery phase Many businesses within this industry are still struggling to rebound from the significant losses incurred during the COVID-19 pandemic, while investor cash flows are increasingly shifting towards other sectors.
- NI I lì Mall ll Jt,
(Source: Author's own synthesis) Figure 4.1.8 Real Estate
The real estate industry, previously viewed as having substantial growth potential, faces significant challenges due to the COVID-19 pandemic Many real estate businesses are struggling with project delays, leading to reduced supply and a decline in demand for office and resort rentals This downturn has been reflected in the financial results of listed real estate companies, which have experienced a notable decrease in both revenue and profit compared to pre-pandemic levels Consequently, the tail risk in the real estate sector has intensified, with fluctuations in performance hovering between 2% and 2.5%, showing minimal positive changes by the end of 2020.
By 2021, as regulations eased social distancing, real estate companies began to operate normally, leading to a gradual decrease in the industry's tail risk Notably, real estate emerged as the top sector for bond issuance in late 2021, with a significant increase of nearly 3% on November 2 However, by the end of Q1 2022, the industry's tail risk surged to 5.2%, the highest since 2015 This rise was attributed to the Tan Hoang Minh Group's unilateral termination of a land purchase contract in Thu Thiem and the FLC Group Chairman's unreported sale of shares Despite some recovery attempts, sustainable growth remains challenging due to market conditions and investor apprehension.
(Source: Author's own synthesis) Figure 4.1.9 Retail
Vietnam's retail industry has faced significant challenges due to the COVID-19 pandemic, with fluctuations from 2019 to 2021 leading to increased tail risks and declining business performance The implementation of social distancing measures and the closure of non-essential shops severely reduced customer demand Additionally, the pandemic adversely affected employment and consumer income, resulting in a substantial drop in demand for retail products.
In early 2022, the retail sector experienced a significant rise in tail risk following four months of reopening, driven by increased production, trading activities, and improved consumer sentiment However, this tail risk began to decline despite ongoing inflationary pressures impacting the retail industry.
NI 1II000000000010111TT TT 0.1 AT wi IÌ 2020 ““ “6/04 2025 Wy 2025