The Fragility-Grievances-Conﬂict Triangle in the Middle East and North Africa (MENA): An Exploration of the Correlative Associations

: The intention of this special issue of Social Sciences is to study state fragility and its relationship with conﬂict and grievances in the post-Cold War Middle East and North Africa (MENA). This article will lay the foundation for such a study by offering a conceptual foundation, data and the identiﬁcation of the correlative associations that are speciﬁc to the MENA region. This article suggests that the relationship between political legitimacy, factionalism of the state, and conﬂict needs special, MENA-speciﬁc emphasis, as this relationship seems more prominently different in the MENA region, compared to the rest of the world. While in the rest of the world, different aspects of state fragility all relate to grievances and conﬂict dynamics, in the MENA region political factionalism has a disproportionate role in the explanation of conﬂict grievances and violence. Moreover, the role of oil dependence, and the impact of external intervention requires attention of specialists of the region.


Introduction
Politically induced human suffering in the post-Cold War era has often been described as something grounded in chaos, opportunistic violence and exploitation, mobilised by personal or group interest of identity groups that fragile states cannot control (Duffield 2001;Gilbert 2003;Heupel and Zangl 2010;Kaldor 1999Kaldor , 2013Kalyvas 2001;Melander et al. 2009;Mello 2010;Münkler 2002). Barbaric violence "is usually unleashed when overarching political or economic systems are either weakened or collapsed" (Duffield 2001, p. 110). "The new wars arise in the context of the erosion of the autonomy of the state and, in some extreme cases, the disintegration of the state" (Kaldor 1999, p. 5). The description of post-Cold War political human suffering often merges some of the old elements of conflict, development and state-building into a holistic context in which the interaction between previously separately studied elements is in the focus.
The core concept of this article and the special issue is the concept of state fragility, and the focus is on interaction between state fragility and grievances, and fragility and conflict. Laying the foundation to such analysis will be done by clarifying the definitions and conceptual associations between the three clusters of variables. Based on the definitions of the core concepts of state fragility, conflict and grievances, this article will select and adjust data for the analysis of the causal associations between the variables of fragility, conflict and grievance. These data are openly available in the University of Bath Research Data Archive at https://doi.org/10.15125/BATH-00951 (Kivimäki 2021a). Finally, this Soc. Sci. 2021, 10, 120 2 of 17 article will do exploratory work by investigating bivariate correlations between such indicators of state fragility, conflict and grievances that make sense from the perspective of the conceptual foundation. The intention of such exploration is to flag out the apparent anomalies of MENA, so that it would be possible for the rest of the articles of this special issue to focus on MENA-specific structures and dynamics. This way we can identify how to complement the more global and general theories of state fragility, in order to understand the phenomenon in the MENA region. Instead of attempting to move from correlative associations to explanatory, causal models simply by using the quantitative data and more sophisticated quantitative methods and models, the intention is to flag the regionally specific associations and leave their explanation to more understanding, qualitative analysis. This way, the explanation of causal complexes is approached through the understanding of the region and its conflicts and dynamics of state fragility rather than approaching full explanations from the perspective of numbers. Yet, the statistical exploration of the correlative associations in this article intends to guide qualitative analysis of this special issue and prevent it from overgeneralizing exceptions into rules.

Conceptual Relationship between State Fragility, Conflict and Grievances
State fragility is often described as consisting of two elements: lack of efficiency and lack of legitimacy of state institutions (Fukuyama 2011;Lemay-Hébert 2009;Marshall and Cole 2014;Marshall and Elzinga-Marshall 2017). Effectiveness of state institutions is emphasised by scholars who follow Weberian tradition and define fixed criteria for capable, inclusive and democratic state institutions (Acemoglu and Robinson 2012;Weber 1946). Others, following the Durkheimian tradition, emphasise the legitimacy side by claiming that, since state institutions are supposed to work for the people, a felt ownership of these institutions is more important than the perfect shape and form of the institutions themselves (Durkheim 1957;Lemay-Hébert 2009). For our purposes there is no need to take sides in favour or either two schools, as they both are right in their definition of state fragility. No matter how legitimate institutions are, if they fail to produce welfare or security, they add little to the quality of lives of the citizens. A country with perfect legitimacy but no effectivity is like a car with an accurate steering, but no engine. Effective, but illegitimate institutions, which people work against and fail to consider their own, are also not useful for the citizens: They are like a car with a powerful engine, but no steering wheel. There is a need, both, for legitimacy and effectiveness of state institutions to be strong: Without one or the other, states are fragile.
The effectiveness aspects of state fragility are conceptually and empirically linked with conflict and grievances in two ways. On the one hand, inability to deliver (economic, political and security effectiveness) in poor countries implies grievances that make the life of citizens or at least groups of citizens, unbearable. This way, the lack of effectiveness is directly linked with grievances and indirectly with conflict, too, as much of conflict violence is motivated by grievances. This indirect link from state inefficiency to conflict has been emphasised in the conflict literature tradition that focuses on the grievance-related motives of violence (Davies 1962;Gurr 1970;Runciman 1966).
The narrative that links a state's social and economic inefficiency with grievances and conflict is simple. We can assume that people generally want to avoid risky and morally repulsive use of violence. Yet, state fragility or failure and the resulting grievances may make life so unbearable that changing this intolerable situation may be a stronger motive than the willingness to avoid violence. In the literature of grievance-based violence, a state's inability to deliver either creates absolute or relative deprivation. In the former case, a state's inability and the resulting grievances lower the threshold of violence (Burton 1993;Collier and Hoeffler 2004;Fisher 1990;Hoadley 1981;Mitchell 1990;Rubenstein 1990) and, consequently, cause conflict. In the case of relative deprivation-a situation where there is a gap between the expected and observed receipt of welfare, income, wealth, political power, or something elsea state's reduced capacity to deliver basic needs, motivates violent change (Gurr 1970(Gurr , 1993Runciman 1966). The literature on relative deprivation has developed further, distinguishing between horizontal and vertical inequalities that can create the feeling of discrepancy between expected and observed receipt of welfare. While the original theories of relative deprivation represented the latter version, and managed to get only spurious empirical support (Brush 1996), the theory of horizontal inequalities as causes of conflict have now a much stronger empirical evidence base (Cederman et al. 2011).
On the other hand, state effectiveness is conceptually and empirically associated with opportunistic violence: A state's inability to enforce rules of political and economic competition may lead to the temptation to aim at private or a group's selfish gains by means of violence (Tilly 1978;Duffield 2001;Kaldor 1999;Mello 2010;Münkler 2002). Thus, the linkage between state fragility and conflict can also be related to the opportunities for violence, not just motives.
In a rational theory of warfare, the linkage between state fragility and violence is based on a narrative in which a state's inability to secure individual citizens forces individuals to seek security from their ethnic, religious, regional or other groups. Consequently, in a situation where a state cannot contain violence, groups will see each other as threats, and this threat is seen the more severe, the stronger other groups are. This leads to a security dilemma, in which groups mobilise against each other and this makes intra-state conflicts likely (Fearon 1995;Fearon and Laitin 2004;Poulton and Youssouf 1998).
While violence requires both motives and opportunities, conflict literature often differentiates between opportunity-based (Tilly 1978) and motive/grievance based theories (Gurr 1970). However, in the 1990s, theories emerged with the focus on gainful, rather than grievance-based motives (Collier and Hoeffler 2004;Tanter 1998). These theories focused on exploitation, looting and pillaging as motives for conflict (Collier and Hoeffler 2004;De Sousa 2000;Kaldor 1999). Implicitly, the distinction between gainful vs. grievancebased theories was based on judgements on legitimacy of motives. Motives related to mere survival or basic needs were considered as legitimate and they were therefore not questioned: Conflict could only be avoided if all groups had access to basic needs (Burton 1993;Fisher 1990;Hoadley 1981;Mitchell 1990;Rubenstein 1990). In these theories, the root cause of conflict is (often developmental) grievance, which links conflict and grievances empirically. This linkage was strengthened in early peace research by the articulation of the concept of structural violence. Such violence does not have an agent-nobody does direct violence to anybody-yet years of human life are lost due to unfair structures that deny groups of people access to basic needs (Galtung 1969;Galtung and Höivik 1971). This way, the focus on lost years of life makes it possible to measure and compare the effects of structural grievances that Galtung then also calls violence and direct violence that we see in conflicts. State's failure to effectively offer all citizens an equal access to basic needs is then simultaneously state fragility, violence and a grievance.
Dealing with the greed-related conflict motives, states needed to focus on the opportunity side of conflict prevention. While states could not simply control opportunities to violence that emanate from serious grievances, as that itself would constitute support for structural violence, states were expected to control gainful greedy motives of violence.
Finally, state capacity is also related to opportunities to non-violent change. The lack of legitimacy of the state's governance of security, politics, economy and social affairs affects perceptions of alternatives of violent change: if citizens perceive no legitimate means of influence, illegitimate violent means often become more attractive (Lichbach and Gurr 1981). Thus, both legitimacy and effectiveness of the state apparatus are conceptually and empirically linked to grievances and violence.
Putting together greed and grievance-based motives and opportunities to gain with violence and opportunities to gain without violence brings us closer to the economics of conflict economics where some of the likely causes of conflict motives seem hindrances to the opportunities to use violence rationally. Poverty and vertical inequality that may give rise to conflict motives, may also render violence "economically unviable" by creating obstacles for the funding or arming violence (Anderton and Carter 2019, pp. 115-18). Furthermore, within this context of conflict economy, we can interpret the relationship of democratic political grievances with conflict also as a question of alternative costs and benefits. Non-violent channels of protest may not relate only to reduced motives of conflict; they may also relate to conflict by offering less costly, more economical, alternatives to violence.

Measuring State Failure, Conflicts and Developmental Grievances
Before revealing how the Middle East and North Africa (MENA) region differs from the rest of the world, there is a need to define the data that such flagging out of differences will be based on. This section will explain the data, assess their validity and reliability and sort out the modifications to existing datasets that have been made for this study to avoid tautological reasoning.
A study that focuses on state fragility will have to operate based on observations of country years, i.e., how state fragility, conflict and human development vary between each country in each year. There are 3960 such observations in our analysis, 432 country years in the MENA area. All countries in the world are included in the analysis from 1995 to 2018, which is the last year with data on state fragility. Countries (often small and those who are not independent) with no data on state fragility are not included even if there were data on the other variables.
There are two main datasets on state fragility: State Fragility Index (SFI) 1995-2018 of the Systemic Peace project (Marshall and Elzinga-Marshall 2017) and the Fund for Peace data for 2006-2020 (Fund For Peace 2020). Both are used extensively, and both consist of indicators that consider both legitimacy and effectivity of state institutions. For this study, the former is more useful as it builds more explicitly on the idea of state fragility consisting of problems with effectiveness and legitimacy and because its data reach almost to the beginning of the post-Cold War era, which, in many senses, is a unique era in which the triangle of fragility, conflict and grievances are aligned in a relatively stable way.
The legitimacy and effectiveness of state institutions can vary between different clusters of issue areas where the state operates. One way to take this into account is to distinguish between security, political, economic and social institutions and study the level of effectivity and legitimacy separately between these clusters of state functions. This is the strategy adopted here, even though all these clusters of issue areas are interlinked. It is useful also to study the profile of state fragility and for that we need to analytically separate between fragility of the security, political, economic and social institutions. It is clear that there can be effective but illegitimate states (Saudi Arabia) or legitimate, but ineffective states (India or perhaps Egypt after the democratic election of Morsi), quite as there can be states that manage their economy in a legitimate and effective manner (Singapore) but have a limited political legitimacy. Different profiles of state fragility/strength may have different kinds of problems with the alleviation of popular grievances and controlling violence and conflict.
The measurement of legitimacy and efficiency of security, political, economic and social governance of the state is based on indicators that are explained in detail in Marshall's global report of 2017 (Marshall and Cole 2014;Marshall and Elzinga-Marshall 2017). Security efficiency in this dataset means the ability of the state to guarantee the safety of citizens and institutions crucial to the wellbeing and safety of citizens. Security legitimacy again means the feeling of justification of the agency, referents (what objects and values are being protected) and methods of the security establishment for the production of security in the state. In the State Fragility Index, security efficiency is measured by the outcomes of a state's security measures. This is done by measuring the sum of annual scores for all wars in which the country is directly involved for each continuous period of armed conflict, then taking into account the interim years of "no war" between periods of armed conflict and then still counting years of peace, or no war, since the end of most recent war period (Marshall and Elzinga-Marshall 2017, p. 52). Our measurement of political violence is based on the best estimate of the number of deaths relating to combat between warring parties or to violence against civilians (Högbladh 2020, p. 5;Sundberg and Melander 2013).

Our main measurement of the level or intensity of violence in a given state and in a given
year is based on the UCDP Georeferenced Event Dataset (Sundberg and Melander 2013), which gives us the number of fatalities in each country during each year from 1995 to 2018, and the World Bank's World Development Indicators data (World Bank 2020) on population. Instead of simply comparing the number of conflict fatalities in China and San Marino, for example, data created for this paper (Kivimäki 2021a) will look at conflict fatalities per population. This indicator gives a better indication on the reality of conflict intensity in a given country and shows the risk for individual citizens to be killed in violence. This is something that can then be correlated with state fragility indicators in that specific country. Data of population is from World Bank's World Development Indicators https://databank. worldbank.org/reports.aspx?source=2&series=SP.POP.TOTL&country=# (accessed on 22 December 2020). Thus, in order to avoid tautological reasoning, one cannot use this aspect of state fragility in correlative analysis of the associations between state fragility and conflict. Security efficiency is naturally considered in all other calculations, though.
Security legitimacy, again, is measured in the State Fragility Index by using state repression as an indicator (Marshall and Elzinga-Marshall 2017, p. 52). Since our definition of political violence/conflict includes one-sided violence against unarmed civilians (Sundberg and Melander 2013), this measure of state fragility can neither be included when investigating the relationship between conflict and state fragility. Thus, I have created an additional State Fragility value: New War State Fragility, that reaches from 0 (no fragility) to 19 (maximum fragility). This value includes all but these two elements of state fragility, and it can then be used for the investigation of the relationship between state fragility and conflict. The full state fragility index with security indicators is used in all other tests of association.
Since MENA is different than the rest of the world in terms of the association between political legitimacy of state institutions and violence, it is useful also to explicate how political legitimacy is measured in this article. Political legitimacy is defined here as popular perception of the inclusion and fairness of the political system. I follow the State Fragility Index and measure political legitimacy as lack of factionalism, lack of political discrimination, lack of political salience of elite ethnicity, lack of polity fragmentation and lack of exclusionary ideology of the ruling elite (Marshall and Elzinga-Marshall 2017, p. 53).
Economic efficiency as an element of state strength (lack of fragility) is defined as the ability of the state to generate and allow private sector to generate wealth and economic prosperity (Marshall and Elzinga-Marshall 2017, p. 53). In the State Fragility Index, it is measured by an indicator based on per capita GDP values. My measurement of developmental grievances is based on UNDP's Human Development Index on the level of human development: The lower the level of human development is, the greater the objective developmental grievances are. UNDP measures human development by using life expectancy, expected years of schooling, mean years if schooling and GNI per capita (PPP) as elements of the indicator (United Nations Development Report 1995). Thus, the two indicators (economic efficiency and developmental grievances) overlap conceptually and cannot be used to study the empirical interaction between developmental grievances and the economic efficiency aspect of state fragility. The indicator of social effectiveness in the State Fragility Index also derives from some of the same indicators as the Human Development Index that we use for our indicator of grievances and cannot, therefore, be used in the study of the relationship between state fragility and grievances. In order to reconcile this, I have created a fragility index that is a sum of all other efficiency and legitimacy indexes, except for the economic efficiency. This will be used for the analysis of the relationship between developmental grievances and state fragility.
Problems of economic legitimacy could be measured by indicators that measure corruption or the extent to which state institutions are extractive and lack transparency. Another possibility would be to look at how much economic performance adds to the development of the national economy by measuring the dependence of foreign trade on the export of primary commodities. While the former seems to correspond to the intuitive idea of economic legitimacy, the latter, adopted by the State Fragility Index, is interesting as it describes the quality of economic performance of the state. The economic management system may be effective in the production of wealth, but if this wealth production is too singularly focused on the production and trade of primary commodities, it does not create the forward link to development (Senghaas 1985). Furthermore, such a development strategy also creates dependencies that make the national governance of welfare and wealth production difficult. The use of the SFI indicator for economic legitimacy is especially useful in the MENA context, given that some of the problems of state fragility, developmental grievances and conflict seem to be causally linked to the dependence of the MENA countries and great powers on oil and other energy resources. This is something where the MENA region differs drastically from the rest of the world.
However, since the legitimacy of the official economic governance is intuitively associated with the state's ability to curb corruption, I will also need to consult indicators of corruption and transparency to reveal the relationship of this element of state fragility with violence and grievances. I will use Transparency International's Corruption Index, which is a scale based on a meta-analysis of expert assessments on "bribery, diversion of public funds, prevalence of officials using public office for private gain without facing consequences, ability of governments to contain corruption and enforce effective integrity mechanisms in the public sector, red tape and excessive bureaucratic burden which may increase opportunities for corruption, meritocratic versus nepotistic appointments in the civil service, effective criminal prosecution for corrupt officials, adequate laws on financial disclosure and conflict of interest prevention for public officials, legal protection for whistle blowers, journalists, investigators when they are reporting cases of bribery and corruption, state capture by narrow vested interests, access of civil society to information on public affairs." (Transparency International 2019, p. 2) However, this index has a uniform methodology only from 2012, while corruption was measured slightly differently before that. Thus, it is not possible to compare the transparency score before and after 2012. Since such comparison is vital for this study, I have made certain methodological arrangements.
On the one hand, I have used the Transparency International's Corruption Index and investigated the two periods 2012-2018 and 1995-2011 separately. This makes some sense especially in the MENA region where year 2012 represents the first year after the turbulent year of the Arab Spring. However, given that our data are annual, this makes both periods very short, and this limits the statistical significance of our findings. Consequently, I have used the anti-corruption rank of countries as an indicator of relative measure of corruption. The methodologies before and after the turn from 2011 to 2012 data are different, and thus, developments cannot be followed from 2011 to 2012 in each country. Yet, both methodologies measure corruption. Therefore, following the corruption rank makes more sense than comparing the index that before 2012 ranges between 0 and 10, and after 2011 between 1 and 100. The corruption rank measure cannot capture global developments as it assumed that the absolute level of corruption does not change in the world (which is a simplifying assumption). Since there is variation in the capture of the data over year (in 1995 the number of countries included is just 41), the rank of each country is divided by the number of countries captured by the data each year. This way, the indicator varies between 1 (the most corrupt country) and 1/n (n = number of countries in that year's data) and gives the highest value to the most corrupt country 1/n and the lowest value to the best country of the n countries of that year.
Thus, I will use the Transparency International's measures for the study of corruption as an element of economic illegitimacy, while also using the SFI measure of foreign dependent economic management as an indicator of illegitimacy of state's economic management. The two indicators are rather independent from each other both globally and in the MENA region. There is hardly any correlation between dependence on trade in primary commodities and corruption, despite the fact that countries with dependency on oil trade could be assumed to be more corrupt in their economic management.
Unilateral international efforts, by great powers such as the US, UK, France and Russia, at dealing with the violent state failure, have often approached the problem of state fragility from a very militaristic standpoint. These efforts have linked fragility with political violence (and punishing the predatory agents of violence), while much less attention has been paid to the grievance side of the triangle. Since one of the obvious intervening variables that could explain the difference of MENA's relationship between state fragility, grievances and conflict compared to the rest of the world, is the greater likelihood of external unilateral intervention, there is a need, also to define how intervention is identified and measured.
In this article, intervention is defined as external military involvement with warfighting, (defined as something that produces battle deaths) as listed by UCDP conflict data (Pettersson et al. 2019), by great powers in originally intra-state conflict or one-sided violence of a fragile state (with New War State Fragility above 6). (For a justification of such choice, see Kivimäki 2019, Chapter 2, and Kivimäki 2021b, Chapter 3). As great powers, we define the five permanent members of the UN of which four-US, UK, France and Russia-have conducted interventions. As a result of this ruling the list of interventions in the world is as follows (Table 1).

Fragility and Grievances in the MENA Region
There is an intuitive understanding of the relationship between state fragility and human development. It is difficult to develop state capacity if the level of health, education and economic development is low. Equally clear is that human development will not thrive in countries where the state apparatus is incapable to facilitate such development.
Crude correlative testing supports this assumption. 1 In general, the negative relationship between human development and state fragility is statistically very significant all through the indicators of fragility. 2 Unsurprisingly, it is very strong between human development and the legitimacy of social governance. There is a strong association between human development on the one hand and economic legitimacy problems (reliance on trade in primary goods), political efficiency problems and security legitimacy, and a moderate association between human development and security efficiency, political legitimacy, and economic legitimacy (All this supports the findings by Acemoglu and Robinson 2012, for example).
Time-lagging human development variable and time-lagging fragility variables reveal that the lack of development predicts fragility equally as much as fragility predicts problems with human development. If we measure economic legitimacy problems with the relative corruption index, we can see that corruption, too, is very strongly and statistically very significantly associated with problems of human security. There, too, it is impossible to say which influences the other more. The same is true, both for the period before and after the turn from year 2011 to 2012 if we measure this with the original Transparency International's corruption indicator.
In general, the negative association between human development and state fragility is weaker in the MENA than in the rest of the world. The overall legitimacy-related fragility as well as lack of legitimacy of governance in social and security affairs were strongly negatively associated with the level of human development. In MENA the statistical effect of human development on state fragility is slightly stronger than the effect of fragility on development. What is especially peculiar is that economic legitimacy problems are positively (even if negligibly) associated with human development (Unlike assumed by Senghaas 1985 in his theory of forward and backward economic linkages). If we analyse economic legitimacy by using corruption as an indicator, MENA is relatively similar to the rest of the world. There is a very significant, very strong association between corruption and human development. This association is getting stronger over time and seems much stronger after the Arab Spring.
The main anomaly of the MENA region is related to the issue of dependence on primary production: oil and its international impact. If we look at this association by looking at how human development varies in general between the four levels of dependence on trade in primary products (in the SFI data), we can see that the entire positive association is based on the low level of human security in the second lowest category of dependence. The entire positive correlation is due to the fact that in category 1 (of categories from 0 to 3) there are two outliers, Egypt in 1995-2011 and Syria after 2015, with low human development and political instabilities that reduced dependence on the trade in oil. So, the real general MENA-level anomaly is that there is no negative correlation between dependence and human development, not that there was a strong positive correlation. It may be that the oil economy can be better mobilised for social protection and human development than economy based on the production of other primary products. It may be that the heavy investments required for oil exploration necessitate greater attention to grievances. The fixed nature of the assets in the oil business motivates the investing states to make greater efforts at avoiding political instability that could compromise the investments. This could be the reason why some scholars have found evidence of the effect of oil in increasing the stability of rentier states, whose stability is consolidated by rents and 1 Testing the variables that indicate fragility, development, violence and intervention by using Shapiro-Wilk W test reveals that none of our variables are normally distributed. Scatterplots of the relationship between these variables also show that many these relationships are not necessarily linear, and data are not equally distributed about the regression line (homoscedasticity). Thus, when article refers to statistical associations between variables, the reference is to Spearman correlations, rather than Pearson correlations. Spearman correlation does not carry any assumptions about the distribution of the data. Thus, it is the appropriate correlation analysis for this data. 2 This paper uses the Quinnipiac University (Politics) standard for the interpretation of correlative relations. An association is significant if p-value is below 0.05, and very significant if it is below 0.02. Correlation is negligible if correlation coefficient is [0. appeasing social protection from the state's oil income (Basedau and Richter 2014). At the same time, many other factors may intervene in the effect of dependence on oil exports on human development, and there have been theories that have suggested a negative impact of oil dependence on human development (through adverse political effects) (Ross 2001). In addition to the MENA anomaly, there is the anomaly of Egypt and Syria, who both are low human development countries with relatively low dependence on trade in commodities (Syria only after 4 years of war and Egypt until the Arab Spring). In Syria, conflict-related sanctions must have been large part of the explanation for the reduction of oil sales, and they were associated with declining human development. In Egypt the Arab Spring reduced reliance on trade in primary products, while eventually then leading to increasing human development.
If we try to trace the reason for the lack of the global association between dependence and lack of human development in the MENA region, we can see that the anomaly is mostly constituted by the difference of oil producing states and more diverse economies. I will look at each year since the beginning of the Arab Spring in 2011 and compare MENA countries with high economic dependence (ecoleg 2 or 3) on trade in primary produce (mainly oil) with countries with low dependence (ecoleg 0 or 1). There, it is possible to see that until Syria's declining oil exports lowered it from the former to the latter group of countries, the average level of human development was almost the same in both groups. At the same time, the higher dependence countries were lifted to a higher human development level once Syria moved from the group to the lower dependence group after sanctions had lowered its economic dependence on the exportation of oil.
Annual variation in the dependence on commodity production within countries also to some extent explains the positive correlation (or absence of correlation if Syria and Egypt are excluded) between the level of dependence and human development in the MENA region. In general, the level of dependence of countries in MENA region varies only a little. There was no variation or no data in six of the MENA countries. Yet, it was the main oil producing countries, Saudi Arabia, UAE, Iraq, Bahrain, Oman and Qatar, 3 where stable human development was a condition to stable oil production. There, stable oil production also afforded social protection and human development. 4 Interestingly, Egypt affected the human development comparison between economically dependent and independent countries in the MENA region by contributing to the anomaly of MENA. Yet Egypt, was also one of those countries whose year-to-year comparison weakened this anomaly: Egypt's economy grew more with the diversification of its economy.
Despite the description of the correlative relationships between economic dependence on trade in primary commodities and human development, numbers do not explain why the MENA region is different in this respect. Why does dependence on primary commodities have this differing effect in MENA compared to the rest of the world remains a puzzle that only qualitative investigations and zooming into the relationship in individual countries can answer.

State Fragility and Violence: Apparent and Real Anomalies in MENA
In general, our data show that fatalities and civilian fatalities per population are very significantly associated with all state fragility indicators, as many of the scholars of resource mobilisation suggest. (This supports the findings by Commission on State Fragility 2018 ;Tilly 1978;World Bank 1988). Conflict intensity predicts the weakening of the state, while state fragility predicts, almost exactly as strongly, conflict intensification. In the MENA region, in general, conflict fatalities and civilian fatalities are less strongly associated with state fragility in the MENA region than in the rest of the world (see Table 2 below). However, what has not been revealed in the previous research is that the relationship between general state fragility and violence is not linear. Figures 1 and 2 consisting of a scatterplot displaying values for variables of New War State Fragility and conflict fatalities per population show the non-linear relationship between fragility and violence in the world (Figure 1) and in MENA region (Figure 2). MENA region, in general, conflict fatalities and civilian fatalities are less strongly associated with state fragility in the MENA region than in the rest of the world (see Table 2 below). However, what has not been revealed in the previous research is that the relationship between general state fragility and violence is not linear. Figures 1 and 2 consisting of a scatterplot displaying values for variables of New War State Fragility and conflict fatalities per population show the non-linear relationship between fragility and violence in the world (Figure 1) and in MENA region (Figure 2).   Soc. Sci. 2021, 10, 120 10 of 17 MENA region, in general, conflict fatalities and civilian fatalities are less strongly associated with state fragility in the MENA region than in the rest of the world (see Table 2 below). However, what has not been revealed in the previous research is that the relationship between general state fragility and violence is not linear. Figures 1 and 2 consisting of a scatterplot displaying values for variables of New War State Fragility and conflict fatalities per population show the non-linear relationship between fragility and violence in the world (Figure 1) and in MENA region (Figure 2).   Instead of a linear increase of fatalities as state fragility increases, it is possible to see, both in the global and in the MENA data, that while fragility does increase the risk of conflict, the main conflict risks are in the categories of New War State Fragility between 9 and 12 in the MENA region and between 9 and 13 (including both) in the whole world. The most conflict prone level of state fragility is the one of Algeria since 2002, Egypt 1996-2016 (excluding 2012, when the state was stronger), Iran 2004-2012 (after which it was stronger), Libya after the ousting of Gaddafi, Saudi Arabia until 2001, Syria before 2008 and after 2012 and Tunisia before 2001. This corresponds to the findings on the relationship between democratic political legitimacy by Hegre and others (Hegre et al. 2001). Yet, political legitimacy is not the only factor that creates this inverse U-curve relationship between state fragility and conflict. It seems that the mobilisation of the most intensive violence requires some state strength, while at the same time strong states with legitimacy also manage to regulate competition and disagreements and contain violence. The inverse U-shape could also be related to conflicts through the economic logic that was revealed already in 1955. Economic growth, which is crucial for the alleviation of developmental grievances, requires some economic inequalities and then perhaps also fragmentation in society, while too much inequality hampers growth as it reduces society's ability to avoid growth disrupting conflict (Kuznets 1955).
The global association is strong between fatalities per population and civilian fatalities per population, on the one hand, and new war fragility and general legitimacy fragility, on the other. It is moderate between conflict fatalities, on the one hand, and the general effectivity fragility indicator, political efficiency and legitimacy fragility, economic efficiency fragility, and social efficiency and legitimacy fragility indicators, on the other. Only weak associations can be found between both fatality indicators and economic legitimacy, while civilian fatalities are also weakly associated with political and economic effectivity fragility. Fragility and violence predict each other almost as well, while it seems that state fragility is a better predictor of civilian fatalities than civilian fatalities of fragility.
Corruption is very significantly associated with both conflict fatalities and civilian fatalities of conflict. The association is strong between conflict fatalities and corruption perceptions index after 2011 and moderate before that, while the association between civilian fatalities and corruption is moderate and strengthening through the whole period. This way, our findings support the mainstream findings on the association between corruption and conflict (Rus 2014). There is no systematic direction of the association: Violence predicts corruption just as corruption predicts violence. Violent international punishments of corrupt leaders tend to increase rather than decrease corruption, and thus, less militaristic measures would be more useful in the fight against corruption (Kivimäki 2021b). While in general the association between violence and corruption is similar in MENA, corruption predicts violence more efficiently there than violence predicts corruption. The association is clearest in the main conflict countries: Yemen, Libya and Syria, but also in Algeria, and Iran. In Iraq the data are missing, while in Egypt, Morocco, Kuwait, Lebanon and Turkey the association is negligible.
In the MENA region, the association between fragility and violence is weaker than in the rest of the world except when it comes to political legitimacy problems, i.e., factionalism, political discrimination, political salience of elite ethnicity, fragmentation and exclusionary ideology of the ruling elite. This association has been theorised by thinkers of the so-called subaltern realism, who have emphasised the need for third world states to consolidate their state power before they start developing checks and balances or humanitarian constraints to such power (Ayoob 1991(Ayoob , 1997Azar and Moon 1988). However, these theories have not discovered how MENA-specific this conflict problem of subnational fragmentation is. This aspect of political governance-state apparatus serving the interests of some groups only-explains 43% of the variation in fatalities of political violence per population in the MENA region (and 16% in the rest of the world), if we assume that the direction is from fragility to conflict rather than the other way around. The overall fragility level is strongly associated with conflict fatalities and moderately with civilian fatalities in the world but only moderately with conflict fatalities and weakly with civilian fatalities in the MENA region. The following table reveals the differences between MENA and the world in detail by showing the non-parametric correlations between the violence and fragility variables (all correlations are statistically very significant, the number of observations for the world is 3906, for the MENA: 432).  Table 2 reveals two stark differences between the region and the rest of the world. One is the greater association between political legitimacy problems and conflict in the MENA area, and the other is the negative association between economic legitimacy fragility (reliance on the trade in primary commodities) and conflict. It seems that primary production, i.e., oil, makes MENA's conflicts different. Oil economies tend to invest more on stability-enhancing social protection and, thus, avoid conflict better. Yet, the relationship between the two cannot be revealed without a fuller analysis of a number of independent, overlapping, mediating, moderating and proxy variables that interact with the effect of oil dependence on development (Ross 2012).
It would be reasonable to expect that US intervention is an intervening variable that somehow explains the variation in the association between primary production and violence. External intervention in general and US intervention, specifically, is associated with increased number of fatalities of conflict per population. In the MENA region, this is especially true. While in the world, there will be more than five times the number of fatalities if the US intervenes in the conflict than if the conflict is allowed to develop without US intervention. In MENA, US intervened conflicts are more than 10 times more fatal. With one-year time lags in data, we can also see that both in the world and in MENA, US intervention predicts increased fatalities more than increased fatalities predict US intervention.
Analysis of the data suggests that in the world the association between primary production and violence is strong if the US intervenes in violence, whereas it is negligible if there is no US intervention. However, the association is strongly negative in the MENA region if there is a context of US intervention, whereas it is weaker if there is no such context. So, US intervention implies a negative relationship between trade in primary trade and conflict in the MENA region, and the opposite implication in the rest of the word! In other words, reliance of countries on ordinary commodity trade predicts violence, especially if US intervenes, but it predicts lower levels of violence in the MENA, especially if the US is militarily involved in the country during the year. If one adds one-year time lags to the variables on violence or those on state fragility, one can see that dependence on trade in primary products predicts violence in the world slightly more than the other way around if US is involved militarily. In MENA, the impact goes both ways with equal strength independent of US intervention.
An investigation of variation over time within countries (panel data models) does not reveal a clear picture of what contributes to the MENA anomaly. A comparison between countries produces an almost equally confusing picture. However, if one takes out Syria as an outlier, we can see that the real MENA anomaly is that the region lacks any correlation between conflict and reliance on export of primary products (whereas globally there is a positive correlation). The negative correlation between the two is explained by the developments in and sanctions against Syria.
The profile of the much stronger association between problems of political legitimacy and violence in the Middle East and North Africa than in the rest of the world is interesting. If we look at the distribution of observations in a two-dimensional graph (see Figure 3) and add a Lowess curve to describe the association, we can see that all major events of violence have taken place in countries with major problems of political legitimacy. The Lowess curve is useful especially in non-parametric strategies for fitting a smooth curve to data points. Since none of the variables used in this study are normally distributed, using the Lowess curve for our descriptive statistics is appropriate. The curve shows that almost all serious violence takes place in countries within the highest political legitimacy problem category. Tunisia is the only country with no political legitimacy issues and yet has experience of major violent events, especially in 2015 and 2016 in the MENA region. Soc. Sci. 2021, 10, 120 13 of 17 An investigation of variation over time within countries (panel data models) does not reveal a clear picture of what contributes to the MENA anomaly. A comparison between countries produces an almost equally confusing picture. However, if one takes out Syria as an outlier, we can see that the real MENA anomaly is that the region lacks any correlation between conflict and reliance on export of primary products (whereas globally there is a positive correlation). The negative correlation between the two is explained by the developments in and sanctions against Syria.
The profile of the much stronger association between problems of political legitimacy and violence in the Middle East and North Africa than in the rest of the world is interesting. If we look at the distribution of observations in a two-dimensional graph (see Figure  3) and add a Lowess curve to describe the association, we can see that all major events of violence have taken place in countries with major problems of political legitimacy. The Lowess curve is useful especially in non-parametric strategies for fitting a smooth curve to data points. Since none of the variables used in this study are normally distributed, using the Lowess curve for our descriptive statistics is appropriate. The curve shows that almost all serious violence takes place in countries within the highest political legitimacy problem category. Tunisia is the only country with no political legitimacy issues and yet has experience of major violent events, especially in 2015 and 2016 in the MENA region. Given the role of Syria in the explanation of the anomaly of the MENA region with regards to the associations between violence and trade dependence, and human development and trade dependence, it would seem natural to test the role of Syria with regards to the MENA anomaly about stronger association between conflict and political legitimacy problems. However, the stronger association between political legitimacy problems and violence is not related to Syria. In fact, without Syria the MENA Region would be even more anomalous in this respect. If we then look at the countries with the highest levels of political legitimacy problems or highest levels of conflict fatalities per population, we can see that no single country dominates in the explanation of the MENA region's stronger association between war and political legitimacy problems. The relationship is Given the role of Syria in the explanation of the anomaly of the MENA region with regards to the associations between violence and trade dependence, and human development and trade dependence, it would seem natural to test the role of Syria with regards to the MENA anomaly about stronger association between conflict and political legitimacy problems. However, the stronger association between political legitimacy problems and violence is not related to Syria. In fact, without Syria the MENA Region would be even more anomalous in this respect. If we then look at the countries with the highest levels of political legitimacy problems or highest levels of conflict fatalities per population, we can see that no single country dominates in the explanation of the MENA region's stronger association between war and political legitimacy problems. The relationship is also clearly mutual between the two variables; there seems to be a typical mutually constituting and reinforcing relationship between the two. Yet, factionalism, political discrimination, political salience of elite ethnicity, fragmentation and exclusionary ideology of the ruling elite predicts conflict fatalities more than the other way around.
The strong association between political legitimacy problems and violence is not primarily because of changes within countries. Variation in legitimacy does not lead to variation in the number of conflict fatalities per population, except in Libya and Yemen, whereas in six MENA countries, there was no variation in political legitimacy, while in other countries the association between the two variables was weak, negligible or nonexistent and in some cases negative (Tunisia, for example). Thus, the reason for the strong association between the two variables was the fact that countries without serious political legitimacy issues did not have serious problems with violence.
Testing the levels of violence between groups of nations in the highest political legitimacy problem category (3 out of categories from 0 to 3) with the rest of countries makes sense, on the one hand, because that divides countries in roughly equal groups, and on the other, because Figure 1 shows that the main problem with violence is with the highest category of political legitimacy problems. By comparing these groups in each year, it is possible to see that the average annual number of conflict fatalities per population was in each year of observation  at least 16 times higher (and often much higher), and on average more than 35 times greater in countries of the highest political legitimacy problem category compared to the rest of MENA countries. Clearly the strong association between the two variables is because of differences between MENA countries rather than between years. Syria, Libya and Iraq with massive political legitimacy problems and extensive political violence compared to Morocco, Kuwait, Tunisia and Oman with political legitimacy and relative peace is what this association is about.

Conclusions
To understand state fragility in the Middle East and Northern Africa, we need to understand to what degree MENA region is similar to the rest of the world, to what degree it is different, and to what degree individual countries different from the regional or global patterns. This article has shown that state fragility predicts developmental grievances and violence, while violence and grievances also predict state fragility. The relationship can be causal to both directions or a relationship of mutual constitution in which part of the interaction is not exogenous but partly conceptual and partly related to our knowledge and interpretations on the related elements. The interpretation of the role of sub-state groups, such as religious, ethnic or regional groups within states has a crucial role especially in the explanation of violence in MENA.
In addition to similarities, the MENA region also has its unique problems related to state fragility. The regional economy is very dependent on the production of oil and other energy resources. It seems clear that the relationship between such reliance and conflict as well as such reliance and development is very different in the MENA region than in the rest of the world. Egypt and Syria are the most different from the rest of the world in this respect. To trace the dynamics of this anomaly and the map the causal complex between oil dependence and conflict/grievance will require analysis that is informed by MENA-specific understandings. Furthermore, the region is different from the rest of the world in the sense that it has been exposed to external intervention more often than the rest of the world. External intervention, especially by the United States, has affected both state fragility and conflict in the region.
Finally, the region has suffered from problems of political legitimacy, especially factionalism in polities. This factionalism and subnational challenges to the state lie much closer to the problem of political violence in the MENA region than similar problems in the rest of the world. It would be possible to claim, on the basis of this paper's exploration, that the relationship between factionalism and conflict is the dominant problems of state fragility in the MENA region.
Numbers can reveal the nature of the difference, as has been demonstrated above, but the understanding of it requires closer qualitative analysis of the different elements, different countries and different historical contexts. This is especially true for the associa-tions that seem to be mediated through interpretations of realities of subnational groups. If we think of the prominence of problems of political legitimacy in the explanation of MENA conflict, for example, it may be important that we do not interpret the relationship simplistically as purely exogenous. If fragmentation predicts conflict, this knowledge may actually strengthen the real reason for conflict in fragmented MENA societies. Resistance of diversity and forced assimilation, as well as the idea that the state belongs to one group and others should assimilate into it, may very well be what constitutes conflict realities in some MENA societies. This, even if correlative analysis suggests that its proxy-fragmentation and discrimination-is behind violence. Thus, we need explanations deeper from the societal dynamics. This is the task of the other articles of this special issue.
Funding: This research was funded by the UK Global Challenges Research Fund project "Conflict and peace-building in the MENA region: is social protection the missing link?" (Grant number AH/T003537/1).

Institutional Review Board Statement:
The study was conducted according to the ethical guidelines of the University of Bath and the project was approved by the university's ethics committee.

Informed Consent Statement: Not applicable.
Data Availability Statement: The data (Kivimäki 2021a) presented in this study are openly available in the University of Bath Research Data Archive, at doi:10.15125/BATH-00951.

Conflicts of Interest:
The author declares no conflict of interest