3.4 Dataset
3.4.2 Main Variables
We have a panel dataset with 50 African Countries, 479 ethnic groups and 50 group of non-ethnic groups, tracked over 39 years (from 1975 to 2014).
downwards. This means that using GREG will tend to bias the results against us and this attenuation bias makes them appear less strong than they are".
Table 3.1: Summary statistics in Country-level Analysis Dataset
Variable Mean Std. Dev. N
Conict 2.025 0.748 1407
Political Rights Quality 2.602 0.686 1923 Oil Production (Lag) 0.157 0.364 1438
Autocracy 0.408 0.492 1923
Population Growth (Lag) 15.662 1.323 1587 Mountainous Terrain 0.113 0.196 1519
Gini Indicator 0.202 0.333 1493
Soil Fertility 0.303 0.204 1795
Est Africa 0.248 0.432 1795
Central Africa 0.173 0.378 1795
Gold Production 0.491 0.5 1325
Diamond Production 0.356 0.479 1553 Economic Growth (Lag) 0.032 0.073 1672
Time Span 4.242 5.338 1923
Table 3.2: Summary statistics in Ethnic-level Analysis Database
Variable Mean Std. Dev. N
Conict 1.598 0.807 9892
Political Rights Quality 2.668 0.609 9988 Oil Production (Lag) 0.194 0.395 9437
Autocracy 0.246 0.431 9988
Population Growth (Lag) 16.134 1.253 9986 Mountainous Terrain 0.103 0.152 9808
Gini Indicator 0.29 0.374 9982
Soil Fertility 0.316 0.185 9970
Est Africa 0.296 0.457 9970
Central Africa 0.194 0.395 9970
Gold Production 0.662 0.473 8222
Diamond Production 0.364 0.481 9836 Economic Growth (Lag) 0.009 0.085 6696
Time Span 2.389 2.555 9988
Table 3.3: Summary statistics in Non-Ethnic-level Analysis Dataset
Variable Mean Std. Dev. N
Conict 1.735 0.818 604
Political Rights Quality 2.635 0.676 1111 Oil Production (Lag) 0.153 0.361 691
Autocracy 0.474 0.5 1111
Population Growth (Lag) 15.223 1.291 754 Mountainous Terrain 0.096 0.176 713
Gini Indicator 0.139 0.287 667
Soil Fertility 0.301 0.208 1003
Est Africa 0.26 0.439 1003
Central Africa 0.14 0.347 1003
Gold Production 0.468 0.499 570
Diamond Production 0.361 0.481 732
Economic Growth (Lag) 0.028 0.07 915
Time Span 3.657 5.033 1111
Dependent Variable
Conict Indicator: as showed in previous section, we have developed a conict indicator by aggregating information from SCAD, ACLED and UCDP. In particular, we have measured a weighted mean of violent event over every year covering the timespan 1975 - 2016. This variable ranges from 1 to 3 (respectively, peace and civil war, passing through armed conict).
Commodities
As broadly showed in economic literature, many conicts in Africa are related to commodities and to a balance of strength and balance of control on resources.
Oil exporter: a dummy variable taking a value of 1 if in a given country and year the fuel exports (in % of merchandise exports) are above 33%. Variable from Fearon and Laitin (2003), but updated with recent data of fuel exports (in % of merchandise exports) from World Bank Indicator (2015).
Oil Production (Lag) per capita: From Humphreys (2005) we take the average amount per capita of oil extracted per day in a given year, measured in millions of barrels per day.
Gold production dummy: takes a value of 1 when there is gold production in a country year, and 0 otherwise. From Lujala et al.(2005).
Diamond production dummy: takes a value of 1 when there is diamond production in a country year, and 0 otherwise. From Lujala et al.(2005).
Oil Gini: developed by Morelli and Rohner (2015), this indicator is a time-varying measure of the unevenness of oil eld distribution across ethnic groups for a given country and year.
Ethnic Oil Concentration: developed by Morelli and Rohner (2015), this variable is created by interacting oil exporter countries with the oil Gini Indicator. It reects the surface of an ethnic group's territory covered with oil and gas as a percentage of the country's total surface covered with oil and gas.
Institutional Variables
We use few institutional variables to contextualize conictual countries. Usually, conict studies use only indicators from Polity IV. Here, instead, we use other Databases.
Political and Civil Rights Quality15: is taken from Freedom House (2016). This score is based on a rating from 1 to 7 for Political Rights and civil liberties, with 1 representing the most free and 7 the least free. We have revised the score, by aggregating this variable into 3 scores of right quality.
Public Expenditure16: From WDI we take data on African general government nal con-sumption expenditure (% of GDP)17includes all government current expenditures for purchases of goods and services (including compensation of employees). It also includes most expenditures on national defense and security, but excludes government military expenditures.
Autocracy: we have created a dummy variable of autocracy from Polity IV(2015). In this in-dicator, democracy variable ranges from -10 (strongly autocratic) to +10 (strongly democratic).
We have created autocracy variable from democracy, so if democracy in previous year in a given country is low, we assign to autocracy the value 1. Otherwise, it is 0.
Population Growth (Lag): from WDI, we measure Population Growth (Lag) from the total population. Total population is based on the de facto denition of population, which counts all residents regardless of legal status or citizenship.
Political, Environmental and Religious Conict: each root of conict is a distinct dichoto-mous variable assuming 1 when the country, or the ethnic or the non-ethnic group is experienc-ing that specic root of conict (namely, political, or environmental or religious conict). These variables have been designed as explained in section 3.4.
Geographical Variables
Est Africa: Dichotomous variables assuming 1 if the country is located in East Africa.
Central Africa: Dichotomous variables assuming 1 if the country is located in Central Africa.
Mountainous terrain: percentage of territory covered by mountains. From Collier et al.
(2009). This variable is broadly covered in economic literature. Fearon and Laitin (2003) are the rst who have found that mountainous terrain is one of the main driver to civil war.
Soil Fertility18: From the Harmonized World Soil Database (Nachtergaele et al., 2008). Their complete global grid of nutrient availability is ranked from 1 (no or slight constraints) to 4
15This variables is really important for our analysis. As we have previously said, often non-ethnic groups protest for political, labour, press freedom. Therefore, this variable helps us to understand whether it exists a link between the escalation of violence and the quality of rights.
16We nd important also military spending. However, this variable has too many missing. This is a shame, since of course it is important to take into account the militarisation process of a country at risk of civil war. , missing data on weapons are related to the fact that Africa has a huge weapons black market coming from all over the world. For instance, after Ghedda's collapse, Libyan weapons have been dispatched to various destinations, such as Syria, Nigeria, Tunisia, Mali and more. In Libya there were 22.000 MANPADS (Man-portable air-defense systems) and 450000 military weapons. Only 5000 MANPADS and 12000 weapons are still in Libya, the others are disappeared. We do not know for sure where these weapons are, but if we have a look at the recent history, we can imagine who are the new owner: many armed groups live above and under Sahara desert. Therefore, it is dicult to have credible data on military expenditures and, aboveall, on weapons market. However, this variable is crucial for predicting civil wars.
17This variable is crucial for our analysis. Whether a government spends its budget for public interest reason may foster a conict. However, public spending is an ambiguous variable since if it is true that more money for public services reduce grievance, it is also true that it is important as well to know how these money are used.
Corruption in developing countries is a concern and corruption may aect good governmental intervention.
18This is an important variable because in Africa the majority of people rely on agriculture for their daily diet and, often, for their income. Moreover, climate change is aecting the fertility of many lands and desertication is a cruel cause to poverty and conict. The link between Soil Fertility and conict is both direct and indirect:
communities are forced to emigrate to nd fertile lands. They might bump into other communities who do not want to share their land. In history, we have several cases of communities ghting each other because one group uses the land for livestock and the other for agriculture. Usually, we interpret these kind of wars as ethnic conicts, because two ethnic groups are competing, but actually it is simply a conict for the survival. Some other communities prefer to escape to cities, already crowed, and this increases tensions in towns or illegal emigration problems (with European countries complaining about emigrates and greedy people proting from this illegal business).
(very severe constraints), and also including categories 5 (mainly non-soil), 6 (permafrost area) and 7 (water bodies). Our dummy takes a value of 1 for categories 1 and 2, categories 3 to 6 get a value of 0, and category 7 is set to missing.
Economic Growth (Lag): from Penn World Tables, we measure the lag of Purchasing Power Parity adjusted GDP per capita at constant prices. When there are missing, we use estimates of growth rate of per capita income provided by the World Development Indicators.