The 1st International Student Research Conference on Economics and Business (SR-ICYREB 2022)

THE RATE OF RETURNS WITH COPULAS

Author: Huynh Thanh Hai1, Vu Thi Hanh Dung, Phan Thi Thuy Duyen, Nguyen Tan Sang
Mentor: Ph.D. Le Thanh Hoa2
University of Economics and Law

ABSTRACT
The paper discusses the dependence structure of the stock market in Vietnam, especially on the Ho Chi
Minh Stock Exchange (HOSE). The data comes from four stocks traded on the HOSE such as SSI, HCM,
VND, and FTS, which belong to the category of financial stocks. over a period of five years from April
21, 2017 to March 28, 2022. The Copulas model is used to determine the dependency structure between
the series of returns, in order to compare the dependencies, analyze the trend of stocks, and assist others
in choosing stocks or sectors to invest in the financial markets. The study investigates the model’s fitness
of marginal distribution using real data of stock closing prices in order to develop the Copulas model.
Nothing that the model parameters are estimated by the Maximum Likelihood Method based on Bayesian
statistics.
Keywords: Copulas; dependence; financial market; rate of returns; stocks.

1
Corresponding author: Huynh Thanh Hai; Tel: +84 909 378832; Email: haiht20413c@st.uel.edu.vn
2
Mentor: Le Thanh Hoa; Tel: +84 937 775398; Email: hoalt@uel.edu.vn
1
 The 1st International Student Research Conference on Economics and Business (SR-ICYREB 2022)

THE RATE OF RETURNS WITH COPULAS

Author: Huynh Thanh Hai, Vu Thi Hanh Dung, Phan Thi Thuy Duyen, Nguyen Tan Sang
Mentor: Ph.D. Le Thanh Hoa
University of Economics and Law

ABSTRACT
The paper discusses the dependence structure of the stock market in Vietnam, especially on the Ho Chi
Minh Stock Exchange (HOSE). The data comes from four stocks traded on the HOSE such as SSI, HCM,
VND, and FTS, which belong to the category of financial stocks. over a period of five years from April
21, 2017 to March 28, 2022. The Copulas model is used to determine the dependency structure between
the series of returns, in order to compare the dependencies, analyze the trend of stocks, and assist others
in choosing stocks or sectors to invest in the financial markets. The study investigates the model’s fitness
of marginal distribution using real data of stock closing prices in order to develop the Copulas model.
Nothing that the model parameters are estimated by the Maximum Likelihood Method based on Bayesian
statistics.
Keywords: Copulas; dependence; financial market; rate of returns; stocks.

1. Introduction
Dependence on financial markets is a major concern, and changing this stocks will result in the change
of other stocks. If conventional measurements are concerned with the numerical dependence of the
entire process, it is not possible to analyze the dependence at different levels of change. Copula's
measurement satisfies the need of determining the degree of dependency between random variables,
reported in stock prices based on their probability distributions.
By measuring dependence, it is able to measure the risks of dependent stocks, allowing investors to
reduce the degree of losses and ensure that financial activities run smoothly, effectively, and securely.
This benefits not only financial institutions but also individual investors. However, there is not many
research focused on this topic in Vietnam. Therefore, investigating the dependence structure among
financial markets in Vietnam is vitally essential and worthy of attention, helping us in understanding
the operation of law and the economy's functioning.
The purpose of this research paper is to use Copulas to evaluate the dependence structure of stocks in a
random portfolio on the HOSE? Consider whether the stocks are independent of each other or not?
(low dependence, medium dependence, or high dependence). The R-studio software is used to
calculate Copulas in order to examine the dependence and correlation of stocks in the same stock
market portfolio. Thus, the author may apply more of its experimental research findings to the
selection and decision-making of sensible investments, therefore reducing undesirable risks.
The authors employ a quantitative technique in the study, especially the Copulas function, by
referencing data from similar studies.
2. Theoretical framework
2.1. Pearson’s linear dependency measurement – Pearson correlation coefficient
Definition 2.1: Pearson correlation coefficient r(X, Y) is only appropriate to measure the degree of
dependence of random variables X and Y when they have a linear relationship. This coefficient is calculated
by:

2
 The 1st International Student Research Conference on Economics and Business (SR-ICYREB 2022)

cov ( X , Y ) E [( X −EX)(Y −EY )]

r ( X ,Y )= =
√ Var (X ) .Var (Y ) √ Var (X ) .Var (Y )
Where the variances being finite, it means that 0 < Var(X), Var(Y) < ∞ .
As X and Y have a nonlinear relationship, r(X, Y) cannot characterize the degree of dependence.
Furthermore, r(X, Y) = 0, we cannot conclude that X and Y are independent unless they have a standard
distribution.
2.2. H – Volume
In R2, rectangle A with A = [ a 1 , a2 ] × [ b 1 , b2 ] is a Cartesian volume of two closed regions, the points (
a 1 , a2 ), (a 2 , b1 ), (a 1 , b2 ) and (a 2 , b2 ) are vertices of rectangle A. Then unit square I 2=I × I with I = [0,1].
A function of two variable H is a function whose domain is a subset of R2denoted DomH and whose value
set RanH is a subset of R.
Definition 1.2.1: Let H be a function of two variable defined on DomH = A1 × A2 where A1 and A2
are two non - empty subsets of R , A = [ a 1 , a2 ] × [ b 1 , b2 ] is a rectangle whose all vertices belong to DomH.
Then, H-volume of A is given by:
V H ( A )=H (a 2 , b2 )−H ( a2 ,b 1)−H ( a1 ,b 2)+ H ( a1 ,b 1).
We see that V H (A ) have similar to property (ii) of P( X 1 ≤ x 1 , X 2 ≤ x 2 ¿ . If we get difference equation
a2 b2
△ a 1 H ( a , b)=H (a 2 , b)−H ( a1 ,b) và △ b 1 H ( a , b)=H (a , b 2)−H ( a , b1) ,
then
b2 a2
V H (A )=△ b 1 △ a 1 H (a , b)
Definition 1.2.2: A two-variable H function is called 2 - increasing if V H (A ) ≥ 0 with every A
rectangle whose vertices belong to DomH. Thus, here we see that an H function defined on A has acquired
properties similar to those of a distribution function.

2.3. N – increasing function

An H function is called n - increasing if it is defined on A=A 1 × A 2 × ...× A n,
Ai=[ai , bi ] ⊂[ 0,1], i=1 ,... , n and satisfies:
V H ( A )= ∑ sgn(u) H (u)≥ 0
u∈ A

As u = (u1 ,u 2 , .... ,u n ¿ moves on the angles of A with ui=a, or ui=bi.

2.4. Copulas
Copulas is considered a joint distribution function or multivariate uniform distribution function
where each marginal is a univariate distribution function or copula as a uniform distribution with one-
dimensional margins. Copula characterizes dependence and only properties that depend between components
in multivariate distributions. It can be combined with any set of univariate marginal distributions to establish
a multivariate distribution. As such, the use of copula allows us to expand the layer of multivariate
distributions from univariate marginal distributions.
Let X and Y be two random variable with probability distribution functions N ( x )=P( X ≤ x) and
G( y )=P(Y ≤ y). Their joint distribution F (x , y )=P(X ≤ x , Y ≤ y )∈[0,1]. With each pair of real
numbers ( x , y ) we can associate them with N ( x ), G( y ) và F (x , y ). Specifically, we set U = N (x ) and V
=G( y ) and are random variables that are evenly distributed above [0.1], i.e.

3
 The 1st International Student Research Conference on Economics and Business (SR-ICYREB 2022)

{ {
0 , u<0 0 , v< 0
N U (u)=P(U ≤ u)= u ,u ∈[0,1] và G V ( v)=P(V ≤ v)= v , v ∈[ 0,1]
1 , u>1 1 , v> 1

2.5. The Copulas families

2.5.1. The Elliptics Copulas
a) Gaussian Copulas (Normal Copulas)
Gaussian copula is a copula in the Ellipstics Copulas family. It is defined as follows:
n −1 −1
C Ga( u1 , u2 , ..., un )=Φ d (Φ (u1 ), ... ,Φ (u d )).
n
Where Φ is the cumulative distribution function of standard normal distribution N(0.1) and Φ d is the
cumulative distribution function of multivariate standard distribution with an average of 0 and a covariance
matrix Σ .
In the bivariate case the copula expression can be written as:
−1 −1
Φ (u) Φ (v)
1
C Ga(u , v)= ∫ ∫ 2π ¿¿
¿¿
−∞ −∞

The following expression is equivalent to the above expression, as demonstrated by Roncalli (2002):

( )
u −1 −1
Φ (v )−ρ1,2 Φ (t )
C Ga(u , v)=∫ dt .
0 √1− ρ 2
12

Thus, we see that Gaussian Copulas with two marginals as Gaussian distributions corresponding to
bivariate Gaussian distributions. If X,Y are two variable with a standard normal distribution with a
correlation coefficient ρ12, thenY ∨ X=x is standard normal distribution with ρ12 x expectations and
2
variance 1− ρ12 .From the conditional distribution formula of the standard Copulas function of two variable
we get:

∂
∂u
C (u , v , ρ12 )=Φ
(
Φ−1 ( v)− ρ12 Φ−1(u)
√ 1−ρ212
,
)
( )
−1 −1
∂ Φ (u)−ρ12 Φ ( v)
C (u , v , ρ12 )=Φ .
∂u √ 1−ρ212
Copula density function:

∂2 φ2 ( Φ−1 ( u ) , Φ−1 ( v ) ; ρ12 ) ;

c ( u , v , ρ12) = C ( u , v , ρ12) =
∂u ∂ v φ ( Φ−1 ( u ) ) φ ( Φ−1 ( v ) )

(
2 ρ 12 Φ−1 (u)Φ−1 (v )−ρ12
)
2
1 (Φ−1 (u)2+Φ−1 (v)2 )
¿ 2
exp 2
.
1−ρ12 2( 1−ρ12 )
In the case of bivariate, the distribution function is represented as the integral of the density function
u v
C (u , v , ρ12)=∫∫ c ( s , t ; ρ12 )dtds .
0 0

4
 The 1st International Student Research Conference on Economics and Business (SR-ICYREB 2022)
b) Student’s Copulas (T Copulas)
Gaussian copula is a copula in the Ellipstics Copulas family. It is defined as follows:
n −1 −1
C Ga( u1 , u2 , ..., un )=Φ d (Φ (u1 ), ... ,Φ (u d )).
n
Where Φ is the cumulative distribution function of standard normal distribution N(0.1) and Φ d is the
cumulative distribution function of multivariate standard distribution with an average of 0 and a covariance
matrix Σ .
In the bivariate case the copula expression can be written as:
−1 −1
Φ (u) Φ (v)
1
C Ga(u , v)= ∫ ∫ 2π ¿¿
¿¿
−∞ −∞

The following expression is equivalent to the above expression, as demonstrated by Roncalli (2002):

( )
u
Φ−1(v )−ρ1,2 Φ−1 (t )
C Ga(u , v)=∫ dt .
0 √1− ρ 2
12

Thus, we see that Gaussian Copulas with two marginals as Gaussian distributions corresponding to
bivariate Gaussian distributions. If X,Y are two variable with a standard normal distribution with a
correlation coefficient ρ12, thenY ∨ X=x is standard normal distribution with ρ12 x expectations and
2
variance 1− ρ12 .From the conditional distribution formula of the standard Copulas function of two variable
we get:

∂
∂u
C (u , v , ρ12 )=Φ
(
Φ−1 ( v)− ρ12 Φ−1(u)
√ 1−ρ2
,
12
)
∂
∂u
C (u , v , ρ12 )=Φ
(
Φ−1 (u)−ρ12 Φ−1 ( v)
√ 1−ρ2
.
12
)
Copula density function:
2 −1 −1
∂ φ 2(Φ (u),Φ (v) ; ρ12);
c (u , v , ρ12)= C (u , v , ρ12)= −1 −1
∂u∂v φ(Φ (u))φ(Φ ( v))

(
2 ρ 12 Φ−1 (u)Φ−1 (v )−ρ12
)
2
1 (Φ−1 (u)2+Φ−1 (v)2 )
¿ 2
exp 2
.
1−ρ12 2( 1−ρ12 )
In the case of bivariate, the distribution function is represented as the integral of the density function
u v
C (u , v , ρ12)=∫∫ c ( s , t ; ρ12 )dtds .
0 0

3. Research method (Alt + 7)

3.1. Method 2-1 (Alt + 8)
1.1. Method 2-2
3.1.1. Content 2-2-1 (Alt + 9)
a) Content 2-2-1-1(Alt + 0)

5
 The 1st International Student Research Conference on Economics and Business (SR-ICYREB 2022)
b) Content 2-2-1 -2(Alt + 0)
3.1.2. Content 2-2-2 (Alt + 9)
4. Results and discussion (Alt + 7)
4.1. Results (Alt + 8)
This section should simply state the findings, without bias or interpretation, and arranged in a logical
sequence. Non-textual elements, such as, figures, charts, photos, maps, tables, etc. to further illustrate the
findings, should also be included if appropriate. In the text, refer to each non-textual element in
consecutively numbered order [e.g., Table 1, Table 2; Chart 1,Chart 2; Map 1, Map 2], and complete with a
heading [title with description goes above the figure, table, chart, etc.].
Table 1.An example of a table. (Alt + B)
An example of a column heading Column A Column B
And an entry 1 4
And another entry 2 5
And another entry 3 6
Source:

Figure 1. Communal house in Kontum (Alt + B)

Source:
Equations and formulae should be typed in Mathtype, and numbered consecutively with Arabic
numerals in parentheses on the right hand side of the page (if referred to explicitly in the text). They should
also be separated from the surrounding text by one space.

(1)
4.2. Discussion (Alt + 8)
This section puts forward further explanation about the meaning, importance and relevance of your
results. Four key elements should be discussed: interpretations (what do the results mean?), implications
(why do the results matter?), limitations (what can’t the results tell us?), recommendations (what practical
actions or scientific studies should follow?).
5. Conclusion
A synthesis of key points and practical and managerial implications are crucial elements of conclusion
section.
6. Appendix

6
 The 1st International Student Research Conference on Economics and Business (SR-ICYREB 2022)
Authors including an appendix section should do so before the References section. Multiple
appendices should all have headings in the style used below. They will automatically be ordered A, B, C etc.
Appendix A. An example appendix
A.1. Example of a sub-heading within an appendix
There is also the option to include a subheading within the Appendix if you wish.

REFERENCES (Alt + D)

[1] Anderson E. et Weitz B. (1989), Determinants of continuity in conventional industrial channel dyads,
Marketing Science, 8, 4, 310-323. (Alt + T)
[2] Nguyen Van A (2010), Hội nhập quốc tế của Việt Nam thế kỷ XX, Publisher XYZ, Ha Noi.
[3] Blau M.P (1964), Exchange and power in social life, New York, John Wiley & Sons. (Alt + T)
[4] Cannon J.P. et Perreault W.D. (1999), Buyer-seller relationships in business markets, Journal of
Marketing Management, 36, 4, 439-460. (Alt + T)
[5] Le Van A (2009), “Tác động của toàn cầu hóa đến tăng trưởng kinh tế: Trường hợp nghiên cứu tại Việt
Nam”, Tạp chí Khoa học Kinh tế, 15, 4, 12-19. (Alt + T)
[6] Westbrook, R. A. (1987). “Product/consumption-based affective responses and post purchase
processes”, Journal of Marketing Research, 24, 3, 258-270. (Alt + T)

Note:
Requirements on submission papers: from 5.000 to 7.000 words including references and appendix
presented on A4 paper, Times New Roman font, font size 11, 2cm margins. The required template is
published on the conference’s website: http://due.udn.vn/vi-vn/sricyreb-2022.