Youth unemployment and political instability: evidence from IGAD member countries

Abstract

Evidence show that IGAD member countries are the most political unstable countries in the world. On the other hand, literatures on the subject reveal that youth unemployment contributes the most towards the political instability across the world. Nonetheless, investigation on the effect of youth unemployment on political instability particularly in IGAD member countries is very scanty. Thus, the objective of the current study is to investigate the effect of youth unemployment on political instability in IGAD member countries. For this purpose, the necessary secondary data is collected from ICRG, WDI and ILO on five selected IGAD member countries. To find out the effect of youth unemployment on political instability; fixed effect model, instrumental variable fixed effect model and one step system GMM estimation on dynamic panel data have been employed. The analysis result revealed that there is a significant effect of youth unemployment on political instability in IGAD member countries. This region specifically needs a sound youth employment policy not only for the sake of youths but also for the relief of government reducing the burden of controlling continuous internal instabilities. Moreover, in region-wise the IGAD countries better to have common youth employment creation policies so that they can manage the internal conflicts arises here and then.

Keywords:

• Youth Unemployment
• Political instability
• IGAD countries

PUBLIC INTEREST STATEMENT

Youth unemployment is a big macroeconomic problem in the world. In addition, political instability has been a cause of concern for many countries around the world and a headache for government irrespective of the state of development or their political regime. A surging youth population combined with unemployment and other factors can lead to violence. Therefore, focusing on the effect of youth unemployment on political instability may have many benefits for the region.

1. Introduction

Unemployment is a worldwide phenomenon and it is not a new issue to discuss, but it is still a serious problem that affects the economic, political, and social performances of nations throughout the world (Azeng & Yogo, 2015). In IGAD¹ countries, despite the recent economic growth rates and positive activities recorded in education and health; a higher rate of youth unemployment and the slow pace with which new jobs are created remain critical challenges of the region (Ahmed, 2017).

In most developing countries, a widespread unemployment among young is often a key reason for political and social movements; these economies are mostly characterized by continuing regional conflicts and difficult political transitions (Bernards & ILO, 2016). IGAD countries are highly characterized by political, social, and economic instabilities, and unstable activities, civil wars, and coup d’etat are their manifestation. The problem is a fifty years of protracted, chronic, and complex conflicts, these are society–state, state–state and society–society conflicts reflecting interracial, inter-religious, inter-occupational tensions (Murad & Alshyab, 2019).

Political, social, and economic instabilities are the main characteristics of countries in the IGAD region. All member countries of the region have struggled with political instability and challenges to the regimes. Like many African states, most peace and security problems in the region emanate from internal problems such as the nature of states and political parties, religious and ethnic tensions, and income inequalities as well as external interferences (Bruce, 2016). Currently, these eight nations of the region are the first ones in the list of the most unstable countries in the world, which shows how much the problem is severe in the region than other parts of the world. Indeed, many African states in conflict are strong in the wrong functions of state, effective only in the maintenance of regime security and safeguarding the interests of political parties and colluding individuals or groups (Haider et al., 2011).

The study by Urdal and Hoelscher (2012) that has been done to conduct the causes of political instability and which is a study on youth unemployment and political violence found that the presence of youth unemployment increases the risk of conflict. Researchers show that large youth populations with youth—unemployment are sometimes linked to outbreaks of violence (Amirali Asha, 2019; Azeng & Yogo, 2015; Christophe et al., 2010; Prince et al., 2018).

On top of these, youth unemployment is a big macroeconomic problem in IGAD countries. Also, the region has eight countries that are highly characterized by civil wars, economic, social, and political instabilities, almost all nations of the region are included in the list of the world top unstable nations. At the same time, two-third of the total population is below the age of 25, and unemployment is the main macroeconomic problem in the region. As such investigating the effect of youth unemployment on political instability is crucial to guide the effort to address political instability in the region. Despite the effort to address the subject youth unemployment and political instability separately efforts made to examine the effect of youth unemployment and political instability in IGAD countries is very scanty. This study analyses the effect of youth unemployment on the political instability of five IGAD countries, for which there is no much research is done jointly, using the 27 years (1992–2018) data by employing a system GMM model. For this purpose, the necessary secondary data is collected from ICRG, WDI and ILO on five selected IGAD member countries. This study adds to the existing empirical literature and lead to a conclusive decision on the impact of youth unemployment on political instability. The main objective of this study is to investigate the effect of youth unemployment on political instability in IGAD countries and to examine the trend of youth unemployment in IGAD countries.

The rest of the paper is organized as follows. The paper presents a review of literature, which includes theoretical literature and empirical literature. Then, it includes the description of the study area, data source, methodology of the study, and model specification. After that, it presents the final results, and it provides empirical analysis and discusses the findings. Finally, it concludes the study with the main findings and recommendations.s

2. Overview of youth unemployment

2.1. Youth unemployment in IGAD countries

Due to a shortage of data, the thesis analyzed selected youth unemployment rates for selected IGAD countries (Ethiopia, Kenya, Uganda, Sudan, and Somalia). Therefore, we have seen the trend of youth unemployment in selected IGAD African countries.

From the above figure, we can simply determine that the youth unemployment rate for Ethiopia and Uganda is lower than the selected IGAD countries in the period from 1992 to 2018. From the period 1992 to 2000, youth unemployment rate in Uganda is relatively lower than that of Ethiopia and from the period 2000 to 2013 the reverse is true.

As we observed from the above figure (Figure 1), we have concluded that the youth unemployment rate of Ethiopia is varying from year to year with different increasing and decreasing rate. For example, during the year 1992 to the year 1995, it increased from 4.4% to 4.8%. Also, the change from year 1999 to 2010 has a great variation; the youth unemployment rate continuously declines (i.e. 5.4% to 3.6%). The change in the year 1995 to 1999 and the year 2010 to 2015 is not significantly varied and the graph looks like a smooth curve. From the year 2011 to 2016, the youth unemployment rate starts to decline from 3.0% to 2.8%.

Figure 1. Trend on Youth Unemployment in IGAD Countries.

The above data shows that the Ethiopian youth unemployment rate was declining in the year 2011 to 2016; it starts to decline from 3.0 to 2.8. It may be because of the great emphasis given by the government to ensure effective and efficient public sector employment creation through labour-intensive urban infrastructure and housing development Hence, these sectors have generated much employment for both semi-skilled and unskilled labour forces. The construction industry is expected to generate employment for about 1,264,598 persons between 2010/11 and 2014/15 (MUDC 2011). Some attention is given to the development of Micro and Small Enterprises (MSEs), which created 542,000 jobs in 2010/11 (Gebre-Selassie et al., 2012).

The above figure shows that the youth unemployment rate in Uganda increases from year to year starting from the year 1992 until the year 2003 from 1.8% to 6.0%, the figure is continuously upward sloping. In these periods, the highest unemployment rate was also recorded in Uganda in 2003 which is 6.0%. Then, the unemployment rate starts to decline from 2004 to 2005 but it lasts only for two years and again it grew from 3.4% to 5.6% from 2006 to 2010. There is an unbelievable reduction of youth unemployment rate from the year 2012 to 2013 (5.3% to 2.7%). Then, from this period to 2018, unemployment rate is more or less constant.

Although youth unemployment in Uganda is low in absolute terms, it is systematically higher than the national average rate for all adults. Even if the youth unemployment rate in Uganda is lower than other IGAD countries, it is more than doubled from 1.78% in 1992 and reached its peak at 6.0% in 2003, shown in Figure 1. However, these trends are not unique to Uganda relative to other countries.

From 1992 to 2013 the unemployment rate was continuously increasing except the year 2004, it is due to employment grew by an average 2.96% per annum between 1992 and 2013, this is marginally below population growth of 3% over the same period, and below labour force growth of 3.1% over the period 1991–2012. The total labour force in Uganda increased from approximately 7.5 million in 1990 to around 10.1 million in 2000, and almost doubled to 14.5 million by 2012 (Mihyo & Mukuna, 2015).

The youth unemployment rate in Kenya is higher than Ethiopia and Uganda but it is relatively lower than Somalia and Sudan. The rate fluctuates from year to year. For example, from 2004 to 2008 youth unemployment rate of Kenya reduced from 20.0% to 18.2%, which is the higher difference (steep slope of the graph).

A significant proportion of the population in Sub-Saharan Africa (SSA) and, in particular, in Kenya is below the age of 25. In 2011, there were more than 8.5 million people between the ages of 15 and 24 in Kenya, comprising 37.4% of the working-age population and rising to over two-thirds when the population under the age of 34 is considered. (ILO, 2014)

The youth unemployment rate for Somalia is higher than the selected IGAD Countries next to Sudan in the period from 1992 to 2018. The youth unemployment rate for the country was steady from 1992 up to 2005 and started fluctuating thereafter until 2008 (from 27.0% to 25.0%).

Since 1991 the break out of the Somali civil war, the youth in Somalia met with numerous problems for their potential development for the last two decades, such as limited employment opportunities, shortage of recognized higher education attainment, and internal security concerns (Yusuf et al., 2019).

The youth unemployment rate for Sudan is higher than all selected IGAD Countries from the period 1992 to 2018. The youth unemployment rate in Sudan was steady from 1992 up to 2003 and started fluctuating thereafter. Then, from the period 2003 to 2008, it declines from 28.2% to 25.0% and again it becomes steady until 2018.

3. Literature review

Without a doubt, unemployment is a threat to political stability in Africa. Most of the time unemployment gives birth to political instability in a country; unemployed person can easily participate in antisocial activities. They consider that government is worthless, which fails to provide them work. In Africa, where the dependence ratio on government is very high, people tend to resort to any means to retaliate to get their issues addressed (Riechi, 2019).

Tard Dahrendorf was one of the earliest writers to establish a relationship between crime and the environment. He believes that people learn crime through imitation or contact with criminals in their environment. Gabriel Tard writes on this:” the behaviour of an urban city can be studied in the form of a social environment and hence, social crimes and deviancies in the urban environment could be analyzed by the expanse of the city, population density and diversity of various cultures in new urban environments. Dahrendof (1976) However, people’s actions, particularly criminal actions are influenced by an environment where the action takes place, expands, and is transferred to other new environments as an imitable behavior (Dahrendorf, 1960).

The relevant theory for this study is the anomie theory as propounded by Emile Durkheim. Durkheim’s anomie theory describes the effects of the social division of labour developing in early industrialism and the rising suicide rate. French philosopher Emile Durkheim’s book The Division of Labour in Society (De la Division du Travail Social) debuted in 1893. It was his first major published work and the one in which he introduced the concept of anomie or the breakdown of the influence of social norms on individuals within a society.

At the time, The Division of Labour in Society is influential in advancing sociological theories and thought. Today, it is highly revered for its forward-thinking perspective by some and deeply scrutinized by others. He believed that the specialized division of labour and the rapid expansion of industrialized society contained threats to social solidarity. Thus, anomie refers to the breakdown of social norms and a condition where those norms no longer control the activities of the members of the society. Without clear rules to guide them, individuals cannot find their place in society and have difficulty adjusting to the changing conditions of life. This in turn leads to dissatisfaction, frustration, conflict, and deviant behaviours (Anthony, 2013).

One of the most popular phenomena that explain why youth involve in violence is Grievance based explanation. Grievances and horizontal inequalities may be better at explaining why conflicts begin, but not necessarily why they persist. It argues that relative deprivation and the grievance it produces fuels conflict. Central to grievance are concepts of inter-ethnic or horizontal inequality. Identity formation is also crucial to intra-state conflict, as it overcomes the collective action problem (Dorn, 2013) Another theoretical framework used in the world is the deprivation 4 theory of Ted Gurr, which provides a classical theory explaining reasons why youth could engage in various forms of violence. After writing a whole book, Gurr et al. (1970) argues that the primary sources of violence are emotive, psychological, non-rational causes of violence, Gurr tacks on this passage about the rational utilities of violence

The theory of deprivation also explains the relationship between insecurity and unemployment. This does not mean that all unemployed young people are potential terrorists but rather suggests considering youth unemployment as an additional push factor in violence. Following Ted Robert Gurr’s theory on relative deprivation (Gurr et al., 1970), violence does not take root in absolute deprivation but rather in relative deprivation. In this regard, specific attention should be paid to the particular patterns of youth unemployment in the MENA region where highly qualified, educated young people suffer more from unemployment than any other group. This widens the gap between individuals’ expectations (job, salary, and lifestyle) and reality within broader segments of the population than merely unemployed lower-class people. Added to the mismatch between education skills and the labour market, the lack of economic opportunities for educated people contributes to exacerbating the gap between expectations and reality and thereby feeds the feeling of relative deprivation.

In a broader perspective, recent research on Arab youth confirmed Gurr’s theory: the drivers of political violence are rooted in the sense of injustice, discrimination, corruption, and abuse by security forces . In this respect, one of the most counterproductive risks of the security approach is to make room for practices that could further push vulnerable individuals into violent extremism (Doorn, 2013).

According to the paper by Azeng and Yogo (2013) large rate of youth unemployment makes countries more unstable at all. The research tried to show the effect of youth unemployment on political instability using a sample covering 24 developing countries over the period 1980–2010, a 30 years panel data. The paper finds that youth unemployment is significantly associated with an increase in the risk of political instability. Finally, the research recommended that large youth unemployment rate associated with socioeconomic inequalities and corruption makes countries more vulnerable to political instability and national insecurity.

The study by (Fakih et al., 2020) examines the microeconomic determinants of youth unemployment in the MENA region using a unique and novel data on young people aged 15–29 from the year 2016. The results show that being a male and graduated from a public school increase the probability of being unemployed. Moreover, job concerns, corruption, and unequal rights in the society are also found to have a positive incidence on unemployment. However, we find that enhancing gender equality in the labor market, education, family codes, and political participation decreases the probability of employment. Similarly, the results indicate that improving economic inclusion in the post-Arab Spring decreases the probability of unemployment.

Apolte and Gerling (2015) investigated the link between armed insurrection and large youth cohorts using cross-country panel data from 169 countries. They develop a model of insurrection markets and integrate the youth bulge as measured by the relative youth cohort size. They find that it is not the demographic structure or the relative size of the youth cohort as such but rather the reality of large youth populations facing significant unemployment that coincides with insurrection. By testing their implications in an empirical model based on cross country panel data and find that the effect of the relative youth cohort size on insurrection outbreaks is moderated by changes in the underlying institutional setting, and more precise changes in the labour-market conditions as approximated by unemployment rates. While statistical analysis may be useful in establishing broad trends, a great deal of criticism has been levelled against how these studies have causally linked violence with unemployment (Oosterom, 2018).

According to Assaad and Krafft (2016), although it is well-established in the literature that unemployment is a labour market insertion problem in the Middle East and North Africa (MENA), the dynamics driving youth unemployment remain poorly understood. Using panel and retrospective data from the Labour Market Panel Surveys in Egypt, Jordan and Tunisia, we are able to substantially improve our understanding of youth unemployment in MENA by studying flows into and out of employment and unemployment. We also decompose trends in the unemployment rates in Egypt, Jordan, and Tunisia over the past decade into the contributions of individuals entering unemployment from outside the labour force and from previous employment, and changes in the duration of unemployment these individuals experience. Models for the incidence and duration of unemployment illustrate the relationship between individuals’ characteristics and their unemployment dynamics.

The study by Collier (2000) shows that large youth cohorts may be a factor that reduces recruitment costs through the abundant supply of labour with low opportunity cost and it increases the risk of armed conflict. If young people are left with unemployment and poverty, they are more likely to join a rebellion as an alternative way of generating an income. The author also recommended that the expansion of higher education as a strategy to reduce the risk of political violence. Since educated men have better income earning opportunities than the uneducated, they would have more to lose and hence be less likely to join a rebellion. Paul Collier, former Director of the Research Development Department of the World Bank, has suggested that relatively large youth cohorts may be a factor that reduces recruitment costs through the abundant supply of rebel labour with low opportunity cost, increasing the risk of armed conflict. According to the opportunity perspective, rebellion is feasible only when the potential gain from joining a rebel or terrorist organization is so high and the expected costs so low that certain individuals will favour joining over alternative income-earning opportunities.

Urdal (2004) examined the effect of Youth bulges on domestic armed conflict. The research hypotheses are tested in an event history statistical model covering a high number of countries and politically dependent areas over the period 1950–2000. The study finds robust support for the hypothesis that youth bulges increase the risk of domestic armed conflict, and especially so under conditions of economic stagnation. He notes that while youth bulges are strongly associated with increased levels of domestic armed conflict, the reasons for this are not easily generalized, cannot be assumed, and must be empirically investigated. However, the combination of youth bulges and widespread unemployment increases the likelihood of violence, particularly when young people cannot easily migrate away from their societies in search of a better life as they did in 19th century Europe.

Amirali Asha (2019) investigates the relationship between the “youth bulge” and political unrest to understand when and why young Iraqis resort to violence or disruption. In Iraq, widespread youth unemployment combines with high levels of political exclusion, sectarian politics, militarization, perceptions of injustice, frustrated aspirations, war-related trauma, and the rapid breakdown and transformation of traditional institutions such as family and tribe. Together, these factors (and others not covered in this review) produce violence in specific moments. A systematic study of how Iraq’s youth “bulge” affects the country’s political dynamics is warranted. Similarly (Sambanis, 2002), the empirical analysis does indeed suggest young male bulges are more likely to increase the risk of conflict in societies where male secondary education is low. This suggests that the availability of large cohorts of poorly educated youth can support armed conflict.

Sylor (2016) undertook a study on the unemployment and implications for social and political conflict in Zimbabwe. The study showed that for the last 3 decades, Zimbabwean youth have been involved as main factors behind the social unrest and violent activities in the country. This paper argues dissatisfaction and frustration of youth especially graduate urban youth are regarded as one of the major threats to social and political instability. The paper also presents various challenges Zimbabwean youth face and their implications to social and political conflict. The paper discussed the major causes of youth unemployment such as: Sluggish investment and growth, weak export performance, population growth rate, geography, poor macroeconomic policy, and the growth path (depend on commodity products). The paper suggests that there should be land reform, but land reform without the creation of youth employment is only leading to the alienation of youth groups that will fight against the establishment.

Thomas (2015) focuses on addressing youth unemployment in Morocco examines the challenges related to youth unemployment and how Morocco, as a country with high rates of youth unemployment, can benefit from international experiences. High youth unemployment has potentially severe implications on overall economic stability and social cohesion. The issue of youth unemployment must be addressed promptly when youth unemployment rates are high. Research points to a strong correlation between youth unemployment and socio-economic and political instability. If prolonged, youth unemployment can lead to negative consequences at both the individual and societal level. This paper suggests that to tackle its youth unemployment problem Morocco can use policies resulting from lessons learned internationally such as active labour market policies, more adequate minimum wage settings, more flexible contracts, initiatives to acquire new skills related to current technological changes; and finally, more public–private partnerships (PPP). The issue of youth unemployment must be addressed promptly when youth unemployment rates are high because of the strong correlation between youth unemployment and socio-economic and political instability. If prolonged, youth unemployment can lead to negative consequences at both individual and social level.

Another study by (Riechi, 2019) addresses the effect of youth unemployment on security in Kenya for the case of Kwale County. The study used both qualitative and quantitative research approaches and the investigation utilized a descriptive research design. Qualitative research design was used and the target population was leaders drawn from the youth, police, Kwale County government officials, national government administration officers in Kwale County, and the religious sector. The study utilized both primary and secondary data. The researcher used questionnaires for the majority of leaders, interviews for key informants, and focused group discussion for the youth leaders. The study has argued that youth unemployment in Kwale County has been securitized with a majority of leaders terming unemployed youth as a threat to security. Therefore, the paper concludes that unemployed youth end up engaging in crime and another effect of unemployment is hopelessness leading to drug and alcohol abuse, similar findings with Dike (2014).

The reviews conducted above indicate that previous studies have made an effort to examine the effect of youth unemployment on political instability. Despite the effort made to address the subject in other regions of the world, to the best of our knowledge no effort is made to address the issue in IGAD member countries. The region has eight countries that are highly characterized by civil wars, economic, social, and political instabilities, almost all nations of the region are included in the list of the world top unstable nations. At the same time, two-third of the total population is below the age of 25, and unemployment is the main macroeconomic problem in the region. As such investigating the effect of youth unemployment on political instability is crucial to guide the effort to address political instability in the region. Previous studies on the subject have found inconsistent result which necessities further investigation on the subject. Besides, this particular concern is not researched in IGAD member countries. Therefore, this paper investigated the effect of youth unemployment on political instability in selected IGAD countries using recent panel data (1992 to 2018).

4. Data and methodology

The study used secondary data sources. The data on youth unemployment is collected from the ILO key indicators of the Labour Market (KILM). And the data from the International country risk guide measures (ICRG) database on political and economic risk, and we use the Risk of Internal Conflict as a measure of political instability (Azeng & Yogo, 2013).

The data for Economic variables such as GDP per capita and annual inflation rate is captured from world development indicators (WDI) from World Bank national accounts data and OECD national accounts data files. For all the rest variables, the data captured from the International country risk (ICRG) annual data. Those are government stability, religious tensions, corruption, and ethnic tensions.

4.1. Empirical model specification

The current study is inspired by the paper of Urdal (2004) and Therese F. Azeng and Thierry U. Yogo (2013). They used a panel data approach to study the impact of various economic, social and political conditions on conflict.

This paper examined the relationship between youth unemployment and political instability using the simplest specification of this model:

(1) $\hspace{0.17em}\mathrm{P}\mathrm{o}\mathrm{l}\mathrm{i}\mathrm{n}{\mathrm{s}}_{\mathrm{i}\mathrm{t}}\hspace{0.17em}={\beta }_0\hspace{0.17em}+{\beta }_1\hspace{0.17em}\mathrm{y}\mathrm{u}{\mathrm{n}}_{\mathrm{i}\mathrm{t}}\hspace{0.17em}+\hspace{0.17em}{\mathrm{X}}_{\mathrm{i}\mathrm{t}}{\lambda }^{\prime }\hspace{0.17em}+\hspace{0.17em}{\mathrm{U}}_{\mathrm{i}\mathrm{t}}$(1)

Polins_it = the measure of political instability for the country i at time t.

Yun_it = the youth unemployment rate for the country i at time t.

U_it = random disturbance term

β₁ = the coefficient of youth unemployment

λ = the coefficient of all other independent variables

X_it = the vector of independent variables including:-

∙ Government Stability

∙ Corruption

∙ Religious Tensions

∙ Ethnic Tensions

∙ GDP per Capita

∙ Annual Inflation Rate

4.2. Measuring political instability

According to the International Country Risk Guide (ICRG) Howell (2013), political risk rating is aims to provide a means of assessing the political stability of countries. This is done by assigning risk points to a preset group of factors, termed political risk components. Those are Government Stability, Socioeconomic Conditions, Investment Profile, Internal Conflict, External Conflict, Corruption, and Military in Politics, Religious Tensions, Law and Order, Ethnic Tensions, Democratic Accountability and Bureaucracy Quality.

From the above 12 political risk components, we used the risk of internal conflict as a measure of political instability. Internal conflict is an assessment of political violence in the country and its actual or potential impact on governance. The highest rating is given to those countries where there is no armed or civil opposition to the government and the government does not treat in arbitrary violence, direct or indirect, against its people. The lowest rating is given to a country involved in an on-going civil war. The risk rating assigned is the sum of three subcomponents, each with a maximum score of four points and a minimum score of 0 points. A score of 4 points equates to Very Low Risk and a score of 0 points to Very High Risk.

The subcomponents are:

• Civil War/Coup Threat

• Terrorism/Political Violence

• Civil Disorder

Therefore the total point of the risk of internal conflict is 12. A score of 12 points equates to very low risk and a score of 0 points to very high risk.

4.3. Scales of all the variables

Political risk components

Table

Sequence	Component	Points (max.)
A	Government Stability	12
B	Religious Tensions	6
C	Corruption	6
D	Ethnic Tensions	6

Source: International Country Risk Guide Annual²

The risk rating assigned to Government stability is the sum of three subcomponents, each with a maximum score of 4 points and a minimum score of 0 points. The sum of the three sub component is a maximum score of 12 and a minimum score of 0 points. A score of 12 points equates to Very Low Risk and a score of 0 points to Very High Risk.

The risk rating assigned to Religious Tensions and Corruption is with a maximum score of 6 points and a minimum score of 0 points. A score of 6 points equates to Very Low Risk and a score of 0 points to Very High Risk.

The risk rating assigned to Ethnic Tensions is with a maximum score of 6 points and a minimum score of 0 points. Lower ratings are given to countries where racial and nationality tensions are high because opposing groups are intolerant and unwilling to compromise. Higher ratings are given to countries where tensions are minimal, even though such differences may still exist.

To determine the relationship between dependent and independent variables, the fixed effects model and the random-effects model, which are the most common linear panel data analysis models were used.

As a linear model, equation (1)(1) $\hspace{0.17em}\mathrm{P}\mathrm{o}\mathrm{l}\mathrm{i}\mathrm{n}{\mathrm{s}}_{\mathrm{i}\mathrm{t}}\hspace{0.17em}={\beta }_0\hspace{0.17em}+{\beta }_1\hspace{0.17em}\mathrm{y}\mathrm{u}{\mathrm{n}}_{\mathrm{i}\mathrm{t}}\hspace{0.17em}+\hspace{0.17em}{\mathrm{X}}_{\mathrm{i}\mathrm{t}}{\lambda }^{\prime }\hspace{0.17em}+\hspace{0.17em}{\mathrm{U}}_{\mathrm{i}\mathrm{t}}$(1) can be estimated using simply ordinary least square (OLS). The main drawback behind OLS is that OLS results are biased if youth unemployment is correlated with the unobserved component of political instability. For instance, political instability could lead to a higher unemployment rate, rather than vice versa. Political instability originates in high uncertainty, which may decrease labour demand and therefore increase unemployment (Azeng & Yogo, 2013). In this case, the effect of youth unemployment could be misleading. OLS results could be therefore biased toward zero and they can underestimate the “true” impact. This leads to the endogeneity bias, the problem of endogeneity occurs when a predictor variable (x) in a regression model is correlated with the error term (e) in the model.

To deal with Endogeneity bias, we resort to IV/GMM estimator. First, we used IV fixed effect estimation and then we make use of one-step moments (GMM)³ estimator. The efficiency gains of this estimator relative to the traditional IV estimator derive from the use of the optimal weighting matrix, the over identifying restrictions of the model, and the relaxation of the i.i.d. assumption. The coefficient that will draw our attention is β1, which should be significantly positive.

GMM allows for most flexible identification and its estimates can be identified by any set of moments from the data as long as you have at least as many moments as you have parameters to estimate and that those moments are independent enough to identify the parameters. (And the parameters are independent enough of each other to be separately identified.) GMM estimator has Good large sample properties. The GMM estimator is strongly consistent and asymptotically normal and it will likely be the best estimator if you have a lot of data.

There are two estimators for GMM dynamic panel data called the differenced GMM model and System GMM model. To decide between difference and system GMM model, the rule stated by Bond et al. (2001) used. First, pooled OLS, for the within effect, and fixed effect models are estimated, and the coefficients for the lagged dependent variable from the pooled OLS estimation is taken as an upper bound estimate and the coefficient in the fixed effect estimation is taken as a lower-bound estimate. Second, one-step and two-step difference GMM model is estimated. The coefficient of the lagged dependent variable, which is Political instability in this case, from the two-step difference GMM estimation, is compared with the coefficients of the previous two estimations. If it is close to or below the fixed effect estimate, it’s suggested that the difference GMM is downward biased and system GMM estimator is preferred.

The GMM model is empirically presented as follows:

(2) $\mathrm{P}\mathrm{o}\mathrm{l}\mathrm{i}\mathrm{n}{\mathrm{s}}_{\mathrm{i}\mathrm{t}}={\beta }_1\mathrm{P}\mathrm{o}\mathrm{l}\mathrm{i}\mathrm{n}{\mathrm{s}}_{\mathrm{i}\mathrm{t}-1}+{\beta }_2\mathrm{y}\mathrm{u}{\mathrm{n}}_{\mathrm{i}\mathrm{t}}+{\beta }_3\mathrm{g}\mathrm{o}\mathrm{v}\mathrm{s}\mathrm{t}{\mathrm{a}}_{\mathrm{i}\mathrm{t}}+{\beta }_4\mathrm{r}\mathrm{e}\mathrm{l}\mathrm{i}\mathrm{g}\mathrm{t}\mathrm{e}{\mathrm{n}}_{\mathrm{i}\mathrm{t}}+{\beta }_5\mathrm{c}\mathrm{o}\mathrm{r}{\mathrm{r}}_{\mathrm{i}\mathrm{t}}+\beta 6\mathrm{e}\mathrm{t}\mathrm{h}\mathrm{t}\mathrm{e}{\mathrm{n}}_{\mathrm{i}\mathrm{t}}+{\beta }_7\mathrm{G}\mathrm{D}\mathrm{P}\mathrm{p}\mathrm{e}{\mathrm{r}}_{\mathrm{i}\mathrm{t}}+{\beta }_8\mathrm{A}\mathrm{n}\mathrm{n}\mathrm{i}\mathrm{n}{\mathrm{f}}_{\mathrm{i}\mathrm{t}}+{\mathrm{t}}_{\mathrm{i}}+{\gamma }_{it}$(2)

(3) ${\epsilon }_{\mathrm{i}\mathrm{t}}+{\mathrm{u}}_{\mathrm{i}\mathrm{t}}={\gamma }_{it}$(3)

Where,

Polins_it -is the measure of political instability for the country i at time t.

Polins_it-1 -is the lag value of political instability

Yun_it -is the youth unemployment rate for the country i at time t.

t—represents time effects

While ${\gamma }_{it}$—represents both country effects (u_it) and the remainder disturbance, which varies over both country and time (ε_it).

4.4. Summary statistics and description

Table 1 summarizes descriptive statistics, which are the mean, standard deviation, minimum and maximum values for the dependent variable, political instability (Polins) and the independent variables; youth unemployment (Youn), government stability (Govsta), religious tensions (Religten), Corruption (Corr), ethnic tensions (Ethten), GDP per capita (GDPper), Annual inflation rate (Annin).

Table 1. Descriptive statistics

Variable	Observations	Mean	Std. Dev.	Min	Max
Political Instability	135	6.27	2.24	0	11
Youth unemp’t rate	135	16.85	11.17	1.78	32.69
Government Stability	135	7.22	2.10	0.66	11.58
Religious Tensions	135	3.21	1.32	0	5
Corruption	135	1.59	.67	.5	3.45
Ethnic Tensions	135	2.34	1.04	0	4.58
GDP Per Head	135	2.68	4.85	−11.89	22.32
Annual Inflation Rate	135	13.83	22.23	−27.78	132.82

Political instability for the five countries used in the empirical analysis averaged 6.2 between the years 1992 to 2018 ranging from 0 point for Somalia to 11 point for Kenya with the standard deviation of 2.2. This instability variation in the region is because of internal factors such as corruption, religious tensions, ethnic tensions, democratic accountability, and other political risk index variables.

The level of youth unemployment in the region is averaged at 16.8% for the period 1992 to 2018. The minimum level of youth unemployment stood at 1.78% for Uganda in 1992 and the maximum value recorded at 32.69% for Sudan in 2000.

The mean value for the variable Government stability is 7.2 for the period 1992 to 2018. The minimum value of this variable is 0.67 point and the maximum value is 11.58 point. The minimum value of government stability is recorded in Somalia in 1992 and the maximum value of this variable stood at Ethiopia in 2000. The variation from the mean is 2.1 point. And, the summary statistics for religious tensions showed that for the period 1992 to 2018 stood at an average of 3.2 point with a standard deviation of 1.3. Also, the minimum and the maximum value for religious tension are ranges from 0 to 5 point.

As per Table 1, the average value of corruption is 1.6 point from the period 1992 to 2018. The ratio ranges from a minimum value of 0.5 point to the maximum value of 3.45 point for Kenya in 2003. The standard deviation of the variable is 0.67 point. As per Table 6.1, the average GDP per head is 2.68% from the period 1992 to 2018. The variable ranges from a minimum value at −11.89% for Ethiopia in 1992 to the maximum value at 22.32% for Somalia in 1994.

According to the summary statistics Table 1, the mean value of the inflation rate in the region is 13.83% for the period 1992 to 2018. The minimum and the maximum value of this variable is −27.78% for Ethiopia in 2010 and 132.82% for Sudan in 1996, respectively. The variation from the mean for the inflation rate is 22.51%. This highest variation from the mean is an indication of macro-economic instability in the region through overall price skyrocketing.

5. Empirical results and discussions

5.1. Regression diagnostics test results

5.1.1. Panel co-integration test

After conducted panel unit root tests, the next step was to establish whether the non-stationary variables are co-integrated or not. Usually, when variables are differenced to attain stationarity, the long-run properties are lost. Co-integration means that there is a long-run relationship between two or more non-stationary variables. Since the dependent variable (Political Instability) was stationary at level (I (0)), it is not necessary to check for co-integration in this particular model.

5.1.2. Endogeniety test

This study used Endogeniety test of Mark E Schaffer, for checking Endogeniety problem in panel data. The null hypotheses of Endogeniety test for endogenous regressors in panel data is variables are exogenous, and the alternative hypothesis is variables are endogenous. The resulting test statistics show that the p-value is 0.000, which is less than 0.05, so we reject the null hypothesis, and there is endogeniety bias.

5.2. Regression results

5.2.1. Fixed effect regression model

The fixed effect regression shows the within effect, which explains only 64.37% of variation on political instability across time and within countries with an overall explanation of 46.41% of variation. The interclass correlation result shows that 88.96% of the variances are due to differences across panels.

Table 2 shows the result of the OLS Fixed effect estimation between Political instability, Youth unemployment, government stability, religious tensions, corruption, ethnic tensions, GDP per capita, and Inflation. The result shows that all variables except corruption, religious tensions and GDP per capita are statistically significant. Government stability and ethnic tensions are statistically significant at 1% level of significance while youth unemployment and inflation are statistically significant at 5% level of significance.

Table 2. Youth unemployment and political instability, OLS fixed effects

Political Instability	Coef.	St.Err.	t –value	p-value	[95% Conf Interval]		Sig
Youth Unemployment	−.168	.069	−2.43	.016	−.305	−.031	**
Gov’t stability	.271	.059	4.62	0	.155	.387	***
Religious tension	−.186	.168	−1.11	.271	−.517	.146
corruption	.362	.194	1.87	.064	−.022	.746	*
ethnic tension	−1.617	.196	8.25	0	1.229	2.006	***
GDP per capital	−.008	.018	−0.44	.664	−.044	.028
Inflation	−.012	.005	−2.27	.025	−.022	−.002	**
Constant	3.548	1.272	2.79	.006	1.031	6.065	***
Mean dependent var		6.273		SD dependent var		2.244
R-squared		0.644		Number of obs		135.000
F-test		31.741		Prob > F		0.000
Akaike crit. (AIC)		378.030		Bayesian crit. (BIC)		401.272

Note: Robust standard errors in parentheses.*** p<.01, ** p<.05, * p<.1

Source: author’s computation in stata

Youth unemployment and Inflation and ethnic tensions have a positive relationship with the political instability. And government stability is positively related to political instability. The result in the table below reveals that when the youth unemployment rate increases by 1%; it leads to political instability point to decrease by 16%. This means an increase in youth unemployment leads to political instability. If all the explanatory variables are set to be zero, Political instability would be 3.5 for a given country. The fixed-effect model could be written as:

In the above estimation, we firstly estimated using a simple OLS fixed effect estimator. It could happen that country with a high level of political instability also experiences a high level of unemployment. In this vein, OLS estimates are biased since the unemployment rate may be correlated with some unobservable. To deal with the endogenous bias, we resort to the instrumental variable approach.

Then, two main instruments are used: the lagged unemployment variable and the first difference of unemployment variable (Azeng & Yogo, 2013). The main justification for this choice is based on the idea of hysteresis in unemployment (Blanchard & Summers, 1986). This concept embodies the idea that the equilibrium unemployment rate depends on the history of the actual unemployment rate. In other words, the unemployment rate at the period t may be highly dependent on the unemployment rate at t-1. One of the main explanations is the destruction of human capital, which in some cases is associated with the discouragement of job seekers.

As it can be seen from Table 3, a one percent increase in Youth unemployment increases political instability on average by 1.43% across time for a given country, other things remaining constant, showing a positive impact of youth unemployment.⁴ The coefficient of youth unemployment is statistically significant at 5% level of significance. While a one percent increase in government stability decreases political instability by an average of 0.19% across time for a given country citrus paribus with a statistical significance level at 1% level of significance.

Table 3. Youth Unemployment and Political Instability, IV Fixed Effects

Political Instability	Coef.	St.Err.	t-value	p-value	[95% Conf Interval]		Sig
Youth Unemployment	−.143	.07	−2.05	.041	−.28	−.006	**
Gov’t stability	.195	.064	3.03	.002	.069	.321	***
Religious tension	−.071	.176	−0.40	.686	−.416	.273
corruption	.385	.201	1.91	.056	−.009	.779	*
ethnic tension	−1.355	.232	5.84	0	.9	1.81	***
GDP per capita	−.028	.022	−1.28	.202	−.07	.015
Inflation	−.017	.006	−2.82	.005	−.028	−.005	***
Constant	4.045	1.287	3.14	.002	1.524	6.567	***
Mean dependent var		6.407		SD dependent var		2.121
Overall r-squared		0.467		Number of obs		135.000
Chi-square		6007.009		Prob > chi2		0.000
R-squared within		6007.009		R-squared between		0.532

Note: Robust standard errors in parentheses. *** p < .01, ** p < .05, * p < .1

Source: Author’s computation in stata

Other things remaining constant, a one percent increase in ethnic tensions increases political instability by 1.35% at 1% level of significance. Annual inflation rate on the other hand reduces political instability on average by 0.1% when it increases by 1% across time for a given country, other things remaining constant. The coefficient of inflation is statistically significant at 1% level of significance. The rest variables which are corruption, religious tensions and GDP per capita are found to be statistically insignificant at 5% level of significance.

5.2.2. Robustness checks

We check the robustness of our results with the use of Dynamic specification on an alternative sample.

5.2.3. Dynamic specification

The analysis presented above omits years for which the measure of political instability is not reported. We treat these non-reports as zeros, and we use a dynamic specification. Table 3 shows that the effect of youth unemployment remains similar to what was obtained from the fixed effect model and the instrumental fixed effect model.

Dynamic panel data estimators the DPD approach (Dynamic Panel Data) approach is usually considered the work of Arellano and Bond (AB) (Rev. Ec. Stud., 1991). It is based on the notion that the instrumental variables approach noted above does not exploit all of the information available in the sample. By doing so in a Generalized Method of Moments (GMM) context, we may construct more efficient estimates of the dynamic panel data model.

The Arellano–Bond estimator sets up a generalized method of moments (GMM) problem in which the model is specified as a system of equations, one per period, where the instruments applicable to each equation differ (for instance, in later periods, additionally aged values of the instruments are available). This estimator is available in STATA as xtbond. The original estimator is often entitled to difference GMM, while the expanded estimator is commonly termed System GMM. The cost of the System GMM estimator involves a set of additional restrictions on the initial conditions of the process generating.

5.2.3.1. Choosing between difference and system GMM models

The coefficient of the lagged dependent variable from the two-step difference GMM estimation is compared with the coefficients from the pooled OLS and fixed effect model to decide between difference and system GMM. First Estimate the fixed effect model with lagged dependent variable. The coefficient of a lagged dependent variable is considered to be a lower bound. And then Estimate the Pooled OLS, and include the lag of the dependent variable, the coefficient of a lagged dependent variable is considered as an upper bound estimate. Finally, Estimate the difference GMM. If the two-step difference GMM estimate coefficient of a lagged dependent variable is close to or below the fixed effect model, this suggests that the former estimate is downward biased because of weak instrumentation, and system GMM should be used. Therefore by calculating coefficients for the lag of political instability for Pooled OLS, Fixed Effect, One step Diff. GMM and Two step Diff. GMM we choose whether differenced GMM or System GMM is better. Estimation from the difference GMM suggests that using system GMM is beneficial in this case.

5.2.3.2. One step system GMM

The one-step system GMM estimation result presented in the table below shows the value of the Hansen test is equal 1 in which case the hypothesis cannot reject the validity of instruments. The Arellano-Bond test for AR (2) is 0.268 also stating there is no serial correlation. The empirical regression result shows that religious tensions, corruption, GDP per capita and inflation are statistically insignificant. The rest of the variables which are the lag of internal conflict, youth unemployment and ethnic tensions are found to be statistically significant at 5% level of significance.

The lag of political instability increases the political instability of IGAD countries by 0.5% on average. In the one step GMM specification the coefficient of youth unemployment is negative and statistically significant at 5% level of significance. As it can be seen from Table 4 a percentage increase in youth unemployment is associated with a 0.04% increase in internal conflict in the short run, at 1% level of significance, on average citrus paribus. Therefore, the results are consistent for all FE model, IV fixed effect model and one-step system GMM model. In all estimation techniques, youth unemployment has a positive effect on political instability. Hence, the region should give attention for youths who are unemployed in order to make politically stable region.

Table 4. Robustness check using one step system GMM regression

Political Instability	Coef.	St.Err.	p-value	[95% Conf Interval]		Sig
L. Political Instability	0.51	3.11	0.002	.191	.843	***
Youth Unemployment	-0.04	3.74	0.000	-.069	-.021	***
Gov’t stability	0.25	4.28	0.000	.138	.371	***
Religious tension	-0.23	1.06	0.290	-.659	.197
Corruption	-0.22	1.07	0.285	-.629	.184
Ethnic tension	-0.82	2.06	0.039	.040	1.60	**
GDP per capita	-0.01	0.69	0.493	-.057	.027
Inflation	-0.00	0.88	0.378	-.016	.006
Arellano-Bond test for AR(1) in first differences				0.101
Arellano-Bond test for AR(2) in first differences				0.268
Hansen OID test				1.000
Sargan test				0.105
Number of obs				108

Note: Robust standard errors in parentheses.*** p < .01, ** p < .05, * p < .1

Source: Author’s computation in stata

For government stability, a 1% increase leads to a 0.25% decrease in political instability on average and citrus paribus. This increment is statistically significant at 1% level of significance. An increase in change in ethnic tensions by 1% increases political instability by an average of 0.82% with a statistical significance level at 5%, other things remaining constant.

We assert that our empirical findings conform to the predictions of the hypothetical theory laid out above. Countries with higher level of youth unemployment rate are more vulnerable for political instabilities. The result of this finding is consistent with the findings of Azeng and Yogo (2013), Apolte and Gerling (2015), Collier (2000), Urdal (2004), Amirali Asha (2019), Sylor (2016), and Thomas (2015), (Riechi, 2019), those papers finds that youth unemployment is significantly associated with an increase in the risk of political instability. However, our study is contrary to the findings of Rezyk (2016) and Enria (2018), the result of these studies revealed that youth unemployment does not lead to political instability.

Therefore, our result also conform the finding that Youth unemployment is significantly and positively associated with political instabilities of nations, and we support the hypothesis that youth unemployment causes political instability. From the result of this study, it is recommended that controlling youth unemployment helps to reduce political instability. Consequently, Countries should give attention for youth’s not only for the sake of youths but also for the relief of government reducing the burden of controlling continuous internal instabilities.

6. Conclusions and policy implications

This paper analyzes the effects of youth unemployment on political instability using the data from 1992 to 2018 from five IGAD countries, which are Ethiopia, Kenya, Uganda, Sudan, and Somalia. The potential endogeneity of the unemployment is addressed using an instrumental variable approach and the one-step system GMM model.

The empirical result obtained in this study indicates that there is a significant effect of youth unemployment on political instability, and we support the hypothesis that youth unemployment causes political instability in IGAD countries. This paper also examined how other social and economic factors can affect political instability. Our results suggest that political instability occurs particularly in countries where youth unemployment, as well as inflation and government instability, are high.

The one-step system GMM was used for regression to handle endogeneity by introducing more instruments to improve efficiency and transform the instruments to make them uncorrelated with the fixed effect. The regression result from the one-step system GMM model shows that the lag of internal conflict increases the internal conflict of IGAD countries. Also, our results suggest that political instability occurs particularly in countries where youth unemployment is high. This study results are conclusive and add to the literature that asserts that government instability is the most critical determinant triggering political instability in IGAD countries.

From the result of this study, it is recommended that controlling youth unemployment helps to reduce political instability. Policies intended to smoothen the political instability in our specific context of internal conflict should work on creating job opportunities for the youths. Also, the population growth should be controlled in line with offering jobs. The overall trend of youth unemployment and political instability shows that political instability is primarily caused by the youths who are unemployed. This region specifically needs a sound youth employment policy not only for the sake of youths but also for the relief of government reducing the burden of controlling continuous internal instabilities. Moreover, in region-wise the IGAD countries better to have common youth employment creation policies so that they can manage the internal conflicts that arises here and then.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The author received no direct funding for this research.

Notes on contributors

Yemareshet Hailu Demeke

Yemareshet Hailu Demeke, the author of this article, is M.Sc in economics holder from Addis Ababa University with specialization of Economic policy analysis. Currently, I am working for Debre Berhan University as lecturer. I am interested to conduct a study on macroeconomic variables.

Notes

1. IGAD-The Intergovernmental Authority on Development is an eight-country trade bloc in Africa. It includes Djibouti, Ethiopia, Kenya, Somalia, Eritrea, Sudan, South Sudan, and Uganda.

2. For more details, see PRS Group (2012) International Country Risk Guide Annual. http://www.prsgroup.com/ICRG_Methodology.aspx

3. GMM- the Arellano–Bond estimator is a generalized method of moments estimator used to estimate dynamic models of panel data.

4. Let recall that the higher is the value, the highest is the risk of political instability.