INTRODUCTION
Problem statement
The Association of Southeast Asian Nations (ASEAN) is a regional organization formed on August 8, 1967, by Indonesia, Malaysia, the Philippines, Singapore, and Thailand, now comprising ten Southeast Asian nations ASEAN aims to promote inter-governmental cooperation and facilitate economic integration among its members, which include Brunei, Cambodia, Laos, Myanmar, and Vietnam The organization focuses on accelerating economic growth, social progress, and sociocultural evolution, while also ensuring regional stability and peaceful conflict resolution Over the years, ASEAN members have deepened their integration, particularly in economic aspects, by removing most tariffs to enhance the flow of goods and services within the region.
The establishment of the ASEAN Economic Community (AEC) in December 2015 marked a significant milestone in regional economic integration, creating a vast market valued at $2.6 trillion and encompassing over 622 million people By adopting a 'smallest common denominator' approach that prioritizes harmonious relations and national sovereignty, ASEAN countries have advanced trade through ambitious economic treaties and free-trade agreements As of 2014, the AEC represented the third-largest economy in Asia and the seventh-largest globally, highlighting its importance in the international economic landscape.
Economic integration within ASEAN presents both opportunities and challenges While it can foster growth and collaboration among member countries, it may also lead to increased costs due to the economic and cultural diversity within the organization.
Economic growth patterns vary significantly among ASEAN countries, with most classified as low-middle income, while nations like Singapore and Brunei enjoy stronger economic positions Post-AEC integration, income inequality may widen, exacerbated by high inflation rates in certain countries This disparity can lead to varying price levels and purchasing power, enabling wealthier nations to acquire more goods from their neighbors Additionally, differing inflation rates influence investment levels, driven by distinct monetary policy responses, potentially causing economic losses in vulnerable sectors Consequently, workers from less stable economies may seek migration opportunities in more prosperous member countries.
The ASEAN economies exhibit significant disparities in development, with high-saving nations like Brunei, Malaysia, and Singapore contrasting sharply with low-saving countries such as Cambodia, Laos, and the Philippines A December 2015 survey by the American Chamber of Commerce in Singapore revealed that multinational companies are considering various "pull factors" for expansion, highlighting the market's appeal and the relative absence of political, corruption, and security risks (Asian Development Bank, 2016).
Vietnam is poised to become a young Tiger in Asia over the next decade, driven by stable economic development and increased participation from foreign individual investors and multinational corporations This surge in foreign capital inflow signals the growing attractiveness of the Vietnamese economy and its financial market New investors require vital information to make informed investment decisions, and it is crucial to assess the level of risk associated with the expected rate of return.
In the context of the Vietnamese financial market, it is essential to implement risk measurement tools that guide investors, especially given the relevance of all industries for investment considerations.
The establishment of the ASEAN Economic Community (AEC) poses a significant threat to various industries in Vietnam, as the country and other ASEAN members exhibit competitive advantages in certain sectors Therefore, it is crucial to assess and evaluate the risks faced by key industries in Vietnam to offer timely recommendations for policymakers.
In light of the opportunities and challenges presented by the establishment of the ASEAN Economic Community (AEC) in December 2015, as well as potential economic agreements like the Trans-Pacific Partnership (TPP) under the new American administration, this study titled "Measuring Market Risk for ASEAN: A Value-at-Risk Approach" has been conducted to assess market risks in the region.
Research objectives
This study is conducted in order to achieve the following two key objectives:
To assess market risk across various industries in Vietnam, Thailand, Singapore, and Malaysia, we will analyze publicly available data spanning approximately 10 years Utilizing advanced techniques such as Value at Risk (VaR) and Conditional Value at Risk (CVaR), we aim to establish a relative level of market risk that reflects the current state of the international financial market in the ASEAN region.
The conventional market risk, referred to as Beta in the capital asset pricing model, is assessed through quantile regression These estimates are subsequently compared to the risk levels of key industries, determined using Value at Risk (VaR) and Conditional Value at Risk (CVaR) techniques.
Research questions
In order to achieve the above mentioned objectives, the following research questions have been raised:
How substantial does the level of the market risk change between the periods of pre-crisis and post-crisis in the 2008/2009 global financial crisis using VaR and CVaR?
What are the currently prevailing levels of the market risk for all industries in selected countries in the ASEAN, being Vietnam, Thailand, Singapore, and Malaysia?
Whether or not the market risk level of all industries obtained from the VaR and CVaR techniques and the conventional Beta are consistent?
Contribution of thesis
Since the 1950s, when Harry Markowitz pioneered risk management, the field has evolved significantly, establishing itself as a distinct subfield within finance Effective risk management now emphasizes the importance of qualitative and organizational factors, requiring sound judgment and an understanding of potential market pitfalls This study aims to measure market risks through innovative approaches like Value at Risk (VaR) and Conditional Value at Risk (CVaR), which are expected to offer crucial insights into risk management practices.
This study introduces the Value at Risk (VaR) concept to Vietnam's securities market by calculating VaR for ten industries over a decade and extending the analysis with Conditional Value at Risk (CVaR) for the same sectors The findings will serve as empirical evidence for the Vietnamese Government to enhance privatization and equitization efforts Additionally, the results offer historical insights that can assist investors in making informed investment decisions.
Second, this study use conventional Beta as a critical factor The values of
The comparison of Value at Risk (VaR) and Conditional Value at Risk (CVaR) with Betas will assess their consistency, providing valuable insights for investors This analysis serves as a guide for employing risk measurement techniques to enhance investment decision-making.
This study highlights the need for new risk measurements in Vietnam, as traditional methods may not fully capture emerging challenges By addressing this gap, the research aims to pave the way for further exploration and innovation in risk assessment practices within the region.
Structure of thesis
This study is constructed as follows The first chapter is Introduction Chapter
Chapter 2 summarizes the literature on risk measurements, focusing on Value at Risk and Conditional Value at Risk approaches, while also reviewing relevant empirical studies Chapter 3 details the data description, research methods, and models used in the study Chapter 4 presents the empirical results, and Chapter 5 concludes with a summary of the main findings and discusses implications based on the results from the previous chapter.
LITERATURE REVIEW
Theoretical
This elective section addresses a fundamental aspect of finance: risk management In many activities, risk is an inherent factor due to uncertainty in outcomes Consequently, a significant portion of finance professionals' responsibilities and the operations within financial departments focus on managing, mitigating, and capitalizing on risk.
Risk is a complex concept that encompasses uncertainty, randomness, and probability related to financial outcomes It is challenging to define risk precisely, as it can refer to both positive and negative random outcomes As noted by Apostolik (2015), various definitions of risk exist, reflecting its multifaceted nature Common interpretations include the likelihood of an undesirable event occurring, the potential magnitude of loss from unexpected events, the probability of unfavorable outcomes, and the impact of adverse results Ultimately, risk encompasses a wide range of perceptions and meanings, making it a subjective yet crucial aspect of decision-making.
Jordio (2007) defined risk can be as the volatility of unexpected outcomes, generally the value of assets or liabilities of interest
Firms face various risks, categorized into business and nonbusiness risks Business risks, which companies willingly take on to gain competitive advantages and enhance shareholder value, relate to the product market and encompass factors such as technological innovations, product design, and marketing strategies Additionally, firms are exposed to macroeconomic risks arising from economic cycles and fluctuations in income and monetary policies, which are beyond their control.
Nonbusiness risks encompass various challenges, including strategic risks arising from significant economic or political shifts, which are challenging to mitigate except through diversification across different business sectors and geographical regions Additionally, financial risks pertain to potential losses in financial markets, such as those resulting from interest rate fluctuations or defaults on financial commitments By optimizing their exposure to financial risks, firms can better focus on effectively managing their core business risks.
This study focuses on measuring market risk, which refers to the potential for loss or gain due to unexpected changes in market prices or rates Market risks are categorized into several types, including interest rate risk, equity risk, exchange rate risk, and commodity price risk, based on the specific risk factor involved Additionally, market risk is differentiated from other financial risks, such as credit risk and operational risk.
There are several techniques of market risk measurement have developed over years However, this study delineate three objective tools that are used
Value at Risk is probably the most widely used risk measure in finance It has become the classic measurement that financial executives use to quantify market risk
The Basel I Capital Accord, established in 1988, marked the introduction of risk-based capital adequacy requirements, focusing on minimum regulatory capital for credit risk This risk arises when borrowers fail to meet their financial obligations to banks Regulatory capital, as outlined in the Accord, is designed to ensure that banks maintain sufficient resources to absorb unexpected losses, thereby promoting financial stability within the banking sector.
In January 2001, the Basel Committee on Banking Supervision (BCBS) introduced the Basel II Capital Accord, which serves as the successor to the Basel I Capital Accord This new framework enhances the previous accord by providing banks with increased flexibility in determining capital reserves based on their specific risk exposure Additionally, Basel II aims to bolster the stability and reliability of the international financial system while promoting advancements in risk management practices.
The Basel I Capital Accord primarily concentrated on establishing minimum regulatory capital requirements In contrast, the Basel II Capital Accord expands this focus by introducing a comprehensive supervisory framework known as the "three pillars."
- Pillar 1 - Minimal regulatory capital requirements;
- Pillar 2 - Supervisory review of capital adequacy;
- Pillar 3 - Market discipline and disclosure;
This paper emphasizes the measurement of credit risk, as detailed in Pillar 1 The section on credit risk measurement provides a concise overview of the key elements outlined in the various Pillars.
Source: Bank of International Settlement
Figure 2.1 Three pillars of Basel II
Pillar 1 - Minimum Regulatory Capital Requirements
For the first pillar of the Basel II Capital Accord the Basel Committee proposed capital requirements associated with three categories of risk:
Market risk refers to the potential decline in an investment's value resulting from fluctuations in market factors The Basel II Capital Accord outlines two primary methods for measuring market risk: the Standardized Approach and the Internal Models Approach.
Operational risk, as defined by Basel II, refers to the potential loss arising from inadequate or failed internal processes, personnel, systems, or external events To assess operational risk, three primary methods can be utilized: the Basic Indicator Approach, the Standardized Approach, and the Advanced Measurement Approach.
Credit risk refers to the potential loss a bank may face if borrowers fail to meet their financial obligations To assess credit risk, banks can utilize various methods, including the Standardized Approach, the Foundation Internal Rating Based Approach, and the Advanced Rating Based Approach The Standardized Approach offers enhanced risk sensitivity compared to Basel I, while the two Internal Rating Based (IRB) approaches leverage banks' internal risk ratings for a significantly more precise assessment of risk.
Pillar 2 - Supervisory review of capital adequacy
The second pillar of Basel II emphasizes the importance of supervisory review in assessing capital adequacy It mandates that national supervisors ensure banks establish an internal capital assessment process and set capital targets aligned with their risk profiles Additionally, it encourages bank management to enhance risk management techniques and integrate them into capital management Supervisors play a crucial role in evaluating the effectiveness of banks' capital adequacy assessments.
In the Netherlands, the Dutch Central Bank (De Nederlandsche Bank, DNB) serves as the supervisory authority overseeing banks' internal processes, ensuring they meet the necessary risk-related requirements.
Pillar 3 - Market discipline and disclosure
The third pillar of the Basel II Capital Accord focuses on market discipline and transparency through enhanced disclosure Its primary objective is to foster improved financial reporting regarding risks, enabling market participants to gain a clearer understanding of banks' risk profiles and the sufficiency of their capital positions.
Empirical studies
This section will present in turn the empirical studies related to VaR, CVaR and conventional Beta,
In recent decades, risk modeling has become an essential component of global wealth management Among various methodologies, Value at Risk (VaR) has emerged as a prominent tool for assessing portfolio risk.
Previous research indicates that correlations between market movements are asymmetric, differing between downturns and upturns Additionally, it has been observed that the tails of return distributions are thicker than those in a normal distribution, as noted by Ang and Chen.
(2002), Boyer et al (1999), Kolari et al (2008), Longin and Solnil (2001)
Recent advancements in trading technology have made high-frequency data widely accessible, leading to the emergence of active market participants like high-frequency traders These traders are characterized by their short investment horizons and reliance on market risk measurement tools High-frequency data, which includes intraday market information such as transaction prices, bid-ask prices, and trading volume, has been the focus of research for over a decade, with early studies primarily examining exchange rates.
Recent studies on equity data, such as those by Adresen et al (2001a), Giot and Laurent (2004), and Fuertes et al (2009), have primarily focused on calculating daily risk measures with a time horizon of less than one day.
Halleib and Pohleier (2012) conducted an empirical study on the Value at Risk (VaR) method, questioning the significance of market capitalization Their research has a broader scope, examining the performance of various VaR models and distributional assumptions across different contexts.
22 estimation time windows Although within a complex study, the author found evidence that market capitalization in important for VaR estimation
Hsu et al (2011) conducted a study to evaluate portfolio risk across six Asian markets—Indonesia, Korea, Malaysia, Singapore, Taiwan, and Thailand—by integrating extreme value theory with traditional Monte Carlo Value at Risk (VaR) simulation The initial focus was on understanding the relationship between stock returns and fluctuations in currency value, which is essential for continuous international portfolio investment assessment.
The continuous rate of return of a portfolio at time t, denoted as \( r_{p,t} \), is calculated using the formula \( r_{p,t} = r_{i,t} + r_{e,t} \) In this equation, \( r_{i,t} \) represents the stock index return in the local currency, while \( r_{e,t} \) indicates the change in the American terms foreign currency exchange rate, which reflects the price of one unit of foreign currency in US dollars This relationship is crucial for understanding the portfolio's performance in relation to both local stock indices and foreign exchange fluctuations Additionally, Value at Risk (VaR) can be mathematically defined within this context.
The Value at Risk (VaR) is defined as the infimum of the set of real numbers \( r \) such that the probability \( P(R_p \geq VaR) = \alpha \), where \( R_p \) represents a sequence of portfolio negative returns over time intervals \( t, t-1, t-2, \ldots, t-h \) The analysis reveals that sample countries exhibit a positive yet weak correlation between stock index returns and fluctuations in currency values Furthermore, variations in foreign exchange rate policies and the extent of government intervention significantly influence the distribution and tail dependence in each nation.
In 2004, Huang and Lin analyzed the forecasting performance of three Value at Risk models, focusing on potential performance enhancements from asymmetry in conditional variance and fat-tailed distributions They evaluated the models using various metrics to assess their accuracy and efficiency The study utilized daily stock index futures prices from SGX-DT and TAIFEX, covering the period from May 7.
1998 to January 31, 2002 Overall, the findings had implications for investors, financial institutions, and futures exchanges
In the financial sector, Value-at-Risk (VaR) is the leading measure for assessing market risk, crucial for both regulators and internal risk managers Despite its popularity, VaR faces criticism for its lack of coherence as a risk measure, as highlighted by Artzner et al (1999) and Acerbi and Tasche (2002) Additionally, VaR fails to account for the statistical properties of significant losses beyond its threshold, known as tail risk To address these limitations, Conditional Value-at-Risk (CVaR) has emerged as a viable alternative for measuring risk.
Conditional Value-at-Risk (CVaR), also known as tail conditional expectation, was introduced by Artzner et al in 1997 and further explored by Inui and Kijima in 2005 Research by Yamai and Yoshiba from 2002 to 2005 delved into the applications of CVaR Additionally, Pattarathammas et al (2008) conducted a study on the estimation of Value-at-Risk (VaR) and CVaR, utilizing nine different models on historical log returns of Asian indices from November 1, 1996, to December 31, 2007 Furthermore, CVaR has gained traction in the insurance industry, as noted by Embrechts et al.
(1997) Bucay and Rosen (1999) used CVaR in credit risk evaluations A case study on application of the CVaR methodology to credit risk was described by Anderson and Urysev (1999)
In a study by Powell (2015), the extreme credit and market risk within Australia's mining industry was analyzed using Value at Risk (VaR) and Conditional Value at Risk (CVaR) to effectively capture tail risk The research compared mining entities to the total of all industries, including the 500 largest companies by market capitalization, during the global financial crisis from 2007 to 2009 Findings revealed that the market risk associated with mining shares, as indicated by both VaR and CVaR, was predominantly higher than that of the broader market.
Numerous studies have sought to evaluate the implications of the Capital Asset Pricing Model (CAPM) by analyzing historical rates of return for various securities alongside the historical return rates of market indices Notably, one of the most recognized studies in this area was conducted by Lintner in 1965, as highlighted by Diacogiannis (1994).
24 whose study was reproduced by Douglas (1968), Jacob (1971), Miller and Scholes
(1972), and Black, Jensen and Scholes (1972), whose methodology has been adopted for the empirical testing of CAPM in ASE, Blume and Friend (1973), and Fama and MacBeth (1973)
Numerous tests of the Capital Asset Pricing Model (CAPM) have focused on the relationship between average asset returns and their betas over specific time intervals, aiming to compare these findings with the CAPM's predictions Initial examinations by Black, Jensen, and Scholes (1972) alongside Fama and MacBeth (1973) utilized a two-stage procedure to conduct their analyses Specifically, Black, Jensen, and Scholes calculated betas from monthly returns of NYSE stocks during the 1926-1930 period, using an equally weighted portfolio of all NYSE stocks Their results indicated that the CAPM did not hold true for the examined timeframe.
Fama and MacBeth (1973) also estimated monthly market returns for all NYSE stocks over 1926-1929, and then they ranked all stocks by beta and formed
In a study analyzing 20 portfolios from 1930 to 1934, researchers calculated average returns and betas, mirroring the methodology of Black, Jensen, and Scholes They then utilized these betas to forecast portfolio returns for the period of 1935 to 1938 The findings revealed that the beta coefficient was statistically insignificant and remained low across various sub-periods Additionally, the analysis indicated that residual risk did not impact security returns, with intercept values significantly exceeding the risk-free rate, ultimately suggesting that the Capital Asset Pricing Model (CAPM) was not applicable in this context.
Though initial empirical studies support the CAPM (Fama and MacBeth
RESEARCH METHODOLOGY AND DATA
Data
This study analyzes daily data from publicly listed companies in four ASEAN countries—Vietnam, Thailand, Malaysia, and Singapore—over a 16-year period from 2000 to 2015 This timeframe encompasses a variety of economic conditions that are anticipated to impact different industries in diverse ways The selection of these countries is based on the availability of the necessary data throughout the research period, ensuring a comprehensive examination of the relevant factors affecting the stock market.
Data sourced from Bloomberg highlights that economic activities can be categorized into 11 distinct sectors according to the Global Industry Classification Standard These sectors encompass Energy, Materials, Industrials, Consumer Discretionary, Consumer Staples, Healthcare, Financials, Information Technology, Telecommunication Services, Utilities, and Real Estate.
Commercial and Professional services Transportation
25 Consumer Discretionary Automobile and Component
Consumer Durables & Apparel Consumer Services
30 Consumer Staples Food & Staples Retailing
Food, Beverage & Tobacco Household & Personal Products
35 Health Care Health Care Equipment & Services
45 Information Technology Technology Hardware & Equipment
The table below summaries the numbers of firms which are selected to measure and estimate the market risk.
Table 3 Daily market price movements in 4 countries (Vietnam, Malaysia, Singapore and Thailand)
No of firms Max (%) Min (%) Aver (%) S.D (%) No of firms Max (%) Min (%) Aver (%) S.D (%)
No of firms Max (%) Min (%) Aver (%) S.D (%) No of firms Max (%) Min (%) Aver (%) S.D (%)
Research methodology – the Models
To address research objectives, following three models/methods are to be considered and proposed to be utilized in this study
This study will utilize the two most popular approaches for estimating Value at Risk (VaR): the Variance-Covariance method and the Historical method.
The equation Pr[∆P ∆t ≤ VaR] = π illustrates that the probability of experiencing a loss equal to or greater than a specified Value at Risk (VaR) over a given time horizon ∆t is represented by π This means that for a defined probability π, losses that meet or exceed the VaR threshold are expected to occur.
Second, the estimates of the market risk using the CVaR techniques which can be expressed in the following formula:
𝐶𝑉𝑎𝑅 𝛼 = 𝐸(𝛿𝑃|𝛿𝑃 ≥ 𝑉𝑎𝑅 𝛼 ) where α is the left tail percentage of the distribution, δP is the absolute value of negative return of P
Quantile regression serves as a valuable tool for estimating conventional beta coefficients Throughout this study, additional econometric techniques will be explored and applied as needed This paper primarily utilizes the widely recognized quantile regression method introduced by Koenker.
Basset Jr (1978), the estimator can be found with following minimization function:
∀𝜏 ∈ (0,1) where the individual return R it and the market return R mt for t = 1, …,n; and the τ–th quantile regression coefficients, ατ and βτ
Measures of correlation between variables are important to practitioners interested in reducing their risk exposure through diversifying their portfolios
Correlation is a measure of the degree to which a value of one variable is related to
The correlation coefficient is a numerical measure that assesses the strength and direction of the relationship between the values of two financial instruments It ranges from -1 to +1, where the sign indicates the direction of movement—positive or negative—while the absolute value signifies the strength of that relationship For instance, a correlation coefficient of 0.5 indicates that one instrument moves in the same direction as the other, but only by half the amount.
A value of zero means that the instruments are uncorrelated, and their movements are independent of each other
Correlation plays a crucial role in Value at Risk (VaR) models, especially in parametric approaches, as it significantly impacts the measurement of a portfolio's variance and volatility For instance, in a simple two-asset portfolio, the overall volatility can be calculated using the individual volatilities of each asset (x and y) along with their correlation The formula for portfolio volatility is expressed as σ port = √(σ x² + σ y² + 2σ x σ y ρ), where σ x represents the volatility of asset x.
𝜎 𝑦 is the volatility of the asset y
The correlation coefficient between two assets uses the covariance between the assets in its calculation The standard formula for covariance is shown below:
The covariance calculation involves dividing the sum of the distances of each return value, x and y, from their mean by the number of observations minus one, which allows us to determine the correlation coefficient.
𝜎 𝑥 𝜎 𝑦 where σ is the standard deviation of each asset
The correlation equation can be applied to multiple instruments, as correlations are typically derived from historical data This aspect is crucial for portfolio construction and analysis, since the risks associated with a portfolio are influenced by the correlations among its components.
A positive correlation among portfolio instruments increases risk, as significant movements in one asset are likely to coincide with similar movements in others, leading to a wider and flatter distribution of returns with higher probabilities of extreme values Conversely, a negative correlation suggests that assets tend to move in opposite directions, thereby reducing overall risk.
In extreme situations like market crashes or significant corrections, asset correlations may lose their relevance as all assets tend to move in unison However, in typical market conditions, leveraging correlations is an effective strategy for mitigating portfolio risk, resulting in a lower Value at Risk (VaR) for diversified portfolios compared to undiversified ones.
While the concept of Value at Risk (VaR) is straightforward, its implementation can be complex This section will outline the key steps involved in calculating VaR effectively.
To calculate the Value at Risk (VaR) for a single asset, one can determine the standard deviation of its returns using either historical or implied volatility For a 95% confidence level, which signifies that 5% of observations fall within the left tail of the normal distribution, the relevant data points are situated 1.65 standard deviations from the mean.
Assume that we have two-asset portfolio with some given information as follows:
Current market price per unit $50 $100
Historical volatility 1.00% (daily) 2.00% (daily) And we determine the market risk for above portfolio, using criteria 99% confidence level (or 2.33 as standard deviation for 01 day) and a holding period of
The number 2.33 represents the standard deviation corresponding to a 99% confidence level, similar to 1.65 In this context, the multiplication factor for a holding period of 10 days is √10 For a holding period of one year, the appropriate multiplication factor is √252, reflecting the 252 trading days in a year.
In the analysis of a two-asset portfolio, we can determine the relationship that allows us to calculate its volatility This calculation is essential for assessing the Value at Risk (VaR) of the portfolio.
VaR port = √𝑤 1 2 𝜎 1 2 + 𝑤 2 2 𝜎 2 2 + 2𝑤 1 𝑤 2 𝜎 1 𝜎 2 𝜌 1,2 where w 1 is the weighting of the first asset w 2 is the weighting of the second asset
𝜎 1 is the standard deviation or volatility of the first asset
𝜎 2 is the standard deviation or volatility of the second asset
𝜌 1,2 is the correlation coefficient between the two assets
The concept of Value at Risk (VaR) is straightforward yet crucial for understanding portfolio risk To accurately assess VaR, one must evaluate each asset's volatility and the historical correlations between asset pairs In portfolios with numerous positions, this task can be complex Importantly, the risk of an individual asset is not solely defined by its return's standard deviation but rather by its contribution to the overall portfolio risk An asset may appear highly risky in isolation, but if its returns correlate positively with other portfolio assets, it may not elevate the overall portfolio's standard deviation, rendering its acquisition riskless Thus, the impact of a new asset on portfolio risk is largely determined by the correlation of its returns with those of existing assets.
Matrices Variance-Covariance Value at Risk
In the previous section, how VaR could be calculated for a two-asseet portfolio Here, the illustration of how this is done using matrices
Consider the following hypothetical portfolio, invested in two assets, as shown as below
The standard deviation of each asset is derived from historical observations of asset returns, which are computed by taking the ratio of closing prices Subsequently, the mean and standard deviation of these returns are calculated using standard statistical formulas.
Hypothesis
In consistence with the research objectives and questions which have previously been discussed, the following research hypotheses have been developed:
H1: Similar levels of market risk exhibited by key industries in selected
ASEAN countries during the research period using VaR rankings of each model
H2: There is an association between undiversified VaR and diversified
H3: There is an association between VaR and CVaR rankings within each model
H4: There is an association between historical and parametric CVaR rankings at industry level
H5: There is an association between undiversified CVaR and diversified
Parametric tests are applied to large data sets and focus on statistical measures like means and standard deviation, under the assumption that the observations originate from a normally distributed population.
Nonparametric tests are ideal for smaller data sets as they do not rely on assumptions about distribution These tests focus on rankings instead of specific statistics like means and standard deviations.
In this study, nonparametric tests are used for two main reasons discussed below:
The dataset is relatively small, consisting of 11 data points for each model, which represent the Value at Risk (VaR) across 11 different industries This allows for a comparative analysis of industry VaRs across various models.
Secondly, this study is more concerned with rankings rather than actual data Different models yield very different actual levels of risk and are calculated on a different basis
Siegel and Castellan (1988), Lee et al (2000) showed a range of parametric tests which are considered suitable for the purposes:
The Spearman Ranking Correlation Test is an effective method for comparing two samples through their rankings, making it ideal for model comparisons This test is also applicable for evaluating both diversified and undiversified rankings, as well as parametric and nonparametric rankings.
The rank correlation coefficient \( r_s \) quantifies the relationship between two ranked data sets, with a value of \( r_s = 1 \) indicating perfect correlation and \( r_s = -1 \) signifying perfect inverse correlation The formula for calculating \( r_s \) is given by \( r_s = 1 - \frac{6 \sum d^2}{n(n^2 - 1)} \).
To assess the significance of differences in ranks among industries, the formula 𝑛(𝑛 2 −1) is utilized, where d represents the rank differences and n denotes the number of industries Statisticians, such as Lee et al (2000), employ the Spearman Rank Correlation Test and suggest using the t-test for analysis Although Siegal and Castellan (1988) argue that the t-test is marginally more effective, they recommend the simpler z-test for ease of use.
√(1−𝑟 𝑠 2 )/(𝑛−2) which has a t distribution with (n-2) degree(s) of freedom
The Krustal-Wallis Test tests variance of rankings where more than 2 populations are involved This is considered for comparison between the 3 rolling window periods
Using the methodology outlined by Lee et al (2000) and Siegel and Castellan
(1988), the test statistic K compares variations in ranking means:
𝑐 𝑖=1 ) − 3(𝑛 + 1) where ni = the number of observation in the ith sample n = n1+n2+n…+nc = total number of observations in the c samples
Ri = sum of the ranks for ith sample
RESEARCH RESULTS AND DICUSSION
VaR and CVaR
The results are presented in the tables below, illustrating both diversified and undiversified approaches The undiversified method reflects the weighted average of individual company Value at Risk (VaR), while the diversified approach accounts for the correlations among all entities within the industry.
The study provides a comprehensive overview of Vietnam by calculating average results over the specified period, as summarized in Table 5 Both historical and parametric approaches are illustrated to enhance understanding of the findings.
Table 5 VaR and CVaR summary over 10-year period in Vietnam
Period (2007-2016) VaR and CVaR 95 percent summary over 10 year
Historical Parametric Historical Parametric Historical Parametric Historical Parametric
Note: Rankings are from 10 (lowest risk) to 1 (highest risk) Source: Author’s calculation
The daily Value at Risk (VaR) values range from 2.8% to 3.7% using the historical method, and from 2.7% to 5.8% with the parametric method Notably, the difference between the lowest and highest historical VaR values is less than that of the parametric VaR In the historical approach, Finance and Real Estate exhibit the same risk level, while the parametric method reveals a 0.6% difference between these two sectors, both categorized as the most risky industries Interestingly, the historical method ranks Utilities as the highest risk at 3.7%, whereas the parametric method places it at the lowest risk Overall, the absolute value differences between the two methods are relatively minor, with the lowest values recorded at 2.8% and 2.7% for historical and parametric methods, respectively.
The analysis of relative rankings reveals notable differences between two approaches The historical method identifies Energy as the highest risk industry, followed by Information Technology and Utilities, while Health Care and Consumer Discretionary are deemed the lowest risk In contrast, the parametric method ranks Real Estate as the highest risk, followed by Finance and Energy, with Health Care, Materials, and Consumer Discretionary occupying the lowest risk quartile throughout the study period.
CVaR consistently surpasses VaR as it focuses on the worst 5% of returns While historical VaR indicates that the Consumer Discretionary sector has an average risk level compared to other industries, CVaR identifies it as one of the riskiest This discrepancy arises from the calculation methods of VaR and CVaR, suggesting that the volatility in the Discretionary sector may be influenced by outliers, significantly impacting the average value of CVaR.
Historical CVaR results for Staples, Consumer, and Information sectors show slight variations compared to VaR The line graph below presents only historical estimates, indicating that the findings are generally consistent with parametric estimates, except for the Information and Health sectors.
Source: Author’s illustration Figure 4.1 Historical VaR and CVaR in Vietnam (2007-2016)
To gain a detailed understanding of Vietnam's financial landscape, it is essential to analyze the smoothed average values of Value at Risk (VaR) and Conditional Value at Risk (CVaR) across different sub-periods.
Next, the in- and post-GFC periods are considered The table shows how the top risk industries change during periods and given methods
Table 6 VaR results over periods: in crisis and post-crisis in Vietnam
Period (2007-2016) VaR 95 per cent in GFC period and post-GFC
In GFC period (2007-2009) In post-GFC period (2010-2016)
Historical Parametric Historical Parametric Historical Parametric Historical Parametric
Note: Rankings are from 10 (lowest risk) to 1 (highest risk) Source: Author’s calculation
From 2007 to 2009, the highest risk industries based on historical Value at Risk (VaR) were Energy, Information Technology, and Finance However, from 2010 to 2016, Utilities emerged as the highest risk sector, followed by Energy in second place and Information Technology in third Throughout both periods, Health Care and Consumer Discretionary consistently ranked as the lowest risk industries This indicates that the risk levels of these sectors have remained stable over the studied timeframes, with the rankings showing little change.
The comparison of ranking methods reveals both similarities and differences in industry risk assessments Notably, Energy, Information Technology, and Finance consistently rank high in risk, while Consumer Discretionary remains the lowest throughout the analyzed periods However, a divergence occurs with Utilities; parametric calculations suggest a moderate risk level, whereas historical methods classify it as the most risky sector Overall, from 2007 to 2016, the relative risk levels among industries appear to remain stable over time.
The analysis reveals a significant decline in the absolute values of Value at Risk (VaR) from the recession period of 2007-2009 to the subsequent years of 2010-2016 During the recession, the stock market's VaR ranged between approximately 3% and 8%, while in the following years, it decreased to a range of about 2% to 5%.
Now we consider the changes in rankings of industries in Vietnam for the GFC period (2007-2009) and the post-GFC period (2010 -2016) under both historical and parametric approaches
Table 7 VaR rankings changes in Vietnam
GFC Change VaR GFC ranking
VaR post- GFC ranking Diff in rank
The figure below will illustrate actual VaR per table 7 This bar indicates the exactly pattern, illustrating the absolute difference between two studying periods
Source: Author’s illustration Figure 4.2 VaR rankings shift in Vietnam
Between 2007-2009 and 2010-2016, all sectors experienced a significant decrease in market risk levels Notably, the Utilities sector, often regarded as a "safe" haven during the Global Financial Crisis (GFC) of 2007-2009, emerged as the riskiest sector in this period.
Vietnam - Changes in VaR between GFC and post-GFC
VaR GFC VaR Post-GFC
The findings reveal that post-GFC, there is no assurance that industry rankings will remain stable or improve, emphasizing the relative market risk levels among different industries.
Table 8 illustrates the shifts in market risk rankings for various businesses in Vietnam from the Global Financial Crisis (GFC) to the post-GFC period Notably, the use of Conditional Value at Risk (CVaR) indicates a significant reduction in extreme losses across industries during these two timeframes, while the overall rankings among industries remain stable.
Table 8 CVaR rankings changes in Vietnam
CVaR Post-GFC Change CVaR GFC ranking
CVaR post- GFC ranking Diff in rank
The figure below will illustrate actual CVaR per table 8 This bar indicates the exactly pattern, illustrating the absolute difference between the GFC and the post-GFC
Source: Author’s calculation Figure 4.3 CVaR rankings shift in Vietnam
The Utilities sector has maintained consistent performance as indicated by its Value at Risk (VaR), despite notable shifts in its ranking Conversely, the Finance industry, once deemed the riskiest during the Global Financial Crisis (GFC), has transformed into a more stable and safer investment option in Vietnam post-GFC.
The Spearman Rank Correlation test is utilized to assess the correlation between the GFC and the post-GFC period in relation to both VaR and CVaR rankings, with the findings detailed in section 4.1.4.
In this section, VaR of Malaysia, Singapore and Thailand will be presented as follows
Beta estimation
The tables below present the estimated beta for two study periods From 2007 to 2009, most beta estimates align positively with market beta, indicating a consistent trend This finding is consistent with the results observed in the subsequent years from 2010 to 2016.
For an overview, most Betas by LAD and OLS estimated are statistically significant and lower than 1 – the market beta When the market return changed by
1 percent, these stock returns would change the same direction with a magnitude of
Energy and Real Estate sectors exhibit betas greater than 1, indicating that a 1 percent change in market returns leads to a more than 1 percent change in their stock returns.
Among industries, Energy and Real Estate are ranked to be as the highest risky industries, besides Industry This imply is seem to align with conclusion form VaR
In the other hand, Beta estimations show that Consumer Discretionary is one of the
Industries such as Energy and Real Estate are identified as having a higher risk compared to the overall market, as indicated by their low rankings in safety assessments This aligns with the findings from the Value at Risk (VaR) and Conditional Value at Risk (CVaR) analyses discussed earlier.
Finance is often surprisingly regarded as one of the lowest risk sectors in the market, contrasting sharply with the conclusions drawn by Value at Risk (VaR) assessments, which categorize finance as a high-risk group.
Table 16 Beta estimates Using CAPM (period 2007 – 2009)
Period (2007-2009) Beta estimations in the GFC period
Weighted-Cap Equally-Cap Weighted-Cap Equally-Cap Weighted-Cap Equally-Cap Weighted-Cap Equally-Cap
Note: Rankings are from 10 (lowest risk) to 1 (highest risk) Source: Author’s calculation
Table 17 Beta estimates Using CAPM (period 2010 – 2016)
Period (2010-2016) Beta estimations in the post-GFC period
Weighted-Cap Equally-Cap Weighted-Cap Equally-Cap Weighted-Cap Equally-Cap Weighted-Cap Equally-Cap
Note: Rankings are from 10 (lowest risk) to 1 (highest risk) Source: Author’s calculation
From 2010 to 2016, equity beta estimates derived from capital-weighted portfolios significantly surpassed those from equally weighted portfolios Notably, the Energy and Real Estate sectors exhibited the highest average beta values, while the Utility, Health, and Finance sectors recorded the lowest averages Additionally, the negative betas observed in the Utility sector indicate that its returns tend to move inversely with market returns.
Comparison in Vietnam
Below is the results obtained by CAPM (quantile regression) and historical VaR in order to compare whether risk level of Vietnamese industries are relatively consistent
Table 18 Comparison between Beta and VaR (period 2007-1016) Comparison between Beta and VaR in the GFC period and the post-GFC period
In the GFC In the post-GFC
Weighted-Cap Equally-Cap Diversified Undiversified Weighted-Cap Equally-Cap Diversified Undiversified
Note: Rankings are from 10 (lowest risk) to 1 (highest risk) Source: Author’s calculation
During the Global Financial Crisis (GFC), Consumer Discretionary was identified as one of the lowest risk sectors according to both Beta and historical Value at Risk (VaR) metrics In addition to Consumer Discretionary, the Finance sector also exhibited low risk, despite being categorized as highly volatile by VaR analysis Conversely, the Energy and Real Estate sectors are typically recognized as the highest risk industries.
From 2010 to 2016, both CAPM and historical VaR indicate that Energy and Real Estate are the highest risk industries, while Health and Consumer Staples are the lowest Interestingly, CAPM classifies Utilities as low risk, contrasting with its designation as the highest volatility sector by diversified historical VaR and an average risk level by undiversified methods Additionally, while VaR ranks Finance as average risk, CAPM identifies it as low risk Overall, Energy consistently ranks as the highest risk industry across both CAPM and VaR, whereas Consumer Staples are considered the "safest" industries.
The CAPM and VaR methods indicate that the energy sector has the highest market risk throughout the study period According to Petro Vietnam Group's 2016 Annual Report, the industry faces significant challenges, including the impact of the financial recession on business operations and living standards The sharp rise in prices of essential goods, particularly fuel, has led to decreased demand as households seek to conserve energy and explore alternative resources Additionally, the influx of new small and medium enterprises has intensified market competition, posing further challenges for retailers aiming to expand nationally.
In the finance industry, two methods yield contrasting outcomes regarding risk assessment The Capital Asset Pricing Model (CAPM) ranks finance as the lowest, while Value at Risk (VaR) indicates that this sector experienced high risk during the Global Financial Crisis (GFC) but transitioned to a "medium" risk category post-crisis Conversely, VaR suggests that sectors such as Consumer Discretionary, Consumer Staples, and Health are relatively "safe" options for risk-averse investors.