VIETNAM NATIONAL UNIVERSITY VNU-HCMC

UNIVSERISTY OF INFORMATION TECHONOLOGY

FINAL CAPSTONE PROJECT

Topic:

Analyze fluctuations and the effects of macroeconomic factors on Gross Domestic

Product (GDP) Growth in developing countries: The case of VietNam, Singapore and
ThaiLan.

Instructor:

Dr. Tran Van Hai Trieu

Group Member:

Full Name Student Code

Nguyen Quoc Khanh 20521452

Pham Thanh Dat 20521461

Bui Thi Thanh 20521908

1
 Mai Hien Nga 20521640

Bui Si Khoa 20521461

Pham Tien Dat 20521176

Dang Thi Kim Yen 20520874

Ho Chi Minh, November 2023

TABLE OF CONTENTS

TABLE OF CONTENTS 2

ACKNOWLEDGEMENTS 2

LIST OF FIGURES 3

1. INTRODUCTION
4

1.1. Research Purpose 4

1.2. General Research Objectives 4

1.3. Specific Research Objectives 4

2. LITERATURE REVIEW
4

2
2.1 Developing Country 4

2.2 The Gross Domestic Product and Macroeconomic Models 4

2.3 Objective of the Study 4

3. RESEARCH METHODOLOGY
4

3.2 Overview data 4

3.2.2. Selection Criteria 4

3.2.3. Data Processing 4

3.3 Hypothesis testing 5

3.4. Regression Model - Ordinary least-square (OLS) 5

3.4 Forecasting 10

3.5.1. Moving Average (MA) Process 11

3.5.2. Auto-Regression (AR) 12

3.5.3 Auto-regressive Integrated Moving Average (ARIMA) Models Lỗi! Thẻ đánh
dấu không được xác định.

3.5.4. Autoregressive Moving Average Model (ARMA) 13

3.5.5. Autoregressive Integrated Moving Average (ARIMA) Process 13

3.4.6. Conceptual Framework of Box Jenkins Methodology 14

Model Identification 16
3
Model Estimation 16

Diagnostic Checking 17

Model Forecasting 18

4. DATA ANALYSIS AND INTERPRETATION 4.1. Data Visualization

19

4.2. Statistical descriptive methods 24

4.3. Hypothesis Testing 28

4.4. Regression methods 30

4.5. Forecasting 36

5. Comparison with Other ARIMA Models CONCLUSION INSIGHTS

42

5.1. Recommendation 42

5.2. Conclusions 42

6. ACKNOWLEDGEMENTS
42

7. PLAGIARISM RESULTS
Lỗi! Thẻ đánh dấu không được xác định.

8. TASK ASSIGNMENTS SHEET

42

9. REFERENCES
42

4
 ACKNOWLEDGEMENTS
In reality, no success is achieved without the support and assistance, whether small or
large, direct or indirect, of others. Throughout this semester, as we enrolled in the
'Business Data Analysis' course, we received immense care and guidance from our
professor, friends, and colleagues both inside and outside the classroom.

With the deepest gratitude, we would like to sincerely thank you, Dr. Trần Văn Hải
Triều, for your dedicated guidance and insightful teaching throughout the course.
Your expertise in business data analysis has been invaluable in deepening our
understanding of this complex subject. Your approach to teaching has not only
enhanced our analytical skills but also inspired us to apply these skills in practical
business contexts.

During the semester, we strived to apply the foundational knowledge we had

accumulated, while learning and researching new information under your mentorship.
We maximized the use of what we had gathered to produce the best possible course

5
project report. However, due to our limited professional knowledge and lack of
practical experience, the report inevitably has shortcomings. We eagerly look forward
to further enriching our knowledge and skills in this field.

Finally, we wish you, Dr. Trần Văn Hải Triều, good health and continued success in
your noble mission of educating and inspiring future generations.

Ho Chi Minh City, November 2023

Student Group Contributors

Nguyen Quoc Khanh - Mai Hien Nga - Pham Thanh Dat - Bui Thi Thanh

Pham Tien Dat - Bui Si Khoa - Dang Thi Kim Yen

LIST OF FIGURES

6
1. INTRODUCTION
1.1. Research Purpose

The pivotal role of Gross Domestic Product (GDP) as a measure of economic status is
universally recognized. For developing countries, understanding the complexity of
GDP growth with macroeconomic factors is crucial. This research focuses on
analyzing the fluctuations and the influence of macroeconomic elements on GDP
growth within the context of developing nations, specifically Vietnam, Singapore, and
Thailand. Through this study, we aim to unravel the complex interdependencies and
to contribute to the formulation of robust economic policies that bolster sustainable
growth.

1.2. General Research Objectives

The general objective of this research is to construct a comprehensive framework that

elucidates the relationship between macroeconomic factors and GDP growth in the
selected developing countries. By doing so, we aim to offer a granular understanding
of how these economies can navigate the tides of global economic dynamics while
fostering domestic economic resilience and growth.

1.3. Specific Research Objectives

The specific objectives of our study are multifold:

1. To quantify the impact of key macroeconomic indicators such as inflation, foreign

direct investment (FDI), trade balance, and governmental policies on the GDP
growth of Vietnam, Singapore, and Thailand.
2. To identify and model the causal relationships between these macroeconomic
variables and GDP growth using contemporary econometric techniques.
3. To assess the vulnerability and responsiveness of these economies to external
economic shocks and policy changes.

7
4. To draw comparisons across the three countries to pinpoint common challenges
and unique opportunities in managing GDP growth.

2. LITERATURE REVIEW
2.1 Developing Country

A developing country, often characterized by a lower industrial base, lower Human

Development Index (HDI), and a lower per capita income compared to developed
countries, faces unique economic challenges and opportunities. The economic
landscape of developing countries is often marked by a higher dependence on
agriculture and primary commodities, a rapidly growing population, and a process of
urbanization and industrialization that is still in its nascent stages.

The development trajectory of such countries is influenced by various factors,

including political stability, infrastructure, education, healthcare, and access to
technology. Developing countries often struggle with issues like poverty, income
inequality, lack of diversified economic structures, and vulnerability to global market
fluctuations.

In terms of macroeconomic management, developing countries face the challenge of

fostering economic growth while managing inflation, unemployment, and budget
deficits. Their economies are more susceptible to external shocks due to factors such
as reliance on a narrow range of exports, volatile commodity prices, and external debt
burdens.

Economic policies in developing countries often focus on accelerating

industrialization, improving agricultural productivity, and fostering human capital
development. Strategies for attracting Foreign Direct Investment (FDI), increasing
participation in international trade, and integrating into global value chains are
common.

In the context of globalization, developing countries are increasingly looking to

leverage their comparative advantages in the global marketplace, while also
addressing internal structural weaknesses. The role of international financial

8
institutions, development aid, and foreign investment becomes crucial in this scenario,
as these countries strive to achieve sustainable economic growth and development.

2.2 The Gross Domestic Product and Macroeconomic Models

The connection between a country's economy and the global economy is invariably
mediated through external economic activities such as foreign investment and
international trade. The topics of Foreign Direct Investment (FDI) and international
trade have consistently captured the interest of numerous scholars, who have
approached these subjects from various perspectives. While empirical research results
on these topics vary, a common theme is the relationship between FDI, international
trade, and national economic growth.

Ali and Hussain (2017) highlighted the role of FDI as a fundamental driver of global
economic integration. They employed correlation and multiple regression analysis
techniques using time series data from 1991-2015 to investigate the correlation
between FDI and Pakistan's economic growth, finding a positive impact of FDI on the
country's economy. They advised the Pakistani government to enhance efforts to
attract FDI to stimulate economic growth.

Mun, Lin, and Man (2008) applied the ordinary least squares method to examine the
relationship between FDI and economic growth in Malaysia. Their findings indicated
a positive correlation, with a 1% increase in FDI corresponding to a 0.046072%
growth in Malaysia's economy. Sokang (2018) adopted a different approach, using the
two-stage least squares method to evaluate FDI's impact on Cambodia's economy
during 2006-2016. He demonstrated a positive influence of FDI on Cambodia's
economic growth, highlighting the benefits of technology transfer, learning by doing,
and labor training. Sokang recommended continued economic reforms in Cambodia
to attract more FDI.

Marobhe (2015) analyzed the impact of FDI on Tanzania's economic growth from
1970-2014, observing a positive effect. He emphasized the role of FDI in the
economic development of countries like Tanzania, through technology transfer,
capital inflow, and skill improvement. He suggested that the Tanzanian government
implement policies to attract FDI, such as tax incentives and infrastructure
improvements.

9
Mohamed Mustafa (2019) examined the contribution of FDI and tourism receipts to
Sri Lanka's GDP from 1978-2016. Using Eviews 10 econometrics software, he
concluded that both variables had a positive and significant long-term effect on Sri
Lanka's GDP.

Similarly, the relationship between international trade and economic growth has been
a focus of scholarly attention. Mogoe and Mongale (2014) assessed the impact of
foreign trade on South Africa's economic growth from 1990Q1-2013Q2, finding that
exports positively affected GDP, whereas imports had a negative influence. They
suggested that exports could expand infrastructure and potentially boost economic
growth.

Javed et al. (2012) used the ordinary least-square technique to study the effects of
international trade on Pakistan's economy from 1973 to 2010. They found that
international trade was crucial for the Pakistani economy, with each unit increase in
exports leading to a 0.32units increase in GDP, and each unit increase in imports
contributing to a 0.18units rise in GDP.

Azeez, Dada, and Aluko (2014) evaluated the impact of international trade on
Nigeria's economic growth using data from 2000 to 2012. Their statistical tests
showed that international trade positively impacted economic growth, with each unit
increase in imports and exports leading to rises in Nigeria's GDP by 0.359 units and
0.635 units, respectively.

Çevik, Atukeren, and Korkmaz (2019) investigated the effect of trade openness on
Turkey's economic growth, highlighting its potential benefits such as efficiency in
resource allocation and technology spillovers. Their research for the period 1950-
2014 showed that trade openness positively influenced economic growth and vice
versa, a relationship they termed “sequential feedback.”

Hye, Wizarat, and Lau (2016) applied the autoregressive distributed lag (ARDL)
cointegration technique and rolling regression method to explore the trade openness-
economic growth nexus in China from 1975-2009. They confirmed a positive
correlation between trade openness and economic growth in both the short and long
term, underlining its importance for sustainable development.

10
These studies reflect a diverse array of approaches and findings regarding the impact
of economic variables, including FDI and international trade, on economic growth
across different countries and regions. (4)

2.3 Objective of the Study

The objective of this study is to analyze the fluctuations and impacts of

macroeconomic factors on the Gross Domestic Product (GDP) growth in developing
countries, focusing specifically on Vietnam, Singapore, and Thailand, based on key
GDP components like Foreign Direct Investment (FDI), Exports (EXP), and Imports
(IMP). This research will explore how FDI, along with export and import activities,
influence economic growth in each of these countries. It's crucial to consider the
unique economic structures and stages of development of Vietnam, Singapore, and
Thailand, as these factors significantly influence their GDP growth patterns and their
responses to macroeconomic changes.

3. RESEARCH METHODOLOGY
3.2 Overview data

In this section, we aim to collect and utilize a diverse dataset to analyze the impact of
macroeconomic factors on GDP growth in developing countries, focusing on
Vietnam, Singapore, and Thailand. The primary variables in the dataset include
Foreign Direct Investment (FDI), Exports (EXP), and Imports (IMP), along with other
indicators related to GDP growth.

3.2.1. Data Source

The dataset be compiled from reputable sources such as the World Bank, the World
Trade Organization (WTO), and the national statistical agencies of the
aforementioned countries. The timeframe for the data will span from 2012 to 2022, to
reflect trends and fluctuations across different economic periods.

11
3.2.2. Selection Criteria

The chosen data must ensure objectivity, accuracy, and completeness. We will
exclude any data that is unreliable or ambiguous in origin. Additionally, the data must
be the most current and accurately reflect the current state of each country.

3.2.3. Data Processing

The collected data will undergo a cleaning process, which includes removing missing
values, correcting errors, and standardizing formats to ensure consistency.
Subsequently, we will analyze trends and relationships between variables using
statistical and machine learning methods.

3.3 Hypothesis testing

In this part of our research, we will employ various hypothesis testing methods to
validate the relationships and effects observed in our data. These methods will
include:

● T-test: Used for comparing the means of two groups, especially useful when
assessing the impact of a variable on GDP growth in two different economic
scenarios or countries.
● F-test: This test will help in comparing the variances between groups, which is
particularly useful in assessing the homogeneity of data across different countries.
● ANOVA (Analysis of Variance) Test: This method will allow us to compare the
means among three or more groups, offering insights into whether different
macroeconomic factors have distinct impacts on GDP growth.
● Chi-Square Test: Ideal for examining the relationship between categorical
variables, such as the classification of countries based on economic tiers and their
corresponding GDP growth patterns.

12
3.4. Regression Model - Ordinary least-square (OLS)

In our report, we have evaluated different regression methods to analyze the

impact of macroeconomic factors on GDP growth within developing countries, with a
focus on Vietnam, Singapore, and Thailand. The choice of the regression method is
pivotal in ensuring the accuracy and reliability of our findings. Here is an integrated
approach based on our evaluation:

Linear Regression:

● This method is beneficial if we were to examine the impact of a single

independent variable on GDP growth. For instance, if we were interested solely in
the effect of FDI on GDP growth, linear regression would provide us with a clear
understanding of the strength of this relationship.

Multiple Regression:

● Our analysis involves multiple independent variables—FDI, Exports (EXP), and

Imports (IMP)—all of which are believed to influence GDP growth. Multiple
regression is the optimal method for such a multifaceted analysis as it allows us to
assess the combined effect of these variables on GDP growth.
● This method will enable us to account for different variables simultaneously and
provide insights into the relative importance of each factor while controlling for
the others.

Choice of Method:

● Considering that GDP growth is a continuous variable and our aim is to

understand its relationship with several predictors, multiple regression emerges as
the most appropriate method for our study.
● Multiple regression will also allow us to identify and address multicollinearity if it
exists among our independent variables, which is crucial for the integrity of our
regression coefficients.

Conclusion:
13
● We conclude that multiple regression is the most suitable model for our analysis. It
will provide us with a comprehensive understanding of how FDI, Exports, and
Imports collectively impact GDP growth in the selected developing countries.
● Logistic Regression, while important in its own right, is not suitable for our
analysis as it is designed for categorical outcomes, not continuous ones like GDP
growth.

In summary, the multiple regression model stands out as our method of choice
for investigating the intricate dynamics of GDP growth and its determinants in our
selected countries. This model aligns with our objective of creating a robust analytical
framework that can inform economic policies and strategic decision-making.

Figure 1. Mutiple Regression

14
Ordinary least-square (OLS) method is used in analyzing the impact of FDI, export
and import on economic growth of Vietnam in the period 2012-2022. Secondary data
is taken from the General Statistics Office of Vietnam. A regression model is built
where independent variable is GDP and three dependent variables are FDI, export and
import.

GDP=F ( FDI ,exp , IMP )

where:

⮚ GDP: Gross Domestic Product

⮚ FDI: Foreign Direct Investment

⮚ EXP: Export

⮚ IMP: Import

The model is expressed as follow:

GDP=β1 + β 2 FDI + β 3 exp+ β 4 MP+u

where:

⮚ β1: Regression constant.

⮚ β2, β3, β4: Coefficients to be estimated. They measure the effects of FDI, EXP
and IMP on GDP, respectively.
⮚ u: stochastic error term.

Four hypotheses are set up based on regression results to test the impact of FDI,
export and import on the economy.

⮚ Hypothesis 1: Whether the population regression function is significant or not

15
⮚ Hypothesis 2: Whether FDI impacts GDP or not

⮚ Hypothesis 3: Whether EXP impacts GDP or not.

⮚ Hypothesis 4: Whether IMP impacts GDP or not.

16
3.4 Forecasting

For forecasting future trends, we will implement time series analysis models, with a
specific focus on the ARIMA (Autoregressive Integrated Moving Average) model.
This model is particularly suited for forecasting economic data due to its ability to
handle various types of time series data, including those with trends and seasonalities.
We will use the ARIMA model to project future GDP growth trends for Vietnam,
Singapore, and Thailand, considering the impact of the studied macroeconomic
factors.

Figure 2. Forecast from ARIMA

The Gross Domestic Product (GDP) is the market value of all goods and services
produced within the borders of a nation in a year.

17
In this report, GDP data obtained from the Kenya National Bureau of statistics for the
years 2012 to 2022 was studied. Python and SPSS 21 statistical software were used to
build a class of ARIMA (autoregressive integrated moving average) models following
the Box-Jenkins method to model the GDP. The ARIMA (2, 2, 2) time series model
was established as the best for modeling the Group GDP according to the recognition
rules and stationary test of time series under the AIC criterion. The results of an in-
sample forecast showed that the relative and predicted values were within the range of
5%, and the forecasting effect of this model was relatively adequate and efficient in
modeling the annual returns of the Group GDP. Finally, we used the fitted ARIMA
model to forecast the GDP of Group Country for the next five years.

Auto-regressive Integrated Moving Average (ARIMA) Models

Autoregressive Integrated Moving Average models (ARIMA models) were

popularized by George Box and Gwilym Jenkins in the early 1970s. It’s an iterative
process that involves four stages; identification, estimation, diagnostic checking and
forecasting of time series.

According to [5], ARIMA models are a class of linear models that is capable of
representing stationary as well as non-stationary. They do not involve independent
variables in their construction, but rather make use of the information in the series
itself to generate forecasts. ARIMA models therefore, rely heavily on autocorrelation
patterns in the data.

ARIMA methodology of forecasting is different from most methods because it does

not assume any particular pattern in the historical data of the series to be forecast. It
uses an interactive approach of identifying a possible model from a general class of
models. The chosen model is then checked against the historical data to see if it
accurately describes the series. Most of the traditional forecasting models therefore,
provide a limited number of models relative to the complex behaviour of many time
series with little guidelines and statistical tests for verifying the validity of the selected
model.

3.5.1. Moving Average (MA) Process

This is a time series model which uses past errors as explanatory variable [19]. Let ut
(t=1,2,3,...) be a white noise process, a sequence of independently and identically

18
distributed (iid) random variables with E(ut )=0 and Var(ut ) = σ 2. Then the qth order
MA model is given as:

y t =μ+ut + θ1 u t−1+ θ2 u t−2 +…+θ q

This model is expressed in terms of past errors and thus we estimate the coefficients
θ j , j=1 , … , q ,and use the model for forecasting. Therefore only q errors will affect the
current level y t but higher order errors do not affect y t . This implies that it is a short
memory model.

3.5.2. Auto-Regression (AR)

According to [6], an autoregressive model of order p, an AR (p) can be expressed as;

y t =c +a1 y t−1 +a 2 y t −2+ …+a p y t − p

Where, ut wn ( 0 , σ 2) .

The model is expressed in terms of past values and therefore, we wish to estimate the
coefficients a j , j=1 , … , p . and use the model for forecasting. In this case, all
previous values will have cumulative effects on the current level y t and thus, it is a
long-run memory model. The ACF(s) therefore does not die out easily since it takes a
longer time to have ACF close to zero.

Partial Autocorrelation Functions (PACF) measures the correlation between an

observation k periods ago and the current observation, after controlling for
observations at intermediate lags (i.e. all lags ¿ k )

PACF (k) = ACF (k) after controlling the effects of ( y t −1 , … , y t−kt 1) . Thus PACF (k)
can be found as the coefficient of y t −kin the regression.

Y t =α 0 +α 1 y t −1 + a2 y t −2+ …+α k−1 y t −k+1 +a k y t −k + μt

→ α k =PACF ( K )

19
Hence the PACF is useful for telling the maximum order of an AR process.

Auto-regressive (AR) models can be coupled with moving average (MA) models to
form a general and useful class of time series models called Autoregressive Moving
Average (ARMA) models. These can be used when the data are stationary.

3.5.4. Autoregressive Moving Average Model (ARMA)

[6] expressed an ARMA (p, q) model as follows:

y t =c +α 1 y t −1+ α 2 y t−2 +…+ α p y t− p +ut +θ1 ut−1 +θ2 ut −2 +…+ θq ut −q

This is a combination of both AR and MA models. In this case therefore, neither ACF
nor PACF can solely provide the information on the maximum orders of p or q.

This class of models can further be extended to non-stationary series by allowing the
differencing of the data series resulting to Autoregressive Integrated Moving Average
(ARIMA) models.

3.5.5. Autoregressive Integrated Moving Average (ARIMA) Process

There are a large variety of ARIMA models [4]. The general non-seasonal model is
known as ARIMA (p, d, q): where p is the number of autoregressive terms, d is the
number of differences and q is the number of moving average terms. A white noise
model is classified as ARIMA (0, 0, 0) since there exists no AR part because y t does
not depend on y t −1, there is no differencing involved and also there’s no MA part
since y t does not depend on e t−1.

For instance, if y t is non-stationary, we take a first-difference of y t so that yt

becomes stationary.

∆ y t = y t − y t −1

∆ y t =c + α 1 ∆ y t−1 +α 2 ∆ y t−2 +…+ α p ∆ y t − p+ θ1 ut −1+ θ2 u t−2 +…+θ q u t−q +ut is an ARIMA (p,
1, q) model.

20
According to [5], A random walk model is classified as ARIMA (0, 1, 0) because
there is no AR and MA part involved and only one difference exists.

3.4.6. Conceptual Framework of Box Jenkins Methodology

The process uses four iterative stages of Modeling that involves; identification,
estimation, diagnostic checking and forecasting (See figure below).

Figure 3. ARIMA forecasting procedure

21
 Figure 4. Stages in the Box-Jenkins iterative approach.

The four stages modeling in the Box-Jenkins iterative approach:

● Model identification: making sure that the variables are stationary, identifying
seasonality in the series, and using the plots of the AutoCorrelation Function
(ACF) and Partial Auto-Correlation Function (PACF) of the series to
identification which autoregressive or movingaverage component should be
used in the model.
● Model estimation: using computation algorithms to arrive at coefficients that
best fit the selected ARIMA model. The most common methods use Maximum
Likelihood Estimation (MLE) or non-linear least-squares estimation.
● Model checking: by testing whether the estimated model conforms to the
specifications of a stationary univariate process. In particular, the residuals
should be independent of each other and constant in mean and variance over
time; plotting the ACF and PACF of the residuals are helpful to identify
misspecification. If the estimation is inadequate, we have to return to step one
and attempt to build a better model. Moreover, the estimated model should be

22
 compared with other ARIMA models to choose the best model for the data.
The two common criteria used in model selection: Akaike’s Information
Criterion (AIC) and Bayesian Information Criteria (BIC) which are defined by:

AIC=2 m−2 ln ( ^L ) , BIC=ln ( n ) m−2 ln ¿)

where ^Ldenotes the maximum value of the likelihood function for the
model, 𝑚 is the number of parameters estimated by the model, and 𝑛 is the
number of observations (sample size). Practically, AIC and BIC are used with
the classical criterion: the Mean Squared Error (MSE).

● Forecasting: when the selected ARIMA model conforms to the specifications

of a stationary univariate process, then we can use this model for forecasting.

Model Identification

A preliminary Box-Jenkins analysis with a plot of the initial data should be run as
the starting point in determining an appropriate model. The input data must be
adjusted to form a stationary series and identify seasonality in the dependent series
(seasonally differencing it if necessary), and using plots of the autocorrelation and
partial autocorrelation functions of the dependent time series to decide which (if any)
autoregressive (AR) or moving average (MA) component should be used in the model

Model Estimation

Model Estimation The parameters of the selected ARIMA (p, d, q) model can be
estimated consistently by least-squares or by maximum likelihood. Both estimation
procedures are based on the computation of the innovations ' from the values of the
stationary variable. The least-squares methods minimize the sum of squares;

❑
min ∑ ε (7)
2

The log-likelihood can be derived from the joint probability density function of the
innovations ε 1 , … , ε T ,that takes the following form under the normality assumption,
ε t N . I . D ( 0 , σ 2 ):

23
 { }
T
−∑ ε (8)
2

−T t=1
f ( ε1 , … , εT ) ⋉ σ exp exp
❑

In order to solve the estimation problem, equations 6 and 7 should be written in

terms of the observed data and the set of parameters,( Θ , ϕ , σ 2 ). An ARMA (p, q)
process for the stationary transformation Z t can be expressed as:

p q
ε t=Z t−σ−∑ ϕ i Z t−1−∑ θ i ε t−1 ( 9 )
i=1 i=1

Then, to compute the innovations corresponding to a given set of observations

(Z 1 , … Z T ) and parameters, it is necessary to count with the starting values
Z 0 , .. Z p−1 , ε 0 , … , ε q−1. More realistically, the innovations should be approximated by
setting appropriate conditions about the initial values, giving to conditional least
squares or conditional maximum likelihood estimators.

Diagnostic Checking

Before using the model for forecasting, it must be checked for adequacy
(diagnostic checking). The model is considered adequate if the residuals left over after
fitting the model is simply white noise and also the pattern of ACF and PACF of the
residuals may suggest how the model can be improved.

Akaike’s Information Criterion (AIC) is one of the most robust methods used in
estimating parameters of an identified model.

AIC=−2 log log L+ 2m

Where; L denotes the likelihood and m is the number of parameters estimated in

the model such that;

m= p+ q+ P+Q

24
 However, not all computer programs produce the AIC or the likelihood L, thus it
is not always possible to find the AIC for a given model. A useful approximation to
the AIC is therefore denoted as;

AIC=n ( 1+log log ( 2 π ) +nlog σ 2+ 2 m)

As an alternative to AIC, the Bayesian Information Criteria (BIC) and the

Schwarz- Bayesian Information Criteria (SBC) are also used as model diagnostics.
The SBC is given by;

SBC=log log σ +(mlogn)/n

Model Forecasting

Model forecasting states the difference between in-sample forecasting and out-
of sample forecasting. In-sample forecasting for instance, explains how the chosen
model fits the data in a given sample while Out-of-sample forecasting on the other
hand, is concerned with determining how a fitted model forecasts future values of
the regressand, given the values of the regressors. To build a reliable model, the
following factors are highly considered in forecasting;

a) The level of accuracy required – forecasts should be prepared as accurately as

possible to facilitate the decision making process especially made on the basis
of the GDP forecasts.
b) Availability of data and information – a wealth of reliable and up-to-date GDP
data results to a reliable model.
c) The time horizon that the GDP forecast is intended to cover. This study for
instance, covered a short run period.

25
4. DATA ANALYSIS AND INTERPRETATION
4.1. Data Visualization

Figure SEQ Figure \* ARABIC 5. GDP Growth Rate (%) - 2012 to 2022
- Forecasting Results:

The visualized data portrays the GDP growth rates of Singapore, Thailand, and
Vietnam from 2012 to 2022. Analyzing the fluctuations:

Singapore experienced a significant dip in 2020, likely due to the COVID-19

pandemic's impact on its economy, heavily reliant on trade and investment. The sharp
rebound in 2021 suggests a strong recovery, possibly due to effective pandemic
management and economic stimulus measures.

Thailand's GDP growth rate shows volatility, with a notable decline in 2020, which
was the most significant dip in the period observed. Thailand's economy, with its
significant tourism sector, was particularly hard hit by the global travel restrictions.
The modest recovery in 2021 and 2022 indicates ongoing challenges in bouncing back
to pre-pandemic levels.

Vietnam shows a consistent growth trend, with the least impact from the pandemic
compared to the other two countries. There are many factors that contribute to
Vietnam being less affected by the COVID-19 pandemic than Singapore and

26
Thailand, including the government's quick response, the application of effective
disease prevention measures, cooperation and consensus of people and organizations
in implementing disease prevention measures, and ways to manage and control
epidemics. The country's swift response to the pandemic, coupled with its growing
manufacturing sector and export economy, likely contributed to this resilience. The
spike in 2022 might reflect the global demand surge as the world economies began to
recover.

These patterns reflect how each country's economic structure, policy responses to
global events, and sectoral composition influence their GDP growth rates. The data
suggests that while Singapore and Thailand were significantly impacted by the
pandemic, Vietnam's economy proved to be more resilient. The subsequent recovery
rates also highlight the effectiveness of the respective government policies and the
underlying economic strengths of each nation.

Figure SEQ Figure \* ARABIC 6. Export Value (USD) - 2012 to 2022

The visualized data shows the export values of Singapore, Thailand, and Vietnam
from 2012 to 2022, expressed in million USD. Here's an analysis of the export
fluctuations for each country:

Singapore's export values exhibit significant resilience, with a notable dip in 2015 and
2016, potentially due to global economic challenges such as reduced global demand
and lower commodity prices affecting trade-dependent economies. A sharp decrease
is observed in 2020, likely due to the COVID-19 pandemic's disruption of global
supply chains. However, a robust rebound in 2021 and further growth in 2022

27
suggests a strong recovery fueled by the global economic rebound and possibly
increased demand for electronics and pharmaceuticals, key export sectors for
Singapore.

Thailand's exports show a gradual increase with fluctuations, experiencing a sharp

decline in 2020, which could be attributed to the global economic slowdown caused
by the pandemic. Thailand's heavy reliance on exports, including automotive and
electronics, might have been impacted by factory shutdowns and global trade
disruptions. The recovery in 2021 and 2022 indicates a possible resurgence in global
demand for its exports.

Vietnam displays a consistent upward trend in its export values, with an impressive
growth trajectory over the decade. The slight dip in 2020 is modest compared to its
peers, reflecting Vietnam's increasing role as a major manufacturing hub and its
successful containment of the pandemic with minimal disruptions to production. The
significant increase in 2021 and 2022 could be due to the shift of manufacturing bases
from China to Vietnam, increased electronics exports, and free trade agreements
boosting trade.

These export trends reflect the varying degrees to which global events impact
different economies, the structure of each country's export sector, and their ability to
adapt to changing.

Figure SEQ Figure \* ARABIC 7. Foreign Direct Investment (FDI) Trends: 2012 to 2022 (US$)

Foreign Direct Investment (FDI) for Vietnam, Thailand, and Singapore from 2012 to
2022.

28
Vietnam: FDI in Vietnam has shown a consistent trend, with values fluctuating
slightly but generally remaining stable in the negative region. It's important to note
that negative FDI does not necessarily mean a lack of investment; rather, it could
indicate that outward investments by Vietnamese entities have surpassed the inward
FDI. FDI has significantly contributed to its growth, particularly in manufacturing,
such as electronics and textiles.

Thailand: Thailand's FDI experienced more volatility, with significant peaks and
troughs throughout the years. Notably, there was a significant increase in 2020, which
then dropped into negative figures by 2022. FDI has played a role in developing
industries like automotive, electronics, tourism, and agriculture, contributing to GDP
growth.

Singapore: Singapore's FDI has also been quite volatile, with negative values
throughout the period. The country's FDI saw a steep decline starting in 2019,
reaching the lowest levels in the last two years of the dataset. As a global financial
hub, Singapore attracts substantial FDI, boosting its GDP through various sectors,
including finance, electronics, chemicals, and services.

The reasons for these fluctuations can be multifaceted, often influenced by global
economic conditions, regional developments, domestic policies, and investor
confidence.

29
 Figure SEQ
Singapore: There wasFigure
a dip\* ARABIC 8. Merchandise
in imports around Imports
2015, by Product group -by
followed 2012atorecovery
2022 and
growth, especially after 2020. The post-2020 increase in imports is likely due to
economic recovery and stimulus efforts following the global downturn caused by the
COVID-19 pandemic. Given its small size and lack of natural resources, Singapore
heavily relies on imports for both production and consumption. Its role as a trade hub
contributes to its GDP.

Thailand: Imports for Thailand showed a decline around 2015, similar to Singapore,
with a gradual recovery until a sharp drop in 2020, likely due to the pandemic's
impact. Following 2020, there was a rapid increase in imports, indicating a strong
economic rebound. Thailand's diverse economy depends on imports that support
various industries, including automotive and electronics. Tourism also influences
consumer goods imports.

Vietnam: Vietnam displayed a consistent increase in imports over the entire period,
with no significant downturns, reflecting continuous economic growth and integration
into global supply chains. The steady rise in imports can be linked to Vietnam's
growing manufacturing sector and increased foreign direct investment. Imports are
crucial for supporting its industrialization and economic development, especially in
the manufacturing sector, much of which is for export.

Factors contributing to these fluctuations include global economic conditions,

domestic economic policies, currency fluctuations, supply chain shifts, and the post-
COVID-19 global economic recovery.

30
4.2. Statistical descriptive methods

Figure 9. Research of Statistical descriptive methods.

Summary of Findings:

1. Exports (EXP):

Singapore's exports have a high mean value, suggesting a strong export sector, with a
noticeable peak in recent years.

● Thailand's exports show less variability, indicating a more stable export

market.
● Vietnam's exports demonstrate substantial growth over the period, as shown
by the range and mean, reflecting the country's increasing integration into
global trade.
2. Foreign Direct Investment (FDI):
● Singapore has an exceptionally high mean for FDI, indicating its
attractiveness as an investment destination. However, the high variance
suggests fluctuation in investment flows, possibly due to policy changes or
global economic conditions.
● Thailand's FDI shows a negative mode value at one point, which may
indicate a year of net outflows or significant investment withdrawals.
● Vietnam's FDI has been growing, with the range indicating increased
investment peaks, possibly due to favorable investment policies.
3. Gross Domestic Product (GDP) Growth:
● Singapore's GDP growth has moderate variability with some years of
contraction, reflecting the impact of global economic trends on this trade-
dependent economy.

31
 ● Thailand's GDP growth demonstrates the highest variance and range,
indicating significant fluctuations, which may be attributed to domestic and
external economic shocks.
● Vietnam's GDP growth has the highest mean growth rate, suggesting robust
economic performance, with less variability pointing towards steady
growth.
4. Imports (IMP):
● Singapore's import values are high, paralleling its export figures, which is
characteristic of its trade-heavy economy.
● Thailand's import figures are consistent with its exports, showing less
variation year over year.
● Vietnam's imports have increased in line with its exports, reflecting the
growth of its economy and increased demand for foreign goods.

Visualization and Conclusion:

Next, let's visualize these statistics to get a more tangible grasp of the data and then
draw some overarching conclusions. We will create a series of plots to illustrate the mean
values and variability of each indicator for the three countries. This will help us visually
assess the central tendencies and the spread of the data over the years.

32
Figure 10. Visualize of statistical descriptive methods.

33
The visualizations above showcase the mean values of each economic indicator
(Exports, FDI, GDP, Imports) for Singapore, Thailand, and Vietnam over the years.
Here are the overarching conclusions drawn from both the statistical analysis and the
visual representations:

1. Exports (EXP):
● Singapore leads in average export values, which aligns with its well-
established global trade connections and diverse manufacturing base.
● Thailand and Vietnam show growth in exports, with Vietnam displaying a
notable increase, which may reflect its emerging market status and
increased competitiveness in international trade.
2. Foreign Direct Investment (FDI):
● Singapore's average FDI is significantly higher than that of Thailand and
Vietnam, highlighting its status as a global financial hub and a preferred
destination for investors.
● Thailand and Vietnam's FDI figures, while lower, still suggest healthy
investment levels, with Vietnam's increasing trend indicating rising investor
confidence.
3. Gross Domestic Product (GDP) Growth:
● Vietnam shows the highest mean GDP growth, suggesting robust economic
expansion and resilience over the years.
● Singapore's GDP growth is moderate but with some volatility, which might
be reflective of its sensitivity to global economic cycles due to its trade-
dependent economy.
● Thailand's GDP growth has been more variable, indicating susceptibility to
both internal and external economic pressures.
4. Imports (IMP):
● Singapore's import figures are high, reflecting its status as a trade hub that
re-exports goods as well as its domestic consumption needs.
● Thailand and Vietnam have lower average import values, but the consistent
growth aligns with their economic development and increasing domestic
demand.

Conclusions:

34
● Singapore's economic indicators reflect its established economic structure, with
high trade volumes and significant FDI inflows contributing to steady GDP
growth, despite occasional downturns.
● Thailand's economy shows more fluctuations, which could be due to a variety of
factors including political changes, regional economic shifts, and global market
conditions.
● Vietnam stands out with its remarkable GDP growth and rapidly increasing trade
figures, indicating successful economic reforms and integration into the global
economy.

These insights suggest that while Singapore remains a powerhouse with its strategic
economic positioning, Vietnam is rapidly catching up, showcasing significant
potential for growth and investment. Thailand, while facing more variability, still
plays a crucial role in the regional economy with its own unique strengths and
challenges. Each country's economic trajectory is influenced by a complex interplay
of domestic policies, global economic trends, and regional developments.

4.3. Hypothesis Testing

Here are the results and conclusions from the hypothesis tests:

Figure SEQ Figure \* ARABIC 11. Result of hypothesis testing.

1. T-test between Singapore and Thailand

35
 ● Reason for test: To determine if there is a statistically significant difference
in GDP growth rates between Singapore and Thailand.
● Result: The p-value is 0.432, which is greater than the significance level of
0.05.
● Conclusion: We fail to reject the null hypothesis; there is no statistically
significant difference in GDP growth rates between Singapore and
Thailand.
2. T-test between Singapore and Vietnam
● Reason for test: To determine if there is a statistically significant difference
in GDP growth rates between Singapore and Vietnam.
● Result: The p-value is 0.024, which is less than the significance level of
0.05.
● Conclusion: We reject the null hypothesis; there is a statistically significant
difference in GDP growth rates between Singapore and Vietnam.
3. T-test between Thailand and Vietnam
● Reason for test: To determine if there is a statistically significant difference
in GDP growth rates between Thailand and Vietnam.
● Result: The p-value is 0.0038, which is less than the significance level of
0.05.
● Conclusion: We reject the null hypothesis; there is a statistically significant
difference in GDP growth rates between Thailand and Vietnam.
4. ANOVA test among Singapore, Thailand and Vietnam
● Reason for test: To determine if there are any statistically significant
differences in GDP growth rates among the three countries.
● Result: The p-value is 0.0125, which is less than the significance level of
0.05.
● Conclusion: We reject the null hypothesis; there are statistically significant
differences in GDP growth rates among the three countries.

Insights:

36
● The ANOVA test indicates that not all countries have the same average GDP
growth rate, but it does not tell us which countries differ from each other. This is
where the T-tests provide additional insights.
● The T-tests reveal that while Singapore's and Thailand's GDP growth rates are not
significantly different from each other, both countries have significantly different
GDP growth rates when compared to Vietnam.
● This suggests that over the period from 2012 to 2022, Vietnam's economy has had
a different trajectory in terms of GDP growth compared to Singapore and
Thailand.
● Policymakers and investors may find this information useful for making decisions
related to economic planning and investment in these countries.

4.4. Regression methods

❖ In VietNam

Figure 12. Result of regression methods in VietNam.

Based on the table above, we have the following result:

GDP = 5,31067 + -2,20644E-10×FDI - 8,2465E-05×Export + 7,53028E-05 ×Import

37
Hypothesis 1: Whether the population regression function is significant or not.

● H0: R2 = 0 (The population regression function is not significant)

● H1: R2
0 (The population regression function is significant)

F-statistic is used to test. With number of observations (n) =11, number of variables
(k) = 3, and significance level (α) = 0.05, the result of F is 0.8785 < F α k−1
n−k = 4.4590.

− Based on the above results, H1 is rejected, therefore, H0 is accepted.

− When we accept H0, the implications of the result could be:

o No significant relationship: If there isn't enough evidence to reject H0, we
cannot conclude that the independent variables (FDI, EXP, and IMP) have a
significant relationship with the dependent variable (GDP) in the constructed
regression model.
o Inability to explain variations: When R^2 (coefficient of determination) equals
0, it might imply that no variation of the dependent variable (GDP) is
explained by the independent variables (FDI, EXP, and IMP). This could
suggest that the model is inadequate to predict or explain the variability in
GDP.

Hypothesis 2: Whether FDI impacts GDP or not.

● H0: β2 = 0 (FDI does not impact on GDP)

● H1: β2 0 (FDI impacts on GDP)

T-statistics are used to test. With number of observations (n) =11, number of variables
(k) = 3, and significance level (α) = 0.05, the result of T is -0.54209 <t n−k
α / 2 = 2.365.

The result indicates that H1 is rejected, thereby confirming that explanatory variable
FDI has no effect on explained variable GDP.

38
Hypothesis 3: Whether EXP impacts GDP or not.

● H0: β2 = 0 (EXP does not impact on GDP)

● H1: β2 0 (EXP impacts on GDP)

T-statistics are used to test. With number of observations (n) =11, number of variables
(k) = 3, and significance level (α) = 0.05, the result of T is -0,62687 <t n−k
α / 2 = 2.365.

The result indicates that H1 is rejected, thereby confirming that explanatory variable
EXP has no effect on explained variable GDP.

Hypothesis 4: Whether IMP impacts GDP or not.

● H0: β2 = 0 (IMP does not impact on GDP)

● H1: 2 0 (IMP impacts on GDP)

T-statistics are used to test. With number of observations (n) =11, number of variables
(k) = 3, and significance level (α) = 0.05, the result of T is 0,5597 < t n−k
α / 2 = 2.365.

The result indicates that H1 is rejected, thereby confirming that explanatory variable
IMP has no effect on explained variable GDP.

⮚ Summarily, the statistical hypothesis test results indicate that, overall, from
2012 to 2022, neither FDI, exports, nor imports had an impact on Vietnam's
economic growth.

❖ In Thailand

39
 Figure 13. Result of regression methods in ThaiLand.

Based on the table above, we have the following result:

GDP = 8,02879+ -2,79683E-10×FDI -3,02878E-05×Export + 1,14601E-05×Import

Hypothesis 1: Whether the population regression function is significant or not.

● H0: R2 = 0 (The population regression function is not significant)

● H1: R2
0 (The population regression function is significant)

F-statistic is used to test. With number of observations (n) =11, number of variables
(k) = 3, and significance level (α) = 0.05, the result of F is 2,2783 < F α k−1
n−k = 4.4590.

− Based on the above results, H1 is rejected, therefore, H0 is accepted.

− When we accept H0, the implications of the result could be:

o No significant relationship: If there isn't enough evidence to reject H0,
we cannot conclude that the independent variables (FDI, EXP, and IMP)
have a significant relationship with the dependent variable (GDP) in the
constructed regression model.

40
 o Inability to explain variations: When R^2 (coefficient of determination)
equals 0, it might imply that no variation of the dependent variable
(GDP) is explained by the independent variables (FDI, EXP, and IMP).
This could suggest that the model is inadequate to predict or explain the
variability in GDP.

Hypothesis 2: Whether FDI impacts GDP or not.

● H0: β2 = 0 (FDI does not impact on GDP)

● H1: β2 0 (FDI impacts on GDP)

T-statistics are used to test. With number of observations (n) =11, number of variables
(k) = 3, and significance level (α) = 0.05, the result of T is -1.7465 < t n−k
α / 2 = 2.365. The

result indicates that H1 is rejected, thereby confirming that explanatory variable FDI
has no effect on explained variable GDP.

Hypothesis 3: Whether EXP impacts GDP or not.

● H0: β2 = 0 (EXP does not impact on GDP)

● H1: β2 0 (EXP impacts on GDP)

T-statistics are used to test. With number of observations (n) =11, number of variables
(k) = 3, and significance level (α) = 0.05, the result of T is -0.3073< t n−k
α / 2 = 2.365.

The result indicates that H1 is rejected, thereby confirming that explanatory variable
EXP has no effect on explained variable GDP.

Hypothesis 4: Whether IMP impacts GDP or not.

● H0: β2 = 0 (IMP does not impact on GDP)

● H1: 2 0 (IMP impacts on GDP)

41
T-statistics are used to test. With number of observations (n) =11, number of variables
(k) = 3, and significance level (α) = 0.05, the result of T is 0.1442 < t n−k
α / 2 = 2.365.

The result indicates that H1 is rejected, thereby confirming that explanatory variable
IMP has no effect on explained variable GDP.

⮚ Summarily, the statistical hypothesis test results indicate that, overall, from
2012 to 2022, neither FDI, exports, nor imports had an impact on Thailand's
economic growth.
❖ In Singapore

Figure 14. Result of regression methods in Singapore.

Based on the table above, we have the following result:

GDP = -23,1562+ 5,15399E-11×FDI -0,00023563×Export + 0,00028334×Import

Hypothesis 1: Whether the population regression function is significant or not.

● H0: R2 = 0 (The population regression function is not significant)

● H1: R2
0 (The population regression function is significant)

42
F-statistic is used to test. With number of observations (n) =11, number of variables
(k) = 3, and significance level (α) = 0.05, the result of F is 2.3511 < F α k−1
n−k = 4.4590.

● Based on the above results, H1 is rejected, therefore, H0 is accepted.

● When we accept H0, the implications of the result could be:

o No significant relationship: If there isn't enough evidence to reject H0, we
cannot conclude that the independent variables (FDI, EXP, and IMP) have a
significant relationship with the dependent variable (GDP) in the constructed
regression model.
o Inability to explain variations: When R^2 (coefficient of determination) equals
0, it might imply that no variation of the dependent variable (GDP) is
explained by the independent variables (FDI, EXP, and IMP). This could
suggest that the model is inadequate to predict or explain the variability in
GDP.

Hypothesis 2: Whether FDI impacts GDP or not.

● H0: β2 = 0 (FDI does not impact on GDP)

● H1: β2 0 (FDI impacts on GDP)

T-statistics are used to test. With number of observations (n) =11, number of variables
(k) = 3, and significance level (α) = 0.05, the result of T is 0,969489371< t n−k
α / 2 = 2.365.

The result indicates that H1 is rejected, thereby confirming that explanatory variable
FDI has no effect on explained variable GDP.

Hypothesis 3: Whether EXP impacts GDP or not.

● H0: β2 = 0 (EXP does not impact on GDP)

● H1: β2 0 (EXP impacts on GDP)

43
T-statistics are used to test. With number of observations (n) =11, number of variables
(k) = 3, and significance level (α) = 0.05, the result of T is -2,162760533< t n−k α/ 2 =

2.365.

The result indicates that H1 is rejected, thereby confirming that explanatory variable
EXP has no effect on explained variable GDP.

Hypothesis 4: Whether IMP impacts GDP or not.

● H0: β2 = 0 (IMP does not impact on GDP)

● H1: 2 0 (IMP impacts on GDP)

T-statistics are used to test. With number of observations (n) =11, number of variables
(k) = 3, and significance level (α) = 0.05, the result of T is 2,330241871< t n−k
α / 2 = 2.365.

The result indicates that H1 is rejected, thereby confirming that explanatory variable
IMP has no effect on explained variable GDP.

⮚ Summarily, the statistical hypothesis test results indicate that, overall, from
2012 to 2022, neither FDI, exports, nor imports had an impact on Singapore's
economic growth.

4.5. Forecasting

4.5.1. Data analysis

The empirical characteristics of the univariate time series data were checked by
obtaining time plots for the data. To gain an insight into univariate processes,
autocorrelation and partial autocorrelation functions (ACF and PACF) were
considered. The ACF measures the ratio of the covariance between observations k
lags apart and the geometric average of the variance of observations (i.e. the variance
of the process when it is stationary, as Var ( Y t ) =Var (Y t −1)¿ .

44
 However, some of the observed autocorrelation between and
were due to both being correlated with intervening lags. The PACF on the other hand
seeks to measure the autocorrelation between Y t and Y t −1 correcting for the correlation
with intervening lags.

The log likelihood ratio test, AIC and the BIC were used for model diagnostic
checks. Adequacy of the model was carried out for all cases through the analysis of
the residuals by use of the Ljung-Pierce Q-statistics. In addition to the residual plots,
the Maximum Likelihood Estimate (MSE) was used to check on the efficiency of the
model. These were facilitated by use of Gretl statistical software.

Figure 15. Data analysis matrix.

4.5.2. Estimation Results

ARIMA Modeling for Vietnam's GDP Time Series Analysis

Stationarity Check: We conducted the Augmented Dickey-Fuller (ADF) test to check the
stationarity of the GDP time series. The test yielded a test statistic of -4.4955 and a p-value of
0.0002. Given that the test statistic is lower than the 1% critical value and the p-value is
significantly low, we reject the null hypothesis, concluding that the series is stationary.

Figure 16. ADF Test results for VietNam GDP.

Model Identification: We plotted the Autocorrelation Function (ACF) and Partial

Autocorrelation Function (PACF) for the GDP series. Based on the PACF, we observed a
significant spike at lag 1, suggesting an AR(1) component. The ACF showed a similar pattern,

45
which led us to consider an MA(1) component. Consequently, we started with an ARIMA(1,0,1)
model.

Figure 17. ACF Plot

Figure 18. PACF Plot.

Model Fitting: We fitted an ARIMA(1,0,1) model to the GDP data. The model fitting
resulted in the following:

● The AR(1) coefficient was not significant (p-value: 0.901).

46
● The MA(1) coefficient was also not significant, and the confidence interval was
extremely wide, indicating high uncertainty.
● The model's AIC was 47.286 and BIC was 48.878, which can be used for comparison
with other models.
● The Ljung-Box test showed a high p-value, indicating that there is no significant
autocorrelation in the residuals.

Figure 19. SARIMAX results.

Comparison with Other ARIMA Models

The above model was compared with different ARIMA models by use of model
selection criteria such as Akaike information criterion, Log likelihood, Hannan-Quinn
and Schwarz criterion, but the above model proved to be relatively robust compared
to other competing models. The results are presented below.

47
 Figure 20. Metrics for all the ARIMA models (1).

The ARIMA(2,0,2) model has the highest Log Likelihood, indicating it may be the

Figure SEQ Figure \* ARABIC 21. Metrics for all the ARIMA models (2).
best fit among those tested.

48
 The complete table with metrics for all the ARIMA models, including Mean
Squared Error (MSE)

The model with the lowest AIC and BIC is ARIMA(2, 0, 2), which has a Log
Likelihood of -16.80, a BIC of 47.99, an AIC of 45.60, a Ljung-Box p-value of 0.771
(suggesting no autocorrelation in residuals), and an MSE of 1.081. According to the
Ljung-Box test, the residuals of this model do not show evidence of autocorrelation,
which is a good indication of model fit.

Best ARIMA Model Based on AIC/BIC (ARIMA(2, 2, 2)):

● Jarque-Bera Statistic: 0.3416 (implies residuals are normally distributed as p-value >
0.05)
● Ljung-Box Statistic: 1.1777 (implies no autocorrelation in residuals as p-value > 0.05)

Best ARIMA Model Based on MSE (ARIMA(2, 0, 2)):

● MSE: 1.0805

Parameters for the Best MSE Model:

● Constant: 6.2966
● AR(1): 0.6247
● AR(2): -0.6830
● MA(1): -0.3209
● MA(2): -0.6789
● Variance (sigma^2): 0.8145

The ARIMA(2, 0, 2) model has the lowest MSE among the tested models, which would
typically indicate it as the best model for forecasting in terms of prediction error. However, the
ARIMA(2, 2, 2) model was selected based on the lowest AIC and BIC values in the paper, which
implies it is the most parsimonious model.

Given these findings, the final model selection would depend on whether we prioritize lower
MSE or lower AIC/BIC values. The paper indicates the ARIMA(2, 2, 2) as the best model, so we
would proceed with it unless the objective strictly requires minimizing MSE.

The ARIMA(2, 2, 2) model can be written as:

49
 Y t =c +0.6247 Y t −1−0.6830 Y t−2−0.3209 et −1−0.6789 e t−2 +e t

Where Y t represents the predicted value of lnGDP, et represents the error term, and c is the
constant term. The coefficients reflect the influence of the past two observations and errors on
the current prediction. The error term is assumed to be normally distributed with a mean of zero
and variance σ 2.

4.5.3. Out-of-Sample Forecasts

The ARIMA(2, 0, 2) model has been fitted to the GDP data, and forecasts for the next
5 years have been generated. Here are the forecasted values along with the lower and
upper confidence intervals:

The ARIMA model's forecasts suggest a varied trajectory for Vietnam's GDP from
2023 to 2027. The model does not necessarily predict a consistent growth trend over
the next five years; instead, the projections fluctuate. It's important to note that these
predictions are based on the assumption that past patterns in the GDP data will persist.
However, in reality, numerous factors, including economic policies, market dynamics,
and global economic conditions, will influence the actual GDP outcomes.
Consequently, while the ARIMA model provides a statistical forecast based on
historical data, it should be used cautiously and in conjunction with other economic
analyses for planning and decision-making.

However, it's essential to acknowledge that these forecasts are extrapolations based
on historical data and statistical modeling. The actual economic outcomes can be
influenced by a myriad of factors, including global economic trends, domestic
policies, and unexpected events. Therefore, while the model provides a valuable
quantitative perspective on potential future growth, it is not immune to the
complexities and dynamics of Vietnam's national economy.

50
 Adjustments in macroeconomic policy, changes in the international economic
environment, or significant socio-political events could lead to deviations from these
predictions. As such, it is prudent for policymakers and economists to remain vigilant
to the risks of economic volatility. They should be prepared to adapt macroeconomic
regulatory measures to ensure economic stability and to revise growth targets as new
data and developments unfold.

In summary, while the ARIMA(2, 0, 2) model's forecasts are promising, signaling

an upward trend in Vietnam's GDP growth for the coming years, they should be used
as one of several tools for economic planning, always considering the potential for
unforeseen changes that could impact the actual trajectory of the economy.

5. CONCLUSION INSIGHTS
5.1. Recommendation

In order to improve the effect of FDI and international trade on Vietnam's economic
growth, some recommendations for policy makers are proposed as follows:

Firstly, preferential policies to attract FDI need to be maintained, implemented

and further developed. Vietnam should have many outstanding preferential and
internationally competitive policies to attract strategic investors, leading multinational
corporations to invest in large-scale projects in Vietnam.

Secondly, FDI policies should be selective, taking quality, efficiency, technology

and environmental protection as the top goals. Vietnam should have policies that
prioritize projects with advanced and clean technology, modern management, high
added value, and pervasive effects, global connection between production and supply
chains.

Thirdly, free trade agreements should be further negotiated and concluded to

create a favorable environment for foreign trade activities. Vietnam needs to continue
affirming and implementing trade liberalization policies through active and full
participation in signing international trade commitments.

51
 Fourthly, the export-oriented economic development policy needs to be
persistently pursued. Vietnam should continue to identify and invest in developing
key commodities with highly competitive advantages to boost export.

Fifthly, benefits from import and export activities need to be enhanced. Vietnam
should increase the added value of exported goods and restrict the export of raw
materials, resources and minerals, control types of goods imported into Vietnam, shift
from import of consumer goods which can be produced domestically to import of raw
materials for production, advanced technology, and domestically unavailable goods or
insufficient production to meet domestic demand.

5.2. Conclusions

This study used the ordinary least-square method to assess the impact of FDI,
international trade (export and import) on economic growth for the period 2012-2022.
The statistical test results show that FDI and international trade are related to
economic growth, however economic variables have different impacts. FDI and
exports have a positive and statistically significant effect on economic growth.
Imports also have a negative but not statistically significant impact on economic
growth. Identification of the above-mentioned directions of impacts helps the
government to have appropriate policies to improve the effect of FDI and
international trade on economic growth. .

6.[5.] ACKNOWLEDGEMENTS
We would like to express our deepest gratitude to everyone who has supported and
contributed to the completion of this report. The invaluable guidance and knowledge
from our teachers, the assistance and support from our friends and colleagues, and the
encouragement from our families have been the pillars of our journey throughout the
research and writing process. This acknowledgment is also a sincere tribute to all
those who have directly or indirectly contributed to the success of this work.

7.[6.] TASK ASSIGNMENTS SHEET

● Trello Dashboard

52
● List of work assignments for team members:

Full Name Assigned Task

Nguyen Quoc Khanh - Present + Do Slide

- Statistical descriptive methods
- Hypothesis Testing
- Forecasting Section

Mai Hien Nga - Do report

- Do slide

Dang Thi Kim Yen - Do report

- Do slide

Bui Thi Thanh - Support Hypothesis Testing

- Support Multiple Regession

53
 Model
- Support Presenting

Pham Thanh Dat - Support Hypothesis Testing

- Support Multiple Regession
Model
- Present + Do Slide
- Do report

Pham Tien Dat - Data Overview

- Hypothesis Testing

Bui Si Khoa - Data Overview

- Statistical descriptive methods

8.[7.] REFERENCES

[1] Dhiraj Jain*, K. Sanal Nair** and Vaishali Jain* (2015) Factors Affecting GDP
(Manufacturing, Services, Industry): An Indian Perspective

[2] Alex Reuben Kira (2013) The Factors Affecting Gross Domestic Product (GDP)
in Developing Countries: The Case of Tanzania

[3] Cenap Ilter (2017) WHAT ECONOMIC AND SOCIAL FACTORS AFFECT
GDP PER CAPITA? A STUDY ON 40 COUNTRIES.
[4] Nguyen, H. (2020). Impact of foreign direct investment and international trade on
economic growth: Empirical study in Vietnam. The Journal of Asian Finance,
Economics and Business, 7(3), 323–331.

54
[5] Box, George E. P. and Gwilym M. Jenkins (1976). Time Series Analysis:
Forecasting and Control, Revised Edition, Oakland, CA: Holden-Day.

[6] Wold H. (1938), A Study in Analysis of Stationary Time Series, Uppsala.

55