Abstract
It is indeed a proven fact that no country, regardless of her abundance of resources can flourish on her own and this has transformed the world into a global village where international trade among countries takes place. It is based on this backdrop that this study was carried out bearing in mind that the objective to be achieved is investigate the impact of foreign direct investment on international market. Foreign Direct Investment, capital formation, exchange rate and the inflation rate formed the exogenous variables for the study. The years of observation for the study is 40 years that span from 1989 to 2019. The relevant theory that underpins the study is David Ricardo’s theory of comparative cost advantage.
The data was sourced from World Development Indicator data bank .The study began the analysis by testing for stationarity and it was found that the variables are stationary at level or after first difference. This mixed result of stationarity led to the adoption of Autoregressive Distributed Lag (ARDL) Model as the estimation techniques for the study. The result of the ARDL Bound test for longrun relationship revealed that the form of relationship is inconclusive.
The findings form the long run estimate revealed that of the four exogenous variables that are used in the model, only FDI and exchange rate are statistically significant and has the expected signs. The diagnostics checks further revealed that the model is fit for any prediction as it is devoid of all the econometrics problems autocorrelation, heterosckedasticty, and model misspecification amongst others. The study observed some limitations in the study and recommendations are made as regards the improvement of these recommendations in future studies. The policy implications are also made for policy and stakeholders in Malaysia.
CHAPTER ONE: INTRODUCTION
1.1 Brief Introduction:
It is indeed a proven fact that no country, regardless of her abundance of resources can flourish on her own and this has transformed the world into a global village where international trade among countries takes place. This is may be cited as the reasons why Malaysia is promoting both regional and bilateral trade ties with other countries through individual and regional groups arrangements (Huan & Skrine, 2020).
This transformation has offered lots of benefits. Among the benefits is that international trade enhance and improve the inflow of foreign direct investment into countries, enhance capital formation and improve exchange rates since it attracts foreign currency. This particular phenomenon has generated a lot of debates among scholars.
This research study will is being carried out bearing in mind that it is to achieve the sole aim of examining the “impacts of Foreign Direct Investment (FDI) on International trade” in Asia by conducting an empirical study on Malaysia economy. The chapter layout for the study will be in five folds with each chapter having its subchapters. Chapter one of the study will be about the introduction of the work while chapter two will be about the Literature review. The third Chapter shall present the methodology while chapter four will present the data analysis and discussion of the study. The study’s findings, summary of the whole work, as well as Recommendation is will be discussed in chapter five.
1.2 Background of the study
It is widely believed that the more a country embraces international trade, the more chances of attracting foreign investors from different countries to bring their funds for investment in that country. This is expected to improve the host country’s exchange rate and capital formation amongst other things. However, inflation is seen as a factor that may impede the benefits that foreign direct investment is supposed to offer any country that is opened to international trade. International trade, if properly maximized has been seen as a veritable tool that can be used to increase the interdependence of markets across nations on exchange of products, technological knowhow and human capital across the borders and in most cases the across the regions. It is in this light that this study seeks to incorporate exchange rate, capital formation and inflation rate as the control variables along with foreign direct investment.
International trade has offered a lot of noticeable increase in flow to countries both in trade and investment in the 21st century. It is on this premise that (Román, Calvo, & Rute, 2012), noticed that geographical reorganization of production has been responsible for favourable economic climate among firms.
In his own view, (Simionescu, 2014), submitted that international trade as well as FDI are two important economic that have different impacts on the economies of each country. FDI is observed by (Ibrahim, 2002), as an important part of an open international economic system. He however noticed that majority of the advantages of FDI are yet to be enjoyed by most of the countries that embraced it.
In the past few years, series of debates, opinions and submissions have emanated among scholars from as regards the benefits that awaits countries that embrace international trade. while contributing to the effect of trade openness on economic growth, (Ay, Kurşunel, & Baoua, 2017) submitted that a lot of opinions emerged among the scholars. While majority of the scholars are in support of the preponderant role played trade openness towards the improvement of property many are indifferent or yet to find any positive effect of trade openness on economic growth.
According to (Xiao & Paul, 2000), as cited by (Koojaroenprasit, 2012), Foreign Direct Investment through several channels contributes to significantly the growth of the host nation. The duo posits that aside the fact that FDI serves as a means of capital formation to host country, it also help in increasing total investment which in turn contributes directly to growth. The macroeconomic areas where foreign direct investment is said to enhance is export, employment, savings and consumption.
The relationship between international trade and inflation has been referred to by Temple (2002) as cited by (Babatunde, 2017), as one of the unsolved puzzles of the modern international macroeconomics. Elsewhere (Bostan, Toderas, & Firtescu, 2018) posits that exchange rate is vital in international trade. Despite the increasing number of studies by reputable scholars as regards the topic, (Nicita, 2013) opined that the main effect of exchange rate on international trade still needed to be explored for a better policy formation in the area of international trade among countries.
1.3 Problem statement
International trade attracts foreign direct investment which offers lots of benefits and advantages that are numerous to mention or exhaust among countries that are fully ready to tap into this advantages. These advantages ranges from transformation in technology knowhow to employment opportunities and lots more.
However, lots of concern has been raised by dependency theorists as regards the dependence of countries on foreign investment. Their concern according to (Adhikary, 2015), is that dependency on FDI has every possibility of affecting the economic growth and income distribution of the host country. This may be rationally acceptable given the fact that such country’s economy may be controlled by foreign investors who have little or no interest in the growth and development of the country. This set the premise for this study to examine the impacts of foreign direct investment on international trade using data on Malaysia economy.
Over the years not too much of literature have been published by renowned scholars in the field of international trade as regards the impacts of foreign direct investment (Incekara, 2012), despite huge contribution that international trade has been projected to have on host country’s economic growth, the review of related and existing literatures revealed that there are lots of gaps to be filled in order to augment the success that has been achieved about these phenomena. Hence, another problem that this study is set to address.
A meticulous literature review about past studies (Ay, Kurşunel, & Baoua, 2017), (Babatunde, 2017) (Ibraheem, 2002) (Koojaroenprasit, 2012) (Simionescu, 2014), have revealed that series of independent studies have been carried out on most of the factors that are assumed to be influencing international trade, but up till date, there is no existing literature or publications that have done an empirical study of this type to investigate the impact of FDI on international trade. Achieving this aim of the study will set out the uniqueness of this study as one of its kind and also set a basis for other researchers to further their study in this area.
It is in this regards that this study amongst other things will be aiming to fill the identified gaps in order to be a source of reference for scholars and students and to contribute to existing literature.
1.4 Research Questions:
In this study, the main question of research is “What is the Impact of foreign direct investment on international trade?” other specific questions that are to be answered in this study are:
 What impact does foreign direct investment has on international trade?
 What impact does capital formation has on international trade?
 What impact does exchange rate has on international trade?
 What impact does inflation rate has on international trade?
1.5 Research Objectives:
The broad objective of this study is to analyze the Impact of foreign direct investment on international trade. There are four other specific objectives to be achieved in this study. They are:
 to examine the impact of FDI on international trade.
 to examine the impact of capital formation on international trade
 to examine the impact of exchange rate on international trade
 to examine the impact of inflation on international trade
1.6 Scope of study
The study will focus on “examining the impact of foreign direct investment (FDI) on international trade” obtaining data on Malaysia economy. Furthermore, the study seeks to examine the impacts on Malaysia as one of the countries in Asia by employing secondary data on the identified variables from World Bank Data bank. The year of observation spans from 19802019 as this is the only time frame for which the relevant data’s accessibility is feasible. Eviews which is one of the most widely used statistical analytical software for time series data will be used to analyse the data.
1.7 Significance of research
Suffice to say that the significance of this study can never be overemphasized, particularly at this point in time when economies across the globe are dependent on each other and at this same time where policy makers are working tirelessly towards removing every barriers that are seen impediment to international trade. The significance of studies like this is not exhaustible but this study will unfold a threefold significance amongst others.
 Being the first of its kind as regards international trade and foreign direct investment, this study will serve as a guide for students of research as well as scholars who may be very much interested in carrying out subsequent research in a bid to contribute to the existing literature about international trade and FDI.
ii The study will be an eyeopener to all economies as well and economists that are yet to tap into the full advantage that international trade can offer by examining the variables such as that needed to be meticulously guided
iii This study is also expected to serves as a policy guide for policy makers as regards the stabilization of exchange rate and combating inflation rate.
1.8 Summary
As set out in the introductory part of this chapter, this study take us through the background to the study in order to show the reason why a study of this type is important to embark upon. In addition to this, the statement of problems with the study’s objectives and research questions are articulately treated in this chapter. This last part of the chapter also made justice to the significance of the study as well as the scope of the study.
2 CHAPTER TWO: LITERATURE REVIEW
2.0 Introduction
In order to support and justify some issues discussed in chapter one as regards the background of the study, problem statement, research questions and research objectives, this chapter is set to review series of related literature as regards the variables that are considered to directly or indirectly impact international trade in Malaysia. Thus, this chapter will thoroughly review all past studies that have been carried out regarding international trade and foreign direct investment with the inclusion of control variables such as capital formation, inflation and exchange rate.
2.1 Dependent Variable
Generally, a dependent variable is a kind of variable whose outcome solely relies on the form of manipulations and the changes that happened to other variables known as either explanatory variable or independent variable. In this study, International trade in Malaysia is considered the most relevant as it is the main variable that this study intends to examine. Hence, section 2.1.1 will comprehensively review the concepts, findings and submissions of researchers regarding international trade.
2.1.1 Dependent Variable: International trade
International trade is a topic that scholars have done a lot to make sense of. There is a varying degree of views as regards its conceptualization. However, the duo of (Ikechi & Anthony, 2020) described it as the sale or purchase of goods and services across national borders. These goods and services can be consumer or capital goods, securities and any form of natural resources. They further opined that the crossborder transactions could be facilitated by barter or exchange of national currencies with the latter being preferred to the former”.
Other scholars have contributed to the conceptualization of international trade. However, their studies can be concluded that international trade, unlike any other trade, is the exchange (through sale or purchase) of goods and services between two or more countries.
Many scholars believed the source of rapid economic growth and development experienced by some Asian countries in the last five decades could be traced to their participation in the international economy (Soi, Koskei, Buigut, & Kibet, 2013). This may not be farfetched because the majority of these countries are exporters of high technology.
The concept of international trade cannot be discussed in isolation. This has led to the submission of (Simionescu, 2014) that the two most important variables that impact globalization are trade and foreign direct investment. He, however, clarified that this impact varies among the countries.
In recent years, many publications and studies have been made regarding international trade; however, literature is scant as regards an empirical study that examines the impact of foreign direct investment on international trade.
The study of (Simionescu, 2014) directly examined the relationship that exists between trade and foreign direct investment. This study was specifically carried out on the G7 countries, which comprises countries such as Japan, Germany, France, Canada, Italy, the United Kingdom and the United States. The study employed a panel data approach and carried out a Granger causality test on panel data from 2002 to 2013. The study’s findings revealed that there is causality between foreign direct investment and international trade in the short run.
In addition to this, the study also found that there is a unidirectional causal relationship between Foreign Direct Investment and international trade among the countries of interest in the long run. To the best of the researcher’s knowledge, there is no specific study conducted in Malaysia regarding the impact of foreign direct investment and international trade.
2.2 Independent Variables
In order to empirically achieve the objectives established in the previous chapter, three crucial variables will be incorporated alongside foreign direct investment as the independent variables for this study. According to the literature review, the variables selected are Foreign Direct Investment, Exchange rate, Capital formation, and inflation rate. The justification for the inclusion of these variables is that they directly or indirectly impact international trade.
2.2.1 Foreign Direct Investment
In order to empirically achieve the objectives established in the previous chapter, three crucial variables will be incorporated alongside foreign direct investment as the independent variables for this study. According to the literature review, the variables selected are Foreign Direct Investment, Exchange rate, Capital formation, and inflation rate. The justification for the inclusion of these variables is that they directly or indirectly impact international trade.
2.2.2 Capital formation
The importance of capital formation can never be overstressed in international trade, which has generated debates and research from reputable scholars. However, literature is a dearth in this area. Even though literature is scant about the relationship or impact of capital formation on international trade, few scholars have made lots of effort to contribute to the literature in this area.
To investigate the linkage between capital formation, foreign direct investment, human capital and economic growth, (Adhikary, 2015) conducted an empirical study by employing the vector error correction (VEC) model. The study revealed a longrun relationship between the dependent and independent variables of interest in the study. Specifically, it was found that capital formation hurts economic growth within the year of observation.
Recently, (Ay, Kurşunel, & Baoua, 2018) conducted an empirical study regarding trade openness within some African countries. They employed a panel cointegration and causality approach on the time series data that consist of 25 years of observation from 1990 to 2014. The results obtained from the study’s analysis revealed a bidirectional causality link between capital formation and trade openness among the 38 African countries under investigation. This further led to their suggestion that African countries should massively improve their investment promotions in order to increase their capital formation and international trade openness.
2.2.3 Exchange rate
According to (Barbosa & Jayme, 2018), the exchange rate is the ratio of an overseas product price level over the local product price level, which is multiplied via the nominal exchange rate. Theoretically, it is expected that a lower domestic price in the host country will lead to healthy exports and also facilitate a trade surplus. Scholars and researchers from different fields have submitted that the exchange rate plays a significant role in the host country’s international trade, the balance of payment and overall economic performance.
However, empirical study regarding exchange rate and international trade has either has been very much neglected by the previous empirical analysis or needed more findings to substantiate the fewer findings of the scholars. Hence, previous scholars’ empirical studies shall be reviewed to establish the kind of relationship among the variables of interest.
In the bid to further contribute to the existing literature as regards exchange rate and international trade, (Nicita, 2013) carried out a study that aimed at investigating the importance of exchange rates on international trade. A panel data were collected on 100 countries from the years (20002009), and it was analyzed through the help of a fixed effects model. The study suggested that attention should be paid to exchange rates in order to maximize the full advantage of international trade. The study also recommend that certain strategies that are multilateral cooperation and are related to the stabilization of exchange rates towards their equilibrium level should be put in place in order to avoid the resurgence of protectionist measures.
The duo of (Kang & Dagli, 2018) analyzed the link between international trade and exchange rate on bilateral data spans from 2015 among seventytwo economies of the world. The study found that there exists a positive relationship between exchange rate and international trade.
2.2.4 Inflation rate
Inflation is a unique concept that is defined according to (Fernando, 2020) as the reduction over time in the purchasing power of a given currency. It is a dynamic phenomenon; hence, finding a single suitable empirical model that fits the conditions of all countries becomes difficult, which has made this concept an empirical question in economic literature. Despite this, the literature on the effect of trade openness on inflation seems negligible.
However, there are some available works of scholars that have empirically contributed to the effort of unravelling the questions that center around inflation. Among the studies that have been carried out to investigate the impact of inflation on international trade is that of (Manual & San, 2019), who found that inflation rate and the other macroeconomic variables included in the study significantly affect international trade in Malaysia.
In his study conducted in Nigeria to investigate the relationship between trade openness and inflation, (Babatunde, 2017) employed the nonlinear autoregressive distributed lag (NARDL) model approach to analyze the data spans from 1980 to 2015. The findings from the study revealed that a timespecific relationship existed between inflation and trade openness. In addition to this, it was found that there is a significant positive longrun relationship between inflation and trade openness. Although, the opposite was found in the short run. Conclusively, the study revealed that trade openness affects inflation in a nonlinear and asymmetric way.
2.3 Theoretical Review
Numerous theories have been propounded in the area in the area of international trade. Among these theories are the theories of absolute advantage propounded by Adam Smith, the theory of comparative cost advantage propounded by David Ricardo, and the HeckscherOhlin theory propounded by Eli Heckscher and Bertil Ohlin. However, this study adopted the theory of comparative cost advantage as it explains some underlining features peculiar to Malaysia as an exporting country.
2.3.1 Theory of Comparative cost advantage
David Ricardo propounded this theory in the 19th century, and it is still instrumental when determining how a nation should utilize its resources. It also helps channel more effort in the production of the actual goods that bring income through export. The theory of comparative advantage premise on the fact that everyone is better off when countries concentrate on the production of goods where they enjoyed comparative cost advantage (Uthman, 2016).
According to (Mojeed, 2011), the theory postulated that a country should concentrate on exporting the goods that can be produced at a relatively cheaper cost and import only those goods that can be produced at a relatively higher cost. It is essential to state that comparative cost advantage is a phenomenon of opportunity cost rather than absolute cost. Thus, a country is said to have a comparative cost advantage over other countries when it has the lowest opportunity cost.
The other advantage of this international trade theory is that it facilitates increased world production, enhances effective utilization of resources, and increases consumption of commodities. In addition to this, this theory offers a reduction in the prices of goods and services due to mass production (Balogun, 2017).
It is worthy of mentioning that this theory, as good as it is, has some limitations. One of these limitations is that there are more than two commodities in the world as propounded by this theory, which implies that three or more commodities render the theory impracticable. Another shortcoming of the theory is that it did not consider the possibility of more than two countries trading simultaneously, which in real life is practicable. Thus, this principle is considered to be more of an abstract which is hard to practicalize.
2.4 Theoretical framework
Figure 2.1 Theoretical Framework proposed in this study
Source: (Mukhtarov, Alalawneh, Ibadov, & Huseynli, 2019; Adhikary, 2015; Kang & Dagli, 2018; Babatunde, 2017)
Figure 2.1 above presents the adopted theoretical linkage that is to be examined in this research. It is from this framework that the hypotheses to be tested is been drawn.
2.5 Hypothesis Development
To have a robust findings and conclusion in this study, the following hypothesis are to be tested:
H_{0}: foreign direct investment has no significant impact on international trade in Malaysia.
H_{1}: foreign direct investment has a significant impact on international trade in Malaysia.
H_{0}: capital formation has no significant impact on international trade in Malaysia.
H_{2}: capital formation has a significant impact on international trade in Malaysia.
H_{0}: exchange rate has no significant impact on international trade in Malaysia.
H_{3}: exchange rate has a significant impact on international trade in Malaysia.
H_{0}: inflation rate has no significant impact on international trade in Malaysia.
H_{4}: inflation rate has a significant impact on international trade in Malaysia.
2.6 Study Gap
A comprehensive review of existing literature regarding this study revealed that the literature is awash with studies on the impact of the independent variables (foreign direct investment, capital formation, and exchange rate and inflation rate) on international trade. It can be concluded that the few scholars that specifically studied this impact did so on other countries that are not Malaysia see (Simionescu, 2014; Xiao, 2009). In addition to this, there is evidence that those who did an empirical study regarding this topic did so by employing panel data. Conclusively, there is no existing literature that has specifically carried out any empirical study that is in anyways related to this in Malaysia. On this premise of the inadequate amount of empirical studies about Malaysia, this present study offers to fill the identified gap.
2.7 Concluding Remarks
As stated in the introductory part of this chapter, the study has extensively reviewed the existing literature from relevant journals regarding the conceptualization of the variables. In addition to this, the study also revealed the findings and methods of some empirical studies. The theory of comparative advantage was adopted as the economic theory that underpins this study, and the chapter also presents the theoretical framework that presents the linkage between the variables. The development of the research hypotheses was formulated in this study in this chapter, and the study gap was also identified.
3 CHAPER THREE: RESEARCH METHODOLOGY
3.1 INTRODUCTION
This chapter of the study is aimed at offering a comprehensive explanation of the compulsory methodologies, a priori expectation as well as the necessary estimation techniques that will be adopted in the data analysis and discussion in the chapter that comes next. The chapter starts with the data source and measurement of variables. In addition to this, the chapter explains the methodology of the study which involves model specification, pretest (unit root test), estimation techniques such linear regression method, Autoregressive Distributed Lag (ARDL) Approach and Bound Cointegration. The last part of the chapter deals with the various diagnostics test that should be conducted when working with a time series data. Hence, the relevance of the diagnostic tests such as Serial correlation LM test, Heteroskedasticity test, Ramsey test and Cusum Stability test to our model shall be revealed.
3.2 Data Source and Measurement of Variables
3.2.1 Data Source
In pursuant to the study’s main objective of investigating the impact of Foreign Direct Investment on International Trade, five different variables have been considered relevant for the analysis. The Dependent variable is international trade while the independent or explanatory variable are Foreign Direct Investment, Capital Formation, Exchange Rate and Inflation rate. They all represents secondary data from Malaysia and it is an annual time series data that covered the period of 40 years (1980 to 2019) on all the variables and they are obtained from World Development Indicators.
3.2.2 Measurement of Variables
The variables are carefully measured in an articulated manner that is devoid of misconception or ambiguity.
The International trade that is the dependent variable in this study is measured by the summation of imports and exports of goods and services as a share of Malaysia’s GDP. Furthermore, Foreign Direct investment in this study is measured by the net inflows of foreign direct investment divided Malaysia’s GDP while exchange rate is measured in terms of official exchange rate in Malaysia. In addition to this, capital formation is measured by the yearly growth rate of gross capital formation estimated with constant local currency of Malaysia and inflation rate is measured by the yearly growth rate of Malaysia’s GDP implicit deflator.
3.2.3 Estimation Techniques
The estimation techniques that can be adopted for achieving a study of this nature shall be highlighted and explained. This estimation techniques are into three phases. They are the preliminary tests the main tests and the diagnostic tests. The preliminary tests are the unit root test which in most cases are Augmented Dickey Fuller (and the Philip Perron Unit Root test. The main tests are the ordinary least squares regression method, the Cointegration techniques, the Vector Error correction mechanism. The diagnostics tests are used to ensure that the result of the study is devoid of any econometric problem such as the heteroscedastic, autocorrelation, multicollinearity
and the model misspecification.
3.2.4 Unit root test
In a timeseries analysis, there is every likelihood that the series has unit root, a problem that is capable of triggering unpredictable results in the analysis. To ensure that this problem is avoided, a test for stationarity is always conducted. There are different techniques of testing for unit root test, however, The Augmented Dicker Fuller for stationary is a unit root test is one of the mostly used method.
In doing this, is always recommended to determine the optimal lag length that is good for the model. The optimal lag length was chosen on the basis of Akaike Information Criterion (AIC) and Swartz Criterion In most social sciences research, the null hypothesis that unit root is present is usually dismissed at 90 and 95 percent confidence level when the pvalue of the result is less than 0.1 and 0.05 respectively.
The hypothesis statement as follows:
H_{0}: β_{t} is nonstationary (β_{t} has unit root)
H_{1}: β_{t} is stationary (β_{t} has no unit root)
(Harris, 1992) opined that when comparing both data, the DickeyFuller test presents the critical value and tstatistics both in absolute term. If the absolute tstatistics are greater than the absolute critical value, H_{0} which is the null hypothesis should be rejected. However, we do not reject H_{0} if the tstatistics in absolute term is lesser than the absolute critical value.
The rejection of H_{0} implies that there is no existence of unit root in the time series data will be verified.
He further posits that in a case where the H_{0} is not rejected, thee the unit root problem exist and it can be overcome by differencing the series at first or second order.
3.2.5 Ordinary Least Square Approach
In the word (Kapur, 2018), Ordinary Least Square regression method basically used to estimate the relationship that linearly exists between the dependent and independent variables. In a case where the independent variables in the model are more than two, Multiple regression method is the best method to use.
The linear multiple regression equation can be expressed as:
Where:
The dependent variable to be predicted.
X_{1} through X_{p} = the independent variables values.
b_{0} = the value of the intercept when all other variables are held constant i.e they are equal to zero.
b_{1} through b_{p} = the estimated coefficients of the explanatory variables.
Econometrically, the model can be formulated as:
INT_{t}= β_{0}+β_{1}FD_{t}+ β_{2}CFN_{t}+ β_{3}EXC_{t} + β_{4}INF_{t} + Ɛ_{ȶ}
Where:
β_{0} represnts the Intercept of the model
β_{1} – β_{4}, represents the parameter to be estimated
t represents the time dimension
Ɛ_{ȶ } represents Stochastic or Disturbance term
3.2.6 Autoregressive Distributed Lag (ARDL) Approach
In a time series data, the result of the unit root test may imply that the study examine the long run relationship that exist between the variables of interest. This becomes necessary whenever the unit root test revealed that all the variables used are nonstationary at level and are in the same order of integration (Sahu & Pandey, 2020). According to (Nasir, Hassan, Nasir, & Harun, 2013), times series data is cointergated if there is presence of unit root at level but stationary at first difference. Hence, to test if there is any likelihood of longrun relationship between the variables, the
3.2.6.1 Bound cointegration test
Cointegration test is an extension of the Autoregressive Distributed Lag approach. This is conducted when working with a time series analysis in order to estimate variables (Y_{t} and X_{t}) that are of the same order of integration I(0) or I(1). This is a test that is usually carried out after the stationariy or nonstationarity of the time series data have been established through the unit root test. In addition to this, cointegration test is employed in order to establish the long run relationship that may exist among the variables.
3.2.7 The error correction mechanism
To examine the short run relationship among series that are integrated of the same order an error correction mechanism (ECM) is employed. In addition to this, the error correction mechanism is basically used to determine the speed of adjustment of the model. T speed of adjustment implies the time at which it will take an economy in time of disequilibrium to attain its equilibrium position. The ECM coefficient must be negative and must also be statistically significant.
3.3 Diagnostic Checking
A time series data is said to be prone to some econometric problems like the autocorrelation, heteroscedasticity, multicollinearity and others that shall be revealed in chapter four. Bearing this in mind, it is important to test if any of these problems’ id present the model for a study.
3.3.1 Multicollinearity
In a case where it is found that there is a very high correlation between two explanatory variables in a model, econometricians conclude that there is existence of multicollinearity in the model. (Farrar & Glauber, 1967), in the study submitted that a model suffers from the existence of multicollinearity if the ttest is insignificant for all parameters, the Ftest is significant but with a very high Rsquared. There are different approaches of detecting multicollinearity in a regression model, the model and their coefficients remain the same.
Another logical way of detecting the problem of multicollinearity in a model is by examining the significance of the effects/impacts of the explanatory variables on the dependent variables. In case there is problem of multicollinearity in the model, the individual effect of the individual variables will be insignificant but the R^{2} will be very high. Thus, the model could have a problem of multicollinearity. Lastly, one of the most widely used methods of detecting multicollinearity in a time series data is the Variance Inflation Factor (VIF).
Traditionally, the practice is to regress an explanatory variable on all other explanatory variables in order to obtain the coefficient of determination R^{2}. This process is to be done on every explanatory variable in order to have all the R^{2}. When this is done, the next line of action is to calculate the Variance Inflation Factor (VIF). Which can be calculated as VIF= I/(1R^{2}). The assumption according to (Kelava, Moosbrugger, Dimitruk, & SchermellehEngel, 2008), is that the higher the R^{2}, the higher VIF value, which can the lead to the conclusion that there is presence of severe multicollinearity between the explanatory variables in the model.
However, the value for the VIF can be conveniently and accurately obtained with the help of Eviews. Hence, the traditional method of obtaining the VIF will be replaced with the adoption of the Eviews value in the analysis chapter where the test will be carried out.
The table below presents the degree of multicollinearity between variables according to the outcome of the VIF. A zero R^{2} will lead to a VIF that is equal to 1 and the explanation for this is that there is absence of multicollinearity between two or more explanatory variables. However, a higher R^{2} to the lever of 0.9 or greater will result in a VIF that is greater than 10 and this implied that there is existence of severe multicollinearity between two or more explanatory variables in the model.
Table 3.1 VIF Degree of collinearity
VIF Range Degree of correlation
VIF = 1 no collinearity
1 ˂ VIF ˂ 5 Fairl collinearity
VIF ˃ 5 to 10 High collinearity
VIF ˃ 10 Severe collinearity
3.3.2 Heteroscedasticity
In a times series analysis, there is always a great chance of the existence of heteroscedasticity in the model. The main assumption of heteroscedasticity is that the variances of the error term are not constant (Schink & Chiu, 1966). There are numerous econometric ways of detecting the problem of heteroscedasticity although, all methods are expected to reveal the same result, particularly with the help of Eviews ceteris paribus However the ARCH test stands out as one of the most generally used approach (Wang, Gelder, Vrijling, & Ma, 2005).
The hypothesis for heteroscedastic is:
H_{0}: there is presence of homoscedasticity in the model
H_{1}: there is presence of heteroscedasticity in the model
Decision Rule: if the pvalue is lesser than 0.05 then reject the null hypothesis. (ARCH) test.
3.3.3 Normality Test
The Jarque Bera normality test which is the most highly adopted normality test is carried out to examine if the variables in the series are normally distributed of not. This test considers the kurtosis and the skewness of the distribution in order to reveal if the distribution of the data is normal or not. This test shall also be carried out in the next chapter in order to confirm the normality of the test.
The hypothesis for autocorrelation is:
H_{0}: the data is normally distributed
H_{1}: the data is not normally distributed
Decision Rule: if the JaqueBerra pvalue is lesser than 0.05 then reject the null hypothesis
3.3.4 Ramsey reset test
The Ramsey Reset test is a frequent test that is always conducted by researchers when working with a times series data and they are using regression analysis approach. This test is carried out in order to examine if the model used in the study is correctly or wrongly specified. Specifically, the test according to Ramsey in 1996, is conducted to examine the issue of omitted variable and functional form of the model.
The hypothesis is developed thus:
H_{0}: the model is correctly specified
H_{1}: the model is wrongly specified
Decision Rule: Reject null hypothesis if the pvalue is lesser than 0.05
3.3.5 CUSUM Stability test
In line with the normality test, it is desirable to know that the series in the model is stable. To examine this, researchers do conduct a Cusum stability test which the duo of (Harish & T.Mallikarjunappa, 2015) refered to as one of the most widely test performed by researchers when examining the stability of a series. They posit that the CUSUMSQ test is also performed in addition to the CUSUM test and these two tests are performed in the same procedure.
The hypothesis developed as:
H_{0}: The series of the model is unstable
H_{1}: The series of the model is stable
Decision rule: Reject null hypothesis of instability if the plot of CUSUM for the model is not with the five percent critical bound. If otherwise, do no reject.
3.4 Conclusion
The methodology of a research work is said to lay a solid foundation for a researcher to build a robust analysis on. Thus, this chapter has revealed the various important tests that are likely to be conducted in the next chapter. This test are categorized into three phases which are, the preliminary, the main and the diagnostics tests. The chapter also revealed the source of the data that will be used for the analysis as well as the measurement of the variables.
4 CHAPTER FOUR: DATA ANALYSIS AND INTERPRETATION
4.1 Introduction
The chapter presents the various analysis that are generated with the help of Eviews 10, in order to achieve the objectives of the study. The analysis for the study was done in three different phases in order to ensure that the findings from the study is devoid of any econometric problem that may render the result a spurious.
The first phase is the pretests phase that accommodates the check for stationarity in the series with the adoption of Augmented DickeyFuller and Phillip Perron amongst another unit root test. While the second phase presents the main estimation techniques as suggested in chapter three. The AutoRegressive Distributed Lag model was adopted as the estimation technique that is used to investigate the impact of FDI on international trade in Malaysia. Furthermore, the last phase of the study presents the Diagnostic checks for the model. The diagnostics check was performed to examine the problem of autocorrelation, heteroscedasticity, multicollinearity e.t.c in the model.
4.2 Stationarity Test
In order to ensure have a robust result when dealing with a time series data, the importance of stationarity check cannot be downplayed. Hence, this study begins the analysis by investigating whether the data that constitutes the model for this study are stationary or not. If the series are stationary, the problem of unit root problem is absent. If otherwise, there is unit root in the model. One unique importance of conducting the unit root test on time series data is that is suggests the appropriate estimation techniques to be adopted in the study (Shrestha & Bhatta, 2018).
This study adopted the two most widely used unit root tests of ADF and Philip Perron in this study. It is widely observed that most time series data are likely to exhibit the occurrence of unit root at level, the step to take in order to ensure that the problem is solved is by converting the nonstationary data to a stationary by taking its difference from the first lag (Greene, 2003).
Table 4.1 Abridged Augmented DickeyFuller and Philip Perron Unit Root Test
ADF Statistics  Phillip Perron  
Level  First Differences  Level  First Differences  
Trend and Intercept  Intercept  Trend and Intercept  Intercept  
INT  0.245700  4.007191 (1)***  0.439127  4.147141(1)*** 
FDI  3.055763  6.779174 (1)***  3.127362  6.895726(1)*** 
CFN  5.259769 (0) ***  6.074679  5.264966(0)***  21.84382 
EXC  2.014412  5.155576 (1)***  2.014412  5.106565(1)*** 
INF  3.931327(0)***  5.671466  3.838360(0)***  9.089108 
Note: *** Indicates 5% level of significance.
Source: Author’s Computation from Eviews 10, 2021
Hypothesis:
H_{0}: INT/ FDI/CFN/EXC/INF has no unit root
H_{1}: GDP/ FDI/ CPI/ NX/UER has unit root
Significance Level: α = 5%
Decision Rule: Reject H_{0} if pvalue is less than α @ 5%. Otherwise, do not reject H_{0}.
The above table is an abridged summary of the Augmented DickeyFuller and Philip Peron Unit root tests in the study. The two methods revealed the same staionarity result which implies that any of the methods could have been sufficient to draw conclusion. From the result obtained using the two methods, it was revealed that CFN and IFN are the two variables that are stationary at level @ 0.05 significance level. However, INT, FDI and EXC are found to have unit root at level. Since all the series in the model must be stationary before a meaningful analysis can be done, the study proceeds to differencing the other variables that have unit root by order one. Having done that, it was revealed that INT, FDI and EXC are now stationary @ first difference and they are significance at α=0.05.
The study can be taken further to the main analysis since it has been confirmed that the series for the study are now devoid of unit root problem. The study will proceed to achieving the aim of the study which is to investigate the impact of FDI on international trade in Malaysia. This will done by adopting ARDL method of analysis.
Table 4.2 Optimum lag criterion
NO of lag  Akaike info criterion  Schwarz criterion  Optimum lag 
2  7.492428*  8.189041*  0.69661* 
3  7.658662  8.582382  0.92372 
4  7.522688  8.678089

1.1554 
5  7.146738  8.53842  1.39168 
Source: Author’s Computation from Eviews 10, 2021
The table above revealed the summarized process of the determination of the optimum lag that is suitable for this model. The condition for optimum lag selection is that such lag must result to the minimum difference of AIC and SC. Thus, the optimum lag length selected for the ARDL model is at lag 2. This takes the analysis further to obtaining the ARDL result in order to examine the existence of long run relationship in the model.
4.3 The Autoregressive Distributed Lag Model Result
Table 4.3 AutoRegressive Distributed Lag Model Result
Dependent Variable: D(INT)  
Method: ARDL  
Variable  Coefficient  Standard. Error  tStatistics  Probability value. 
C  39.75203  12.69307  3.131789  0.0050 
D(INT(1))  0.137797  0.211054  0.652900  0.5209 
D(INT(2))  0.473810  0.245955  1.926415  0.0677 
D(FDI(1))  2.097804  1.503800  1.395002  0.1776 
D(FDI(2))  0.930123  1.324201  0.702403  0.4901 
D(CFN(1))  0.118872  0.201680  0.589412  0.5619 
D(CFN(2))  0.010636  0.142171  0.074814  0.9411 
D(EXC(1))  10.14924  8.709237  1.165342  0.2569 
D(EXC(2))  32.09514  12.85348  2.497000  0.0209 
D(INF(1))  2.622492  1.359320  1.929267  0.0673 
D(INF(2))  1.632807  1.122220  1.454980  0.1605 
INT(1)  0.060377  0.063803  0.946301  0.3548 
FDI(1)  1.702097  1.769927  0.961676  0.3472 
CFN(1)  0.140387  0.260339  0.539247  0.5954 
EXC(1)  14.25153  4.639054  3.072076  0.0058 
INF(1)  4.778690  1.750930  2.729229  0.0126 
Source: Authors computation. Eviews 10
The ARDL result presented above requires that we investigate if the model does not exhibit any of the econometric problems that are peculiar to timeseries analysis. Hence, the study investigated the existence of heteroscedasticity, autocorrelation, specification error and CUSUM stability on the ARDL’s result presented above. Following these findings which shall be extensively discussed in the Diagnostics phase of this chapter, it was found that the ARDL result above is devoid of all the econometric problems. Having confirmed this, the next procedure of the techniques is to determine if there is a longrun relationship among the variables of interest which are proxied by INT, FDI, CFN, EXC and INF. This shall be done with the help ARDL Bound Cointegration Test presented below.
4.3.1 ARDL Bound Cointegration Test Model
Table 4.4 ARDL Bound Cointegration Test Model
Wald F statistics = 3.049964
k=5 n=40 

Level of Significance.  Lower bound I(0)  Critical value:
Narayan (2005) 
Upper Bound I(1)  
1%  3.94  6.48 
5%  2.65  5.64 
10%  2.12  3.76 
Source : Author’s computation. Eviews 10
Hypothesis:
H_{0}: INT=FDI=CFN=EXC=INF=0
H_{1}: INT≠FDI≠CFN≠EXC≠INF≠0
Decision Rule: Reject H_{0} if test statistic is greater than upper critical value, otherwise do not reject H0.
The Wald test’s Fstats is 3.049964 which is greater than the (2.65) lower bound test critical value at 5% level of significance but it is lesser than the (5.64) Upper Bound test critical level. This implies that the investigation of the long run relationship among the variables of interest in the study is inconclusive. Although, the associated probability value of the FStatistics stands at 0.0318 which means that it is significant @ 5%. Therefore, the null hypothesis of no cointegration can still be rejected and it is concluded that there is evidence of longrun relationship among the variables. The next line of action is to determine the longrun relationship of the variables of interest in section 4.2.2 below.
4.3.2 The LongRun Relationship among INT and other exogenous variables
Table 4.5 long run relationship between INT and its determinant
Dependent Variable: INT(1)  
Included observations: 39 after adjustments  
Variable  Coefficient  Std. Error  tStatistic  Prob. 
C  26.56082  33.92052  0.783031  0.4390 
FDI(1)  7.147768  3.481293  2.053193  0.0478 
CFN(1)  0.081762  0.405432  0.201667  0.8414 
EXC(1)  34.76499  8.818594  3.942237  0.0004 
INF(1)  1.753567  3.125654  0.561024  0.5785 
Source: Author’s Computation from Eviews 10, 2021
4.3.3 Econometric model
The C in the model depicts the constant which implies that taking all other exogenous variables fixed, international trade in Malaysia will on average increase by 26% although, it is insignificant due to its high probability value.
The FDI in the model is used to proxy Foreign Direct Investment. Its estimated coefficient value is 7.06 which infer that by holding CFN, EXC and INF constant, a percent increase in the inflow of Foreign Direct Investment to Malaysia will result to an estimated 7.06 percent increase in Malaysia’s international trade. Statistically, this result is significant at 5%. In the case of EXC, it was found that all things being equal among the other exogenous variables, a unit increase in exchange rate will exerts a positive impact on Malaysia international trade.
The probability value of CFN and INF which are used to proxy capital formation and inflation rate respectively are insignificant in the model. It is necessary to also interpret what impact they would have exerted on Malaysia international trade according to the result of the analysis.
For CFN, its estimated coefficient value is 0.08. This implies that if FDI, EXC and INF remains zero during the period under investigation, a per cent increase in capital formation in Malaysia will negatively impact the Malaysian international trade. In addition to this, the estimated coefficient of INF is 1.75. This can be interpreted that if all other variables in the model are held constant, a percent increase in the rate of inflation in Malaysia will negatively impact the success of Malaysia in the international market. The major findings as regards the result presented above shall be presented in the next chapter.
The R^{2} in the model is 0.3737 which implies that the explanatory variables for the model can only explain 37% of the whole happenings in the Malaysian international trade during the period under investigation. As stated before, it is expedient to investigate the existence of the econometric problems in this model. This will thus lead to the last phase of the chapter which is the Diagnostics phase.
4.4 Diagnostics Tests
4.4.1 Autocorrelation Test
The BreuschGodfrey LM Test is carried out in order to investigate if the model is not being affected by the problem of serial correlation. The null hypothesis is that “no autocorrelation” which will be rejected at 5% significance level. The table below revealed the serial correlation
Table 4.6 BreuschGodfrey LM Test
BreuschGodfrey Serial Correlation LM Test:  
Fstatistic  0.777523  Prob. F(2,19)  0.4736  
Obs*Rsquared  2.799152  Prob. ChiSquare(2)  0.2467  
Source: Author’s Computation from Eviews 10, 2021
Decision: the null hypothesis of no autocorrelation is not rejected since the probability value that is associated with the Fstat is 0.4736 greater than the acceptable 5% significance level. Therefore, the study concluded that the model is devoid of autocorrelation problem.
4.4.2 Heteroscedasticity Test
The ARCH heteroscedasticity Test is carried out in order to investigate if the ARDL model is not being affected by the problem of homoscedasticity. The null hypothesis is that “no heteroscedasticity” which will be rejected at 5% significance level. The table below revealed the result of the ARCH test
Table 4.7 ARCH heteroscedasticity Test result
Heteroskedasticity Test: ARCH  
Fstatistic  3.034845  Prob. F(2,32)  0.0621  
Obs*Rsquared  5.580269  Prob. ChiSquare(2)  0.0614  
Source: Author’s Computation from Eviews 10, 2021
Decision: the null hypothesis of no heteroskedasitiy is not rejected since the probability value that is associated with the Fstat is 0.0621 and greater than the acceptable 5% level of significance. Thus, the study concluded that the ARDL model is devoid of heteroskedasticty problem.
4.4.3 Specification Bias Test
The study adopt Ramsey Reset test to examine whether the model is correctly specified or not. The null hypothesis is that the ARDL model is linear and correctly specified.
Table 4.8 Ramsey RESET Test
Ramsey RESET Test  
Value  df  Probability  
tstatistic  1.866987  20  0.0766  
Fstatistic  3.485639  (1, 20)  0.0766  
Likelihood ratio  5.944304  1  0.0148  
Source: Author’s Computation from Eviews 10, 2021
Decision: the null hypothesis of linearity and correct specification is not rejected since the probability value that is associated with the Fstat is 0.0766 and greater than the acceptable 5% level of significance. Thus, the study concluded that the ARDL model is linear and correctly specified.
4.4.4 CUSUM and CUSUMSQ
The Cusum diagnostic test is carried out to investigate if the ARDL model did not suffer from structural break. The nonrejection of the CUSUM hypothesis will mean that the blue line falls within the two red lines at 5% level of significance (Idenyi, Ifeyinwa, Obinna, & Promise, 2016).
Figure 4.1 CUSUM (5%)
Figure 4.2 CUSUMSQ (5%)
Source: Author’s Computation from Eviews 10, 2021
Conclusion: from the figures above, it is concluded that the model is stable and suitable for forecasting by structural break.
4.5 Conclusion
This chapter adopted the ARDL method of estimation to analyze the model. The adoption for the ARDL method amongst other estimation techniques was as a result of the unit root result. The Bound test conducted to examine the long run relationship amongst the variables revealed an inconclusive result. In the long run, foreign Direct Investment and Exchange rate are the two explanatory variables that significantly impact Malaysia’s international trade during the period under investigation. The diagnostics tests shows that the model is devoid of all econometric problems. The next chapter will anchor the comprehensive discussion of the major findings.
5 CHAPTER FIVE: DISCUSSIONS, RECOMMENDATIONS AND POLICY IMPLICATIONS
5.1 Introduction
The aim of the study is to investigate the impact of foreign Direct Investment on International trade in Malaysia during 1980 to 2019. This chapter is set to comprehensively discuss the results of the analysis done chapter four. The discussions will be done in order to reveal the major findings from the analysis. The decision as regards the hypothesis formulated in the study shall be revealed during the discussions. This chapter will further reveal the policy implications of the study in order to guide the policymakers as well as the stakeholders in the Malaysia to formulate the right decision as regards the observed variables. In addition to these, the chapter will highlight and subject the limitations of the study to further studies by making helpful recommendations.
5.2 Summary of the Major Findings
The researcher, bearing in mind that stationarity test is paramount in a time series analysis conducted unit root test by adopting the two most widely used methods of Augmented DickeyFuller and Phillip Perron unit root tests. The two methods revealed the same result of a mixed stationarity as some variables are stationary at level while the remaining two are stationary after first difference. This mixed stationarity result implies that ARDL model is the best estimation technique to adopt for the analysis. The optimum Lag length used in the model is at lag two. This has been extensively discussed during the analysis.
The ARDL model procedure was carried out and it was found that there is a long run relationship among the variables during the period under investigation. And lastly, the diagnostics checks on the model revealed that the model yield a robust result as it suffers from none of the econometrics problem.
5.3 Major Findings
5.3.1 Foreign Direct Investment
The long run investigation of the relationship between the endogenous and exogenous variables revealed that Foreign Direct Investment exerts a positive relationship on international trade during the period. The justification for this result is that FDI comes along with technological knowhow which is expected to enhance the production capacity of the local industries. If this is achieved, Malaysia will have sufficient goods available for exportation in the international market. The findings is accurately supported by the study of (Mukhtarov, Alalawneh, Ibadov, & Huseynli, 2019) who also employed ARDL method of estimation to investigate the impact of FDI on international trade and found that Foreign Direct Investment has a positive and statistically significant longrun impact on international trade. The null hypothesis formulated as regards FDI and international trade was rejected since the associated probability value is 0.0478.
5.3.2 Exchange rate
From the ARDL analysis in the previous chapter, it was revealed that exchange rate exerts positive and a statistically significant long run relationship with international trade during the period under review. This finding is backed by the theoretical postulations that a decrease in exchange rate in the host country will lead to increase in export of goods and service. The findings are consistent with that of (Kang & Dagli, 2018) who analysed the link between international trade and exchange rate and found that there exists a positive relationship between exchange rate and international trade. The tstatistic shows that the coefficient of exchange rate is significant at 5% significance level. Hence, the null hypothesis formulated as regards exchange rate and international trade is rejected. Capital formation and inflation rate insignificant in the model despite the correct negative signs.
5.4 Limitations of the Study
To conduct a study of this type without any limitations will be very rare. Hence the limitations that are assumed to mitigate the full success of this study will be highlighted for further studies.
The first among the limitations that the researcher deemed fit to have limited the achievement of the study is the measurement of the dependent variable which is international trade. Based on the review of scholarly articles, there is no generally acceptable means of measuring this variable. This implies that the support or contradiction of the findings in relation to other study might have been influenced by the way the variable is measured in the study as against how it is measured in other studies.
One more limitation that the researcher deemed fit to also highlight for future researchers is that the happenings in international trade cannot really be captured by using just a country’s economy as a case study.
5.5 Recommendations
The highlighted limitations in the previous section were highlighted by the researcher in other for future researchers to improve it. This however led to the following future recommendations belwo.
Owing to the fact that there is no generally acceptable way for measuring a country’s international trade, the study suggest that the average of exports and imports should be tried in the future study as against taking either the import or the exports.
One more recommendation that is derived from the observed limitations above is that a cross sectional data analysis should be tried in the future study as against a timeseries data in order to have a better result on the happenings in the international trade.
The estimate obtained from the multiple regression analysis revealed that foreign direct investment and exchange rate significantly impact the Malaysian stock market during the period under investigation.
Exchange rate stability is assumed to strengthen investors’ confidence which may lead to the attraction of more foreign direct investment which could be channelled through the Malaysian stock market. Based on this revelation, the study implores policymakers to promote policies towards exchange rate stability and attraction of foreign direct investment.
Furthermore, the study implores policy makers in Malaysia to improving the trading system on stock market since it is found that pumping more money to the economy will not enhance the stock market performance but a stable exchange rate will.
5.6 Policy Implications
The long run analysis in the previous chapter revealed that foreign direct investment and exchange rate exerts positive and significant relationship with international trade.
In the case of Foreign Direct Investment, policy makers are advised to formulate policies that are geared towards attracting foreign Direct Investment as this will expand the productive capacity and technological knowhow of the manufacturing industries in the country. Once there are enough locally produced goods and services for export, Malaysia will gain more from the international trade.
In addition to this, the policy makers, especially the monetary policymakers in Malaysia should formulate policies that will relatively lower the value of the Malaysian Ringgit as this will promote export and in turn, stabilize the exchange rate in the long run.
Policy makers should ensure that the huge sum of foreign direct investment into Malaysia economy should be channeled at increasing the capital formation since this will lead to improvement in productive capacity and then promote export. In addition to this, a close monitoring should be given to the aspect of inflation as this will negatively impact all attempts made towards international trade.
5.7 Conclusion
The chapter carefully discussed the findings of the ARDL analysis. A longrun relationship was found among the variables during the review period. The two variables that are significant to the study are the FDI and Exchange rate. The findings are supported by previous research work and the null hypothesis for FDI and Exchange rate are rejected based on the tstatistics and probability value at 5%. The assumed limitations found in the study are highlighted for future studies to accommodate. The limitations led to the recommendations made in this chapter and the policy implications are articulately discussed in order to guide the policy makers, researchers and students of knowledge.
Appendices
Appendix A: EView Output
DATA FOR THE ANALYSIS
YEAR  INT  FDI  CFN  EXC  INF 
1980  112.5872  3.813716  18.83021  2.176883  6.674920 
1981  110.8621  5.057831  17.19071  2.304125  9.700000 
1982  110.4586  5.212562  14.37437  2.335392  5.818900 
1983  108.0163  4.153743  7.634970  2.321250  3.704235 
1984  106.6298  2.349424  5.474078  2.343642  3.897273 
1985  104.6831  2.226631  19.73156  2.483042  0.346459 
1986  106.4978  1.762689  13.44882  2.581442  0.737003 
1987  111.9196  1.313417  2.599023  2.519638  0.290008 
1988  122.6242  2.039636  25.94572  2.618783  2.556519 
1989  136.6891  4.293264  22.97641  2.708842  2.813201 
1990  146.8883  5.298123  21.39160  2.704875  2.617801 
1991  159.3114  8.136330  29.49786  2.750067  4.358333 
1992  150.6112  8.760474  3.416797  2.547383  4.767228 
1993  157.9405  7.482854  22.77511  2.574095  3.536585 
1994  179.9049  5.829614  12.65663  2.624257  3.724971 
1995  192.1132  4.710245  25.77860  2.504404  3.450575 
1996  181.7670  5.035343  5.780323  2.515943  3.488559 
1997  185.6651  5.136241  11.23564  2.813192  2.662515 
1998  209.4922  2.997751  43.04433  3.924375  5.270342 
1999  217.5709  4.921467  3.862111  3.800000  2.744561 
2000  220.4068  4.038429  29.82470  3.800000  1.534740 
2001  203.3646  0.597029  9.319520  3.800000  1.416785 
2002  199.3562  3.166124  7.908375  3.800000  1.807872 
2003  194.1951  2.920942  1.511222  3.800000  1.089676 
2004  210.3743  3.507873  6.860418  3.800000  1.421271 
2005  203.8546  2.734393  2.461324  3.787092  2.975071 
2006  202.5771  4.727194  8.577085  3.668177  3.609236 
2007  192.4655  4.686888  10.54063  3.437569  2.027353 
2008  176.6683  3.280787  1.867939  3.335833  5.440782 
2009  162.5590  0.056692  9.688507  3.524503  0.583308 
2010  157.9448  4.268664  23.88030  3.221087  1.622852 
2011  154.9377  5.074455  4.520445  3.060003  3.174471 
2012  147.8418  2.829056  18.27939  3.088801  1.663571 
2013  142.7210  3.494302  4.854900  3.150909  2.105012 
2014  138.3122  3.141268  2.549172  3.272860  3.142991 
2015  131.3701  3.270949  6.738290  3.905500  2.104390 
2016  126.8990  4.471319  4.430964  4.148301  2.090567 
2017  133.1552  2.935792  6.284135  4.300441  3.871201 
2018  130.4304  2.315063  1.674166  4.035130  0.884709 
2019  123.0002  2.495618  3.919804  4.142470  0.662892 
ADF UNIT ROOT
Int @ level
Null Hypothesis: INT has a unit root  
Exogenous: Constant, Linear Trend  
Lag Length: 0 (Automatic – based on SIC, maxlag=9)  
tStatistic  Prob.*  
Augmented DickeyFuller test statistic  0.245700  0.9896  
Test critical values:  1% level  4.211868  
5% level  3.529758  
10% level  3.196411  
*MacKinnon (1996) onesided pvalues.  
Augmented DickeyFuller Test Equation  
Dependent Variable: D(INT)  
Method: Least Squares  
Date: 05/29/21 Time: 09:10  
Sample (adjusted): 1981 2019  
Included observations: 39 after adjustments  
Variable  Coefficient  Std. Error  tStatistic  Prob. 
INT(1)  0.010768  0.043827  0.245700  0.8073 
C  8.008632  6.657440  1.202960  0.2368 
@TREND(“1980”)  0.302292  0.141509  2.136204  0.0395 
Rsquared  0.136084  Mean dependent var  0.266999  
Adjusted Rsquared  0.088088  S.D. dependent var  9.768201  
S.E. of regression  9.328054  Akaike info criterion  7.377733  
Sum squared resid  3132.453  Schwarz criterion  7.505700  
Log likelihood  140.8658  HannanQuinn criter.  7.423646  
Fstatistic  2.835348  DurbinWatson stat  1.425347  
Prob(Fstatistic)  0.071861  
int @ first diference
Null Hypothesis: D(INT) has a unit root  
Exogenous: Constant  
Lag Length: 0 (Automatic – based on SIC, maxlag=9)  
tStatistic  Prob.*  
Augmented DickeyFuller test statistic  4.007191  0.0036  
Test critical values:  1% level  3.615588  
5% level  2.941145  
10% level  2.609066  
*MacKinnon (1996) onesided pvalues.  
Augmented DickeyFuller Test Equation  
Dependent Variable: D(INT,2)  
Method: Least Squares  
Date: 05/29/21 Time: 09:11  
Sample (adjusted): 1982 2019  
Included observations: 38 after adjustments  
Variable  Coefficient  Std. Error  tStatistic  Prob. 
D(INT(1))  0.624773  0.155913  4.007191  0.0003 
C  0.143232  1.511938  0.094734  0.9251 
Rsquared  0.308458  Mean dependent var  0.150135  
Adjusted Rsquared  0.289249  S.D. dependent var  11.04224  
S.E. of regression  9.309280  Akaike info criterion  7.351097  
Sum squared resid  3119.857  Schwarz criterion  7.437285  
Log likelihood  137.6708  HannanQuinn criter.  7.381762  
Fstatistic  16.05758  DurbinWatson stat  1.923166  
Prob(Fstatistic)  0.000295  
fdi @ level
Null Hypothesis: FDI has a unit root  
Exogenous: Constant, Linear Trend  
Lag Length: 0 (Automatic – based on SIC, maxlag=9)  
tStatistic  Prob.*  
Augmented DickeyFuller test statistic  3.055763  0.1310  
Test critical values:  1% level  4.211868  
5% level  3.529758  
10% level  3.196411  
*MacKinnon (1996) onesided pvalues.  
Augmented DickeyFuller Test Equation  
Dependent Variable: D(FDI)  
Method: Least Squares  
Date: 05/29/21 Time: 09:12  
Sample (adjusted): 1981 2019  
Included observations: 39 after adjustments  
Variable  Coefficient  Std. Error  tStatistic  Prob. 
FDI(1)  0.411694  0.134727  3.055763  0.0042 
C  1.948607  0.780251  2.497411  0.0172 
@TREND(“1980”)  0.018867  0.021567  0.874798  0.3875 
Rsquared  0.207248  Mean dependent var  0.033797  
Adjusted Rsquared  0.163207  S.D. dependent var  1.623905  
S.E. of regression  1.485490  Akaike info criterion  3.703170  
Sum squared resid  79.44048  Schwarz criterion  3.831136  
Log likelihood  69.21181  HannanQuinn criter.  3.749083  
Fstatistic  4.705727  DurbinWatson stat  1.845104  
Prob(Fstatistic)  0.015292  
fdi @ first difference
Null Hypothesis: D(FDI) has a unit root  
Exogenous: Constant  
Lag Length: 0 (Automatic – based on SIC, maxlag=9)  
tStatistic  Prob.*  
Augmented DickeyFuller test statistic  6.779174  0.0000  
Test critical values:  1% level  3.615588  
5% level  2.941145  
10% level  2.609066  
*MacKinnon (1996) onesided pvalues.  
Augmented DickeyFuller Test Equation  
Dependent Variable: D(FDI,2)  
Method: Least Squares  
Date: 05/29/21 Time: 09:13  
Sample (adjusted): 1982 2019  
Included observations: 38 after adjustments  
Variable  Coefficient  Std. Error  tStatistic  Prob. 
D(FDI(1))  1.113302  0.164224  6.779174  0.0000 
C  0.071895  0.266700  0.269573  0.7890 
Rsquared  0.560746  Mean dependent var  0.027988  
Adjusted Rsquared  0.548545  S.D. dependent var  2.446126  
S.E. of regression  1.643562  Akaike info criterion  3.882805  
Sum squared resid  97.24671  Schwarz criterion  3.968994  
Log likelihood  71.77330  HannanQuinn criter.  3.913471  
Fstatistic  45.95720  DurbinWatson stat  2.064174  
Prob(Fstatistic)  0.000000  
cfn @ level
Null Hypothesis: CFN has a unit root  
Exogenous: Constant, Linear Trend  
Lag Length: 0 (Automatic – based on SIC, maxlag=9)  
tStatistic  Prob.*  
Augmented DickeyFuller test statistic  5.259769  0.0006  
Test critical values:  1% level  4.211868  
5% level  3.529758  
10% level  3.196411  
*MacKinnon (1996) onesided pvalues.  
Augmented DickeyFuller Test Equation  
Dependent Variable: D(CFN)  
Method: Least Squares  
Date: 05/29/21 Time: 09:14  
Sample (adjusted): 1981 2019  
Included observations: 39 after adjustments  
Variable  Coefficient  Std. Error  tStatistic  Prob. 
CFN(1)  0.867634  0.164957  5.259769  0.0000 
C  8.449542  5.060475  1.669713  0.1037 
@TREND(“1980”)  0.146042  0.209083  0.698486  0.4894 
Rsquared  0.434548  Mean dependent var  0.583334  
Adjusted Rsquared  0.403134  S.D. dependent var  18.83993  
S.E. of regression  14.55518  Akaike info criterion  8.267574  
Sum squared resid  7626.717  Schwarz criterion  8.395541  
Log likelihood  158.2177  HannanQuinn criter.  8.313488  
Fstatistic  13.83295  DurbinWatson stat  2.011643  
Prob(Fstatistic)  0.000035  
Cfn @ first difference
Null Hypothesis: D(CFN) has a unit root  
Exogenous: Constant  
Lag Length: 1 (Automatic – based on SIC, maxlag=9)  
tStatistic  Prob.*  
Augmented DickeyFuller test statistic  6.074679  0.0000  
Test critical values:  1% level  3.621023  
5% level  2.943427  
10% level  2.610263  
*MacKinnon (1996) onesided pvalues.  
Augmented DickeyFuller Test Equation  
Dependent Variable: D(CFN,2)  
Method: Least Squares  
Date: 05/29/21 Time: 09:14  
Sample (adjusted): 1983 2019  
Included observations: 37 after adjustments  
Variable  Coefficient  Std. Error  tStatistic  Prob. 
D(CFN(1))  1.744141  0.287117  6.074679  0.0000 
D(CFN(1),2)  0.200593  0.168524  1.190291  0.2422 
C  0.839589  2.862705  0.293285  0.7711 
Rsquared  0.737293  Mean dependent var  0.015424  
Adjusted Rsquared  0.721839  S.D. dependent var  32.98603  
S.E. of regression  17.39714  Akaike info criterion  8.628093  
Sum squared resid  10290.46  Schwarz criterion  8.758708  
Log likelihood  156.6197  HannanQuinn criter.  8.674141  
Fstatistic  47.71083  DurbinWatson stat  2.071654  
Prob(Fstatistic)  0.000000  
exc @ level
Null Hypothesis: EXC has a unit root  
Exogenous: Constant, Linear Trend  
Lag Length: 0 (Automatic – based on SIC, maxlag=9)  
tStatistic  Prob.*  
Augmented DickeyFuller test statistic  2.014412  0.5755  
Test critical values:  1% level  4.211868  
5% level  3.529758  
10% level  3.196411  
*MacKinnon (1996) onesided pvalues.  
Augmented DickeyFuller Test Equation  
Dependent Variable: D(EXC)  
Method: Least Squares  
Date: 05/29/21 Time: 09:15  
Sample (adjusted): 1981 2019  
Included observations: 39 after adjustments  
Variable  Coefficient  Std. Error  tStatistic  Prob. 
EXC(1)  0.202760  0.100655  2.014412  0.0515 
C  0.502569  0.237362  2.117312  0.0412 
@TREND(“1980”)  0.009128  0.005606  1.628220  0.1122 
Rsquared  0.101302  Mean dependent var  0.050400  
Adjusted Rsquared  0.051374  S.D. dependent var  0.239756  
S.E. of regression  0.233516  Akaike info criterion  0.002670  
Sum squared resid  1.963068  Schwarz criterion  0.130636  
Log likelihood  2.947937  HannanQuinn criter.  0.048583  
Fstatistic  2.028973  DurbinWatson stat  1.548021  
Prob(Fstatistic)  0.146234  
exc @ first difference
Null Hypothesis: D(EXC) has a unit root  
Exogenous: Constant  
Lag Length: 0 (Automatic – based on SIC, maxlag=9)  
tStatistic  Prob.*  
Augmented DickeyFuller test statistic  5.155576  0.0001  
Test critical values:  1% level  3.615588  
5% level  2.941145  
10% level  2.609066  
*MacKinnon (1996) onesided pvalues.  
Augmented DickeyFuller Test Equation  
Dependent Variable: D(EXC,2)  
Method: Least Squares  
Date: 05/29/21 Time: 09:16  
Sample (adjusted): 1982 2019  
Included observations: 38 after adjustments  
Variable  Coefficient  Std. Error  tStatistic  Prob. 
D(EXC(1))  0.848845  0.164646  5.155576  0.0000 
C  0.040986  0.040258  1.018077  0.3154 
Rsquared  0.424736  Mean dependent var  0.000524  
Adjusted Rsquared  0.408756  S.D. dependent var  0.316226  
S.E. of regression  0.243154  Akaike info criterion  0.060950  
Sum squared resid  2.128454  Schwarz criterion  0.147139  
Log likelihood  0.841948  HannanQuinn criter.  0.091615  
Fstatistic  26.57996  DurbinWatson stat  1.970197  
Prob(Fstatistic)  0.000009  
inf @ level
Null Hypothesis: INF has a unit root  
Exogenous: Constant, Linear Trend  
Lag Length: 0 (Automatic – based on SIC, maxlag=9)  
tStatistic  Prob.*  
Augmented DickeyFuller test statistic  3.931327  0.0199  
Test critical values:  1% level  4.211868  
5% level  3.529758  
10% level  3.196411  
*MacKinnon (1996) onesided pvalues.  
Augmented DickeyFuller Test Equation  
Dependent Variable: D(INF)  
Method: Least Squares  
Date: 05/29/21 Time: 09:19  
Sample (adjusted): 1981 2019  
Included observations: 39 after adjustments  
Variable  Coefficient  Std. Error  tStatistic  Prob. 
INF(1)  0.575057  0.146275  3.931327  0.0004 
C  2.159078  0.805200  2.681419  0.0110 
@TREND(“1980”)  0.030342  0.024250  1.251237  0.2189 
Rsquared  0.301386  Mean dependent var  0.154155  
Adjusted Rsquared  0.262574  S.D. dependent var  1.833654  
S.E. of regression  1.574623  Akaike info criterion  3.819712  
Sum squared resid  89.25972  Schwarz criterion  3.947678  
Log likelihood  71.48438  HannanQuinn criter.  3.865625  
Fstatistic  7.765311  DurbinWatson stat  2.037723  
Prob(Fstatistic)  0.001571  
inf @ first difference
Null Hypothesis: D(INF) has a unit root  
Exogenous: Constant  
Lag Length: 1 (Automatic – based on SIC, maxlag=9)  
tStatistic  Prob.*  
Augmented DickeyFuller test statistic  5.671466  0.0000  
Test critical values:  1% level  3.621023  
5% level  2.943427  
10% level  2.610263  
*MacKinnon (1996) onesided pvalues.  
Augmented DickeyFuller Test Equation  
Dependent Variable: D(INF,2)  
Method: Least Squares  
Date: 05/29/21 Time: 09:20  
Sample (adjusted): 1983 2019  
Included observations: 37 after adjustments  
Variable  Coefficient  Std. Error  tStatistic  Prob. 
D(INF(1))  1.550181  0.273330  5.671466  0.0000 
D(INF(1),2)  0.173447  0.162418  1.067906  0.2931 
C  0.242252  0.270935  0.894133  0.3775 
Rsquared  0.693723  Mean dependent var  0.098900  
Adjusted Rsquared  0.675707  S.D. dependent var  2.853027  
S.E. of regression  1.624705  Akaike info criterion  3.886134  
Sum squared resid  89.74864  Schwarz criterion  4.016749  
Log likelihood  68.89348  HannanQuinn criter.  3.932182  
Fstatistic  38.50539  DurbinWatson stat  1.976604  
Prob(Fstatistic)  0.000000  
PHILIP PHERON
INT @ LEVEL
Null Hypothesis: INT has a unit root  
Exogenous: Constant, Linear Trend  
Bandwidth: 2 (NeweyWest automatic) using Bartlett kernel  
Adj. tStat  Prob.*  
PhillipsPerron test statistic  0.439127  0.9824  
Test critical values:  1% level  4.211868  
5% level  3.529758  
10% level  3.196411  
*MacKinnon (1996) onesided pvalues.  
Residual variance (no correction)  80.31931  
HAC corrected variance (Bartlett kernel)  105.5194  
PhillipsPerron Test Equation  
Dependent Variable: D(INT)  
Method: Least Squares  
Date: 05/29/21 Time: 09:21  
Sample (adjusted): 1981 2019  
Included observations: 39 after adjustments  
Variable  Coefficient  Std. Error  tStatistic  Prob. 
INT(1)  0.010768  0.043827  0.245700  0.8073 
C  8.008632  6.657440  1.202960  0.2368 
@TREND(“1980”)  0.302292  0.141509  2.136204  0.0395 
Rsquared  0.136084  Mean dependent var  0.266999  
Adjusted Rsquared  0.088088  S.D. dependent var  9.768201  
S.E. of regression  9.328054  Akaike info criterion  7.377733  
Sum squared resid  3132.453  Schwarz criterion  7.505700  
Log likelihood  140.8658  HannanQuinn criter.  7.423646  
Fstatistic  2.835348  DurbinWatson stat  1.425347  
Prob(Fstatistic)  0.071861  
INT @ FIRST DIFFERENCE
Null Hypothesis: D(INT) has a unit root  
Exogenous: Constant  
Bandwidth: 6 (NeweyWest automatic) using Bartlett kernel  
Adj. tStat  Prob.*  
PhillipsPerron test statistic  4.147141  0.0024  
Test critical values:  1% level  3.615588  
5% level  2.941145  
10% level  2.609066  
*MacKinnon (1996) onesided pvalues.  
Residual variance (no correction)  82.10151  
HAC corrected variance (Bartlett kernel)  95.07331  
PhillipsPerron Test Equation  
Dependent Variable: D(INT,2)  
Method: Least Squares  
Date: 05/29/21 Time: 09:22  
Sample (adjusted): 1982 2019  
Included observations: 38 after adjustments  
Variable  Coefficient  Std. Error  tStatistic  Prob. 
D(INT(1))  0.624773  0.155913  4.007191  0.0003 
C  0.143232  1.511938  0.094734  0.9251 
Rsquared  0.308458  Mean dependent var  0.150135  
Adjusted Rsquared  0.289249  S.D. dependent var  11.04224  
S.E. of regression  9.309280  Akaike info criterion  7.351097  
Sum squared resid  3119.857  Schwarz criterion  7.437285  
Log likelihood  137.6708  HannanQuinn criter.  7.381762  
Fstatistic  16.05758  DurbinWatson stat  1.923166  
Prob(Fstatistic)  0.000295  
FDI @ LEVEL
Null Hypothesis: FDI has a unit root  
Exogenous: Constant, Linear Trend  
Bandwidth: 1 (NeweyWest automatic) using Bartlett kernel  
Adj. tStat  Prob.*  
PhillipsPerron test statistic  3.127362  0.1144  
Test critical values:  1% level  4.211868  
5% level  3.529758  
10% level  3.196411  
*MacKinnon (1996) onesided pvalues.  
Residual variance (no correction)  2.036935  
HAC corrected variance (Bartlett kernel)  2.184583  
PhillipsPerron Test Equation  
Dependent Variable: D(FDI)  
Method: Least Squares  
Date: 05/29/21 Time: 09:22  
Sample (adjusted): 1981 2019  
Included observations: 39 after adjustments  
Variable  Coefficient  Std. Error  tStatistic  Prob. 
FDI(1)  0.411694  0.134727  3.055763  0.0042 
C  1.948607  0.780251  2.497411  0.0172 
@TREND(“1980”)  0.018867  0.021567  0.874798  0.3875 
Rsquared  0.207248  Mean dependent var  0.033797  
Adjusted Rsquared  0.163207  S.D. dependent var  1.623905  
S.E. of regression  1.485490  Akaike info criterion  3.703170  
Sum squared resid  79.44048  Schwarz criterion  3.831136  
Log likelihood  69.21181  HannanQuinn criter.  3.749083  
Fstatistic  4.705727  DurbinWatson stat  1.845104  
Prob(Fstatistic)  0.015292  
FDI @ FIRST DIFFERENCE
Null Hypothesis: D(FDI) has a unit root  
Exogenous: Constant  
Bandwidth: 2 (NeweyWest automatic) using Bartlett kernel  
Adj. tStat  Prob.*  
PhillipsPerron test statistic  6.895726  0.0000  
Test critical values:  1% level  3.615588  
5% level  2.941145  
10% level  2.609066  
*MacKinnon (1996) onesided pvalues.  
Residual variance (no correction)  2.559124  
HAC corrected variance (Bartlett kernel)  2.060043  
PhillipsPerron Test Equation  
Dependent Variable: D(FDI,2)  
Method: Least Squares  
Date: 05/29/21 Time: 09:23  
Sample (adjusted): 1982 2019  
Included observations: 38 after adjustments  
Variable  Coefficient  Std. Error  tStatistic  Prob. 
D(FDI(1))  1.113302  0.164224  6.779174  0.0000 
C  0.071895  0.266700  0.269573  0.7890 
Rsquared  0.560746  Mean dependent var  0.027988  
Adjusted Rsquared  0.548545  S.D. dependent var  2.446126  
S.E. of regression  1.643562  Akaike info criterion  3.882805  
Sum squared resid  97.24671  Schwarz criterion  3.968994  
Log likelihood  71.77330  HannanQuinn criter.  3.913471  
Fstatistic  45.95720  DurbinWatson stat  2.064174  
Prob(Fstatistic)  0.000000  
CFN @ LEVEL
Null Hypothesis: CFN has a unit root  
Exogenous: Constant, Linear Trend  
Bandwidth: 2 (NeweyWest automatic) using Bartlett kernel  
Adj. tStat  Prob.*  
PhillipsPerron test statistic  5.264966  0.0006  
Test critical values:  1% level  4.211868  
5% level  3.529758  
10% level  3.196411  
*MacKinnon (1996) onesided pvalues.  
Residual variance (no correction)  195.5569  
HAC corrected variance (Bartlett kernel)  197.7408  
PhillipsPerron Test Equation  
Dependent Variable: D(CFN)  
Method: Least Squares  
Date: 05/29/21 Time: 09:23  
Sample (adjusted): 1981 2019  
Included observations: 39 after adjustments  
Variable  Coefficient  Std. Error  tStatistic  Prob. 
CFN(1)  0.867634  0.164957  5.259769  0.0000 
C  8.449542  5.060475  1.669713  0.1037 
@TREND(“1980”)  0.146042  0.209083  0.698486  0.4894 
Rsquared  0.434548  Mean dependent var  0.583334  
Adjusted Rsquared  0.403134  S.D. dependent var  18.83993  
S.E. of regression  14.55518  Akaike info criterion  8.267574  
Sum squared resid  7626.717  Schwarz criterion  8.395541  
Log likelihood  158.2177  HannanQuinn criter.  8.313488  
Fstatistic  13.83295  DurbinWatson stat  2.011643  
Prob(Fstatistic)  0.000035  
CFN @ FIRST DIFFERENCE
Null Hypothesis: D(CFN) has a unit root  
Exogenous: Constant  
Bandwidth: 18 (NeweyWest automatic) using Bartlett kernel  
Adj. tStat  Prob.*  
PhillipsPerron test statistic  21.84382  0.0001  
Test critical values:  1% level  3.615588  
5% level  2.941145  
10% level  2.609066  
*MacKinnon (1996) onesided pvalues.  
Residual variance (no correction)  282.2915  
HAC corrected variance (Bartlett kernel)  31.79622  
PhillipsPerron Test Equation  
Dependent Variable: D(CFN,2)  
Method: Least Squares  
Date: 05/29/21 Time: 09:24  
Sample (adjusted): 1982 2019  
Included observations: 38 after adjustments  
Variable  Coefficient  Std. Error  tStatistic  Prob. 
D(CFN(1))  1.452374  0.148650  9.770448  0.0000 
C  0.799636  2.801404  0.285441  0.7769 
Rsquared  0.726156  Mean dependent var  0.015951  
Adjusted Rsquared  0.718549  S.D. dependent var  32.53780  
S.E. of regression  17.26193  Akaike info criterion  8.586080  
Sum squared resid  10727.08  Schwarz criterion  8.672269  
Log likelihood  161.1355  HannanQuinn criter.  8.616746  
Fstatistic  95.46165  DurbinWatson stat  2.177608  
Prob(Fstatistic)  0.000000  
EXC @ LEVEL
Null Hypothesis: EXC has a unit root  
Exogenous: Constant, Linear Trend  
Bandwidth: 0 (NeweyWest automatic) using Bartlett kernel  
Adj. tStat  Prob.*  
PhillipsPerron test statistic  2.014412  0.5755  
Test critical values:  1% level  4.211868  
5% level  3.529758  
10% level  3.196411  
*MacKinnon (1996) onesided pvalues.  
Residual variance (no correction)  0.050335  
HAC corrected variance (Bartlett kernel)  0.050335  
PhillipsPerron Test Equation  
Dependent Variable: D(EXC)  
Method: Least Squares  
Date: 05/29/21 Time: 09:24  
Sample (adjusted): 1981 2019  
Included observations: 39 after adjustments  
Variable  Coefficient  Std. Error  tStatistic  Prob. 
EXC(1)  0.202760  0.100655  2.014412  0.0515 
C  0.502569  0.237362  2.117312  0.0412 
@TREND(“1980”)  0.009128  0.005606  1.628220  0.1122 
Rsquared  0.101302  Mean dependent var  0.050400  
Adjusted Rsquared  0.051374  S.D. dependent var  0.239756  
S.E. of regression  0.233516  Akaike info criterion  0.002670  
Sum squared resid  1.963068  Schwarz criterion  0.130636  
Log likelihood  2.947937  HannanQuinn criter.  0.048583  
Fstatistic  2.028973  DurbinWatson stat  1.548021  
Prob(Fstatistic)  0.146234  
EXC @ FIRST DIFFERENCE
Null Hypothesis: D(EXC) has a unit root  
Exogenous: Constant  
Bandwidth: 4 (NeweyWest automatic) using Bartlett kernel  
Adj. tStat  Prob.*  
PhillipsPerron test statistic  5.106565  0.0002  
Test critical values:  1% level  3.615588  
5% level  2.941145  
10% level  2.609066  
*MacKinnon (1996) onesided pvalues.  
Residual variance (no correction)  0.056012  
HAC corrected variance (Bartlett kernel)  0.049323  
PhillipsPerron Test Equation  
Dependent Variable: D(EXC,2)  
Method: Least Squares  
Date: 05/29/21 Time: 09:25  
Sample (adjusted): 1982 2019  
Included observations: 38 after adjustments  
Variable  Coefficient  Std. Error  tStatistic  Prob. 
D(EXC(1))  0.848845  0.164646  5.155576  0.0000 
C  0.040986  0.040258  1.018077  0.3154 
Rsquared  0.424736  Mean dependent var  0.000524  
Adjusted Rsquared  0.408756  S.D. dependent var  0.316226  
S.E. of regression  0.243154  Akaike info criterion  0.060950  
Sum squared resid  2.128454  Schwarz criterion  0.147139  
Log likelihood  0.841948  HannanQuinn criter.  0.091615  
Fstatistic  26.57996  DurbinWatson stat  1.970197  
Prob(Fstatistic)  0.000009  
INF @ LEVEL
Null Hypothesis: INF has a unit root  
Exogenous: Constant, Linear Trend  
Bandwidth: 3 (NeweyWest automatic) using Bartlett kernel  
Adj. tStat  Prob.*  
PhillipsPerron test statistic  3.838360  0.0248  
Test critical values:  1% level  4.211868  
5% level  3.529758  
10% level  3.196411  
*MacKinnon (1996) onesided pvalues.  
Residual variance (no correction)  2.288711  
HAC corrected variance (Bartlett kernel)  2.008220  
PhillipsPerron Test Equation  
Dependent Variable: D(INF)  
Method: Least Squares  
Date: 05/29/21 Time: 09:25  
Sample (adjusted): 1981 2019  
Included observations: 39 after adjustments  
Variable  Coefficient  Std. Error  tStatistic  Prob. 
INF(1)  0.575057  0.146275  3.931327  0.0004 
C  2.159078  0.805200  2.681419  0.0110 
@TREND(“1980”)  0.030342  0.024250  1.251237  0.2189 
Rsquared  0.301386  Mean dependent var  0.154155  
Adjusted Rsquared  0.262574  S.D. dependent var  1.833654  
S.E. of regression  1.574623  Akaike info criterion  3.819712  
Sum squared resid  89.25972  Schwarz criterion  3.947678  
Log likelihood  71.48438  HannanQuinn criter.  3.865625  
Fstatistic  7.765311  DurbinWatson stat  2.037723  
Prob(Fstatistic)  0.001571  
INF @ FIRST DIFFERENCE
Null Hypothesis: D(INF) has a unit root  
Exogenous: Constant  
Bandwidth: 3 (NeweyWest automatic) using Bartlett kernel  
Adj. tStat  Prob.*  
PhillipsPerron test statistic  9.089108  0.0000  
Test critical values:  1% level  3.615588  
5% level  2.941145  
10% level  2.609066  
*MacKinnon (1996) onesided pvalues.  
Residual variance (no correction)  2.618133  
HAC corrected variance (Bartlett kernel)  3.020584  
PhillipsPerron Test Equation  
Dependent Variable: D(INF,2)  
Method: Least Squares  
Date: 05/29/21 Time: 09:26  
Sample (adjusted): 1982 2019  
Included observations: 38 after adjustments  
Variable  Coefficient  Std. Error  tStatistic  Prob. 
D(INF(1))  1.374350  0.147074  9.344631  0.0000 
C  0.294860  0.270607  1.089624  0.2831 
Rsquared  0.708082  Mean dependent var  0.085445  
Adjusted Rsquared  0.699973  S.D. dependent var  3.034982  
S.E. of regression  1.662403  Akaike info criterion  3.905602  
Sum squared resid  99.48906  Schwarz criterion  3.991790  
Log likelihood  72.20643  HannanQuinn criter.  3.936267  
Fstatistic  87.32213  DurbinWatson stat  1.921276  
Prob(Fstatistic)  0.000000  
ESTIMATION TECHNIQUES
ARDL WITH LAG 2
Dependent Variable: D(INT)  
Method: Least Squares  
Date: 05/30/21 Time: 08:45  
Sample (adjusted): 1983 2019  
Included observations: 37 after adjustments  
Variable  Coefficient  Std. Error  tStatistic  Prob. 
C  39.75203  12.69307  3.131789  0.0050 
D(INT(1))  0.137797  0.211054  0.652900  0.5209 
D(INT(2))  0.473810  0.245955  1.926415  0.0677 
D(FDI(1))  2.097804  1.503800  1.395002  0.1776 
D(FDI(2))  0.930123  1.324201  0.702403  0.4901 
D(CFN(1))  0.118872  0.201680  0.589412  0.5619 
D(CFN(2))  0.010636  0.142171  0.074814  0.9411 
D(EXC(1))  10.14924  8.709237  1.165342  0.2569 
D(EXC(2))  32.09514  12.85348  2.497000  0.0209 
D(INF(1))  2.622492  1.359320  1.929267  0.0673 
D(INF(2))  1.632807  1.122220  1.454980  0.1605 
INT(1)  0.060377  0.063803  0.946301  0.3548 
FDI(1)  1.702097  1.769927  0.961676  0.3472 
CFN(1)  0.140387  0.260339  0.539247  0.5954 
EXC(1)  14.25153  4.639054  3.072076  0.0058 
INF(1)  4.778690  1.750930  2.729229  0.0126 
Rsquared  0.547955  Mean dependent var  0.338961  
Adjusted Rsquared  0.225066  S.D. dependent var  10.02949  
S.E. of regression  8.828993  Akaike info criterion  7.492428  
Sum squared resid  1636.973  Schwarz criterion  8.189041  
Log likelihood  122.6099  HannanQuinn criter.  7.738017  
Fstatistic  1.697040  DurbinWatson stat  1.986194  
Prob(Fstatistic)  0.129787  
WALD TEST RESULT
Wald Test:  
Equation: Untitled  
Test Statistic  Value  df  Probability 
Fstatistic  3.049964  (5, 21)  0.0318 
Chisquare  15.24982  5  0.0093 
Null Hypothesis: C(12)=C(13)=C(14)=C(15)=C(16)=0  
Null Hypothesis Summary:  
Normalized Restriction (= 0)  Value  Std. Err.  
C(12)  0.060377  0.063803  
C(13)  1.702097  1.769927  
C(14)  0.140387  0.260339  
C(15)  14.25153  4.639054  
C(16)  4.778690  1.750930  
Restrictions are linear in coefficients. 
LONGRUN RELATIONSHIP
Dependent Variable: INT(1)  
Method: Least Squares  
Date: 05/30/21 Time: 09:05  
Sample (adjusted): 1981 2019  
Included observations: 39 after adjustments  
Variable  Coefficient  Std. Error  tStatistic  Prob. 
C  26.56082  33.92052  0.783031  0.4390 
FDI(1)  7.147768  3.481293  2.053193  0.0478 
CFN(1)  0.081762  0.405432  0.201667  0.8414 
EXC(1)  34.76499  8.818594  3.942237  0.0004 
INF(1)  1.753567  3.125654  0.561024  0.5785 
Rsquared  0.373790  Mean dependent var  157.4786  
Adjusted Rsquared  0.300119  S.D. dependent var  36.81368  
S.E. of regression  30.79793  Akaike info criterion  9.811981  
Sum squared resid  32249.42  Schwarz criterion  10.02526  
Log likelihood  186.3336  HannanQuinn criter.  9.888503  
Fstatistic  5.073730  DurbinWatson stat  0.191382  
Prob(Fstatistic)  0.002575  
DIAGNOSTICS TESTS:
BreuschGodfrey Serial Correlation LM Test:
BreuschGodfrey Serial Correlation LM Test:  
Fstatistic  0.777523  Prob. F(2,19)  0.4736  
Obs*Rsquared  2.799152  Prob. ChiSquare(2)  0.2467  
Test Equation:  
Dependent Variable: RESID  
Method: Least Squares  
Date: 05/30/21 Time: 08:58  
Sample: 1983 2019  
Included observations: 37  
Presample missing value lagged residuals set to zero.  
Variable  Coefficient  Std. Error  tStatistic  Prob. 
C  7.682522  14.90892  0.515297  0.6123 
D(INT(1))  0.263979  0.499163  0.528845  0.6030 
D(INT(2))  0.259836  0.381478  0.681130  0.5040 
D(FDI(1))  1.082519  1.809803  0.598142  0.5568 
D(FDI(2))  0.874049  1.528303  0.571909  0.5741 
D(CFN(1))  0.132991  0.245199  0.542382  0.5939 
D(CFN(2))  0.101233  0.165108  0.613135  0.5471 
D(EXC(1))  6.048793  11.17162  0.541443  0.5945 
D(EXC(2))  4.576050  13.51781  0.338520  0.7387 
D(INF(1))  0.986154  1.718316  0.573907  0.5728 
D(INF(2))  0.756579  1.286701  0.587999  0.5635 
INT(1)  0.037852  0.077645  0.487505  0.6315 
FDI(1)  1.449788  2.232495  0.649403  0.5239 
CFN(1)  0.044806  0.271277  0.165168  0.8706 
EXC(1)  1.288234  4.804527  0.268129  0.7915 
INF(1)  0.967162  1.966828  0.491737  0.6285 
RESID(1)  0.294219  0.574589  0.512051  0.6145 
RESID(2)  0.562587  0.454650  1.237409  0.2310 
Rsquared  0.075653  Mean dependent var  2.21E15  
Adjusted Rsquared  0.751395  S.D. dependent var  6.743255  
S.E. of regression  8.924041  Akaike info criterion  7.521869  
Sum squared resid  1513.132  Schwarz criterion  8.305559  
Log likelihood  121.1546  HannanQuinn criter.  7.798156  
Fstatistic  0.091473  DurbinWatson stat  2.083255  
Prob(Fstatistic)  0.999996  
ARCH HETEROSKEDASTICITY TEST
Heteroskedasticity Test: ARCH  
Fstatistic  3.034845  Prob. F(2,32)  0.0621  
Obs*Rsquared  5.580269  Prob. ChiSquare(2)  0.0614  
Test Equation:  
Dependent Variable: RESID^2  
Method: Least Squares  
Date: 05/30/21 Time: 09:00  
Sample (adjusted): 1985 2019  
Included observations: 35 after adjustments  
Variable  Coefficient  Std. Error  tStatistic  Prob. 
C  40.40191  14.96749  2.699311  0.0110 
RESID^2(1)  0.178748  0.165938  1.077203  0.2894 
RESID^2(2)  0.316024  0.165623  1.908096  0.0654 
Rsquared  0.159436  Mean dependent var  46.50941  
Adjusted Rsquared  0.106901  S.D. dependent var  58.89081  
S.E. of regression  55.65412  Akaike info criterion  10.95801  
Sum squared resid  99116.21  Schwarz criterion  11.09132  
Log likelihood  188.7651  HannanQuinn criter.  11.00403  
Fstatistic  3.034845  DurbinWatson stat  1.992594  
Prob(Fstatistic)  0.062106  
RAMSEY REST TEST
Ramsey RESET Test  
Equation: UNTITLED  
Specification: D(INT) C D(INT(1)) D(INT(2)) D(FDI(1)) D(FDI(2)) D(CFN(  
1)) D(CFN(2)) D(EXC(1)) D(EXC(2)) D(INF(1)) D(INF(2)) INT(1)  
FDI(1) CFN(1) EXC(1) INF(1)  
Omitted Variables: Squares of fitted values  
Value  df  Probability  
tstatistic  1.866987  20  0.0766  
Fstatistic  3.485639  (1, 20)  0.0766  
Likelihood ratio  5.944304  1  0.0148  
Ftest summary:  
Sum of Sq.  df  Mean Squares  
Test SSR  242.9527  1  242.9527  
Restricted SSR  1636.973  21  77.95111  
Unrestricted SSR  1394.021  20  69.70103  
LR test summary:  
Value  
Restricted LogL  122.6099  
Unrestricted LogL  119.6378  
Unrestricted Test Equation:  
Dependent Variable: D(INT)  
Method: Least Squares  
Date: 05/30/21 Time: 09:01  
Sample: 1983 2019  
Included observations: 37  
Variable  Coefficient  Std. Error  tStatistic  Prob. 
C  41.68943  12.04737  3.460458  0.0025 
D(INT(1))  0.086384  0.201464  0.428781  0.6727 
D(INT(2))  0.509736  0.233370  2.184241  0.0410 
D(FDI(1))  1.732879  1.435367  1.207272  0.2414 
D(FDI(2))  1.349967  1.272200  1.061128  0.3013 
D(CFN(1))  0.211790  0.197096  1.074551  0.2954 
D(CFN(2))  0.035742  0.136713  0.261436  0.7964 
D(EXC(1))  12.06051  8.298856  1.453273  0.1617 
D(EXC(2))  36.57909  12.38930  2.952475  0.0079 
D(INF(1))  2.576974  1.285607  2.004480  0.0587 
D(INF(2))  1.625072  1.061181  1.531381  0.1413 
INT(1)  0.076914  0.060979  1.261327  0.2217 
FDI(1)  1.003111  1.715011  0.584901  0.5652 
CFN(1)  0.282329  0.257650  1.095787  0.2862 
EXC(1)  15.99623  4.485134  3.566499  0.0019 
INF(1)  4.768987  1.655692  2.880360  0.0093 
FITTED^2  0.044450  0.023808  1.866987  0.0766 
Rsquared  0.615046  Mean dependent var  0.338961  
Adjusted Rsquared  0.307083  S.D. dependent var  10.02949  
S.E. of regression  8.348714  Akaike info criterion  7.385825  
Sum squared resid  1394.021  Schwarz criterion  8.125977  
Log likelihood  119.6378  HannanQuinn criter.  7.646764  
Fstatistic  1.997141  DurbinWatson stat  2.124657  
Prob(Fstatistic)  0.072140  
Appendix B: Student meeting record
Appendix C: Research ethics Checked list
Appendix D: Turnitin Report
Appendix E: Student review record