Comprehensive Report: Analyzing the Relationship of FDI and Exports

This report presents an analysis of the relationship between Foreign Direct Investment (FDI) and exports, utilizing statistical methods such as confidence intervals, t-tests, and correlation and regression analyses. The study examines data from 1995, 2007, and 2016, comparing high-income (H) and low-income (L) countries to determine the impact of FDI on export performance. The analysis includes descriptive statistics, independent sample t-tests, and Pearson's correlation coefficients to assess the strength and direction of the relationship between FDI per capita, export per capita, and YSC (likely a variable related to economic output or development). Regression analysis is employed to model the influence of FDI on exports, considering the whole sample and separately for H and L countries. The results indicate a positive correlation between FDI and exports, with varying degrees of impact across different country groups and time periods, highlighting the evolving economic dynamics. The findings are presented in a series of tables, including data on confidence intervals, hypothesis testing, correlation coefficients, and regression results, providing a comprehensive overview of the study's methodology and conclusions.

1 | P a g e
The relationship between FDI and exports
Student Name: Student ID:
Unit Name: Unit ID:
Date Due: Professor Name:
1

2 | P a g e
Table of Tables
Table 1: Confidence Interval for year 2007 (Appendix table 8 and table 9).......................3
Table 2: Hypothesis testing data for years 1995, 2007, 2016 (Appendix table 10)..............4
Table 3: Correlation coefficient between FDI, EXPORT and YSC at 99% level in 2007......6
Table 4: Correlation coefficient between FDI, EXPORT and YSC at 95% level in 2007 6
Table 5: Regression result data for year 1995...........................................................8
Table 6: Regression result data for year 2007...........................................................9
Table 7: Regression result data for year 2016.........................................................10
Table 8: SPSS DATA FOR DESCRIPTIVE ANALYSIS FOR QUESTION A (95% C.I). .15
Table 9: SPSS DATA FOR DESCRIPTIVE ANALYSIS FOR QUESTION A (99% C.I). .17
Table 10: SPSS DATA FOR INDEPENDENT T-TEST FOR QUESTION B.................20
Table 11: SPSS DATA FOR CORRELTION FOR WHOLE SAMPLE- QUESTION C....22
Table 12: SPSS DATA FOR CORRELTION FOR H GROUP SAMPLE- QUESTION C
..................................................................................................................22
Table 13: SPSS DATA FOR CORRELTION FOR L GROUP SAMPLE -QUESTION C
..................................................................................................................23
Table 14: SPSS DATA FOR REGRESSION ANALYSIS FOR WHOLE SAMPLE -
QUESTION D (I) FOR 1995............................................................................23
Table 15: SPSS DATA FOR REGRESSION ANALYSIS FOR WHOLE SAMPLE -
QUESTION D (I) FOR 2007............................................................................23
Table 16: SPSS DATA FOR REGRESSION ANALYSIS FOR WHOLE SAMPLE -
QUESTION D (I) FOR 2016............................................................................24
Table 17: SPSS DATA FOR REGRESSION ANALYSIS FOR WHOLE SAMPLE -
QUESTION D (II) FOR 1995..........................................................................24
Table 18: SPSS DATA FOR REGRESSION ANALYSIS FOR WHOLE SAMPLE -
QUESTION D (II) FOR 2007..........................................................................25
Table 19: SPSS DATA FOR REGRESSION ANALYSIS FOR WHOLE SAMPLE -
QUESTION D (II) FOR 2016..........................................................................25
Table 20: SPSS DATA FOR REGRESSION ANALYSIS FOR WHOLE SAMPLE -
QUESTION D (III) FOR 1995.........................................................................26
Table 21: SPSS DATA FOR REGRESSION ANALYSIS FOR WHOLE SAMPLE -
QUESTION D (III) FOR 2007.........................................................................26
Table 22: SPSS DATA FOR REGRESSION ANALYSIS FOR WHOLE SAMPLE -
QUESTION D (III) FOR 2016.........................................................................26
Table 23: SPSS DATA FOR REGRESSION ANALYSIS FOR WHOLE SAMPLE -
QUESTION D (IV) FOR 1995.........................................................................27
Table 24: SPSS DATA FOR REGRESSION ANALYSIS FOR WHOLE SAMPLE -
QUESTION D (IV) FOR 2007.........................................................................27
Table 25: SPSS DATA FOR REGRESSION ANALYSIS FOR WHOLE SAMPLE -
QUESTION D (IV) FOR 2016.........................................................................28
Table 26: SPSS DATA FOR REGRESSION ANALYSIS FOR WHOLE SAMPLE -
QUESTION D (V) FOR 1995...........................................................................28
Table 27: SPSS DATA FOR REGRESSION ANALYSIS FOR WHOLE SAMPLE -
QUESTION D (V) FOR 2007...........................................................................29
Table 28: SPSS DATA FOR REGRESSION ANALYSIS FOR WHOLE SAMPLE -
QUESTION D (V) FOR 2016...........................................................................29
2

3 | P a g e
In the entire study the variables used with their units were, (FDI): FDI stocks (millions
US$); EX: exports (millions US$); POP: population (thousands).
ANS: A:
Table 1: Confidence Interval for year 2007 (Appendix table 8 and table 9)
FDI per capita Exports per capita
Whole sample H countries
L
countries
Whole sample H countries L countries
2007
95%
Upper
18505.8052 41587.3883 1529.7144 8297.1925 17103.6120 1321.2137
Lower 1728.4247 2974.1071 459.0664 3764.3745 8678.7850 449.7310
2007
99%
Upper
21255.9890 48300.5897 1711.0649 9040.2216 18568.3297 1468.8288
Lower -1021.7591 -3739.0942 277.7159 3021.3454 7214.0672 302.1158
The confidence interval for mean FDI and mean Export was calculated using the
formula where alpha signifies the level of significance and was the
samples mean (countries available here were taken as sample data), was the population
mean and was the sample size. Here population was taken as all the countries of the world.
The sample means for FDI per capita were 10117.115, 22280.747 and 994.390 respectively
for whole sample, H countries and L countries in 2007. The sample means for exports per
capita in 2007 for the above three strata were respectively 6030.7835, 12891.198 and
885.472.
The confidence interval (C.I) for sample mean statistically assesses the population
mean. Here the C.I was found at 5% and 1% level of significance to assess the population
3

4 | P a g e
data. The C.I gives the range within which population data will lie with a chance of 95% and
99%. The data clearly indicated that for 99% inclusion of population data the range of the C.I
increased.
The large lengths of the confidence intervals were due to the high range of the data
and level of significance. From standard deviation data of H countries (Appendix table 9) it
becomes clear that disparity of FDI per capita was very high in H countries compared to the
L countries, this also had an effect on the length of the confidence interval for the whole
sample.
Standard deviation (Appendix table 9) and C.I ranges for per capita Export in 2007
signified that disparity of export per capita was very high in H countries compared to L
countries. But there were few countries in the entire sample with very low export per capita
in 2007.
ANS: B:
In the analysis below the calculation used YSC for 2005 in the analysis for 2007 and YSC
2010 in the analysis for 2016.
Table 2: Hypothesis testing data for years 1995, 2007, 2016 (Appendix table 10)
1995 2007 2016
FDI per capita
Mean difference 5322.7852 21286.3573 36277.4488
Std error difference 2722.3931 8160.8246 13638.007
t-ratio 1.955 2.608 2.66
Exports per capita
Mean difference 4580.3125 12005.7262 10326.9783
Std error difference 690.9532 1796.6810 1191.9415
t-ratio 6.629 6.682 8.664
4

5 | P a g e
The null hypothesis was assumed as,
: There is no significant difference in mean FDI per capita and mean Exports per capita
between the H and L groups for each of the three years, 1995, 2007 and 2016.
Independent sample t-test (two tailed) was used to analyse the data and to check the validity
of the null hypothesis at 5% level of significance. This test statistic was ideal as sample size
was roughly balanced between H and L group. Here it was also assumed that that FDI per
capita and Export per capita were collected from normally distributed population.
FDI per capita:
The value of the statistic for differences in sample means between H and L groups in 1995
was 5322.7851 with significance of 0.055 whereas in 2007 it was 21286.3572 with
significance of 0.011 and in 2016 it was 36277.4488 with significance 0.010. Hence form the
calculated data it was interpreted that the difference in FDI per capita has increased between
the two groups of countries. In 2016 the difference in mean was at maximum. The values of
the t statistic indicated that the null hypothesis was rejected for 1995, 2007 and 2016.
Export per capita:
The mean difference for export per capita for the year 1995 was 4580.3125 with p value of
0.000, for 2007 the mean difference was 120005.7261 with 0.000 as significance whereas in
2016 the mean difference was 10326.9783 with 0.000 as significance. Export per capita was
higher in H countries compared to L countries in 1995 and 2007 and 2016 and the p values
were less than 0.025 indicated that the null hypothesis was rejected. The calculated t values
were all in the critical region which reflected huge disparity of exports between H and L
countries. The null hypothesis got rejected in 1995, 2007 and in 2016. The L countries were
lagging behind the H countries in Export per capita in 1995 and 2007 and more in 2016.
5

6 | P a g e
ANS: C:
Table 3: Correlation coefficient between FDI, EXPORT and YSC at 99% level in 2007
FDI per capita Exports per capita YSC
Whole sample
FDI per capita 1 0.659 0.231
Exports per capita 0.659 1 0.389
YSC 0.231 0.389 1
H countries
FDI per capita 1 0.638 0.055
Exports per capita 0.638 1 -0.177
YSC -0.177 0.055 1
L countries
FDI per capita 1 0.415 0.496
Exports per capita 0.415 1 0.307
YSC 0.496 0.307 1
Table 4: Correlation coefficient between FDI, EXPORT and YSC at 95% level in
2007
FDI per capita Exports per capita YSC
Whole sample
FDI per capita 1 0.659 0.231
Exports per capita 0.659 1 0.389
YSC 0.231 0.389 1
6

7 | P a g e
FDI per capita Exports per capita YSC
H countries
FDI per capita 1 0.638 0.055
Exports per capita 0.638 1 -0.177
YSC -0.177 0.055 1
L countries
FDI per capita 1 0.415 0.496
Exports per capita 0.415 1 0.307
YSC 0.496 0.307 1
Karl Pearson’s correlation coefficient tests the null hypothesis where population
correlation coefficient is tested based on sample correlation coefficient. It also helps to find
confidence interval for population correlation coefficient. The maximum value (+1) of
Pearson’s correlation implies perfect positive correlation between the bivariate data whereas
minimum value (-1) signifies perfect negative correlation between the bivariate data.
Bivariate data set were considered to be uncorrelated if Pearson’s correlation coefficient is
zero. Pearson’s correlation is calculated by where are the standard
deviations of two of bivariate data set and is the covariance between the given data
set.
For the whole sample, FDI per capita and export per capita exhibited high positive
correlation whereas YSC had low level of positive correlation with FDI (pc) and Export (pc).
The trend implied that the countries with high FDI per capita also had high Export per capita.
7

8 | P a g e
The H group of countries had correlation coefficient of 0.638 between FDI per capita and
export per capita whereas for L countries the coefficient was 0.415 which was less than that
of whole sample. For the L group of countries correlation coefficient between YSC and FDI
per capita was 0.496. This signified that export per capita and YSC had almost equal
correlation with FDI per capita. The H group of countries reflected a very low positive
correlation for YSC with FDI per capita where the correlation coefficient was 0.055. The
correlation coefficient values were similar for 95% and 99% confidence levels.
ANS: D:
Regression analysis is a statistical representation of the relation between a response variable
(dependent) and predictor (independent) variables. The impact of the predictor variables on
the response variable is measured in the analysis. Diagrammatic representation of bivariate
data with trend line (mean line) gives the line of best fit. Method of least squares is used to
find the line of best fit. Here log linear regression model was used for regression analysis.
Table 5: Regression result data for year 1995
Whole sample
(i)
Whole sample
(ii)
H
countries
(iii)
L
countries
(iv)
Whole sample
(v)
Intercept
Coefficient 2.291 1.791 6.652 1.648 2.397
(t-ratio) 6.972 5.712 8.295 4.348 7.134
LN_FDI per
capita
Coefficient 0.749 0.582 0.415 0.543 0.475
(t-ratio) 13.176 9.18 5.135 6.629 7.228
8

9 | P a g e
Whole sample
(i)
Whole sample
(ii)
H
countries
(iii)
L
countries
(iv)
Whole sample
(v)
YSC
Coefficient 0.199 -0.161 0.216 0.123
(t-ratio) 4.374 -2.125 4.230 2.595
Dtype
Coefficient 1.157
(t-ratio) 3.551
R2 0.719 0.781 0.5 0.696 0.816
Adjusted R2 0.714 0.775 0.463 0.679 0.808
Table 6: Regression result data for year 2007
Whole sample
(i)
Whole sample
(ii)
H
countries
(iii)
L
countries
(iv)
Whole sample
(v)
Intercept
Coefficient 1.542 1.553 7.071 1.751 2.415
(t-ratio) 3.986 4.063 6.022 3.71 5.688
LN_FDI per
capita
Coefficient 0.793 0.689 0.347 0.618 0.542
9

10 | P a g e
(t-ratio) 15.485 8.462 2.761 5.917 6.360
Whole sample
(i)
Whole sample
(ii)
H
countries
(iii)
L
countries
(iv)
Whole sample
(v)
YSC
Coefficient 0.092 -0.108 0.091 0.054
(t-ratio) 1.62 -1.253 1.444 1.011
Dtype
Coefficient
(t-ratio)
R2 0.779 0.787 0.221 0.697 0.823
Adjusted R2 0.776 0.781 0.164 0.681 0.815
Table 7: Regression result data for year 2016
Whole sample
(i)
Whole sample
(ii)
H
countries
(iii)
L
countries
(iv)
Whole sample
(v)
Intercept
Coefficient 0.460 0.617 6.959 0.809 1.993
(t-ratio) 0.910 1.328 6.044 1.216 3.813
LN_FDI per
capita
Coefficient 0.869 0.650 0.222 0.642 0.459
10

11 | P a g e
(t-ratio) 13.872 7.960 2.158 5.358 5.373
Whole sample
(i)
Whole sample
(ii)
H
countries
(iii)
L
countries
(iv)
Whole sample
(v)
YSC
Coefficient 0.182 0.002 0.127 0.131
(t-ratio) 3.752 0.030 2.302 2.929
Dtype
Coefficient 1.313
(t-ratio) 4.305
R2 0.739 0.784 0.152 0.662 0.832
Adjusted R2 0.735 0.778 0.089 0.643 0.824
I. The dependent variable was taken as ln(Export per capita) whereas independent variables
were taken as ln(FDI per capita) for the years 1995, 2007 and 2016. The second column
of the tables (table 4 - 6) under the heading ‘whole sample (i)’ has the regression analysis
data for three years. Trend lines for the year 1995, 2007 and 2016 were respectively
,
and
.
The initial value for the dependent variable ln(Export per capita) for 1995 was 2.291
with t value of 6.972 where p value was 0.000 which was less than 0.025 (two tailed) and
11

12 | P a g e
hence the hypothesis of equality in mean ln(Export per capita) and mean ln(FDI per
capita) was rejected. The coefficient of the predictor ln(FDI per capita) variable was
0.749 which implied that for one unit increase in (FDI per capita) the dependent variable
increased with a factor of 0.749 under significance 0.000 and t value of 13.176. The
adjusted R square of 0.719 implied that 71.9% variance of the dependent variable was
explained by independent variable. The dependency on the independent variable
increased in 2007 as correlation coefficient of ln(FDI per capita) increased to 0.793 with
p value of 0.000. The constant term reduced to 1.542 and was significant where the p
value was 0.000. This trend followed in 2016 as constant reduced to 0.46 with a p value
of 0.366 which was highly insignificant. The coefficient of ln(FDI per capita) 0.869 with
t value of 13.872 and p value of 0.000 was highly significant and implied the increasing
dependency of ln(Export per capita) on the independent variable. In 2007 the
independent variable explained 77.6% variance of the dependent ln(Export per capita)
which was evident from adjusted R square.
II. The regression analysis was done for dependent variable ln(EXpc) and independent
variables ln(FDIpc) and YSC for all the years. The regression lines for 1995, 2007
and 2016 were ,
and
.
The constant term was 1.791 with p value of zero which was the initial value of
ln(EXpc) at origin (ln(FDIpc) =0). Coefficient of 0.582 and 0.199 for the independent
variables ln(FDIpc) and YSC was noticed in 1995. The independent variable
explained 77.5% of variance of the dependent ln(EXpc) which was 78.1% in 2007.
The dependency on ln(FDIpc) increased in 2007 with a coefficient of 0.689 and
12