GROUP RESEARCH

TOPIC: FACTOR AFFECTING UNEMPLOYMENT RATE IN 5 AFRICAN

COUNTRIES (BOTSWANA, DJIBOUTI, EQUATORIAL GUINEA, LESOTHO,
SOUTH AFRICA)

QUANTITATIVE RESEARCH METHODS (MGT646)

M1BA2425A

GROUP 5:

NAME STUDENT’S ID

1. Muhamad Amzar bin Zulkifli 2020896788

2. Aini Nadia Nurfaiz binti Mohd Faizal 2020602166

3. Muhamad Ameen bin Saifulzari 2020602598

4. Noraslinda binti Kamaruzaman 2020477604

5. Nur Syazwani binti Shamsuddin 2020897488

6. Siti Hapija binti Basir 2020495356

7. Wan Muhammad Farkhan bin Wan Aman 2020448858

 CHAPTER 1

1.0 RESEARCH BACKGROUND

People who are unemployed are those who do not have a job, are seeking for work, or
are setting up a new enterprise. While the unemployment rate is a comparison of the number
of employed and jobless people during a specific time period given as a percentage (Abdul
Majid et al., 2019) . The unemployment rate is a useful indicator of the imbalances between
the supply and demand for labour, shedding light on the untapped labour pool. The
unemployment rate is one of the most often generated and utilised labour market indicators due
to its great analytical value and significance for evaluating the labour market (Gammarano,
Kapsos and Walsh, 2019).

There are anticipated to be 120.5 million unemployed workers in 2021, up from 114.6
million in 2019. Sub-Saharan Africa is the main cause of this growth; by 2021, it is predicted
to have 99.4 million unemployed workers there, compared to 21 million in northern Africa.
According to the same prediction, the total labour underutilization rate for all of Africa, 24.7
percent for northern Africa, and 21.5 percent for sub-Saharan Africa is anticipated to be around
22 percent in 2021 (Shawa, 2020). Besides, Africa has the world's youngest population, and
the number of African young people is predicted to grow dramatically in the coming years.
This opens up the possibility of reaping the advantages of the demographic dividend, but only
if young people have the appropriate skills required in growth-enhancing areas and can find
profitable jobs (Geneva, 2022). South Africa has a big population, estimated to be 61 million
(0.76% of the total world population) according to latest figures. Southern Africa had the
highest subregional unemployment rate of 35.3 percent in the fourth quarter of 2021, up from
34.9% the previous period. It was the highest rate of unemployment since similar records began
in 2001 according to World Data Bank in South Africa in 2021.

The trend of unemployment in Botswana from 1995 to 2020 has been fluctuating, with
the rate in 1995 being 15.88% and increasing until 2003 (23.79%). This is because the 2001
budget speech reports a further drop in unemployment to around 15.8 percent in 2000. 2016
(Siphambe). The trend then falls to 16.17% in 2009, as inflation accelerates, and workers may
supply labour in the short term due to higher wages (Andreas Hornstein, 2020). Following that,
the trend of the unemployment rate soars to 24.93% in 2020, owing to the country's rapid
growth in these years, which enabled it to move from a position of severe poverty to being one
of the richest in the region, and is one of the few in Sub-Saharan Africa now classified as a
middle-income country (Siphambe, 2016).

From 1995 to 2020, the trend in Djibouti continues to climb, and the unemployment
rate is considered high, at 28.39% in 2020. Djibouti, according to the study, has few natural
resources and little industry. All of these factors contribute to its reliance on foreign aid to
support its balance of payments and finance development projects. Unemployment remains a
major issue with a rate of 60%. (Dianah (2021)

Equatorial Guinea's unemployment rate has also risen from 1995 to 2020, owing to an
increasing population, and researchers claim that people are migrating for better opportunities
(Karen, 2017). Lesotho's unemployment rate has been fluctuating, declining from 1995 to
2019, but rebounding in 2020, with the main reason being the impact of the lockdown. Finally,
South Africa has the highest unemployment rate in Africa. This is due to the fact that low
education levels and barriers to hiring foreign workers weigh on the supply side of the labour
market, as businesses struggle to find qualified workers. Wages are also driven up by strong
labour unions and wage-bargaining agreements that lack cross-sector coordination on the
demand side (Naidoo, 2021).
2.0 PROBLEM STATEMENT

As stated by Mohamed and Ibrahim (2018), the unemployment and poverty rate of
Djibouti still stay on the very critical level although their government has attempted many
measures including forming agencies to promote the entrepreneurships, Strategic Document
for Reducing the Poverty (SRDP) in the year of 2003 and National Initiative of Social
Development (NISD) in 2007. Damane and Sekantsi (2018), explained that the massive
unemployment issue has been of big trouble to economic practitioner, policy-makers in all
expanding countries. Because of this, this has become a hot topic in any debate of politics for
more than decades. Lesotho is also currently facing issues as the country has been enduring
very tremendous unemployment rates which range between 23 and 28 percent over the previous
10 years. According to Sechele (2021), the ranking of Botswana as one of the best countries in
diamond production and sales in the globe and overall economic growth did not significantly
impact on employment creation. In the last quarter of 2020, the overall rate of unemployment
in this country shown a record of 24.5 percent. Onuoha and Agbede (2019) added that in
Equatorial Guinea, unemployment rate has reduced by 22.22% and 364.58% due to the long-
run elasticities of infrastructure and expenditure on education. However, this has also triggered
an increment on unemployment rate by 22.22% and 364.58 because of the long-run defence
and elasticities of health expenditure. Based on the statement of Ferreira and Rossouw (2016),
for South Africa to evade the potentially uncontrollable unemployment crisis in which it has
long been stuck, the country requires a receptive and excellent economic policy network.
Source: World Bank, Unemployment Rate by Country 2022

Source: International Labour Organization, ILOSTAT database.

Data as of June 2022

 The first bad impact of unemployment is a drop in consumer spending. According to
Ganong and Noel (2016) through their statement in the journal study entitled “How Does
Unemployment Affect Consumer Spending?”unemployment has led to a large but short-lived
drop of income, producing a need for liquidity. Drop of income has influenced the unemployed
to have less spending power. Even though some of them received Unemployment Insurance
(UI) to compensate on their living, they still maintain their spending to be on the low level even
after re-employment as an effort to reconstruct their financial safeguard. And for those who
runs out of the UI benefits, their spending power drops lower than the people who still have
their UI.

As we know, when the unemployment rate is high it can lead to the increasing of criminal
cases such as murder, robbery and many more. Thus, a contributing factor in these high crime
rates is the level of unemployment in various nations. In Sub-Saharan African nations,
unemployment can be seen as a significant contributing cause to crime. United Nations figures
show that over 464.000 people were killed from homicide in 2017 (UNODC, 2019). Although
this number provides less information, there are actually differences between countries and
even continents. After the United States of America, Africa has the second-highest murder rate
in the world which the rate is between 35% to 37%. Meanwhile, Asia is 2.3%, Europe is 3%
and Oceania is 2.8%. And we can see the big gap between Africa and the other 3 continents
(Ayang, Timbi 2021). Besides that, another year also stated that unemployment can be seen as
a significant contributing cause to crime in African nations. This is because according to the
World Bank Data in 2019, 5.39% is average unemployment. Nevertheless, it demonstrates the
differences between the many parts of the world. For instance, the Middle East and North
Africa countries have a rate of 11.52%, meanwhile the OECD countries indicate a rate of
5.17%. With a rate of 6.18%, Sub-Saharan Africa (SSA) is slightly above the global average.
 Next, poverty is also one of the bad impacts that contributed from unemployment. As
explained by Isa et. al, (2019) through the journal study titled “, Unemployment could lead
towards poverty in many ways. One of them is that the unemployment crisis will directly
impact the poverty rate income with the consumption rate of poverty if one household holds
liquidity constraints where current consumption is heavily supported by the current income.
However, if that particular household does not deal with liquidity pressure, where current
income does not greatly support the current consumption, then the escalation in poverty in the
long run will occurs due to the upsurge of unemployment but does not entirely effect on short-
term.
3.0 RESEARCH OBJECTIVE

i) General Objectives

The general objective is to examine the relationship between factors that affect
unemployment rate and the independent variables in Botswana, Djibouti,
Equatorial Guinea, Lesotho, South Africa from 1995-2020.

ii) Specific Objectives

(1) To examine the significant relationship between factors that affect

unemployment rate and the population growth, natural resources, foreign direct
investment, inflation, and exchange rates in Botswana, Djibouti, Equatorial
Guinea, Lesotho, South Africa from 1995-2020.

(2) To investigate the short-run or long-run equilibrium relationship between the

factors that affect unemployment rate and the the population growth, natural
resources, foreign direct investment, inflation, and exchange rates in Botswana,
Djibouti, Equatorial Guinea, Lesotho, South Africa from 1995-2020.
 4.0 QUESTION OF THE STUDY

I. Is there any significant relationship between unemployment and at least one of the
independent variables in Botswana, Djibouti, Equatorial Guinea, Lesotho and South
Africa from year 1995-2020
II. Is there any significant relationship between population growth on unemployment in
Botswana, Djibouti, Equatorial Guinea, Lesotho and South Africa from year 1995 –
2020.
III. Is there any significant relationship between natural resources rent and unemployment
in Botswana, Djibouti, Equatorial Guinea, Lesotho and South Africa from year 1995 –
2020.

IV. Is there any significant relationship between FDI inflows and unemployment in
Botswana, Djibouti, Equatorial Guinea, Lesotho and South Africa from year 1995 –
2020.

V. Is there any significant relationship between inflation and unemployment in Botswana,

Djibouti, Equatorial Guinea, Lesotho and South Africa from year 1995 – 2020.

VI. Is there any significant relationship between exchange rates and unemployment in
Botswana, Djibouti, Equatorial Guinea, Lesotho and South Africa from year 1995 –
2020.

VII. Is there any short-run or long-run equilibrium relationship between unemployment and
at least one of the independent variables in Botswana, Djibouti, Equatorial Guinea,
Lesotho and South Africa from year 1995 – 2020.

VIII. Is there any short-run or long-run equilibrium relationship between population growth
on unemployment in Botswana, Djibouti, Equatorial Guinea, Lesotho and South Africa
from year 1995 – 2020.

IX. Is there any short-run or long-run equilibrium relationship between natural resources
rent and unemployment in Botswana, Djibouti, Equatorial Guinea, Lesotho and South
Africa from year 1995 – 2020.
 X. Is there any short-run or long-run equilibrium relationship between FDI inflows and
unemployment in Botswana, Djibouti, Equatorial Guinea, Lesotho and South Africa
from year 1995 – 2020.

XI. Is there any short-run or long-run equilibrium relationship between inflation and
unemployment in Botswana, Djibouti, Equatorial Guinea, Lesotho and South Africa
from year 1995 – 2020.

XII. Is there any short-run or long-run equilibrium relationship between exchange rates and
unemployment in Botswana, Djibouti, Equatorial Guinea, Lesotho and South Africa
from year 1995 – 2020
5.0 HYPOTHESES OF THE STUDY

H0: There is no significant relationship between all independent variables and

unemployment rate in African countries.

H1: At least one independent variable has significant relationship with unemployment

rate in African countries.

H0: There is no significant relationship between increase of population and

unemployment rate in African countries.

H1: There is significant relationship between increase of population and

unemployment rate in African countries.

H0: There is no significant relationship between natural resources and unemployment

rate in African countries.

H1: There is significant relationship between natural resources and unemployment

rate in African countries.

H0: There is no significant relationship between Foreign Direct investment and

unemployment rate in African countries.

H1: There is significant relationship between Foreign Direct investment and

unemployment rate in African countries.

H0: There is no significant relationship between inflation and unemployment rate in

African countries.

H1: There is significant relationship between inflation and unemployment rate in

African countries.

H0: There is no significant relationship between exchange rate and unemployment rate

in African countries.
H1: There is significant relationship between exchange rate and unemployment rate

in African countries.

H0: There is no short run and long run equilibrium significant relationship between all

independent variables and unemployment rate in African countries.

H1: At least one independent variable has short run and long run equilibrium

significant relationship with unemployment rate in African countries.

H0: There is no short run and long run equilibrium significant relationship between

increase of population and unemployment rate in African countries.

H1: There is short run and long run equilibrium significant relationship between

increase of population and unemployment rate in African countries.

H0: There is no short run and long run equilibrium significant relationship between

natural resources and unemployment rate in African countries.

H1: There is short run and long run equilibrium significant relationship between

natural resources and unemployment rate in African countries.

H0: There is no short run and long run equilibrium significant relationship between

Foreign Direct investment and unemployment rate in African countries.

H1: There is short run and long run equilibrium significant relationship between

Foreign Direct investment and unemployment rate in African countries.

H0: There is no short run and long run equilibrium significant relationship between

inflation and unemployment rate in African countries.

H1: There is short run and long run equilibrium significant relationship between

inflation and unemployment rate in African countries.

H0: There is no short run and long run equilibrium significant relationship between
 exchange rate and unemployment rate in African countries.

H1: There is short run and long run equilibrium significant relationship between

exchange rate and unemployment rate in African countries.

6.0 SIGNIFICANT OF THE STUDY

For the last few decades, the unemployment rate has been one the favorite subject of
study. Unfortunately, none of the earlier studies combined together the case of unemployment
in 5 African countries with the highest rate of unemployment from 1995 to 2020. This paper is
about finding what make the unemployment rate in these 5 African countries so high by
analyzing how increase in population, foreign direct investment (FDI), exchange rate, inflation
and natural resources production in South Africa, Botswana, Djibouti, Lesotho, and Equatorial
Guinea over the past two decades influences the unemployment rate. This is done by examining
the changes in unemployment rate and the independent variables from 1995 to 2020.
Unemployment has been eroding these countries’ productivity and growth. Unmistakably, they
need to find the cause to the root so that their policy maker is able to construct a well-informed
and effective action against unemployment.

The results of this research could be highly significant and beneficial for the policy
maker as they'll be able to revise a plan to avoid high unemployment rate. Direct monetary
support is one of the common acts to deal with unemployment and save the unemployed from
falling into poverty (Rahman, 2020). Government’s policy maker has the authority and power
to form an act to maintain the unemployment rate at a minimum which will ensure increase in
country’s GDP.
 CHAPTER 2

1.0 Theory

Professor Arthur Cecil Pigou bought the first theory of unemployment and introduced
the Pigou Effect. The Pigou Effects reflects the relationship between employment, wealth,
consumption and output during the deflation period. It states that when prices are deflation,
employment and output will increase because there is an increase in wealth and consumption.
(Pigou, 1933). There is a conflict between market efficiency and market failure especially in
the scope of labor market workers. The labor market is closely related to the economy which
makes its performance give more impact for most people than any other market (Solow, 1980).
The classical theory that was introduced and analyzed by Pigou and Solow argues that the labor
market consists of demand and supply of labor (Mouhammed, 2012). Demand for labor is a
derived demand that comes from the declining portion of the marginal product of labor. The
demand curve has a negative relationship with real wage. The supply of labor comes from
worker's choice whether to spend part of their time working or not. Next, the supply of hours
worked is a positive function of the real wage. If the real wage increases, workers supply more
hours of work. In stability, demand and supply of labor are crossed-over at a clearing point that
determines the equilibrium real wage rate and full employment. According to John Maynard
Keynes, employment is based on the principle of effective demand (Keynes, 1936).

2.1 Review literature

2.1.1 Population

The population growth rate is a calculation of density or abundance trends.

(Richard and Hone,2016). The study will look at the impact of population growth and
unemployment in Zanzibar from 1990 to 2020. Population growth has been shown to
have a significant impact on Zanzibar's unemployment rate. To test the hypothesis, the
researchers used the Augmented Dickey Fuller (ADF) test. Researchers discovered a
positive, significant, and long-term relationship between population growth and
unemployment. The effect of population growth on unemployment in Zanzibar
demonstrates that both unemployment and population growth contribute significantly
to GDP (Othman Ali et al, 2021).
 Population growth has been shown to have a significant impact on Somalia's
unemployment rate. The study employs a regression model, which demonstrates that
population growth and unemployment have a positive, short-term relationship. Nearly
three-quarters of Somalia's population is under the age of 30. Most were born after the
demise of the Siad Barre regime in 1991 and have only known conflict and violence. If
the majority of the generation has not received an education, a job, or a stable life
(Sagal, 2021).

In Nigeria, a geometric increase in population without a corresponding increase

in economic activity results in a geometric increase in the country's unemployment rate.
Using the VECM, the results show that the population has a short-term, positive, and
significant relationship with Nigeria's unemployment rate. The country's population
continues to grow as a result of the high birth rate and polygamous family structure,
particularly in the north. More importantly, the problem of border control, which allows
people to enter the country without restrictions, especially in the north, increasing the
country's immigrant population, all contribute to the country's rising unemployment
rate (Babatunde and Olayinka, 2017).

The greater the population, the more labour is required in an economy and the
higher the output. In Indonesia, the Johansen Juselius co-integration test was used to
determine the existence of a long run equilibrium relationship between the variables
studied. These findings imply that unemployment, population growth, GDPpc, and
inflation in Indonesia have a long-run relationship. As a result, the series are co-
integrated, and a long run equilibrium relationship exists. The study also states that
there are positive and significant relationships. The qualitative dimension of a
population is related to the quality of human capital embodied in a workforce. Human
capital with higher levels of quality will be more productive in terms of contributing to
the economy (Mahdawi et al, 2021).

Overpopulation in India has resulted in unemployment because there are fewer

jobs to support a large number of people. Unemployment increases crime, such as theft,
because people want to feed their families and provide for their basic needs. The ADF
unit root test was used to determine the order of variable integration. The Wald Test
was used to investigate the long and short term equilibrium of the variables. Population
growth and unemployment have been found to have a positive, significant, and long-
 term relationship. Furthermore, the study's regression analysis results show that
population plays a significant role in India's increased level of unemployment. (Tahir
et. al, 2015).

2.1.2 Natural Resources

The study indicated a relationship between the natural resources and

unemployment in Pakistan. The study has been examined with the data covering the
period from 1980-2016 by using the ADF unit root test and the ARDL bound testing
method. Natural resources have an inverse and significant relationship with
unemployment, according to long-term studies. The results demonstrate natural
resources have a short-term, positive, and significant relationship with Pakistan's
unemployment rate using the VECM. This is because agglomeration of natural
resources is crucial for lowering unemployment in Pakistan. Other aspects of the
agglomeration economy include regional economic policy, the accessibility of natural
resources, and the spatial distribution of Pakistan's labour force. The Pakistani
government should make plans for the discovery of more natural resources so that
natural resource agglomeration can be improved, further lowering Pakistan's
unemployment rate (Ali et al, 2018).

According to the study, there is a relationship between the natural resources and
unemployment in Indonesia. The study has been examined using data from the time
period from 2007-2016 by using the partial least square method. The result of the study
shows that the natural resource share fund has no significant effect on unemployment
rate. The development of oil and gas projects and the high labour costs for work in the
oil and gas sector are enticing for job seekers from neighbouring regions, so the amount
of labor-intensive employment that develops in producing regions is growing. This is
due to infrastructure programmes being still largely concentrated in physical
development, which is probably not directed at employment development, the mining
sector still accounting for the majority of projects that create labor-intensive
employment in producing regions, and the development of mining projects being the
main source of temporary employment. (Majid et al, 2019).

The article studies the relationship between the natural resources and
unemployment in African countries. The study has been examined using data from the
time period from 2007-2016 by using the pooled mean group (PMG) estimator and
utilised stochastic frontier analysis (SFA). The result of the study shows that natural
resources endowment has a negative and significant relationship with the
unemployment level. This is because by implication, the unemployment rate among
SSA declines as resource rent rises. Long-term, this may be the capacity of resource-
rich nations to convert or translate their endowment of natural resources into economic
performance through the expansion of output, creation of jobs, and improvement of the
region's potential for growth. (Akinyele et al, 2022).

Based on the study, it is clear that there is a relationship between natural

resources and unemployment based on research published in the journal "Natural
Resource Rents and Unemployment in Oil Exporting Countries." Natural resource
availability and economic development have an antagonistic relationship, with
countries with limited resources experiencing slower economic growth than those with
abundant resources. Because the ARDL is applicable regardless of stationarity level
and can be used in cases of small samples, it has been used to investigate data from the
years 1991 to 2016 for this study. From another journal entitled “Further Investigation
of Natural Resources and Economic Growth: Do Natural Resources Depress Economic
Growth?”, it shows that the richness of natural resources had positive effects on
economic expansion. Resources will be diverted from the tradable sector to the non-
tradable sector as a result of an increase in resource rents, which causes higher
government spending on the non-tradable sector. As a result, there will be a de-
industrialization of the economy, rising unemployment, and declining output. This is
known as Dutch Disease (Omojolaibi, 2016). There are also the long- and short-run
coefficients between unemployment and natural resources. In the long-run, there is a
positive impact on the unemployment rate, while GDP growth rate has a negative
impact on unemployment. According to the findings, oil and natural gas rentals have
considerable, positive effects on unemployment. The level of unemployment is
positively correlated with gross capital formation. The majority of the variables have
short-term statistical significance of zero. This is because the availability of natural
resources will encourage industry job growth and reduce unemployment.
 The study entitled “Effect of Natural Resources Share Fund on Economic
Growth, Unemployment and Poverty in Producing Regions in Indonesia”, indicates that
using capital expenditures as a medium, analytical findings about the impact of
Indonesia's natural resource share fund (NSF) on economic development and
unemployment will be obtained. This study also uses partial least squares to calculate
the indirect and direct effects that produced oil and gas in Indonesia between 2007 and
2016. It shows that natural resources have a positive and significant effect on economic
growth but don’t have any significant effect on the unemployment rate. This study also
has a long- and short-run. In the long run, it is clear that natural resource yields have a
major impact on the unemployment rate, but in the short run, it has zero significant
effect. From the findings, Okun's Law's prediction of an inverse connection between
unemployment rate and economic growth is not supported by substantial evidence
(Sadiku, 2015). This is due to the high proportion of new immigrants who are job
seekers, which causes a decrease in work opportunities and an increase in the
unemployment rate.

2.1.3 Foreign Direct Investment

According to (Mustafa Alalawneh and Azizun Nessa, 2020), there is no effect

of foreign direct investment on unemployment in the short term and a negative effect
in the long term in six countries in the Middle East and North Africa: Egypt, Jordan,
Lebanon, Morocco, Tunisia, and Turkey. The study examined panel data from 1990 to
2018, with three economic models used to assess the long-run impact of FDI on
unemployment using the Fixed Effect Model (FEM) and Random Effect Model (REM),
as well as determining the causal link in the short term using Panel VAR (Granger
causality tests). In the long term, the FDI coefficient is negative; if FDI grows by 1%,
the unemployment rate (UNE) decreases by 0.092%, assuming other variables remain
stable. This result is consistent with the hypothesis and assumptions that see foreign
direct investment working to achieve economic savings for workers, as represented by
increases in their real wages and productive capabilities, as well as the creation of
vertical and horizontal integrations and the encouragement of supportive project
investments, resulting in the creation of new job opportunities. Due to the weak
unidirectional causal link from unemployment to FDI and the high unidirectional
causation from export to FDI, these results also suggest that foreign direct investment
does not have an immediate impact on unemployment. This outcome confirms that
increased exports have a major effect on foreign direct investment. Exports may serve
as a catalyst for increased FDI when combined with other elements like tax benefits,
less red tape, and efficient administration.

Besides, FDI and its dependent variable in Malaysia have a relationship that is
both long-term and short-term (Hamidah Muhd Irpan et al., 2016). This study employed
yearly data from 1982 to 2012, and the analytic method used was Autoregressive
Distributed Lags (ARDL). According to the study's findings, in the long term, LIFDI
and LFL have a significant impact on the unemployment rate, with a 1% rise in both
items lowering the unemployment rate in Malaysia by 0.255% and 0.625%,
respectively. The short-run dynamic produces negative and significant outcomes. The
absolute magnitude of the coefficient of the error-correction term suggests that short-
run adjustment offsets approximately 55% of the disequilibrium in the unemployment
rate each year. This suggests that a low unemployment rate is followed by changes in
the other variables in the next period.

According to (Elsa Widia et al., 2019), there is a relationship between FDI and
unemployment rates in four ASEAN countries: Indonesia, Malaysia, Singapore, and
Thailand. This study used a series of Vector Error Correction Model (VECM) analyses
to examine the influence of FDI on unemployment rates in each ASEAN 5 countries,
both in the long and short term. This analysis relied on yearly data spanning the years
1980 to 2017. The study's findings explain why the variable impact is felt more in the
long run. In the long run, the unemployment rate has a negative and considerable
influence on FDI in Indonesia, Singapore, and Thailand. With a large coefficient, it is
clear that the unemployment rate significantly affects foreign direct investment.
However, short-term estimations produce inconsistent results that, on average, reveal
insignificant outcomes. With the exception of Thailand, the impact of the
unemployment rate on FDI is minimal compared to the long term. Additionally, the
short-term effects of FDI on the unemployment rate are insignificant. Therefore, this
study demonstrates that FDI does have a positive influence on the economy, but
unfortunately, the benefit is only seen over the long run.
 There were also studies about how FDI affects the unemployment rate in Middle
East and North Africa (MENA) countries which are Egypt, Jordan, Lebanon, Morocco,
Tunisia, and Turkey. The outcome of the studies showed that high FDI reduces the
unemployment rate, for male and females in the long term (Alalawneh, 2020). They
also revealed that there is no strong relationship in the short term between FDI and
unemployment in its various forms. Meanwhile, there is a strong relationship between
FDI and exports according to the three economic models. This study used the Fixed
Effect Model (FEM) and Random Effect Model (REM). In addition, they also use
Panel VAR (Granger causality tests) to finding the causal relationship in the short
term.

Another study that was conducted in South Africa Development Community

(SADC) also agrees on the significant impact of FDI on unemployment rate. Youth
unemployment in the SADC region is high even though the region receives the largest
percentage of FDI inflow (Evans, 2018). The sector that has a huge potential to improve
youth unemployment in Africa is the agricultural sector as it is the main economic
activity in the region and also requires a basic level of skills. Other than more
investment, employment and foreign exchange, FDI can also benefit host countries
from spillovers from productivity gain resulting from the exchanging of knowledge and
technology from overseas investors to local firms and workers, cheaper prices and
efficient allocation of resources (Demena, 2019). This study was conducted using the
Feasible Generalized Least Squares (FGLS-Parks) technique.

2.1.4 Inflation

Based on the major findings done through the study entitled “Inflation and
Unemployment Nexus in Nigeria: Another Test of the Phillips Curve”, it is summarised
that there exist a positive and significant relationship between inflation rate and
unemployment. The implication of this result is that there is evidence of stagflation in
Nigeria. This is proven by the model used in this study which is Ordinary Least Square
(OLS) analysis according to Orji and Okafor (2015).

On the other hand, the result from the findings done through the method known
as Ramsey RESET test in the journal “Inflation and Unemployment Relationship In
 The Long Run: A Comparative Study” displays that there is negative relationship
between inflation and unemployment in USA, Germany, and France in the long-run for
each country test estimation explained by Mustafa (2017).

Based on the model analysis of Ordinary Least Square (OLS) regression

constructed in obtaining the result of the study in the journal entitled “Impact of
Inflation and Unemployment on Economic Growth in Ten Selected Member’s States
of Economic Community of West Africa States”, it shows that an increase of one unit
on unemployment rate would contribute about certain level of increase in inflation rate
in the short-run when there is trade-off but in the long-run when trade-off
disappearance, a unit increase on unemployment rate would lead to a decreases about
certain level of inflation rate which proves that both unemployment and inflation has a
significant relationship on the long-term (Jelilov, Obasa, & Isik, 2016).

Furthermore, the results of the findings on model used which is Error Correction
Model (ECM) show that inflation effect negatively on unemployment while money
supply and exchange rate were discovered to hold positive influence on unemployment.
There is distinction along the equilibrium part which is corrected at 65 percent speed
of adjustment each year even though unemployment and inflation have long-run
association (Okafor, Chijindu, & Ugochukwu, 2016).

Also, based on the results shown through the Ordinary Least Squares (OLS)
regression model, these two independent variables which are inflation and economic
growth have a significant effect on the dependent variable, unemployment. Based on
theoretical exchange, the linear regression model with two independent variables
demonstrated that the relation expressed in this model is statistically significant. The
model from the study displays that the negative relationship between unemployment
and inflation as well as showing us the Philips curve, also holds valid for the Republic
of Macedonia regarding the time period from 1998 to 2013 (Aqifi & Duka, 2016).

2.1.5 Exchange rate

Atif (2019) revealed this paper estimated the impact of the real effective
exchange rate (REER) on Pakistan's unemployment rate from 1991 to 2015. The
Autoregressive Distributive Lag (ARDL) Model approach was used based on the
findings of stationarity tests. The ARDL model's bounds test showed the cointegration
of the variables. Then, based on the ARDL model, the long-term results revealed that
while GDP growth had no effect on unemployment, other factors such as the money
supply, trade openness, and REER appreciation did. Beside that, High convergence
speed of 67% was estimated, and the lagged ECM term's negative and significant
coefficient indicated the long-term relationship's existence. The study's key finding is
that there is no strong correlation among both REER appreciation and unemployment
in Pakistan; rather, appreciation has a negative impact on unemployment, while growth
in Pakistan had little of an impact on it throughout the study's data period. The results
have implications for policy, showing that while REER realignment towards
equilibrium is required to achieve macroeconomic objectives, REER depreciation does
not lower unemployment.

Ku (2018) exposed the problems by concentrating on 12 EU nations between

1994 and 2017. The information demonstrates a strong correlation between real
exchange rate RER and unemployment. Based on the finding, unemployment rises
(falls) when the RER appreciates or depreciates. Besides that, there have a positive
significant relationship between RER and unemployment. The author also indicate that
RER have a short-term effect along with unemployment.

Furthermore, Usman and Elsalih (2018) also discovered a study on the real
exchange rate and unemployment. The authors used linear and nonlinear
Autoregressive Distributed Lag (ARDL) models to study the pass-through of the real
exchange rate (RERT) to unemployment in Brazil for the period 1981/January-
2015/November. As the result they found the significant relationship between two
variables which is linear in short-run and nonlinear in the long-run.

Raji (2019) revealed the issue how actual exchange rates and inflation affect
unemployment. If changes in the real exchange rate and inflation affect unemployment
levels in Nigeria and whether inflation modifies the real exchange rate's impact on
unemployment. Therefore, The Generalized Method of Moments (GMM) technique is
used by the author to reduce the endogeneity of the study's variables. The study show
that the real exchange rate and inflation have a positive significant in short and long
run.
 Furthermore, Mpofu & Nikolaidou (2018) revealed the effect of exchange rate
volatility on employment increase for the period 1995:03-2015:02 in South Africa. The
research employing the ARDL co-integration method revealed that real exchange rate
volatility has a significant stalling effect on the growth of manufacturing employment.
To conclude that, it has a significant positive relationship between exchange rate and
unemployment. Thus, there have relationship in a long run.
DEPENDENT VARIABLE: Unemployment rate in Africa countries.

INDEPENDENT VARIABLES:

1) Increase of population

Year Botswana Djibouti Equatorial Lesotho South Africa

Guinea

1995 2.320880809 0.632478856 44.1014433 4.06747678 9.360533943

1996 1.248209312 0.664007194 55.54560425 4.42866983 10.30309396

1997 1.381343281 0.676909113 60.39751917 4.03080585 8.772342832

1998 0.850202049 0.947348472 43.29494565 4.62753146 6.639169107

1999 1.356367529 0.391087418 62.00105683 3.37253239 6.504001198

2000 2.374904358 0.382079268 56.14016853 3.45870924 9.926358007

2001 1.160004172 0.415974591 57.97764193 3.6129472 8.571841955

2002 0.909346565 0.454277724 42.68974498 4.6497299 8.468764712

2003 1.928985432 0.623059168 57.97764193 5.46817367 8.464808887

2004 3.221841886 0.573154706 56.14016853 3.85504574 10.36752994

2005 2.339752245 0.529249755 59.58415793 3.42595235 12.75513394

2006 6.300446181 0.616385692 56.14016853 3.06843906 13.29129614

2007 7.157616266 0.523310117 49.77586237 5.05256764 13.94259338

2008 3.884623518 0.790439022 47.69416096 5.45904126 18.55844908

2009 2.213216086 0.743212074 29.84970106 5.90302784 10.62920898

2010 2.080608591 0.838156181 37.98484876 3.93439969 13.0709945

2011 1.859884943 0.88846349 41.48334567 3.88951381 16.06830425

2012 2.175497767 0.893774805 39.70623681 4.69813331 13.77940114

2013 1.34946191 0.522615829 34.78195147 5.08948538 11.93041513

2014 0.909144872 0.788998443 29.07344267 5.59052228 10.27552428

2015 0.790999719 0.66101449 18.00896168 6.21135693 6.333754644

2016 0.752329813 0.484687693 16.04665436 7.27048445 6.515658233

2017 0.866281934 0.573709927 22.96992744 6.29839663 8.102472642

2018 0.809852818 0.267564658 30.48183253 3.36827251 8.642011355

2019 0.627107705 0.331863659 27.18096837 3.86721027 7.795553255

2020 0.678482793 0.299293543 23.35242084 5.09557239 6.7563406

2) Natural Resource rent

Year Botswana Djibouti Equatorial Lesotho South Africa

Guinea

1995 2.320880809 0.632478856 44.1014433 4.06747678 9.360533943

1996 1.248209312 0.664007194 55.54560425 4.42866983 10.30309396

1997 1.381343281 0.676909113 60.39751917 4.03080585 8.772342832

1998 0.850202049 0.947348472 43.29494565 4.62753146 6.639169107

1999 1.356367529 0.391087418 62.00105683 3.37253239 6.504001198

2000 2.374904358 0.382079268 56.14016853 3.45870924 9.926358007

2001 1.160004172 0.415974591 57.97764193 3.6129472 8.571841955

2002 0.909346565 0.454277724 42.68974498 4.6497299 8.468764712

2003 1.928985432 0.623059168 57.97764193 5.46817367 8.464808887

2004 3.221841886 0.573154706 56.14016853 3.85504574 10.36752994

2005 2.339752245 0.529249755 59.58415793 3.42595235 12.75513394

2006 6.300446181 0.616385692 56.14016853 3.06843906 13.29129614

2007 7.157616266 0.523310117 49.77586237 5.05256764 13.94259338

2008 3.884623518 0.790439022 47.69416096 5.45904126 18.55844908

2009 2.213216086 0.743212074 29.84970106 5.90302784 10.62920898

2010 2.080608591 0.838156181 37.98484876 3.93439969 13.0709945

2011 1.859884943 0.88846349 41.48334567 3.88951381 16.06830425

2012 2.175497767 0.893774805 39.70623681 4.69813331 13.77940114

2013 1.34946191 0.522615829 34.78195147 5.08948538 11.93041513

2014 0.909144872 0.788998443 29.07344267 5.59052228 10.27552428

2015 0.790999719 0.66101449 18.00896168 6.21135693 6.333754644

2016 0.752329813 0.484687693 16.04665436 7.27048445 6.515658233

2017 0.866281934 0.573709927 22.96992744 6.29839663 8.102472642

2018 0.809852818 0.267564658 30.48183253 3.36827251 8.642011355

2019 0.627107705 0.331863659 27.18096837 3.86721027 7.795553255

2020 0.678482793 0.299293543 23.35242084 5.09557239 6.7563406

 3) Foreign Direct Investment

Year / Botswana Djibouti Equatorial Lesotho South Africa

Country Guinea
Name

1995 70413220.76 3218527.917 126924170.7 24174395.03 1248424933

1996 71182310.71 3263542.294 376180404.2 29073745.83 816389273.8

1997 100109748.7 3100365.179 53451528.1 32675626.14 3810543923

1998 95318134.92 3173513.541 274768116.7 28350349.46 550338596

1999 36675065.78 3212901.12 154296173.9 22037254.59 1503332454

2000 57171341.35 3286049.482 154499589.9 32402238.3 968831356

2001 30680571.21 3392958.626 940741215.7 29695295.78 7270344986

2002 407997392.5 3432346.206 323391507.5 28390464.72 1479804589

2003 418040676.1 14224542.96 689779766 43946224.79 783136092.3

2004 390889251.9 38543559.85 340914468.5 55669445.72 701422007.6

2005 420854175 22203341.19 769146185.2 27441449.23 6522098178

2006 486913109.1 108287709.4 469506014.6 24322286.69 623291744.3

2007 494634226.2 195351140.3 1242731087 75616001.72 6586792253

2008 520917902.3 227654582.2 -793872332.8 11009973.06 9885001293

2009 208699298.2 96859684.56 1636219625 91349553.1 7624489974

2010 218379837.6 36501032.52 2734000000 9512771.55 3693271715

2011 293208060.5 79000230.7 1975000000 61173319.21 4139289123

2012 146084155.2 109998255.7 985256411.5 56653435.23 4626029122

2013 67136806.16 286004467.7 582948701.9 50429031.34 8232518816

2014 515184471 152998238.8 167875182.6 94459059.36 5791659020

2015 378554181.5 143832186.4 233325072.8 206514459.9 1521139945

2016 142522598.3 159997974.4 53998812.9 79234548.91 2215307020

2017 260575129.4 164929862 304827249.5 42168784.11 2058579911

2018 285955061.9 170000000 396077780 40874071.93 5569462350

2019 93607130.03 174999015.3 452287112.1 35732593.41 5116098443

2020 31807196.16 158230011 529894746.1 28001666.92 3153552569

4) Inflation

Year Botswana Djibouti Equatorial Lesotho South

Guinea Africa

1995 10.5 4.9 19.9 10 8.8

1996 10.1 3.5 4.5 9.1 7.3

1997 8.9 2.5 3 8.5 8.6

1998 6.5 2.2 7.9 7.8 7

1999 7.8 -0.4 0.4 8.6 5.1

2000 8.5 1.2 4.8 6.1 5.3

2001 6.6 1.9 8.7 6.9 5.7

2002 8 1.4 7.6 12.6 9

2003 9.2 1.6 7.3 7.2 5.9

2004 7 2.4 4.2 5 1.4

2005 8.6 3.3 5.6 3.4 3.4

2006 11.6 3.5 4.5 6.1 4.7

2007 7.1 5.9 2.8 8 7.1

2008 12.6 8.7 4.7 10.7 11

2009 8.1 5.6 5.7 7.4 7.2

2010 6.9 2.5 5.3 3.5 4.2

2011 8.5 5.2 4.8 5 5

2012 7.5 4.2 3.4 6.1 5.6

2013 5.9 1.1 3.2 4.9 5.7

2014 4.4 1.3 4.3 5.4 6.1

2015 3.1 -0.8 1.7 3.2 4.6

2016 2.8 2.7 1.4 6.6 6.3

2017 3.3 0.6 0.7 4.4 5.3

2018 3.2 0.1 1.3 4.8 4.6

2019 2.7 3.3 1.2 5.2 4.1

2020 1.9 1.8 4.8 5 3.3

2021 7.24 1.19 2.90 6.05 4.61

5) Exchange Rate

Year Botswana Djibouti Equatorial Lesotho South Africa

Guinea

1995 2.772207 177.721 499.1484 3.627085 3.627085

1996 3.324197 177.721 511.5524 4.299349 4.299349

1997 3.650763 177.721 583.6694 4.607962 4.607962

1998 4.22588 177.721 589.9518 5.528284 5.528284

1999 4.624395 177.721 615.4733 6.109484 6.109484

2000 5.102241667 177.721 710.207977 6.939828333 6.939828333

2001 5.841416667 177.721 732.3976933 8.609180833 8.609180833

2002 6.327791667 177.721 693.7132265 10.54074667 10.54074667

2003 4.949666667 177.721 579.8974262 7.564749167 7.564749167

2004 4.693833333 177.721 527.3380323 6.4596925 6.4596925

2005 5.11675 177.721 527.2583626 6.359328333 6.359328333

2006 5.8303 177.721 522.4256249 6.771549167 6.771549167

2007 6.139408333 177.721 478.6337185 7.045365 7.045365

2008 6.826858333 177.721 446.0000414 8.261223333 8.261223333

2009 7.155141667 177.721 470.2934233 8.473674158 8.473674158

2010 6.793625 177.721 494.7942622 7.321221961 7.321221961

2011 6.838233333 177.721 471.2486257 7.261132132 7.261132132

2012 7.640525 177.721 510.5563385 8.209968627 8.209968627

2013 8.398908333 177.721 493.8996239 9.655056069 9.655056069

2014 8.976083333 177.721 493.7573299 10.85265557 10.85265557

2015 10.12899167 177.721 591.211698 12.75893088 12.75893088

2016 10.90115833 177.721 592.6056151 14.70961089 14.70961089

2017 10.34741667 177.721 580.6567496 13.32380142 13.32380142

2018 10.199975 177.721 555.4464584 13.23392647 13.23392647

2019 10.75586667 177.721 585.9110132 14.44842705 14.44842705

2020 11.45624167 177.721 575.5860045 16.45910539 16.45910539

 CHAPTER 3

METHODOLOGY

3.0 INTRODUCTION

Chapter three designates theoretical framework, research design, data collection method, data
processing and data analysis used to achieve the objectives of the study. Research is usually
an approach to solving research problems effectively.

3.1 THEORETICAL FRAMEWORK

Increase in population (IP)

Othman Ali et al (2021),Sagal, (2021) and

(Babatunde and Olayinka, 2017

Natural resources (NR)

Ali et al, (2018), Majid et al, (2019) and

Akinyele et al, (2022)

Foreign Direct Investment (FDI)

Unemployment rate (UR)
Hamidah Muhd Irpan et al., (2016)
Keynes (1936), Solow (1980)
Elsa Widia et al., (2019) and
and Mouhammed (2012)
Alalawneh, (2020)

Inflation rate (IR)

Orji and Okafor (2015), Mustafa (2017)

Jelilov, Obasa, & Isik, (2016)

Exchange rate (EXR)

Atif (2019), Ku (2018) and

Usman and Elsalih (2018)

Independent variable Dependent variable

Figure 3.1: Theoretical Framework

3.2 RESEARCH DESIGN

This study uses research design as the fundamental directions. This study has used
quantitative research. Quantitative research involves the collection of numerical data to answer
specific research questions. In this case, the study is to examine the relationship between the
independent variables which are increase in population as a measured by total population
(Othman Ali et al (2021),Sagal, (2021) and Babatunde and Olayinka, 2017)), natural resources
(NR) as a measured by total natural resources rent (% of GDP) (Ali et al, (2018), Majid et al,
(2019) and Akinyele et al, (2022)), Foreign Direct Investment (FDI) as a measured by African
countries FDI (Hamidah Muhd Irpan et al., (2016) Elsa Widia et al., (2019) and Alalawneh,
(2020)), inflation rate (IR) as a measured by inflation, consumer prices (Orji and Okafor (2015),
Mustafa (2017), Jelilov, Obasa, & Isik, (2016)), and exchange rate (ER) as a measured by
official exchange rate (Atif (2019), Ku (2018) and Usman and Elsalih (2018)) and the
dependent variable which is unemployment rate in African countries.

Besides that, researchers collect data on predetermined instruments to yield statistical

data. It is a more structured data collection technique. Quantitative research not only provides
the summary of the information on the characteristic, but it is also useful in tracking the trend.
There are three type of research design: exploratory research, descriptive research, and causal
research (Pawar, 2020). According to Pawar (2020), exploratory research is used for research
where no design study is done before. Later investigation can be best understood to get
knowledge through this design. The study used for explanation whether future study is possible
or not and data can be used for further development for more research. Descriptive research
includes phenomena being researched and characteristics of population. To describe internal
validity does not require characteristics of population. It used for statistics of data, average and
frequencies (Pawar, 2020). Meanwhile, causal research relates with understanding of
phenomenon with statements "If A, then B". According to assumptions and norms one can
made certain changes in this kind of research. The explanation of test by hypothesis seeks by
majority of scientists like dependent variable, independent variable, variation in one
phenomenon, variation in other phenomenon (Pawar, 2020).
3.3 DATA COLLECTION METHOD

In order to study the effects the variables have on the unemployment rate in African
countries, this study had obtained data on the indicators of increase in population, natural
resources, foreign direct investment, inflation rate, and exchange rate from the World Bank
database. For the study, data researchers used the panel data from the year 1995 to 2020 which
consists of 26 observations. The unemployment rate as a subject in this research is a dependent
variable. we evaluated all independent variables to see how the independent variable influences
the dependent variable.

3.4 DATA PROCESSING

Data Processing (DP) is the process of extracting information from data by organizing,
indexing, and manipulating it. Here, information refers to valuable relationships and patterns
that can aid in the resolution of problems of interest. DP capability and efficiency have
historically improved with technological advancement. Researchers constantly check through,
update, and edit the data to ensure accurate data is selected for analysis. To begin, referencing
several previous studies confirms that the indicators researchers choose to use are fully
supported and proven by previous researchers. While collecting data, researchers recheck the
data and indicators used are consistent with previous research. Because some of the World
Bank data contained missing observations, researchers had to make some changes to ensure
that all observations were made in the same years. Even after the researchers enter the data into
E-views, the figures are double-checked to ensure that there are no mistakes. This complements
the accuracy of the data before performing the analysis.

3.5 DATA ANALYSIS

3.5.1 EVIEWS

EViews is a software package that includes data analysis, regression, and forecasting
tools. It is a regression package for econometric analysis that has been "canned."
EViews is object-oriented in design. Each object type has its own set of 'views' and
procedures that are used in EViews. Until recently, EViews was a command-line-only
programme, and all advanced features, such as Kronecker products, eigenvector
solution, and singular value decomposition, were still processed via the command line.
In addition, the command line mode logs all the steps in an analysis. The graphical user
interface (GUI) is convenient, but there is no record of your session, and you must return
to command line mode to use advanced features (Jianguo, 2015).

Our estimation support ranges from simple tools like single and multiple equation linear
and nonlinear least squares, ARMA, instrumental variables, and exponential smoothing
to more specialized estimators like Generalized Method of Moments, univariate and
multivariate GARCH, Markov switching, nonstationary regression, vector
autoregression and vector error correction, and state-space estimation. And that's just
to get started. EViews even has advanced tools for analyzing both stationary and
nonstationary panel data (Miami,2017).

EViews is a new version of a set of tools for manipulating time series data that were
originally developed for large computers in the Time Series Processor software.
MicroTSP, which was released in 1981, was EViews' immediate predecessor. Despite
the fact that EViews was developed by economists and the majority of its applications
are in economics, nothing in its design limits its usefulness to economic time series.
EViews can handle even large cross-sectional project. EViews provides simple visual
methods for entering data series from the keyboard or disk files, creating new series
from existing ones, displaying, and printing series, and performing statistical analysis
of series relationships (Dr. Paczkowski,2017).
3.5.2 MULTIPLE LINER REGRESSION

This study also used the Multiple Liner Regression model to observe the connection
between dependent variable (Y) and 5 others independent variable (Xi) (Anghelache,
2014). This will explain how independent variable affects dependent variable’s result.
The equation of the model is as follows:

Y = β0 + β1X1i + ……… βkXki + µi

Our main objective to use multiple liner regression rather than simple liner regression
is because this study uses 5 independent variables for prediction and analysis. The
equation for this study is as follows:

URt = β0 + β1IPt + β2NRt + β3FDIt + β4IRt + β5EXRt

UR refers to unemployment rate in the 5 African countries. Β is the coefficient used to

explain the measures it will influence UR. IP is population, NR refers to natural
resources, FDI is foreign direct investment, IR stands for inflation rate, and EXR is
exchange rate.

The main distinctive of the function of multiple liner regression is that the parameter
(βk) should be linear in the model. It also shouldn’t have any relationship with the
independent variables. The fractional regressions coefficient are β1 and β2 with two
predictor variables in the model of multiple linear regression. It proves the influence of
the independent variables (Xi) to the dependent variable (Y), considering other
variables constant.

3.5.3 F-TEST STATISTIC

(Kissell, 2017)stated that F-Test is in regression analysis, the F-test is used to test the
hypothesis that all model parameters are zero. It is also used in statistical analysis to
compare statistical models that have been fitted with the same underlying components
and data set to determine the best fit. It is easier to evaluate data and analyse a P-value
when the F-test statistics are determined. This study indicates that there is any
relationship between independent variables towards dependent variables (Barokah,
2020). Thus, based on the p-value, it must not more than 2.5 or the researchers will
decide to reject the null hypothesis and it means that there is no significant of
independent variables on the dependent variable (Siegel, 2022).

3.5.4 T-TEST STATISTIC

An example of a statistical test used to compare the means of two groups is the t test. It
is one of the statistical hypothesis tests that is most frequently employed in research on
pain. A statistical inference can be made using either parametric or nonparametric
approaches. A statistical approach known as "parametric techniques" involves defining
the probability distribution of probability variables and drawing conclusions about the
distribution's parameters. Nonparametric approaches are used when it is impossible to
specify the probability distribution. When the samples meet the requirements of
normality, equal variance, and independence, the para-metric approach known as the T
test may be applied. T tests fall into two categories. When comparing two groups that
are independent of one another, one may use the independent t test, and when
comparing groups that are dependent on one another, one can use the paired t test.
Because they examine the distinction between two paired outcomes, paired t tests can
be characterised as a sort of t test for a single sample. The findings would change very
little if there was no difference between the two treatments, and the difference in sample
means for a paired t test would be close to zero (Tae Kyun Kim, 2015).

3.5.5 DIAGNOSTIC CHECKING

The model mentioned above may experience the econometric difficulty. Therefore, the
researcher needs to run a number of hypothesis tests to determine and confirm if the
model is free from multicollinearity, autocorrelation, and heteroscedasticity concerns.
Additionally, the researcher will need to do a normality check and test for model
specifications.

3.5.5.1 MODEL SPECIFICATION AND NORMALITY TEST

This study stated that in place of empirical or methodological factors, the model
specification should be based on theoretical ones (Lewis-Beck, 2019).
However, it is crucial that the model estimates utilised in this work depend on
the accurate model specifications. Thus, it can encounter issues if it misspecify
a model. When one or more relevant variables are omitted from the model, the
model specification can be erroneous (Marriott, 2022). On the other hand, the
normality assumption suggests that the data were taken from a population that
had a regularly distributed population. Some characteristics of a normal
distribution exist (Khatun, N. 2021). The normality assumption suggests that
each independent sample should have a normal distribution when comparing
two samples, for instance (Fatih Orcan, 2020). Departure from the normality for
any of the independent sample indicates that the parametric tests should not be
used (Rietveld & van Hout, 2015) since the type I error rate is affected (Blanca,
Alarcon, Arnua, et al., 2017; Cain, Zhang, & Yuan, 2017). That is, parametric
tests are resilient in terms of type I error rate (Demir et al.,2016), and type I
error rate increases as the distribution of the groups diverges from one another
(Blanca et al., 2017). The test for normalcy should be conducted separately for
each independent sample. It is incorrect to check the dependent variable's
normality throughout the entire sample without taking the grouping variable
(the independent variable) into account. The normalcy of exam scores for male
and female students should be checked individually, for instance, if a researcher
intends to compare exam scores between male and female students. The
normality assumption is broken if one of the groups has a regularly distributed
population while the other does not (Fatih Orcan, 2020).

3.5.5.2 MULTICOLLINEARITY

According to Noora (2020), when a number of variables are included in a

multiple linear regression analysis that are substantially associated with both the
dependent variable and one another, this is known as multicollinearity. Some of
the important variables under research become statistically insignificant due to
multicollinearity. Besides that, the correlation between predictors is the one
signs of multicollinearity in the research or analysis. For intense, if the
correlation is not measured, the following signs of multicollinearity exist when
the predictor coefficients differ between models (Jamal,2017). In addition, there
cannot be any linear dependence between the predictors, such as when one
variable is a multiple of another variable, in order to prevent multicollinearity.
Since the variables in this situation are totally correlated, using both variables
in a regression model would not add any value because one variable fully
explains the other variable. There is no multicollinearity if there is no
connection between the independent variables.

3.5.5.3 AUTOCORRELATION

ANOVA and linear regression are two examples of parametric statistical

techniques that make the assumption that the errors of the models employed in
the analysis are independent of one another. In the context of time-series
research designs, the mistakes are considered to be autocorrelated or dependent
when this presumption is not met. The assumption of independent errors
inherent in many parametric statistical studies may not be met in time-series
designs since they require collecting data from a single participant at several
periods in time rather than from multiple participants at one moment in time. If
no corrective steps are done when this happens, the results of these studies and
the inferences that might be formed from them are likely to be inaccurate.
Mathematics can express how similar a time series is to a delayed version of
itself over successive time periods by using autocorrelation. The term
"autocorrelation" refers to a correlation between two time series comparisons
where the same time series occurs twice, once in its original form and once more
with one or more-time intervals between them. Additionally, it analyses how a
variable's present and past values relate to one another (Tim,2021).

3.5.5.4 HETEROSCEDASTICITY

Heteroscedasticity is a situation where the variance of the dependant variable

changes across the data based on The Institute for Statistic Education in the year
of 2014. Heteroscedasticity make analysis to be more difficult due to a lot of
methods in regression analysis are based on a hypothesis variance.

A collection of odd variables is considered heteroscedastic if there exist sub-

populations that hold contrasting variabilities from one another.
 It is a huge concern if heteroscedasticity become a possible existence in any of
the application of regression analysis which includes variance analysis, as the
existence of heteroscedasticity can make void on the statistical tests of
significance that expect that the modelling mistakes are unrelated and normally
distributed and their variances do not differ with the being modelled effects.

Heteroscedasticity usually tend to happen with cross sectional data and this
element may occur under different cases such as in the case of measurement,
data gathering on the income and food expenditures of massive number of
families, and if there are subpopulation differences and any other interactions
repercussions.

Heteroscedasticity could lead on a few consequences. One of them includes

OLS estimators remain unbiased and logical as no explanatory variables is
related with the error term. Not only that, the presence of heteroscedasticity also
causes the standard error to be biased which influence bias in test statistics and
confidence intervals.

To detect heteroscedasticity, there are two ways can be performed which the
first one is done in the informal way called graphical method and the second
one is done through formal tests such as The Glesjer LM Test, The Park LM
Test, and White’s Test.

However, heteroscedasticity can be reduced through three different cases which

consists of Generalized Least Squares, Heteroscedasticity-Consistent
Estimation Methods and the last one is Weighted Least Squares (Kiula, 2014).

3.6 CONCLUSION

As a conclusion, Ordinary Least Square (OLS) estimator will be applied in Chapter 4 on the
data analysis after diagnosing both methodology and data the researcher decided to bring into
play.