Tables 1 and 2 show the descriptive statistics and correlation matrix. Tables 1a, b, and c present the descriptive statistics of all terrorized countries (model 1), very low terrorized countries (model 2), and the highest level of terrorism (model 3) in SSA. The correlation test findings for independent variables of models 1, 2, and 3 are reported in Table 2 in the form of a matrix. Given the range of absolute values (-0.0098 to 0.4193) in all the models, the study can conclude no multicollinearity issue between our explanatory variables based on the rule of thumb. These values fall below the 0.80 benchmarks (Prodan, 2013).
Table 1a
Descriptive statistics of all terrorized countries
Variables
|
Observations
|
Mean
|
St. Deviation
|
Min.
|
Max.
|
FDI
|
483
|
3.763851
|
9.001702
|
-36.56
|
98
|
MEX
|
483
|
1.905383
|
1.490026
|
.2
|
17.3
|
NTA
|
483
|
1.603115
|
4.87491
|
0
|
6.570883
|
GDPG
|
483
|
4.591843
|
3.972806
|
-36.39
|
33.63
|
MRT
|
483
|
1.297143
|
2.655839
|
0
|
20.51
|
EYS
|
483
|
9.148157
|
2.973194
|
.46
|
19.24
|
Table 1b
Descriptive statistics of very low terrorized countries
Variables
|
Observations
|
Mean
|
St. Deviation
|
Min.
|
Max.
|
FDI
|
168
|
3.991845
|
7.000932
|
-6.4
|
40.2
|
MEX
|
168
|
1.960119
|
1.677614
|
.6
|
17.3
|
NTA
|
168
|
.7872984
|
.8097533
|
0
|
3.73767
|
GDPG
|
168
|
4.109286
|
5.451109
|
-17.67
|
19.68
|
MRT
|
168
|
.5069643
|
.8437681
|
0
|
5.53
|
EYS
|
168
|
9.845298
|
2.681846
|
.46
|
19.24
|
Table 1c
Descriptive statistics of highly terrorized countries in SSA
Variables
|
Observations
|
Mean
|
St. Deviation
|
Min.
|
Max.
|
FDI
|
315
|
3.642254
|
9.91335
|
-36.56
|
40.2
|
MEX
|
315
|
1.87619
|
1.38158
|
.2
|
17.3
|
NTA
|
315
|
2.038217
|
1.696697
|
0
|
3.73767
|
GDPG
|
315
|
4.849206
|
4.526035
|
-36.39
|
19.68
|
MRT
|
315
|
1.718571
|
3.152212
|
2.13
|
5.53
|
EYS
|
315
|
8.776349
|
3.057056
|
.46
|
19.24
|
Table 2
Variables
|
FDI
|
MEX
|
NTA
|
GDPG
|
MRT
|
EYS
|
Model 1. All terrorised countries in SSA
|
FDI
|
1.0000
|
|
|
|
|
|
MEX
|
0.0692
|
1.0000
|
|
|
|
|
NTA
|
0.0652
|
0.0558
|
1.0000
|
|
|
|
GDPG
|
0.1879
|
-0.1008
|
-0.0676
|
1.0000
|
|
|
MRT
|
-0.0499
|
-0.1358
|
0.3121
|
0.0217
|
1.0000
|
|
EYS
|
0.0957
|
-0.2409
|
0.1639
|
-0.0435
|
0.0424
|
1.0000
|
Model 2. Very low terrorized countries in SSSA
|
FDI
|
1.0000
|
|
|
|
|
|
MEX
|
0.4193
|
1.0000
|
|
|
|
|
NTA
|
0.1089
|
0.2901
|
1.0000
|
|
|
|
GDPG
|
-0.1248
|
0.1309
|
-0.0345
|
1.0000
|
|
|
MRT
|
-0.1539
|
-0.1226
|
0.0916
|
-0.0899
|
1.0000
|
|
EYS
|
-0.1563
|
-0.3399
|
-0.1525
|
-0.0098
|
-0.0229
|
1.0000
|
Model 3. Highly terrorized countries in SSA
|
FDI
|
1.0000
|
|
|
|
|
|
MEX
|
-0.0879
|
1.0000
|
|
|
|
|
NTA
|
0.0736
|
0.0116
|
1.0000
|
|
|
|
GDPG
|
0.3406
|
-0.0744
|
-0.1348
|
1.0000
|
|
|
MRT
|
-0.0383
|
-0.1601
|
0.2679
|
0.0237
|
1.0000
|
|
EYS
|
0.1770
|
-0.2027
|
0.3500
|
-0.0444
|
0.1031
|
1.0000
|
Table 3 present the empirical results for model 1 (all terrorised countries), model 2 (low and very low terrorised counties), and model 3 for countries with a very high impact of terrorism. The table shows the combined Hausman test results and the related p-values of the parameters. The results established that the Hausman test fails to oppose the long-run homogeneity constraint at conventional significance levels, confirming the PMG estimation's suitability in all models. The values are greater than 0.05, and thus, the PMG is prescribed. However, the study has estimates of MG and DFE in the table, but we report only the PMG.
The long-run coefficient for military expenditure (MEX) in the PMG model has a significant negative impact on FDI in all the models. MEX is not an essential determinant of FDI in the SSA terrorised countries in the short run. The outcome suggests that the rise in military expenditure in the countries under study negatively affected FDI inflow. The study’s findings align with the neoclassical view that military expenditure crowds-out investments (Aziz & Khalid, 2017; Kolawole et al., 2018). The low inflow could be due to the diversion of economic resources (supposedly used for providing basic infrastructure that will facilitate the inflow of FDI) from the real sector of the economy to the defense sector. Moreover, others argued that high military expenditure in free-conflict countries could cause a fear of future arms conflicts, hindering foreign capital flow from abroad (Aziz & Asadullah, 2017; Aziz & Khalid, 2017). The anxiety over a safe return from investment could negate the inflow of foreign direct investment in the economy that spends hugely on defence without issues of terrorism or conflicts.
Although the terrorism coefficient (NTA) is a negative and significant determinant of FDI in the long run for models 1 and 3, the terrorist attack is not an essential determinant of FDI in the countries with low terrorist activities (model 2). In the short run, the effect of NTA on FDI is insensitive in both models. This shows that terrorist activities affect the inflow of FDI in countries with high levels of terrorism. It is in line with the expectation that terrorism will negate the inflow of FDI in the terrorised states.
Also, to examine the joint effect of military expenditure and terrorism on FDI, the study introduced the interaction of military expenditure and the number of terrorist attacks (MEX*NTA). The long-run result of the interaction term revealed that the negative effect of MEX is overturned to positive by MEX*NTA. The coefficient values showed that the interaction terms are positive and significant in models 1 and 3. But the values are irrelevant in countries with a shallow level of terrorism. This showed that amid terrorism, military expenditure could boost the confidence of foreign investors to invest in the economy. This agrees with the findings of Aziz and Asadullah (2017); Lee (2017) that military expenditure amid conflicts affected foreign direct investment positively. Li (2006); Nigh (1985) also stated that when preparing to invest abroad, foreign investors do not only consider the basic structures of the host country but are also concerned with the assurance of a safe return on investment, which led to concerns about political uncertainty.
The other explanatory variables, such as GDP growth (GDPG), mineral rent (MRT), and expected years of schooling (EYS), are statistically significant determinants of foreign direct investment in the long run for both models. This is consistent with the theory and empirical findings. For example, Mah (2010); Boateng et al., (2015); Suleiman, Kaliappan and Ismail (2015) argued that economic growth supports foreign direct investment influx. They alleged that level of income is an essential determinant of FDI inflow despite terrorism challenges. Regarding the mineral rent, Bokpin et al. (2015) Suleiman, kalliappan and Ismail (2015) reported a positive effect of mineral resources on FDI In Sub-Saharan Africa. The studies settled that mineral resources are the main factor attracting FDI inflow in the region. Also, Miningou and Tapsoba (2020) opined that education level (proxied by expected years of schooling) enhances foreign direct investment inflow. Moreover, in line with the rule of thumb, all the models' error correction terms are statistically significant, negative, and less than1. This also shows the rate of convergence adjustment to long-term equilibrium.
Table 3
Panel ARDL Regression for the Terrorized Countries in SSA.
Independent Variable
|
Model 1: All terrorized countries in SSA
|
Model 2: Very low terrorized countries in SSA
|
Model 3: Highly terrorized countries in SSA
|
MG
|
PMG
ARDL 1.2,2,1,2,1
|
DFE
|
MG
|
PMG
ARDL 1,0,2,1,2.1
|
DFE
|
MG
|
PMG
ARDL 1,2,2,1,2,1
|
DFE
|
Long-run coefficients
|
|
|
|
|
|
|
MEX
|
-2.540(-1.66)*
|
-0.221(-2.58)**
|
0.664(0.45)
|
-1.500(1.57)
|
-5.361(3.85)***
|
0.766(0.55)
|
-1.213(-0.99)
|
-1.949(-6.02)***
|
-1.076(-1.76)*
|
NTA
|
-9.436(-1.41)
|
-0.303(-2.29)**
|
2.759(2.56)*
|
-1.689(-1.49)
|
-0.218 (0.50)
|
-1.622(-1.05)
|
-7.097(-1.08)
|
− .285(3.18)***
|
-0.917(-2.11)**
|
MEX*NTA
|
1.611(-1.75)*
|
0.350(3.17)***
|
0.213(0.26)
|
3.051(2.05)**
|
1.62(1.09)
|
-0.425(-0.49)
|
6.931(1.02)
|
0.620(3.97)***
|
0.064(0.18)
|
GDPG
|
2.287(1.10)
|
0.009(2.88)***
|
0.072(0.23)
|
0.868(1.63)
|
0.621(4.90)***
|
0.043(-0.28)
|
2.042(1.85)*
|
0.139(2.27)**
|
0.368(2.35)**
|
EYS
|
2.535(0.92)
|
0.431(1.97)**
|
0.109(0.20)
|
-2.248(-1.50)
|
0.303(8.96)***
|
0.079(0.22)
|
-0.896(-0.38)
|
0.633(9.61)***
|
0.290(0.98)
|
MRT
|
-4.788(-0.80)
|
0.105(2.15)**
|
-0.465(0.63)
|
-1.460(-0.93)
|
0.349(2.98)***
|
-0.569(-0.44)
|
3.950(0.93)
|
0.227(3.85)***
|
0.038(0.15)
|
Speed of adjustment (ECT)
|
-0.572(-2.52)**
|
-0.245(-2.88)***
|
-0.282(-6.30)***
|
-0.604(-3.82)***
|
-0.288(-2.78)***
|
-0.425(-7.75)***
|
-0.607(-1.78)*
|
-0.902(-5.32)***
|
-0.974(-4.85)***
|
Short-run Coefficients
|
|
|
|
|
|
|
∆ MEX
|
-2.819(-1.76)*
|
-0.949(-1.59)
|
0.029(0.10)
|
2.349(1.82)*
|
0.196(0.48)
|
2.163(4.41)***
|
-1.033(-0.91)
|
-0.290(-0.15)
|
1.320(1.48)
|
∆ NTA
|
-0.461(-1.05)
|
-0.585(-1.47)
|
0.042(0.20)
|
1.343(1.72)*
|
0.196(0.48)
|
2.163(4.41)***
|
-1.176(-1.04)
|
-0.124(0.16)
|
0.068(0.22)
|
MEX*NTA
|
1.075(0.65)
|
0.083(0.18)
|
0.641(4.46)***
|
0.434(2.22)**
|
0.521(1.74)
|
1.305(8.87)***
|
-3.019(-1.09)
|
-1.787(-0.88)
|
0.053(0.13)
|
∆GDPG
|
0.962(1.22)
|
0.421(3.87)***
|
0.352(4.89)***
|
-0.023(-1.06)
|
-0.063(-1.40)
|
-0.017(-0.51)
|
-3.920(-0.77)
|
0.568(4.06)***
|
0.573(4.71)***
|
∆ EYS
|
0.232(0.15)
|
1.764(1.68)
|
0.412(0.64)
|
5.463(0.75)
|
4.231(0.94)
|
-0.243(-0.11)
|
1.142(0.23)
|
-1.915(-1.63)
|
-0.586(-0.48)
|
∆ MRT
|
-8.578(-0.29)
|
6.643(1.73)*
|
0.410(1.04)
|
1.912(1.04)
|
8.470(1.01)
|
-0.066(0.16)
|
-1.181(-0.95)
|
-1.384(-0.52)
|
0.252(0.66)
|
CONSTANT
|
-0.056(-0.01)
|
0.368(3.07)***
|
-0.499(-0.28)***
|
3.861(1.65)*
|
2.045(2.76)***
|
3.635(2.86)**
|
-4.190(-1.86)**
|
0.645(3.61)***
|
-0.858(.31)
|
Hausman Test
|
0.79(0.9923)
|
0.77(0.9929)
|
6.38(0.1119)
|
3.23(0.7801)
|
1.41(0.9650)
|
1.89(0.9895)
|
No. of countries
|
23
|
23
|
23
|
8
|
8
|
8
|
15
|
15
|
15
|
Observations
|
437
|
437
|
437
|
144
|
144
|
144
|
285
|
285
|
285
|
Note: The dependent variable is Foreign Direct Investment. Figures outside the parenthesis are the Z-values while those in the parenthesis are p-values. *** and ** represents 1% and 5% significant levels respectively.
|
Marginal Effect
The marginal effect of military expenditure on foreign direct investment was measured at various levels of terrorist attacks: at mean, minimum, and maximum. The mean level of models 1 (all countries) and models 3 (countries with a high level of terrorism) demonstrate a significant positive effect. The outcomes revealed a 1% change in military expenditure is associated with an increase of 0.07% in foreign direct investment inflow in all terrorised countries and the countries with a higher level of terrorism in the SSA. The mean result of the countries with low terrorism is not significant. At the median and maximum level, 1% changes in military expenditure would lead to an increase in FDI by approximately 0.06% and 0.08%, respectively, except for the countries with a low level of terrorism.
Table 4
Marginal Effect of the Interaction Terms Between Military Expenditure and The Number of Terrorist Attacks on Foreign Direct Investment
Interaction
|
Model
|
Marginal Effect
|
Coefficient
|
MEX*NTA
|
All countries
|
Mean
|
0.0665 *** (0.043)
|
Median
|
0.0572** (0.523)
|
Maximum
|
0.0756*** (0.143)
|
MEX*NTA
|
Very low terrorized countries
|
Mean
|
0.1605(0.025)
|
Median
|
-1.054 (0.034)
|
Maximum
|
0.0763 (0.261)
|
MEX*NTA
|
Highly terrorized countries
|
Mean
|
0.0681 *** (0.059)
|
Median
|
0.0567 *** (0.096)
|
Maximum
|
0.0791*** (0.072)
|
Notes: parentheses are standard errors. *, ** and *** indicates significant at 10%, 5% and 1% respectively. |
Panel Quantile Regression
Tables 5, 6, and 7 show the panel quantile regression results. The outcome confirmed that the negative impact of military expenditure differs noticeably and has a substantial effect on foreign direct investment at higher (75th and 90th percentiles). This empirical finding shows that the result is much stronger at higher quantiles. The variable specifying the number of terrorist attacks also reveals a similar pattern of strong negative effect at the higher quantile, except for the panel of countries with a very low impact of terrorism (Table 6), which is statistically not significant at different quantiles. The coefficient of the interaction term of military expenditure and number of terrorist attacks (MEX*NTA) is positive and powerful in all terrorized countries (Table 5) and the panel of countries with a high level of terrorism (Table 7). Thus, when there is increased military expenditure amid terrorism. With high military spending, terrorised countries would gain the confidence of foreign investors,' and thus, the FDI would increase (Aziz & Khalid, 2017). However, the effect differs at different quantiles, with a strong impact at the highest quantile. The impact of the interaction term (MEX*NTA) appeared insignificant at all quantiles in the countries with a very low level of terrorism in Table 6. Similarly, the expected years of schooling and mineral rents are positive and significant in all the panels. The estimated variables disclose that higher quantile levels of the coefficients have the highest values, in line with the literature, where better human and material resources promote foreign direct investment inflow.
The marginal effect, where the number of terrorist attacks is evaluated at the minimum level shows an insignificant result throughout the quantiles in all the models. At the mean level, findings reveal a significant positive effect in the 10th, 25th, 75th, and 90th quantiles in all countries and countries with a higher level of terrorism. However, the results demonstrate the most elevated significance at the higher quantiles (75th and 90th ). The mean result for countries with a low level of terrorism is insignificant. At the maximum level, the marginal effect is positive with a higher level of significance at the highest quantile (75% and 90%) in all countries and countries with a higher level of terrorism. The outcome suggest that military expenditure would enhance foreign direct investment inflow if countries faced a higher level of terrorist activity. This is in line with our expectations and findings (Aziz and Khalid, 2017; Lee, 2017).
Table 5
Panel Quantile Regression Result for all countries
Independent Variable
|
Panel Quantile Regression
|
|
Q-10
|
Q-25
|
Q-75
|
Q-90
|
MEX
|
-0.342
(-0.78)
|
-0.050
(-1.84) *
|
-0.178
(-2.19) **
|
-0.342
(-3.63) ***
|
NTA
|
-0.022
(-1.25)
|
-0.048
(-1.89) *
|
-0.205
(-2.80) ***
|
-0.241
(5.72) ***
|
MEX*NTA
|
0.009
(1.73) *
|
0.081
(2.05) **
|
0.291
(3.94) ***
|
0.568
(7.15) ***
|
GDPG
|
0.762
(5.90) ***
|
0.379
(2.69) ***
|
0.275
(3.08) ***
|
0.501
(2.83) ***
|
EYS
|
0.191
(2.32) **
|
0.157
(8.86) **
|
0.313
(8.22) ***
|
0.084
(5.05) ***
|
MRT
|
0.067
(1.83) *
|
0.084
(1.72) *
|
0.066
(1.98) **
|
0.180
(5.98) ***
|
No. of countries
|
23
|
23
|
23
|
23
|
Observations.
|
483
|
483
|
483
|
483
|
Marginal Effect
|
|
|
|
|
Mean
|
0.0765**
|
0.4302***
|
0.2245***
|
0.0323**
|
Median
|
0.0533
|
0.2522
|
0.1411
|
0.0524
|
Maximum
|
0.3113
|
0.6201*
|
0.2331**
|
0.0032**
|
Notes: The dependent variable for the panel quantile is FDI. The figures in parenthesis are the t-values. ***, ** and * represent 1%, 5% and 10% levels respectively. |
Table 6
Panel Quantile Regression Result for countries with a low level of terrorism
Independent Variable
|
Panel Quantile Regression
|
|
Q-10
|
Q-25
|
Q-75
|
Q-90
|
MEX
|
-0.342
(-1.78) *
|
-0.050
(-1.84) *
|
-0.178
(-2.19) **
|
-0.342
(-3.63) ***
|
NTA
|
0.022
(0.25)
|
-0.048
(-0.87)
|
-0.205
(-1.23)
|
-0.241
(-1.42)
|
MEX*NTA
|
0.009
(0.33)
|
0.081
(1.05)
|
0.291
(1.44)
|
0.568
(1.45)
|
GDPG
|
0.762
(2.90) ***
|
0.379
(3.69) ***
|
0.275
(3.08) ***
|
0.501
(2.83) ***
|
EYS
|
0.191
(1.82) *
|
0.157
(3.22) **
|
0.313
(3.45) ***
|
0.084
(4.05) ***
|
MRT
|
0.067
(1.78) *
|
0.084
(1.76) *
|
0.066
(2.02) **
|
0.180
(3.96) ***
|
No. of countries
|
8
|
8
|
8
|
8
|
Observations.
|
168
|
168
|
168
|
168
|
Marginal Effect
|
|
|
|
|
Mean
|
0.1023
|
0.2377
|
0.6005
|
0.3405
|
Median
|
0.0031
|
0.0352
|
0.0211
|
-0.0401
|
Maximum
|
0.4211
|
0.0621
|
0.3015*
|
0.0722
|
Notes: The dependent variable for the panel quantile is FDI. The figures in parenthesis are the t-values. ***, ** and * represent 1%, 5% and 10% levels respectively. |
Table 7
Panel Quantile Regression Result for countries with a high level of terrorism
Independent Variable
|
Panel Quantile Regression
|
|
Q-10
|
Q-25
|
Q-75
|
Q-90
|
MEX
|
-0.207
(-0.28)
|
-0.79
(-2.37) **
|
-0.72
(-2.38) **
|
-0.182
(-3.49) ***
|
NTA
|
-0.022
(-1.25)
|
-0.021
(-1.98) *
|
-0.214
(-2.43) **
|
-0.809
(3.52) ***
|
MEX*NTA
|
0.009
(1.73) *
|
0.011
(2.35) **
|
0.203
(2.75) ***
|
0.572
(5.08) ***
|
GDPG
|
0.762
(5.90) ***
|
0.126
(3.12) ***
|
0.688
(9.95) ***
|
0.015
(4.56) ***
|
EYS
|
0.191
(2.32) **
|
0.163
(2.43) **
|
0.205
(3.71) ***
|
0.106
(2.45) ***
|
MRT
|
0.067
(1.83) *
|
0.607
(1.98) **
|
0.118
(2.23) ***
|
0.634
(2.90) ***
|
No. of countries
|
15
|
15
|
15
|
15
|
Observations.
|
315
|
315
|
315
|
315
|
Marginal Effect
|
|
|
|
|
Mean
|
0.0035*
|
0.1437**
|
0.0215***
|
0.0232***
|
Median
|
0.0814
|
0.6542
|
0.1226
|
0.0016
|
Maximum
|
0.1852*
|
0.0581**
|
0.0203***
|
0.52524***
|
Notes: The dependent variable for the panel quantile is FDI. The figures in parenthesis are the t-values. ***, ** and * represent 1%, 5% and 10% levels respectively. |
Robustness Checks
While examining the robustness of the results, this study estimated different models using the deaths from terrorist attacks (DTA) as a measure of terrorism instead of terrorist attacks (NTA). The Hausman test result of the robust model also supported PMG; hence, we reported PMG results in Table 8. The table presents the PMG (NTA) comparison with PMG (DTA) for all the models. The outcome discovered that all the PMG variables (DTA) are significant and rightly signed in PMG (NTA). Most importantly, the long-run coefficients of terrorism (NTA and DTA) are substantial and similarly marked, as expected in all the models.
Table 8
Comparison of PMG (NTA) Models and PMG (DTA) Models.
Independent Variable
|
Model 1: All Countries
|
Model 2: Very low terrorized Countries
|
Model 3: Highly terrorized Countries
|
|
PMG(NTA)
ARDL 1.2,2,1,2,1
|
PMG(DTA)
ARDL 1,2,2,0,1,1
|
PMG(NTA)
ARDL 1,0,2,1,2.1
|
PMG(DTA)
ARDL 1,2,2,0,0,1
|
PMG(NTA)
ARDL 1,2,2,1,2,1
|
PMG(DTA)
ARDL 1,2,2,1,1,1
|
MEX
|
-0.221(-2.58)**
|
-0.762(-3.79)***
|
-5.361(3.85)***
|
-0.986(-1.74)*
|
-1.949(-6.02)***
|
-2.035(-7.48)***
|
NTA/DTA
|
-0.303(-3.29)***
|
-0.279(-3.24)***
|
-0.218 (0.50)
|
-0.023(-0.75)
|
-0.285(2.18)**
|
-0.514(-5.41)***
|
MEX*NTA / MEX*DTA
|
0.350(3.87)
|
0.374(5.09)***
|
1.62(1.09)
|
0.306(1.41)
|
0.620(3.17)***
|
0.827(4.49)***
|
GDPG
|
0.009(0.88)
|
1.691(1.94)*
|
0.621(4.90)***
|
0.748(2.03)**
|
0.139(1.77)*
|
0.138(2.35)
|
EYS
|
0.431(11.07)***
|
0.227(6.62)***
|
0.303(8.96)***
|
0.422(4.93)***
|
0.633(9.61)***
|
0.032(0.44)
|
MRT
|
0.005(0.15)
|
0.027(0.43)
|
0.349(2.98)***
|
0.660(5.09)***
|
0.227(3.85)***
|
0.177(3.46)***
|
Speed of adjustment (ECT)
|
-0.245(-2.88)***
|
-0.308(-3.15)***
|
-0.288(-2.78)***
|
-0.781(-2.24)**
|
-0.902(-5.32)***
|
-0.967(-5.26)***
|
Short-run Coefficients
|
|
|
|
|
|
∆ MEX
|
-0.949(-1.59)
|
-0.133(-0.32)
|
0.196(0.48)
|
-1.127(-0.68)
|
-0.290(-0.15)
|
-0.600(-0.34)
|
∆ NTA/DTA
|
-0.585(-1.47)
|
-0.357(-0.79)
|
0.196(0.48)
|
0.287(0.78)
|
-0.124(0.16)
|
2.363(1.71)*
|
MEX*NTA/MEX*DTA
|
0.083(0.18)
|
-0.135(-0.75)
|
0.521(1.74)
|
0.059(0.16)
|
-1.787(-0.88)
|
0.516(0.84)
|
∆GDPG
|
0.421(3.87)***
|
0.489(3.92)***
|
-0.063(-1.40)
|
0.044(0.92)
|
0.568(4.06)***
|
0.681(4.94)***
|
∆ EYS
|
1.764(1.68)
|
0.322(0.78)
|
4.231(0.94)
|
4.734(0.95)
|
-1.915(-1.63)
|
-1.406(-1.27)
|
∆ MRT
|
-6.643(-1.73)*
|
3.860(0.77)
|
8.470(1.01)
|
7.454(-1.00)
|
-1.384(-0.52)
|
9.418(0.83)
|
CONSTANT
|
0.368(3.07)***
|
0..293(3.91)***
|
2.045(2.76)***
|
7.221(2.39)**
|
0.645(3.61)***
|
4.895(3.38)***
|
No. of countries
|
23
|
23
|
8
|
8
|
15
|
15
|
Observations
|
414
|
414
|
152
|
152
|
285
|
285
|