Political violence: why conflicts can result from sub-Saharan African socioeconomic conditions

Abstract

Purpose

This study investigates the extent to which, and in what ways and capacities, the incidence of adverse economic conditions burdening the masses, on the macro-level, increases the propensity for the generation of political instability/violence.

Design/methodology/approach

Drawing on data from a cross-section of 25 Sub-Saharan African (SSA) countries for the period 2005–2019, fixed effects (FE) and generalized method of moments (GMM) estimations are used to determine the nature and significance of the independent variable (economic condition), complemented by three control variables, on the dependent WGI-defined variable political stability scored on the basis of a continuum from −2.5 (most unstable) to +2.5 (most stable). For the link between political instability and socioeconomic conditions, the study employs a construct derived from frustration-aggression and relative deprivation theory.

Findings

The study links socioeconomic adversity to political instability in the context of SSA. In addition, larger populated countries exhibit a greater propensity to political instability than smaller populated countries. In contrast, foreign direct investment (FDI) appears to have no real effect, positive or negative, on political stability.

Practical implications

Poor living conditions seem to be strongly associated with a high risk of political violence in SSA. To buoy socioeconomic status, poverty alleviation needs be elevated into a key initiative in the decision-making agenda, at all levels of governance, with real targeted strides achieved in terms of enhancement of the standard of living of the masses. In addition, policies that control population need to be inaugurated hand-in-hand with welfare measures and a more equitable balancing of the distribution of resources in the society.

Originality/value

Given the high regional incidence of civil strife and violence, combined with a dearth of research of an empirical nature on political risk in SSA, this study provides a largely ignored and useful context on SSA apart from studies on the incidence of violence that consider the developing countries as a monolithic whole.

Citation

Fagbemi, F. and Fajingbesi, A. (2022), "Political violence: why conflicts can result from sub-Saharan African socioeconomic conditions", Journal of Business and Socio-economic Development, Vol. 2 No. 2, pp. 153-164. https://doi.org/10.1108/JBSED-12-2021-0178

1. Introduction

Adverse socioeconomic conditions, a major threat to political stability and a drag on sustainable development, serve as a causal factor of political violence/conflict in modern society in both developed and developing nations (Coccia, 2018). Adverse socioeconomic conditions stem from the absence of social justice and manifest in terms of an uneven distribution of economic resources to the detriment of the standard of living of the masses (Piazza and Von Hippel, 2014). The “security first” view represents a significant approach in the containment of political strife that often is counter-productive: generally engendering an exacerbation rather than an amelioration of civil strife with violence spiraling out of control in conflict-burdened states. Notwithstanding an important move in mainstream literature on political development away from “security first,” any initiatives that fail to address structural issues (such as distributional injustice and inability to meet basic needs) that lie at the heart of violence in many developing regions make an attempt to have a conflict-free society a Sisyphean task. Unassuaged socioeconomic discontent usually results in protracted violence consonant with the “rooted-in-poverty hypothesis” (Piazza, 2006). Hence, without an elaborate explanation of what provokes or triggers socioeconomic discontent, any exiguous improvements in political stability in sub-Saharan Africa (SSA) would almost certainly prove ephemeral.

Social conflict is seen as response by oppressed or down-trodden segments of the population to perceived social injustices and inequalities, poverty, unemployment and underdevelopment usually attributable to governmental ineffectiveness. Spiraling social conflict can shatter fragile political systems. Such violence engendering political instability can be viewed from the perspective of relative deprivation theory: the tension that arises from a difference between the “ought to” and “is” in terms of collective value satisfaction – a gap that precipitates violent acts (Gurr, 1971). In SSA, poverty is pervasive due to governance deficiencies attributable to incompetence, mismanagement and corruption across countries in the region (Mbaku, 2020). Unfortunately, policy-makers pay scarce heed to numerous arguments in the literature on political development that posit adverse living conditions as being a significant driver of political instability (Richardson, 2011; Azeng and Yogo, 2013; Břeň et al., 2019). Ensuing social cleavage opens up centrifugal fault lines tearing up social cohesion and the body-politic in SSA which leads to spiraling violence that cannot be effectively suppressed, in the long-term, by policing or military means. Unfortunately, it would appear that socioeconomic polarization characterizes the process of political development in SSA. Intense civil strife, spilling over into internal armed conflict, have, for the most part, eluded peace initiatives. Arguably, however, ensuring a violence-free society is a condition precedent for achieving sustainable development goals (SDGs). For, the stability of the polity serves as the essential component of modern development.

It would be inappropriate to overlook the role that socioeconomic inequality plays in exacerbating tensions and conflicts in vulnerable and weak economic regions, populated by marginalized elements of society hailing from least developed countries in SSA, where a medley of adverse socioeconomic factors reinforce and perpetuate violence (Willett, 2001). However, empirical evidence is needed to substantiate the impact of socioeconomic conditions on political stability in the context of SSA in addition to gaining fruitful insights into how amelioration of socioeconomic conditions could serve to minimize outbreaks of violence in SSA. To suggest, however, that civil strife uniformly thrives and proliferates in the absence of high standards of living would be a gross over-simplification. In this vein, identification of predictors of civil strife and armed insurgencies warrant further research to permit the assessment of how these predictors interact with a view to ascertaining differentiation across regions in terms of manifestations of patterns of violence (Krueger and Malečková, 2009).

Some previous studies have argued that undemocratic and socially polarized environments (Miguel, 2007), youth unemployment (Urdal, 2012; Azeng and Yogo, 2013), as well as poverty and declining income (Piazza and Von Hippel, 2014) give rise to civil strife in developing countries. However, extant literature typically treats developing countries as a whole, rather than focusing on intra-regional dynamics in specific geographical area like SSA. Such contextual focus is especially warranted with respect to SSA inasmuch as acute fragility characterizes political stability in most African countries with increased incidents of political instability/violence particularly pronounced in SSA (Fagbemi and Asongu, 2021). A comprehensive overview of the socioeconomic drivers of political unrest in SSA would help policy-makers (especially governments) formulate policies designed to combat the underlying causes rather than outward manifestations of civil strife.

In this vein, this present study expands upon insights derived from the literature with a view to fine-tuning the impact of adverse socioeconomic conditions on political instability in SSA. A central goal of the present study is to test the hypothesis through analysis of empirical evidence that low living standards are associated directly with the high incidence of political instability in SSA. Income per capita (purchasing power parity (PPP)) is used as an indicator of socioeconomic level (living standards) in a cross-section of 25 SSA countries between 2005 and 2019. The methodological approach consists of a combination of ordinary least square (OLS) fixed effects (FEs) and the two-step system generalized method of moments (GMM) estimations. These techniques offer the prospect of useful insights into causation inasmuch as reverse causality can be garnered through the instrumental variable procedure.

The rest of the paper is structured as follows: In section two, a literature review is presented followed by an overview of the research methodology in section three; section four presents the empirical results, while the final section (section five) offers concluding remarks and recommendations.

2. Literature review

Notwithstanding that the predominant focus of the literature in political economy and development is on how political instability adversely affects the economic performance of a country (Fosu, 2003; Aisen and Veiga, 2006), several studies clearly link economic conditions as a salient causal factor of political instability: the more abject the former, the greater the scale of the latter (Aisen and Veiga, 2011). In Addy et al. (2021), informed by the crisis management strategic approach of Driscoll and Kraay, the main thrust of the research focused on analyzing how political persecution could retard economic and social development in a cross-section of countries of SSA over the period 2000–2014. In passing, however, these scholars, using a GMM methodology for parametric estimation, linked skewed distribution of wealth (captured by GINI coefficients), illegal narcotics usage, high unemployment rates and homelessness to violence.

In contrast to that linking of economic performance to political stability, collection of empirical evidence on the determinants of political instability has not been less de rigueur in developmental literature. Analyzing data drawn from a cross-section of Colombian provinces between 2000 and 2014, using a methodology later mimicked by Addy et al. (2021), Poveda and Carvajal (2019) likewise demonstrated that skewed distribution of wealth (captured by GINI coefficients), narcotics trading, high unemployment rates and displacement of population generate violence.

Some scholars link persistently high youth unemployment with incidence of violence. Urdal (2012) maintains that the risk of political instability is high if young job-seekers, left with no other alternative but unemployment and poor living conditions, channel their frustration into political violence as an outlet for perceived marginalization. In this context, if joining a rebel group offers an alternative avenue for self-fulfillment in place of income generation, large segments of the youth populations, drawn from generation-Y and generation-Z, may exhibit a high propensity to engage in terrorist activities in place of a conflict-free political orientation. Rebellion is more widespread in a society that fails to address the issue of youth unemployment and poverty (Colino, 2012). In this view, falling income is the key precursor to armed uprisings in the state (Miguel, 2007).

Furthermore, Azeng and Yogo (2013) examine the impact of youth unemployment on political instability in 24 developing countries between 1980 and 2010. Output from FEs regression with instrumental variables revealed that youth unemployment engenders political instability. In particular, extremely high youth unemployment rates, coupled with socioeconomic inequalities and pervasive corruption, precurse political instability and insecurity.

According to the UN General Assembly, social injustice, poverty, income inequality and underdevelopment ratchet up the risk of armed conflict and terrorism (United Nations, 2016). Břeň et al. (2019) argue that social factors (such as educational level, standard of living, social inequality), economic factors (such as GDP, inflation and unemployment) and security-political factors (such as crime rates, incidence of corruption and government ineffectiveness) results in political instability that generates armed conflict and terrorism. Regressing these social, economic and political-security factors (independent variables) against incidence of terrorism (the dependent variable) using data from 162 countries in the year 2017, these researchers show that socioeconomic factors are most explanatory of the incidence of terrorism. Although not causal, interestingly, Burgoon (2006) observes a positive correlation between terrorism and population density.

Tahir (2020) examined the causes of terrorism in 94 countries between 2005 and 2016. Findings indicate that low per capita income and political instability significantly contribute to occurrence of terrorism. Thus, it is suggested that, in order to rein in efficiently manifestations of violence, governments, beset by substantial incidence of terrorism, should adopt policies designed to increase the income of the population rather than merely employing measures enhancing political stability. Focusing exclusively on Nigeria, Isife (2020), while showing how ethnic and religious cleavage, military intervention in politics and poor governance created ideal conditions for political instability to flourish, focuses on socioeconomic hardship as the key driver of internecine conflict. Zeroing in on the incidence of terrorism in the Sahel, Coccia (2018) isolates income inequality, in tandem with high population growth rates, as one of two key determinants, along with what he dubs as “psycho-social risk factors,” that explain the rising incidence of terrorism in the region. Likewise, Fagbemi et al. (2021) confirm that adverse socioeconomic conditions strongly drive political instability/violence.

That the causes of political instability have not been accorded much scholarly attention in the literature, in the specific context of SSA in particular, warrants an exploration in depth as to the causal bases associated with the high risk of civil strife in the region. Clear identification of the factors that influence the occurrence of violence and instability, particularly in SSA, would offer policy-makers a pathway to undertake meaningful reforms designed to mitigate political risk of endemic violence in the region.

3. Research methodology

3.1 Theoretical framework

Pioneered by the sociologists Betty A. Nesvold, Feierabend L. K, James Cavies and John Dollard (Ademola, 2006), frustration/aggression theory, abstracting identity group mobilization and conflict resulting from the inability to achieve certain human development needs, represents the theoretical linchpin of this study. According to Azar (1990), social conflicts arise as a consequence of the failure of social groups or classes to attain basic material expectations. Frustration/aggression theory is premised on the notion that acts of violent are manifestations, in violent form, of unattained socioeconomic development on the personal level by all individuals across a stratum or strata of society. In this context, if the expected gratification was not met, aggressive tendencies would build up within that stratum or strata across society (Ademola, 2006). Using insights derived from the cognate theory of relative deprivation, this stratum or strata coalesces into an “out-group” that strikes at the “in-group” perceived as the source of the frustration with violence often directed at governmental authorities that are perceived by this “out-group” as sanctioning the status quo in which a large mass of people are effectively dispossessed of economic and political rights or at least perceive themselves to be so (Gurr, 1971). In the logic of this theory, aggressiveness displayed by the perceived “out-group” against the “in-group” functions as a cause-effect phenomenon: frustration “causes” aggression (albeit not all aggression is caused by frustration and not all frustration necessarily results in violence: it is an issue of propensity rather than determinism).

When there is a discrepancy between what people seek but have not attained on the one hand and what seemed attainable on the other, the tendency to transfer such frustration into violence indubitably escalates. The greater the frustration experienced in the process of goal attainment, the higher the propensity for aggression. In line with this, widespread disaffection/injustice within a society often engenders political instability that serves as an enabling environment for violence/terrorism in a state.

3.2 Econometric model and techniques

The functional relationship between socioeconomic conditions and political stability is formed based on the theoretical argument inspired by Gurr (1971) and Azar (1990) in relation to frustration/aggression and relative deprivation theory, such theory can be elaborated mathematically as,

(1)POLT=f(ICOM, X)

Here, POLT denotes the indicator of political stability, while ICOM represents the socioeconomic condition indicator (income per capita). X covers other factors that could possibly influence political stability.

Hence, the effect of socioeconomic conditions on political stability is examined econometrically through the following model

(2)POLTit= C+ δICOMit+ X′+μi+ εit

Eq. (2) summarizes the normal OLS FEs estimation procedure. Furthermore, the below equations in level (3) and first difference (4) represent the summary of the standard system of the GMM estimation approach.

(3)POLTit= C+ δICOMit−τ+ ∑h=13γhXh, it−τ+πi+μi+εit

(4)POLTit− POLTit−τ=δ(ICOMit−τ− ICOMit−2τ)+∑h=13γh(Xh, it−τ−Xh,it−2τ) +(μi−μi−τ)+εit−τ

where, POLTit is the level of political stability in country i at period  t, C is a constant, X is a vector of control variables (population, foreign direct investment (FDI), net inflows (percentage of GDP) and GDP growth), τ denotes the estimate of auto-regression which is one for the specification, πi indicates the country-specific effect, μi is the time-specific constant and εit represents the error term.

Given the linear nature of Eq. (2), the output can be estimated with the use of OLS. However, the main disadvantage with OLS is that results can be biased if income per capita PPP is found to be correlated with the unobserved component of the indicator of political stability. In this case, the estimated outcome could be spurious. As a result, the true impact may be underestimated. To deal with the problem of endogeneity bias, the two-step system GMM estimator is adopted. The efficiency gains of GMM estimator are obtained through (1) optimal weighting matrix, (2) a Sargan test (for overidentifying restrictions of the model) and (3) a relaxation of the assumption of independent and identically distributed. The estimate that is mostly essential to note is δ, and it is expected to be positive. For, system GMM instances are characterized by small “T” but large “N” panels (i.e. where cross-sectional observations exceed those of the time period as in the case in this study).

Some preliminary overview of the data is in order. In the model, population, FDI, net inflows (percentage of GDP), and income per capita PPP are in logarithm form, while the indicator of political stability and GDP growth are not as they are given in rates. The study employs raw panel data.

3.3 Data description and sources

In the study, political stability (POLT) encapsulates the view of the public concerning the stability of the political system in the countries, whereas income per capita (ICOM) reflects living conditions of a given population. Inclusion of the control variables incorporated in the model is justified by the literature:

1. population (POP) has been proved to have an influence on the level of political instability in a country (Burgoon, 2006; Urdal, 2012; and Coccia, 2018);

2. GDP growth (annual %) explains the level of the economic performance in the economy

3. FDI, measured in terms of net inflows as a percentage of GDP, impacts human capital development (Fagbemi and Osinubi, 2020).

In alphabetical order, the 25 sub-Saharan African countries included in the study utilizing data over the period 2005–2019 consist of: Angola, Botswana, Burkina Faso, Cameroon, Congo, Congo DR, Cote d'lvoire, Gabon, Gambia, Ghana, Guinea, Guinea Bissau, Kenya, Liberia, Malawi, Mali, Mozambique, Niger, Nigeria, Senegal, Sierra Leone, South Africa, Tanzania, Togo and Uganda. The description and sources of data are specifically presented in Table 1.

4. Empirical results and discussion

4.1 Descriptive statistics

In Table 2, respective features of the series are reported. It is indicated that mean income per capita equals 3779.61 associated with a political stability value of −0.99. Maximum and minimum values of political stability are 1.10 and −2.18 respectively. Income per capita spans $18058.06 (maximum) and $518.84 (minimum) annually. The total number of observations is 353. Regarding correlation analysis in Table 3, population and, to a greater degree, income per capita have a direct relationship with political stability to a statistically significant extent. As for other variables, FDI, to a scale less than that for population, is also directly associated with political stability (albeit not to a statistically significant extent) whereas GDP appears nearly uncorrelated with political stability.

4.2 Panel unit root and diagnostic test

Prior to the main analysis, unit root tests were conducted to know the order of integration of the used variables in the model. Given that Levin and Lin (LL) presupposes balanced panel data before it can be considered appropriate in any study, since it is less restrictive and more suitable, Im, Pesaran and Shin (IPS) is employed. In addition, in relation to Maddala and Wu (1999), ADF, Fisher and Phillips–Perron Fisher form of unit root test was carried out revealing an order of integration among the series of I (0) and I (1). Although, arguably, not needed given the use of GMM, the results of the test of unit root are nonetheless detailed in Appendix (See Table A1) to provide further insight into the stalwartness of the variables. For robustness checking, various diagnostic tests conducted attest to the validity, efficiency and reliability of the estimated outcomes and they certainly suffice as points of reference for subsequent research related to political violence. These tests appear in the lower part of Table 4.

4.3 OLS fixed effects and dynamic two-step GMM estimation

Table 4 reports the study's main results. The aim of this section is to test the assertion, with support from the literature (Isife, 2020; Tahir, 2020), that socioeconomic conditions of the population (as proxied by income per capita) vary inversely with the stability of the polity in a given economy. A valid comparison between the FEs and system GMM estimations is crucial if a lower bound for the impact of income per capita on political stability is to be identified. The FE estimates may suffer from simultaneity in bias with the effect of rendering the estimated coefficients smaller (in absolute terms) than that would otherwise be the case. In practice, GMM results, where sufficiently girded by strong instruments, exhibit no simultaneity in bias and, accordingly, provide credible confirmation of the inverse relationship between socioeconomic condition and political stability. Results are readily comparable. The analysis suggests that magnitude of effects of income per capita on the polity varies across procedures employed (FE and GMM), but results indicate the same sign. This implies that a low level of income buoys the probability of incidents of civil conflicts. Findings indeed support the detrimental effect of poverty on political stability. At least with respect to the cross-section of SSA countries surveyed in this study, estimates generated by this analysis validate the view that fragile socioeconomic conditions underlie social and political unrest consonant with conclusions drawn by Miguel (2007) and Břeň et al. (2019). It could be adduced that unmet societal expectations regarding the distribution of economic opportunity and the potential for socioeconomic advancement can stimulate civil unrest from perceived social injustices raising the inference that the occurrence of political violence can be largely attributed to the absence of social justice and poorly distributed economic resources in most SSA countries.

Another important dynamic borne out by the contemporaneous relationship between the key variables of interest in the analysis is that political violence spawns further political violence: the level of political stability in the preceding period is critical to explaining the level of political stability in the current period.

Among the control variables employed, population is found to affect political stability adversely, to a statistically significant extent, supporting the notion that the greater the number of people left with nothing more than prospects of unemployment and poverty, the higher the risk of political violence (Urdal, 2012). Inasmuch as there is a well-known inverse relationship between rate of population growth and socioeconomic level, the overall high growth rate of populations would exacerbate the inverse relationship between economic condition of the population and political instability – increasing the already high risk of political violence. Of the remaining two variables (GDP and FDI), only the former indicator manifests a significant relationship with political stability: with the larger the GDP, the smaller the risk of political instability. Most likely, the significance of GDP reflects the considerable potential of larger economies to outperform smaller economies in terms of generation of economic opportunities from which large segments of the marginalized work-force could stand to benefit – in turn, enhancing political stability.

In contrast, notably, FDI has insignificant coefficients across regressions. This may be the result of insufficient levels of FDI on a scale large enough to generate substantial employment opportunities. Alternatively, FDI may be misdirected toward capital-intensive extractive industries that hold little or inferior employment opportunities and, finally, the stimulus of FDI may have been diluted or siphoned off by high transaction costs in the guise of corruption diverting public monies to the private bank accounts of a well-placed few in government.

5. Concluding remarks

An understanding of factors that cause political instability has recently attracted widespread scholarly interest. In contrast with the trajectory of most discourses on the causes of political violence that considers developing countries as a whole, however, this study, employing OLS FEs and dynamic two-step GMM estimation, analyzes the impact of socioeconomic conditions on political instability in SSA. With a view to contributing to that debate in the context of SSA. This debate holds significance not only on an academic level but also in terms of governmental policy-making.

In view of the findings, it is possible to argue that changes in policy could be triggered by socioeconomic shocks in the form of inflammatory reactions to unmet socioeconomic aspirations by the masses who, are denied effective recourse to the ballot-box, perceive their government at fault and seek, through violence, to institute change. In this guise, socioeconomic conditions appear to play a significant role in the level of political stability. Contextually, if low income levels exacerbate political instability, poor living conditions, therefore, collocate strongly with the high risk of civil strife, characterized by varying degrees of violence, in SSA. Abject socioeconomic living conditions create critical political instability risk. Conversely, however, improvements in living conditions, ceteris paribus, tend to strengthen the political stability; however, high population growth rates undercut these benefits. Hence, apart from achieving sustainable growth, policy-makers ought not to ignore incentives designed to facilitate meaningful population control.

In reflecting on the vulnerability of political systems to socioeconomic hardship borne by the masses, policy-makers ought to recognize and to be acutely aware that programs that improved social welfare strengthen the effective functioning of governments by creating enabling environments for political stability. Although challenging, policy-makers need to address – and redress – the uneven distribution of economic resources as a strategy to palliate the suffering and marginalization caused by that imbalance. Thus, the study suggests that promoting equity, transparency and accountability in the distribution of common wealth is critical to political stability. In addition, to ensure the sustainability of stable political systems, strategic efforts should be made toward institutionalizing good governance structures across all levels. Essentially, for realization of long-term political stability, impactful policies enhancing the social and economic condition of the masses need to be elevated to the forefront of political agendas by governments with real, not fanciful or cosmetic, change instituted.

With respect to this study, while the analysis is solid with the conclusions drawn from it are robust, it is important to keep in mind some limitations that might be addressed in future studies. Obviously, economic growth is only one indicator of socioeconomic well-being which ought to be encapsulated by a vector of indicators. In addition, the cross-section of countries included in the dataset might be expanded and it might behoove to elicit variations in intra-SSA variable dynamics (e.g. Islamic versus non-Islamic countries, democratic versus non-democratic countries) in terms of propensity for instability and violence stemming from this complex refinement in the articulation of socioeconomic well-being. Finally, additional estimation techniques such as panel ARDL could offer additional insights into the discourse.

Table 1

Description and sources of data

Variable	Code	Description and measurement	Source
Political stability	POLT	It measures perceptions of the likelihood of political instability and/or politically-motivated violence, including terrorism. It is rated on a scale as appropriate −2.5 to +2.5	World Governance Indicators (http://info.worldbank.org/governance/wgi/)
Income per capita (PPP)	ICOM	It is gross national income in purchasing power parity (PPP) divided by mid-year population	World development indicator World Bank (2020)
GDP growth (annual %)	GDP	It is the annual percentage growth rate of GDP at market prices based on constant local currency. Aggregates are based on constant 2015 US  dollars	World development indicator World Bank (2020)
Population	POP	It is based on the de facto definition of population, which counts all residents regardless of legal status or citizenship. The values shown are mid-year estimates	World development indicator World Bank (2020)
Foreign direct investment (FDI), net inflows (% of GDP)	FDI	Representing net inflows—new investment inflows less disinvestment—in the reporting economy from external (foreign) investors	World development indicator World Bank (2020)

Table 2

Summary statistics

	INCOM	GDP	FDI	POP	POLT
Mean	3779.61	4.75	5.46	26,091,609	−0.99
Median	2134.52	5.30	2.63	17,114,770	−1.06
Maximum	18058.06	20.72	103.34	1.96E+08	1.10
Minimum	518.84	−20.60	−6.37	1344931.00	−2.18
Std. Dev	4045.10	3.77	10.92	33,615,070	1.06
Skewness	1.90	−1.21	5.50	3.03	0.36
Kurtosis	5.51	10.47	40.16	13.35	1.90
Jarque–Bera	306.34	908.05	22098.52	2114.86	25.61
Probability	0.00	0.00	0.00	0.00	0.00
Sum	1334203.00	1677.66	1928.66	9.21E+09	−347.59
Sum Sq. Dev.	5.76 E+09	4990.20	41991.21	3.98E+17	397.07
Observations	353	353	353	353	353

Table 3

Correlation matrix

Variable	INCOM	GDP	FDI	POP	POLT
INCOM	1.00
GDP	0.15**	1.00
FDI	0.13**	0.05	1.00
POP	0.21	0.15	−0.14**	1.00
POLT	0.52***	0.01	0.08	0.11**	1.00

Note(s): ** and *** represent the level of significance at 5 and 1%, respectively

Table 4

OLS fixed effects and dynamic two – step system GMM

	OLS fixed effects	GMM estimation
POLT (lag)	–	0.11*** [12.71]
ICOM	0.11*** [4.20]	0.31** [2.98]
GDP	0.13** [3.08]	0.20* [2.32]
FDI	0.11 [1.31]	0.21 [0.13]
POP	−0.41** [-3.04]	−0.02* [-2.19]
Constant	0.12** [3.12]	0.15* [2.23]
R-square	0.66
Observations	353	289
No. of countries	25	25
No. of instruments	–	23
Diagnostic test
Huasman test (p-value)	0.03
Pesaran CD (p-value)	0.27
Heteroscedasticity (p-value)	0.13
Hansen OIR (p-value)		0.28
A – Bond AR(1) (p-value)		0.03
A – Bond AR(2) (p-value)		0.14
Sargan test (p-value)		0.21

Note(s): *,** and *** represent the level of significance at 10%, 5 and 1%, respectively

Table A1

Panel unit root test

Variable	Im, Pesaran and Shin (W-stat)	ADF Fisher (Choi Z-stat)	Phillips – Perron Fisher (Choi Z-stat)
POLT
Level	−11.51***	1.79	2.91
First difference	–	−13.34***	−12.82***
ICOM
Level	−1.41	−1.10	−0.63
First difference	−6.90***	−6.72***	−7.90***
GDP
Level	−8.16***	−6.19***	−7.36***
First difference	–	–	–
FDI
Level	1.32	1.33	2.16
First difference	−11.36***	−13.38***	−12.05***
POP
Level	1.29	2.80	1.67
First difference	−14.01**	−11.58***	−13.69***

Note(s): ADF indicates augmented Dickey – Fuller. ** and *** represent the level of significance at 5 and 1%, respectively. “–” represents the absence of the first difference