List of Figures
List of Tables
2.1 Growth Instability
2.2 Conflicts, Economy and Demographics
3 Methodology and data
List of Figures
Figure 2.1: Youth bulges and conflict
Figure 2.2: Youth dependency's influences on growth volatility
Figure 4.1: Regional distribution of years with conflict from 1975-1990 and youth dependency 1975
List of Tables
Table 4.1: Regional break down of GDP growth development 1960-1990
Table 4.2: Baseline regressions
Table 4.3: Regressions on growth volatility including conflict indicators
Table 4.4: Regressions on growth volatility including change in conflicts indicators
Table 4.5: Robustness checks for change in conflict indicators
In cities in six West African countries I saw [...] young men everywhere-hordes of them. They were like loose molecules in a very unstable social fluid, a fluid that was clearly on the verge of igniting. (Kaplan, 1994)
With his article "The Coming Anarchy" Kaplan ignited a discussion over the threat posed by youth cohorts to the civilized world. More than 15 years later there is still no clear evidence whether large populations of young men are a main factor in determining conflict risk or not. This paper now tries to connect the topic of demographically induced violence with another contemporary topic of development economics: growth volatility.
Simple examples are enough to become aware of the importance of growth stability because already small growth rates - if stable over time - can lead to a tremendous development. A steady growth of annually 2% leads to a doubled GDP over the course of 35 years. 4% annual growth, a reasonable growth rate for many developing countries (The World Bank, 2011b), more than triples the GDP in 30 years. While many developing countries had satisfying steady growth rates after the Second World War until the mid 1970s, development crashed afterwards, especially in Latin American and African countries. If growth rates had been stable, today we would not speak of the 'East Asian Miracle' because many other countries could have developed a similar level of welfare. However, mostly the Asian countries experienced steady progress. Thus economists tried to analyse possible underlying reasons for a country's growth instability. Dani Rodrik argued that weak institutions and latent social conflict were vital determinants of the economic breakdowns in the mid 1970s (Rodrik, 1999). Since his article many other economists have investigated the role of democracy and institutions on growth and volatility and underlined the importance of this relationship (e.g. Acemoglu et al., 2001; Easterly, 2001a).
Still, one particular flaw of Rodrik's explanation was the fact that the Asian dummy, included to control for special characteristics of Asia, remained significant throughout his regressions. Thus although institutions and social conflict were significant influences, they could not explain the East Asian Miracle (Rodrik, 1999). An interesting paper which claimed to have changed this was written by Brant Liddle (Liddle, 2003). In his working paper he bridged two former mostly unconnected literatures, population-economic growth and growth stability. Due to the introduction of youth dependency into his regressions the Asian dummy showed no longer relevant significance (Liddle, 2003). However, this was probably caused by multicollinearity since he included three regional dummies into his regression with only a few out of 94 countries not being covered by them. Hence in the extended published version in the Journal of International Development (so far only available online) he did not put further emphasis on the Asian dummy (Liddle, 2010). Nonetheless his findings are interesting and will be the basis for the following analysis.
Liddle offered two theoretical explanations why youth dependency (the level of age dependency in his published version as well) should matter in determining growth volatility. His first hypothesis was that low dependency ratios and therefore fewer dependents in a society allowed for a more flexible way of adjusting policy to external shocks. Secondly, high dependency rates or growing dependency rates could have added to latent domestic social conflict and by that may have influenced volatility as described by Rodrik (Liddle, 2010). A third explanation could be that high shares of young population not only added to latent social conflict but provided the basis for steady violent outbreaks and by that also impaired economic soundness.
Violent conflicts can induce high costs onto a country's economy (Bozzoli et al., 2010) and while it is not undisputed whether high youth dependency has a causal effect on the onset of civil war it is interesting to examine whether violent conflicts might have been the channel through which the youth dependency Liddle found to be important influenced growth stability. Therefore this paper tries to analyse the hypothesis that higher youth dependency increased the risk of conflict and that the increased number of conflicts led to a rise in growth volatility. The evidence shown in this seminar paper seems to support this hypothesis. Before presenting it, the next part will give deeper background information on the existing literature about conflict, demography and growth. Afterwards the dataset will be introduced and the regressions results will be presented.
2.1 Growth Instability
While several famous articles of development economics focused on average growth rates or on absolute levels of development and the underlying growth constraints ; Easterly et al., 1993 pointed out that growth rates were highly unstable although the fundamental country characteristics did not change much over time and thus should not be simply averaged for policy implications (Easterly et al., 1993; Rodrik, 1999). They argued that this low persistence of growth rates over the decades was caused by external shocks (Easterly et al., 1993). Further research on this topic led to the calculation of country specific break years in which the growth trend changed significantly (Pritchett, 2000). Partly similar earlier ideas of takeoffs (Rostow, 1991) and the accompanying idea of Big Pushes in foreign assistance were disputed in the literature. Still, the greater volatility after the mid 1970s is obvious and was therefore focus of further research.
In 1999, Dani Rodrik added new insights to the discussion about growth instability through his article 'Where did all the growth go?' (Rodrik, 1999). He rejected the underlying importance of external shocks for this volatility since countries like Brazil or South Korea experienced shocks of similar altitude while their growth pattern diverged. Classical macroeconomic factors like investment shares furthermore remained stable while indicators for human capital, health status and policy quality even improved (Easterly, 2001b). Rodrik therefore included social indicators like democracy measured by the Freedom House Index, institutional quality based on the International Country Risk Guide and the ethnic cleavages proxied by an index of ethnolinguistic fragmentation. He found significant influences of all of these social indicators on the growth breakdowns after 1975. This held true for various subsamples and different proxies. For a test of robustness he included classical explications like trade openness, government GDP share or government debt but these did not add explanatory power and entered insignificantly. Social indicators thus seem to be a major cause determining economic performance by impeding or enabling adequate macroeconomic responses to crises (Rodrik, 1999). After having published his work, other authors tested the importance of institutions for the absolute level of development / their influence on growth via mitigating the effects of ethnic cleavages and found evidence in these approaches as well (Acemoglu et al., 2001; Easterly, 2001a).
A new aspect to this topic added Liddle with his paper combining demography with the growth volatility issue (Liddle, 2010). As Rodrik he analysed the underlying factors of changes in growth rates. He examined how population and especially different age structures of populations affect the resilience of an economy to shocks. Earlier literature, trying to connect demography with GDP growth, concentrated on the effects of population size on growth rates or per capita outputs, which led to no significant findings. Liddle stated that more recently other specific links were found by the decomposition of aggregated demographic indicators into separated components. He summed up the current state of research by concluding that there exists evidence indicating that a growing working force of a country fosters economic growth while increases of the dependent population do not. So the demographic transition into a more mature society produces a 'demographic gift' (Bloom and Williamson, 1998 in: Liddle, 2010). Adding to this research he examined the effect of age dependency levels and their rates of change onto the economic stability after the mid 1970s. Including youth indicators into regressions, otherwise similar to Rodrik's, youth dependency changes entered significantly with a negative sign - meaning that the change to a lower dependency increases economic stability - while the actual level was insignificant. Other variables showed similar signs, significance and magnitude as in Rodrik's regressions. Unlike the case of youth dependency, aged dependency levels in 1975 did enter significantly in his regressions. He attributed this to a second demographic dividend, which might provide a higher capital per worker and incentives for capital accumulation (Liddle, 2010). He further analysed whether lower fertility might influence economic growth via higher investment or savings rates but refused this thesis based on his results. Thus, he presented the two hypotheses mentioned in the introduction that either less dependents enable more flexible reaction to crises or lower the potential of social conflict.
A link, which would be theoretically appealing, might be the relationship between youth cohorts and the increased risk of civil conflicts. Via the indirect route over conflicts youth dependency could hinder economic adaption to new challenges. This sounds somehow similar to Liddle's hypotheses but connects them both and provides a specific channel through which this relationship could function. Therefore, in the following chapter the interrelationships between conflicts and the economy as well as the background of demographic influence on the risk of violent conflict will be illustrated.
2.2 Conflicts, Economy and Demographics
Conflicts are inherent in our societies and emerge from the political and economical structure of them (Keen, 1997). In the extreme case, they occur as international wars between states or as civil wars within a country, where civil wars account for the majority of onsets and victims (SIPRI, 2010). Because of the latter and since the literature about demography and conflict focuses on civil wars as well (Urdal, 2004), international conflicts will not be considered in this paper. Furthermore, although conflicts may also appear in the form of uprisings, gang violence or other social disturbances and there exists evidence that high cohort sizes of young men do have an effect on their appearance (Urdal and Hoelscher, 2009), this aspect will not be in the scope of the paper.
The most apparent consequences of civil conflicts are the over 16 million battle deaths since 1945 (Fearon and Laitin, 2003). Besides the direct human suffering, channels through which the economies of countries ravaged by conflicts are affected are social disorder and increased transaction costs, lowered public expenditures for output-enhancing measures and depletion of capital and human capital stock (Collier, 1999). Another possible effect, which is difficult to quantify, is an alteration of preferences due to the course of war (Bodea and Elbadawi, 2008,Voors et al., 2010).
Estimations of the effects on the economy vary greatly between different studies. Bozzoli et al., 2010 summarized various preceding studies on the cost of conflicts and showed that for the same conflict (in this case the armed conflict in Sri Lanka) estimated costs varied from 2 to 22 billion US Dollar depending on the definition and scale of war related costs. Likewise it is unclear what determines if countries after conflict soon return to their old growth path, if they even recover the losses through increased growth or whether they stay behind due to the long-term consequences of previous warfare. Collier, 1999 found that even if a war is short (one year) growth is impaired for the following 5 years by about 2.1 % - a number quite similar to his estimate of a 2.2% lower GDP growth rate during conflict. Other authors found dramatically lower results. Imai and Weinstein, 2000 concluded that in the worst case civil conflicts cost 1.25% of yearly growth rate (Imai and Weinstein, 2000 in: Bozzoli et al., 2010) while Gupta et al., 2004 calculated an impact of -2.17% annual growth mainly through the effects of increased military expenditures (Gupta et al., 2004 in: Bozzoli et al., 2010). Bozzoli et al., 2010 therefore concluded that the evaluation of economic costs of warfare is still in its infancy and many studies are lacking comprehensiveness and consistency.
Besides decreasing an economy's growth potential, conflict may also increase its growth volatility. A fact, which would confirm the hypothesis that youth dependency is affecting growth volatility via civil conflicts. Bodea and Elbadawi, 2008 constructed a model covering the impacts of political violence on economic growth based on an approach from Abadie and Gardeazabal, 2008 who analysed the effects of terrorism on economic activities. Starting from a simple AK model they included political violence as an innovation process which destroys share of the capital stock. Assuming a constant relative risk aversion utility for the consumers they came to the conclusion that political violence does hamper growth and growth stability (Bodea and Elbadawi, 2008).
In addition to the wide literature about economic costs of conflicts, there also exists a diverse literature regarding economic causes for the onset of war. In political science discussions, conflict was long thought to be mainly motivated by grievances due to inequalities or exclusion. Economic analysis of this subject then rejected most of this view because indicators of grievances like ethnic fragmentation and religious or linguistic discrimination by the state showed insignificant in regressions ran regarding the risk of civil wars (Fearon and Laitin, 2003). A well-known paper by Collier and Hoeffler tested grievance explanation as well and contrasted it with the argumentation that not grievance but greed is one of the main causes of conflict (Collier and Hoeffler, 2004). They argued that opportunities during the war are of paramount importance for the onset of warfare and the recruitment of new troops. A typically example of opportunities is the extraction of natural resources like diamonds by conflict parties in West Africa or cocaine in Colombia but donations from diasporas or help from hostile governments can also provide 'greed'. Important for the existence of opportunities are furthermore the alternatives conflict participants have; e.g. do they face unemployment or what are their schooling possibilities. Collier and Hoeffler, 2004 measured grievance similar to the approach by Fearon and Laitin, 2003 through political rights (measuring exclusion), economic inequality and ethnic and religious fractionalization. Likewise Fearon and Laitin, 2003, Collier and Hoeffler, 2004 found no support for the grievance perspective because only one indicator, ethnic dominance, entered significantly into the baseline model. Instead they found strong evidence for the 'greed' perspective of civil wars. Primary commodity exports, diasporas and per capita income measuring costs of rebellion all added significantly to the risk of civil war outbreak. This coincides with a recent survey made by B0as, Tiltnes, and Flat0, 2010 regarding the underlying reasons for joining rebel movements or gangs (B0as, Tiltnes, and Flat0, 2010 in: The World Bank, 2011a). They found that 40% respectively 46% of all respondents joined them due to unemployment or idleness. Collier and Hoeffler, 2004 thus concluded, that motive is not unimportant for war participants but that perceived grievances are not closely connected to the measured indicators like inequality or political rights.
So how does that fit to youth dependency as a potential causation of conflicts? Fearon and Laitin, 2003 as well as Collier and Hoeffler, 2004 included variables covering the share of young people, more precisely the proportion of young male people in their regressions. Still none of them found them to be significant although the indicators entered with the expected signs. They argued that other factors like GDP per capita, the extortion of natural resources or military advantage determine the risk of civil conflict. In fact, for a long time
 While some recent studies found evidence (Collier et al., 2009; Urdal, 2006), older studies often rejected the importance of youth cohorts (Collier and Hoeffler, 2004). Furthermore in a recent article by Goldstone et al. they do not consider demographic variables in their predictive system (Goldstone et al., 2010). More on this in the next chapter.
 One of the earliest sources of this phrase: (The World Bank, 1993).
 Youth dependency is defined as (population aged 0-14) / (population aged 15-64)
 Age dependency is defined as (population aged over 65) / (population aged 15-64)
 E.g. Barro, 1991 regarding general influences on the GDP growth rate between 1960-1985 or Mankiw et al., 1992 explaining different states of development in the world through the use of the Solow-model.
 Basis for Pritchett's article was a working paper at the World Bank in 1997 to which Rodrik refers in his article 1999 (Pritchett, 1997).
 Easterly, 2006e.g. rejected the theory of 'Big Pushes' in development assistance and likewise denied the prevalent existence of takeoffs. However, this does not mean that he was not aware of the fact that growth decreased dramatically after the 1970s in most of the countries (see Easterly, 2001b for an analysis of this observation). Hausmann et al., 2005 analysed cases of accelerations (not similar to the definition of Easterly's takeoffs) and possible causes. But their conclusions might be led astray by an error in their dataset (Jong-A-Pin and de Haan, 2008).
 According to Rodrik using Pritchett's specific break years did not change the results.
 In contrast to the conclusions based on this model, Klomp and de Haan, 2009 could not affirm a significant influence of conflict onto growth volatility. In a GMM model they included four factors covering different aspects of political instability. While the two factors describing regime or government instability entered significantly, the two factors characterizing internal conflict and revolutions respectively riots and protesting showed no significance at standard levels (Klomp and de Haan, 2009).
 Military advantage is describing the capabilities of the government to counter the rebels and by that somehow the perceived chance of success of a rebellion. Collier used geographically dispersion, mountainous terrain and social fractionalization as proxies.
- Quote paper
- Robert Messerle (Author), 2011, Conflicts, Demography and the Economic Volatility in Developing Countries, Munich, GRIN Verlag, https://www.grin.com/document/179979