Why do governments engage in civilian killings?

In a paper published in the Journal of Defense and Peace Economics (2008, vol. 19(2)) by colleague Sang Hoo Bae and Attiat F. Ott, we have investigated theoretically and empirically the predatory behavior of government: the case of mass killing. As such journal papers go; the model and the empirical analysis are not for general consumption. Nonetheless, the story it tells is not difficult to follow and the conclusions it arrives at is worth highlighting.
Before I summarize our findings, I need to acknowledge a contribution to this topic made by Hugo Slim in his book: Killing Civilians: Method, Madness and Morality of War (Colombia University Press, 2008), reviewed by the Economist (February 16th, 2008, p.92). Although I have yet to read the book, the Economist’s review touches on a few of the issues we have addressed on our paper on mass killing. Our findings may go some way to answer the basic question raised: why kill civilians? Let me begin by citing few relevant statistics.
In the 20th century there were 109.7 million civilian killings by the state. This amounted to 4.35% of world population. In the 19th century, these were 19.4 million deaths or 1.65% of the world population, an increase of over 500 percent. According to the World Development Report 2005, most of the violent conflicts leading to civilian killings took place in low income developing countries, with 38% of these taking place in Africa.
Violent conflicts leading to mass killings of civilians fall into three categories: military conflict between states; between states and non-state groups; and between factions within a state. When we talk about civil war we are referring to the third category commonly referred to as “intra state” war. In this type, either the state (government) or a rebel group is the initiator. Examples of civil war include Angola civil war lasting 27 years with more than one million deaths, repeated wars in Sri Lanka with 500,000 deaths, Rwanda with over a million death to name a few. Mr. Slim gave reasons for civilian death including “a desire to exterminate an entire group of purportedly inferior beings; a lust for power and domination, necessity or plunder”, (The Economist, p.92). A student of violent conflicts especially civil wars can find not one but many causes for killing civilians. Sang Hoo Bae and myself being economists, we explain mass killings by the state by modeling the choice of a ruler of a country in which there are two distinct groups of populations (divided along ethnic, religion or other elements) by characterizing the decision as a three stage process. We investigate the ruler’s (assumed belonging to one group) options: engage in mass killings of the other group (initiate civil war); resolve the conflict by forming a coalition government; or do nothing.
Think of the current civil war in Kenya that started following the December 27 presidential elections. The ruler, President Mawai Kibaki is said to have lost the election to the opposition but refused to relinquish to Daila Odinga the presidency. The ruler, Mawai Kibaki options are those we have investigated in our paper. Our model solves for the optimal choice. We show that which option the ruler will choose depends on the “probability to remain in office which is derived from his political power; on the expected wealth from attacking the opposition group and the cost of the attack.” The cost can be of two types: military expenditures and cost of outside sanctions. In the Kenyan case, the flurry of activities by outsiders (the non-state group) including the US secretary of state, the former UN Secretary General, Kofi Annan may raise the cost and might tilt the choice towards a coalition government.
Using data on civilian mass killings over the period 1816 – 1997, we attempted to identify those factors that accounted for the choice of the mass killing option. But what mass killing constitutes? And who compile the data? The data is derived from the Correlates of War (COW) project. COW gives information on conflicts with more than 1000 battle related deaths. In the conflict studies, 1000 battle related deaths have been taken to signify mass killing. In our study we use this number and also redo the statistical analysis with 10,000 battle related deaths as defining a mass killing episode.
The statistical results of testing our model for the civilian mass killing that occurred between 1816 and 1997 give insight into the question raised earlier: the ruler’s choice of mass killing or as a Mr. Slim has put it: why kill civilians? Our findings reveal that “the length of executive tenure (how long the ruler held the office) plays a very significant role in civil war killings. In addition, it identified ethnicity, as a contributing factor, in that the “more fragmented the population – many ethnic groups – the less likely is mass killing and vice versa”. The results that were not unexpected had to do with spending on the military conflict and the income of the country. High military spending increases the probability of killings; the second the higher the income of the country the less likely the ruler is to engage in civilian mass killing.
The civil war in Kenya will hopefully come to an end before it belongs to the COW records where the civilian death rate reaches 1000. It would be a tragedy not only for the civilian population who are paying a heavy price for democracy but also for the world at large. The international community should persuade, cajole even compensate (bribe?) the ruler to restore democracy by opting for the power sharing arrangement for a limited period (to save face) and to restore the presidential office to the opposition if it turned out that Mr. Odinga was indeed the true winner.