03-2023-volatility-correlation-covid-ukraine

JEL Classification: G10, G19

Quy Vo, School of Economics, Finance, and Accounting, International University, Vietnam National University Ho Chi Minh City https://orcid.org/0000-0001-9933-9300

Hanh Trang Tran, School of Economics, Finance, and Accounting, International University, Vietnam National University Ho Chi Minh City

THE CORRELATION OF THE VOLATILITY OF FUTURE MARKET AND STOCK MARKET-A STUDY IN VIETNAM WITH 2 STAGES: COVID-19, AND GEOPOLITICS - UKRAINE INVASION

Received 15 June 2023; accepted 27 June 2023; published 03 July 2023

Abstract. Since the Vietnamese derivatives market was established, futures contracts as the representative financial instrument trading in the derivatives market. The research investigates the correlation between the spot market and the futures markets in Vietnam in which VN30 and VN30 Futures Index was chosen as research sample. Additionally, the deposit interest rate with different terms was also involved to declare the correlation between these two markets. The ARMA/GARCH model was employed to examine the volatility of the spot market, futures market, and the deposit interest rate (overnight, spot week, one-month, three-month,sixmonth,nine-month). The VAR model was employed next step to explore the correlation as well as the impact on each other. The results indicate that the tight relationship between the futures and the spot market while a negative relationship between the deposit interest rate and the spot/future market was found Especially, the examining term has been divided into 2 special stages: the Global Health Crisis (Covid-19: 2020-2021) and the Geopolitics-Ukraine invasion (2022-present) to clearly observe the correlation among three research factors.

Keywords: Volatility, Spot Market, Future Market, ARMA/GARCH,VAR,Covid-19, Geopolitics-Ukraine invasion.

In August 2017, the VN30 future contracts were officially traded with the base VN30 index. A lot of existing paper indicated the two-side impact in both positive and negative aspects between the underlying stock markets versus derivative markets in which available for trading in Vietnam is future markets. The report from Vietnam National Institute for Finance (Vu Chi Dung, 2022)

Vietnamese stock market is considered one of the frontier markets but in 2022, Vietnam has outperformed with the GDP growth rate of 8.02% (VNA, 2023), the remained one-number inflation (4.31%, tradingeconomics). By the end of February 2023, the record of Vietnam domestic stock trading accounts demonstrated nearly seven million accounts including both individual and proprietary. The amount only counted for 7 percent of total Vietnamese population while many assessments have been made and considered that the market of Vietnam as a potential market based on GDP, inflation rate, and especially stock market and worth for investment since the valuation via P/E index is quite affordable.

Introduction Background Research clearly stated the expression of this interrelationship based on five main aspects: (i) Derivatives markets contribute to the efficiency of price formation on the underlying stock markets; (ii) The liquidity of the underlying equities markets is promoted by the derivatives markets; (iii) Underlying stock markets bring condition to a better performance of derivatives markets and for investors in hedging portfolio and seeking for profit; (iv) Derivatives markets can be used to manipulate the underlying stock markets and vice versa; (v) The consequence of the derivative markets break can affect directly the underlying equities markets and the whole financial system. On the other hand, from 2018 to 2021, a downward trend was recorded about the Vietnam deposit rates due to the implementation of monetary policies with the purpose of reducing the interest rate to stimulate economic growth. Then, in 2022, the rate soared up significantly due to the rise of inflation and attracted a large amount of money.

Research Problem

While a lot of newspapers mentioned the significant rise of the deposit interest rate, the remarkably fall of the VN30 Index and the Futures trading volatility simultaneously was also noticed and reported by plenty of financial channels, newspaper ( https://vnexpress.net/). This inverse effect raised a question that whether the deposit interest rate affects the volatility of the spot/futures markets and whether the relationship between the two markets is highlighted clearly by the volatility of deposit interest rate? Since year 2005, Vietnamese stock market has suffered chronological events that affected directly on its volatility which are the Global Financial Crisis (GFC) in 2007-2008, pre and post - Covid-19 (2020-2021), and the Russo-Ukraine war. The emergence of these events led to the fall of the market and reflected the fear of investors and the risks that the stock market may face. Limited research exists on the impact of Ukraine invasion on Vietnam’s stock market volatility compared to studies on the effects of GFC, pandemic, and geopolitical events on global stock market. This study aims to investigate the market volatility and the correlation among these markets under three events objectively.

The expectation of this research would be about the strongly emphasizing on the two-way impacts from the underlying market and futures market through the results of the applied models. Also, the findings from VAR model and relating analyses may help in declaring the role of futures trading in price discovery for the underlying asset in the spot market. Moreover, the findings may bring about meaningful implications to potential investors.

Literature Review

1.1. Relevant Theoretical Frameworks

The linkage between the underlying market and the futures market, and the impact of the deposit interest rate on these markets can be explained by the arbitrage pricing theory developed by the economist Stephen Ross in 1976. The theory implies that investors can generate a risk - free profit by exploiting price discrepancies between two or more markets. The highlighted point is the possibility of arbitrage when the futures price of an asset is remarkably above its spot price. Statistical arbitrage deals with simplest financial instruments like stocks, futures, and currencies (John C. Hull, 2002; Ernest P. Chan, 2021). The relationship between the deposit interest rate and the stock market return or index can be explained based on financial intermediation theory and asset pricing model. According to Financial Intermediation Theory, the deposit interest rate and stock market are interconnected through the functioning of financial institutions. Stock market investments can be a portion of deposits (Anjan V. Thakor & Arnoud W. A. Boot, 1960). Expected returns, risk considerations, and regulatory requirements are considered as factors influencing investing in stock decisions. Higher interest rates may encourage saver to deposit more

R _ {\mathrm {i t}} = \log \left(\frac {C P _ {\mathrm {i t}}}{C P _ {\mathrm {i , t - 1}}}\right)

In which: $ CP\_{\\mathrm{i t}} $ represents the closing price of index at time t, $ R\_{\\mathrm{i t}} $ is the change in index i at time t.

C P _ {\mathrm {i t}}

R _ {\mathrm {i t}}

* * * T h eo pt ima l l a g l e n g t h f or A R (r) an d M A (s) u se d in t hi s m o d e l w er e d e t er min e d b y u s in g A C F an d P A C F in d i ces .

\left(1 - \sum_ {i = 1} ^ {t} \beta_ {i} B ^ {i}\right) R _ {t} = \left(1 + \sum_ {i = 1} ^ {S} \gamma_ {i} B ^ {i}\right) \varepsilon_ {t}

(2) In which: the backshift operator is $ B^{i} $ . $ \\beta\_{i} $ and $ \\gamma\_{i} $ are the coefficients from the ARMA model (the AR and MA parts). The residual is represented by $ \\varepsilon\_{t} $ . After running the model, the Ljung-Box portmanteau （Q）test was applied to confirm the ARCH effect for the dataset. The expectation in this study is that the residual would be independently and identically distributed.

B ^ {i}. \beta_ {i}

\gamma_ {i}

\varepsilon_ {t}

(3)

Q = n (n + 2) \sum_ {j = 1} ^ {m} \frac {1}{n - j} p ^ {2} (j) \rightarrow x _ {m} ^ {2}

m : t h e d e g r eeo f f r ee d o m t o m o d e l s t a t i o na r i t y^{'} s t es t .

p ^ {2} (j)

$ p^{2} ( j ) $ : the estimated autocorrelation of lag j. As the null hypothesis (H0) for expectation the series follows a white noise process, implying no ARCH effects was rejected when the Ljung-Box test indicated the p-value higher than 5%. Thus, the research jumped to the next step which is the application of ARMA(r,s)-GARCH (p, q) model to eliminate the conditional variance of VN30 future.

\widetilde {\sigma} ^ {2} = \alpha_ {0} + \sum_ {i = 1} ^ {q} \beta_ {i} \varepsilon_ {t} + \sum_ {i = 1} ^ {p} y _ {i} \sigma_ {t - i} ^ {2}

(4) 2.3.2. The application of the forecast error variance decomposition in the market volatility spillover estimation with VAR-a vector autoregression model: The VAR model is employed in examining the relationship among variables in the model with their market indices. Then, the optimal lag length was estimated by using AIC, HQIC, and SBIC indices and two lags for models with control variables have been indicated to be suitable for the VAR application. The VAR model: $ \\Delta Y\_{t}=\\alpha+\\beta\_{1}\\Delta Y\_{t-1}+\\beta\_{2}\\Delta Y\_{t-2}+\\dots+\\beta\_{p}\\Delta Y\_{t-p}+\\varepsilon\_{t} $

\Delta Y _ {t} = \alpha + \beta_ {1} \Delta Y _ {t - 1} + \beta_ {2} \Delta Y _ {t - 2} + \dots + \beta_ {p} \Delta Y _ {t - p} + \varepsilon_ {t}

Where: $ Y\_{t}=Y\_{1 t},Y\_{2 t},\\dots,Y\_{n t} $ : Endogenous variables

Y _ {t} = Y _ {1 t}, Y _ {2 t}, \dots , Y _ {n t}

$ \\beta\_{i} $ : Matrix with coefficients associated to lag i

\beta_ {i}

$ \\alpha : $ Vector with coefficients associated to the intercepts.

\varepsilon_ {t}

Table 4 Lagrange - Multiplier Test & Jarque - Bera tes | Equation | Chi2 | Df | Prob>Chi2 | | --- | --- | --- | --- | | VN30 | 832.86 | 2 | 0.00000 | | Vn30fpc | 6.60E+05 | 2 | 0.00000 | | Spot Week Rate | 1.30E+04 | 2 | 0.00000 | | Onrate | 5590.382 | 2 | 0.00000 | | Onemrate | 1274.64 | 2 | 0.00000 | | Sixmrate | 214.193 | 2 | 0.00000 | | Ninemrate | 5478.988 | 2 | 0.00000 | | All | 6.90E+05 | 14.00 | 0.00000 | $ \\varepsilon\_{t} $ : Vector with innovations. * * * Although the VAR stability test showed all variables under the entire period and geopolitics event satisfied the VAR condition, when conducting VAR model for variables under the sub-period - Covid-19, the var model is unstable due to the instability of VN30 index and VN30 future index variables. Thus, Var is conducted with difference of order 1 of these two variables. Moreover, after running VAR and conducting test, the three month deposit interest rate has been eliminated to guarantee there is no autocorrelation in the equation. The result of postestimation - test is below.The Lagrange - multiplier test was conducted to assess whether there is any autocorrelation in the residual of the VAR model. At lag 2, we cannot reject the null hypothesis of no correlation since it is greater than 5 percent. Moreover, the VAR stability test showed the result that all variables satisfied the VAR conditions and lay inside the circle. The below table illustrates the VAR stability test and Granger Causality test result. Table 5 Varstability Test | Eigenvalue | Modulus | | --- | --- | | 0.9971767 | 0.997177 | | 0.990976 | 0.990976 | | 0.9514373 | 0.951437 | | 0.8655907 | 0.865591 | | 0.797424 | 0.797424 | | 0.6221738 | 0.622174 | | 0.5574171 | 0.557417 | | -0.4324334 | 0.432433 | | -0.4159855 | 0.415986 | | -.2443302+.02016336i | 0.245161 | | -.2443302-.02016336i | 0.245161 | | 0.2237208 | 0.223721 | | -0.14248 | 0.14248 | | -0.04799673 | 0.047997 | | All The Eigenvalues Lie Inside the Unit Circle. | | | VAR Satisfies Stability Condition. | | \*Source: Author's elaboration By using Granger Causality, the reason impact has been illustrated obviously among tested variables. Through the Table 6, the underlying market (VN30) does not have causality relationship with the most variables while VN30, and one month deposit interest rate does great Granger cause the future market (vn30fcp). The test also illustrates that two markets (VN30, vn30fcp) do granger cause the six month deposit interest rate. Table 6 | Equation | Excluded | Chi2 | Df | Prob>Chi2 | | --- | --- | --- | --- | --- | | Vn30 | Vn30fp | 4.3943 | 2 | 0.111 | | Vn30 | Spot Week Rate | 0.49321 | 2 | 0.781 | | Vn30 | Onrate | 0.02163 | 2 | 0.989 | | Vn30 | Onemrate | 1.8779 | 2 | 0.391 | | Vn30 | Sixmrate | 0.43906 | 2 | 0.803 | | Vn30 | Ninemrate | 0.28191 | 2 | 0.869 | | Vn30 | All | 11.943 | 12 | 0.450 | | Vn30fp | Vn30 | 46.16 | 2 | 0.0000 | | Vn30fp | Spot Week Rate | 0.62359 | 2 | 0.7320 | | Vn30fp | Onrate | 0.46758 | 2 | 0.7920 | | Vn30fp | Onemrate | 5.0227 | 2 | 0.0810 | | Vn30fp | Sixmrate | 0.34423 | 2 | 0.8420 | | Vn30fp | Ninemrate | 0.34556 | 2 | 0.8410 | | Vn30fp | All | 57.789 | 12 | 0.0000 | Granger Causality Test * * * | Onrate | Vn30 | 0.19119 | 2 | 0.909 | | --- | --- | --- | --- | --- | | Onrate | Vn30fcp | 0.02521 | 2 | 0.987 | | Spot Week Rate | Vn30 | 0.15301 | 2 | 0.926 | | Spot Week Rate | Vn30fcp | 0.11168 | 2 | 0.946 | | Onemrate | Vn30 | 3.4727 | 2 | 0.176 | | Onemrate | Vn30fcp | 4.1642 | 2 | 0.125 | | Sixmrate | Vn30 | 7.9857 | 2 | 0.018 | | Sixmrate | Vn30fcp | 10.611 | 2 | 0.005 | | Ninemrate | Vn30 | 4.2073 | 2 | 0.122 | | Ninemrate | Vn30fcp | 0.4012 | 2 | 0.818 | \*Source: Author's elaboration This test analysis is performed by fitting the vector autoregressive model (VAR) to the time series with L time lags as below: Let $ X ( t ) \\in R^{d x 1} $ for $ t=1,\\ldots,t $ be a dimensional multivariate time series.

X (t) \in R ^ {d x 1}

t = 1, \dots , t

X (t) = \sum_ {\tau = 1} ^ {L} A _ {\tau} X (t - \tau) + \varepsilon (t)

In which: $ \\varepsilon (t) $ is a white Gaussian random vector, $ A\_{\\tau} $ is a matrix for every $ \\tau $ . A time series $ X\_{i} $ is called a Granger cause of another time series $ X\_{j}. $

X _ {i}

X _ {j}

Finally, the impulse and response functions are used with applying the Cholesky forecast error variance decomposition. Results 3.1. Estimating The Market Volatility Using ARMA/GARCH Table 7 below presents the estimated results for entire period (08.2018-04.2023) which illustrated the ARCH-GARCH effects existed in all variables. It implied that the the market volatilities are affected by both volatilities and shocks from events in previous period. The results also were presented by variance constant graph to show the market volatilities across the variables. Table 7 Volatility in the entire period (2018-4/2023 | Variance Equation | RVN30 | RVN30F | Reamrate | Resw | Reonrate | Re3mrate | Resixmrate | Reninmrate | | --- | --- | --- | --- | --- | --- | --- | --- | --- | | ARCH(α) | 0.134\*\*\* | 0.331\*\*\* | 0.481\*\*\* | 0.499\*\*\* | 0.534\*\*\* | 0.564\*\*\* | 0.161\*\*\* | 0.0863\*\*\* | | | (0.0396) | (0.0311) | (0.0344) | (0.0454) | (0.0339) | (0.0389) | (0.0275) | -0.0114 | | GARCH(β) | 0.553\*\*\* | 0.618\*\*\* | 0.784\*\*\* | 0.547\*\*\* | 0.292\*\*\* | 0.462\*\*\* | 1.187\*\*\* | 1.446\*\*\* | | | (0.20) | (0.04) | (0.03) | (0.0288) | (0.0203) | (0.0376) | (0.1120) | -0.0386 | | Constant | 0.00292\* | 0.00000126 | -0.00121\*\*\* | 0.000052 | 0.00133\*\*\* | 0.000330\*\*\* | -0.00100\*\*\* | -0.000604\*\*\* | | | (0.001720) | (0.000001) | (0.000142) | (0.000063) | (0.000078) | (0.000092) | (0.000292) | -0.0000352 | | (α+β) | 0.68700 | 0.9490000 | 1.26500 | 1.04600 | 0.82600 | 1.02600 | 1.34800 | 1.5323 | | Standard errors in parentheses | | | | | | | | | The row of $ (\\alpha+\\beta) $ indicates the mean -reversion speed process of each variable. The number showed the value which is smaller than 1 has mean -reversion progress and the smaller value is, the faster the process is. Thus,the results show overnight rate has mean -reversion process (other type of deposit rate witnessed no mean-reversion) and the fastest one is VN30 index with the lowest value of 0.687 and the lowest one is VN30 futures index with value of 0.949. By this amount,it can be said that the return on VN30 futures index volatility is affected by shock in the long term.

(\beta \text {o f} 0. 7 8 _ {4})

Wiki Hub

Explorer

03-2023-volatility-correlation-covid-ukraine

Graph View