“We are poor not because of a lack of aid; we are poor because aid does not reach the targeted populations.”
– Timbuktu secondary school teacher[1]
In 2012, the Muslim Tuareg minority in northern Mali rebelled against the federal government. The ensuing civil war precipitated a coup d’état that deposed the president and ended only after intervention from France, Mali’s former colonizer. This paper principally argues that in cases like that of Mali, foreign aid can prompt the onset of civil war because of differential access to aid rents among different members of the population. I shall modify the Acemoglu-Robinson economic model of coups and revolutions to show an unequal yet peaceful society “tipping over” and falling prey to civil war due to elite capture of foreign aid.[2]
This modified model predicts that both pre-existing income inequity and unequal allocation of aid among inhabitants of the recipient country increase the chances of violent civil conflict. Its results are then used to suggest optimal policies that avert civil conflict by redistributing resources to marginalized groups in society. The amount of redistribution needed to avoid conflict increases if it is easier for marginalized groups to revolt, or if the inequality in the distribution of income is greater.
Section I of this paper argues that aid stolen by societal elite can trigger the onset of civil war. Section II presents the modified Acemoglu-Robinson model as well as its predictions on the impact of aid on civil conflict. Finally, Section III explains the applications and implications of the model.
I. Aid Stealing and Civil Conflict
Humanitarian aid has emerged as one of the principal instruments through which governments, corporations, and people in the developed world have extended support to conflict-stricken countries. The Organization for Economic Cooperation and Development (OECD), for example, estimates that its member countries spent more than $474 billion on foreign aid in 2012, accounting for both governmental and private contributions. However, aid transfers from the developed to the developing world have been subject to significant criticism over the past forty years.[3]
I focus my model to showcase one such criticism: the adverse impact of aid stealing and elite capture of humanitarian aid in inequitable societies with governmental failure. A corrupt government can easily abuse the fact that aid distribution depends on its bureaucracy, and disproportionately allocate aid to groups that support the regime. Furthermore, aid recipients often have multiple donors, causing a collective action problem in which differing principles and levels of scrutiny from donor countries lead to poor administration among recipient governments. Therefore, in a country with civil conflict and a corrupt government, armed factions could seize a large proportion of foreign aid.
Donors may seek to circumvent a corrupt government by channeling aid to non-governmental organizations (NGOs). However, even ignoring the inefficiencies in the operations of NGOs, the impact of these organizations on total aid disbursement pales in comparison to that of government institutions. The United Nations Research Institute for Social Development estimated that, historically, the percentage of overseas aid channeled to NGOs has never exceeded six percent.[4] Hence, addressing the inefficiencies in aid distributed through the government apparatus is the primary challenge in evaluating the long-term impact of foreign aid.
Illicit gains from aid rents enrich both state and insurgent forces, fattening the wallets of individual commanders and supporting larger strategic goals. The state, upon capturing aid material, can use aid to support its armed forces and constituents, simultaneously freeing resources to fund counterinsurgency operations. Insurgents, on the other hand, presumably lacking the resources available to the state, use the aid to directly fund their war efforts. This “capturing” of aid rents by insurgents may take the form of the aid that does reach the marginalized group represented by insurgent groups. The captured aid may also be material stolen from government warehouses or delivery systems in areas where insurgents are stronger. Potential gains from capturing aid for both sides explains why up to eighty percent of aid is lost en route to its intended recipients.[5]
Insurgents do not just use these materials directly–they also profit from trading them. The influx of foreign aid is accompanied by the establishment of businesses dealing exclusively in stolen or secondhand aid goods. Other establishments may provide services to large numbers of aid workers: one can think of restaurants, brothels, and drug rings as some of these supporting “businesses.”[6] Paul Henze, for example, went so far as to call Djibouti the “free trade area” of the Horn of Africa because of the vast amount of smuggled aid materials traded there.[7]
Mali is a case study that supports the hypothesis presented in this paper. The United Nations noted that the country, long hailed as a model low-income democracy, displayed characteristics that led to the 2012 civil war: untrammeled executive power, little executive accountability, and the regional divide between southern and northern Mali.[8] The Malian government has been accused of giving disproportionate amounts of aid to favored ethnic groups, discriminating against a subset of the Tuareg.[9] Crucially, it was the same group of Tuareg that allied with Al-Qaeda and rebelled against the federal government in 2012.
II. A Model for Aid-Driven Conflict
This section develops a formal model of aid-driven conflict, based on the Acemoglu-Robinson model used in political economy to explain coups and revolutions. In 2001, Acemoglu and Robinson explained the conditions under which developing societies with inequity face revolutions or civil conflict. Further, they proposed policies of transfers from the incumbent, elite social group to the oppressed segment of society that could prevent the latter from revolting or starting an insurgency. This paper formulates a model that extends the Acemoglu-Robinson formulation to the sphere of foreign aid, accounting for the impact of elite capture of aid material in exacerbating civil conflict. The model then proposes policies which the aid-receiving state can adopt to transfer enough resources to the marginalized group in order to avoid conflict and revolution.
I assume that there is a developing country with two population groups–the governmental group (G), which is the incumbent power and enjoys a greater share of the country’s resources, and the insurgent group (I), which is locked out of political power and access to state resources. I normalize the total population of this country to be 1, and take the average income of the country to be γ. Assume that the governmental group accounts for a fraction δ of the population.
I now incorporate the fact that developing countries like Mali are often unequal, with the ruling class controlling a disproportionate share of income. The governmental group G probably controls a disproportionate share of the country’s income. I call this share of the income θ1, such that θ1 > δ. This indicates that the group G controls a larger share of the country’s income than its proportion of the nation’s population. It follows that the group I controls a 1-θ1 share of the nation’s income.
Hence, the average income of a person belonging to group G is:
y^G= (Θ1γ)/δ
Similarly, the average income of a person belonging to group I is:
y^I = [(1-θ1)γ]/1-δ
I can now include aid in the model. Let us assume that the country receives a total aid income of A from the outside world every time period. The research on elite capture of foreign aid can be modeled by factoring in the disproportionate amount of aid income captured by the governmental group G. I may call the share of aid income seized by group G as θ2, where θ2 > δ, again signaling the inequity in distribution of aid.
Accounting for aid, I write the average incomes of the two groups as follows:
y^G= (Θ1γ +Θ2A)/δ
y^I = [(1-θ1)γ + (1-θ2)A]/1-δ
II.1 When will Group I revolt?
Given the levels of inequity in the distribution of both average income and aid income, represented by θ1 and θ2 respectively, Group I faces an optimal strategy problem of deciding when to revolt. In order to predict the group’s decision, I make two further assumptions:
- Conflict is Costly: Group I will incur costs from challenging the authority of the state. The costs of civil war are clear from the death tolls and economic destruction witnessed in conflicts ranging from Darfur to Yugoslavia. This cost to group I is modeled by introducing “L,” the cost each member of Group I bears for revolting.
- Aid consumption is not forward-looking: I assume that all the aid received by members of groups I or G is consumed in the same period it is received, with no savings. Additionally, aid inflows stop when a revolution occurs. Hence, when a revolution occurs, both sides wish to capture the added income in the economy made up by previously sent food, industrial, or military aid. I assume that refugee camps and other forms of wartime humanitarian/medical aid are economically insignificant. This indeed was the case in Mali, where countries such as the United States suspended aid in the wake of the coup.[10]
Hence, if Group I revolts and succeeds at capturing all the resources of the economy, the income of the average member of Group I would be:
y^IRevolution = [(1-L)γ]/1-δ
Group I only revolts if y^IRevolution>y^I, where y^I is what the member of I would have received absent a revolution. Hence, the condition for revolting is:
[(1-L)γ]/1-δ > [(1-θ1)γ + (1-θ2)A]/1-δ
–> (1-L)γ > (1-θ1)γ + (1-θ2)A
–> -Lγ > -γθ1 + (1-θ2)A
–> L > [θ1γ – (1-θ2)A]/γ
–> Loss to insurgents < (Government share of income – Insurgent share of Aid) / Average Income of the Economy
This equation offers a simple indication of what leads marginalized groups to become insurgents. An increase in either θ1 or θ2 raises the chance of a civil conflict. It is not therefore just inequity in the pre-existing average income that causes conflict–it is also a disparity in the distribution of foreign aid.
II.2. Optimal Transfer Policy
The elite in group G can preempt and avoid a revolution by offering economic incentives to members of the opposing group I. To model this, I introduce lump sum tax payments in this system. While analyzing the impact of taxation, I assume that the government in power has the capacity to tax the populace, and a system to redistribute tax income between the two groups in the country. If I let the tax rate be τ, the income of the i^th person in the economy will be
y(i,aftertax) = (1-τ)y(i) + τγ
Note that I am assuming the government does not tax aid–relaxing this assumption changes the result later. In a setting where the members of the powerful group, G, set taxes, their optimal choice of shall be zero, as this will maximize y^G. In this case, the analysis will be the same as above, and Group I will revolt if y^IRevolution>y^I.
Now, for a given τ, the members of Group I revolt if:
y^IRevolution>y^Iaftertax
–> [(1-L)γ]/1-δ > (1-τ){[(1-θ1)γ + (1-θ2)A]/1-δ} + τγ
The members of Group G now set τ so that revolution is just avoided, i.e.
y^IRevolution=y^Iaftertax
–> [(1-L)γ]/1-δ = (1-τ){[(1-θ1)γ + (1-θ2)A]/1-δ} + τγ
–> τ(γ-{(1-τ){[(1-θ1)γ + (1-θ2)A]/1-δ}}) = [(θ1-L)γ – (1-θ2)A)]/1-δ
–> τ([(θ1-δ)γ – (1-θ2)A]/1-δ) =[(θ1-L)γ – (1-θ2)A)]/1-δ
–> τ(θ1-δ) = θ1-L
–> τ = (θ1-L)/(θ1-δ)
Again, this simple dynamic equation for holds important results. If the cost of a revolution (L) decreases, the government must respond by redistributing more wealth to members of Group I in order to avoid conflict. The government must also similarly increase redistribution if inequality (θ1) increases.
As a final modeling exercise, I now assume that the government is also taxing aid.
y(i,aftertax) = (1-τ)y(i) + τ(γ+A)
In this case, the revolution condition y^IRevolution>y^Iaftertax implies
(1-L)γ/1-δ > (1-τ){[(1-θ1)γ + (1-θ2)A]/1-δ} + τ(γ+A)
As before, Group G sets y^IRevolution=y^Iaftertax to stave off the revolution.
–> (1-L)γ/1-δ = (1-τ){[(1-θ1)γ + (1-θ2)A]/1-δ} + τ(γ+A)
–> τ(γ+A – {[(1-θ1)γ + (1-θ2)A]/1-δ}}) = [(θ1-L)γ – (1-θ2)A)]/1-δ
–> τ([(θ1-δ)γ – (θ2-δ)A]/1-δ) =[(θ1-L)γ – (1-θ2)A)]/1-δ
–> τ = [(θ1-L)/(θ1-δ)] + [(θ2-L)/(θ2-δ)]
–> τ = [2(θ1θ2 + δL)]/[(θ1-δ)(θ2-δ)]
This equation for τ says that, ceteris paribus, a decrease in the cost of revolution–L–or an increase in any kind of inequality–inequality of income, measured by Θ1, or inequality of aid income, measured by Θ2–would necessitate an increase in redistribution. This simplified model assumes redistribution as a single payout from each member of group G to each member of group I. A government may alternatively structure the payouts across a span of time, hence gradually eroding the income gap between the two groups and averting a rebellion by group I.
III. Discussion and Conclusion
The crisis in Mali may be viewed as more than just a clash between a powerful government and a disaffected regional insurgency. The persistent income inequality between the north and south in Mali may have been compounded by the unequal distribution of foreign aid controlled by the government in Timbuktu. The contribution of this model to literature on civil conflict in countries like Mali stems from the term Θ2, both in predicting when groups revolt and in evaluating how much redistribution is needed to prevent revolt. As the inequality of aid distribution (Θ2) increases, the model predicts that civil conflict will be more likely, and that greater redistribution will be needed to avert civil war. This finding has important policy implications, as it suggests that the Malian government’s discrimination against the Tuareg in the distribution of aid may have been partly responsible for the civil war.
The predictions of this paper align with those of the model established by Scott on the “moral economy” of the peasant. Scott put forth two major arguments–first, that peasants are risk-averse rational economic actors, and second, that they rebel when they feel exploited.[11] Both of these arguments are nested in the assumptions made in Section II. The rebels of group I are indeed rational and risk-averse economic actors–they only rebel when the gains from rebellion outweigh the expected losses from taking up arms against the government. The second argument is also seen in the way members of Group I rebel when a combination of Θ1 and Θ2 makes them feel oppressed and exploited enough. We may claim that the Tuareg rebelled when a combination of income inequality (Θ1) and aid income inequality (Θ2) displaced their moral economy sufficiently. If this moral economy were restored via the redistribution policies discussed in section II, the Tuareg may no longer feel the need to rebel, and as risk-neutral rational agents, continue to live under the federal government.
A major advantage of this model is that it is agnostic about the level of analysis conducted. The specific identity of opposing groups does not change the model, provided one group starts off controlling a disproportionate share of resources. Returning to the Mali example, Groups G and I may be generalized to refer to various competing groups. This flexibility in the model could, for example, allow us to explain the 1994 war between the powerful Ifoghas Tuareg and the rival, subservient Imghad Tuareg. The Ifoghas, the most elite clan within Tuareg society, had the support of the Timbuktu government and a larger share of both pre-existing income and government-supplied aid material (Lecocq 2010). They could hence be seen as group “G,” and the poorer, less resource-rich Imghads could be thought of as group “I.”
It should be noted that this model does not condemn humanitarian aid outright. Aid is predicted to cause civil conflict only when the combination of Θ1 and Θ2 is sufficient to decrease the expected loss to rebelling insurgents. Even if aid is disproportionately distributed–i.e. Θ2 exceeds δ–the country may not slide into civil conflict if Θ1 is not large enough. On the other hand, aid may actually cause conflicts in otherwise peaceful countries if Θ1 by itself is insufficient to cause conflict and if Θ2 reduces the expected losses to insurgents enough to spark rebellion. Identifying the types of countries where the additional effect of Θ2 increases the likelihood of conflict is a key empirical puzzle posed by the model moving forward.
While this model does not seek to address the impact of foreign aid shocks, I can see evidence for empirical research on this issue from the results. Aid shocks have important real-world consequences: the massive reduction in foreign aid received by Mali between 1984 and 1989 caused a Tuareg rebellion in 1990.[12] In this model, this “aid shock” can be replicated as taking A to now become zero. This value of A would suggest that the rebels, in this case the Tuareg, would rise in rebellion when the expected loss (L) would be less than Θ1. Civil conflict may hence be far more likely than it was when there was aid material in the economy, and L had to be less than [(Θ1γ-(1-Θ2)A]/γ for group I to rebel. Foreign aid shocks hence lower the bar for civil conflict to occur, making it easier for marginalized groups such as the Tuareg to wage war against the federal government.
Policymakers face a unique dilemma in this situation. Too much foreign aid, given a combustible combination of Θ1 and Θ2, could stoke the flames of rebellion among marginalized groups denied their share of foreign aid. On the other hand, cutting off foreign aid could provoke conflict too, due to the impact of the aid shock described above. Policymakers in donor countries should hence evaluate and weigh two criteria. The first criterion is whether the combination of pre-existing income inequality (Θ1) and unequal aid distribution through the recipient government’s machinery (Θ2) can indeed provoke conflict in an otherwise peaceful developing society. The second criterion is to estimate to what extent aid can be cut without provoking conflict via the mechanism of a foreign aid shock. Balancing the two criteria–i.e. not supplying so much foreign aid as to allow the marginalized group to rebel, but not cutting aid enough to cause an aid shock–may yield an optimal outcome.
Three avenues for future research may be pursued. The first relates to agency and decision-making within opposing groups in developing societies. In this model, for instance, I assumed that each opposing group is organized and can be mobilized for immediate action. Furthermore, I assumed that every member of groups G and I has the same decision-making power. The spontaneous mobilization of groups and equal decision-making ability for each group member is an evidently unrealistic assumption. A more comprehensive model would either subdivide groups into collective decision-making units or construct an individual-level framework to predict conditions for civil conflict. Second, future research may focus on strategies to encourage the optimal transfer strategies discussed in Section II. While redistribution could effectively reduce the chances of civil conflict, governments in aid receiving countries may be unwilling to enact such transfer policies. A better model would account for government and leader preferences and predict if different developing countries would be able to enact the required redistribution policies. Finally, it should be noted that some forms of aid, e.g., food, may be redistributed more easily. However, redistributing materials like vehicles, oil and water may be more difficult. A more sophisticated model may thus incorporate the ease of divisibility of the aid material.
In all, the noble intentions behind foreign aid may be undone by the income inequality pervasive in many developing societies, compounded by the unequal distribution of the aid itself. Although the amount of aid given to a country could increase the purchasing power of its people to an extent, the increase in inequality could offset this gain in purchasing power by prompting civil conflict. International development agencies and policymakers would do well to consider local conditions and inequality when constructing aid programs in the future.
Dhruv Aggarwal (’16) is a senior in Jonathan Edwards College.
Works Cited
“Analysis: Mali’s Aid problem.” IRIN News, August 5, 2013. http://www.irinnews.org/report/98528/analysis-mali-s-aid-problem.
Agg, Catherine. “Trends in Government Support for Non-Governmental Organizations: Is the “Golden Age” of the NGO Behind Us?” Civil Society and Social Movements Programme, United Nations Research Institute for Social Development 23 (2006).
Crawford, Jamie. “U.S. suspends aid to Mali in wake of coup.” CNN. Last modified March 27, 2012. http://www.cnn.com/2012/03/26/world/africa/mali-us-aid-suspended/.
Henze, Paul. “The primacy of economics for the future of the horn of Africa.” In The Horn of Africa, edited by Charles Gurdon, 18-24. New York: St. Martin’s Press, 1994.
Klute, Georg , and Trutz von Trotha. “Roads to Peace: From Small War to Parasovereign Peace in the North of Mali.” In Healing the Wounds: Essays on the Reconstruction of Societies after War, edited by Marie-Claire Foblets and Trutz von Trotha, 109-143. Portland: Hart Publishing, 2004.
Lecocq, Baz. 2010. Disputed Desert: Decolonisation, Competing Nationalisms and Tuareg Rebellions in Northern Mali. Vol. 19, Afrika-Studiecentrum: Brill.
LeRiche, Matthew. “Unintended Alliance: The Co-option of Humanitarian Aid in Conflicts.” Parameters: U.S. Army War College 34 no. 1 (2004).
Nielsen, Rich, Michael Findley, Daniel Nielsen, Zachary Davis, and Tara Candland. “Do Foreign Aid Shocks Cause Violent Armed Conflict?” American Journal for Political Science 55, no. 2 (2011) :219-232.
Polman, Linda. The Crisis Caravan: What’s Wrong with Humanitarian Aid?. New York: Picador, 2011.
Scott, James C. The Moral Economy of the Peasant: Rebellion and Subsistence in Southeast Asia. New Haven: Yale University Press, 1976.
United Nations. Security Council. Report of the Secretary-General on the situation in Mali. S/2012/894 (28 November 2012). Available from http://www.un.org/en/ga/search/view_doc.asp?symbol=S/2012/894.
Williamson, Claudia. “Exploring the failure of foreign aid: The role of incentives and information.” Review of Austrian Economics 23, no. 1 (2009).
Endnotes
[1] “Analysis: Mali’s Aid problem,” IRIN News, August 5 2013, <http://www.irinnews.org/report/98528/analysis-mali-s-aid-problem>.
[2] The particular formalization for the Acemoglu-Robinson model used in this paper is derived from Ben Olken’s publicly available class notes for his Political Economy class at MIT (Course 14.75).
[3] Claudia Williamson, “Exploring the failure of foreign aid: The role of incentives and information,” Review of Austrian Economics 23, no. 1 (2009).
[4] Catherine Agg, “Trends in Government Support for Non-Governmental Organizations: Is the ‘Golden Age’ of the NGO Behind Us?” Civil Society and Social Movements Programme, United Nations Research Institute for Social Development 23 (2006).
[5] Linda Polman, The Crisis Carvan: What’s Wrong with Humanitarian Aid? (New York: Picador, 2011).
[6] Matthew LeRiche, “Unintended Alliance: The Co-option of Humanitarian Aid in Conflicts,” Parameters: U.S. Army War College 34, no. 1 (2004).
[7] Paul Henze, “The primacy of economics for the future of the horn of Africa,” in The Horn of Africa, ed. Charles Gurdon (New York: St. Martin’s Press, 1994), 18-24.
[8] United Nations, Security Council, Report of the Secretary-General on the situation in Mali, S/2012/894 (28 November 2012), available from http://www.un.org/en/ga/search/view_doc.asp?symbol=S/2012/894.
[9] Georg Klute and Trutz von Trotha, “Roads to Peace: From Small War to Parasovereign Peace in the North of Mali,” in Healing the Wounds: Essays on the Reconstruction of Societies after War, ed. Marie-Claire Foblets and Trutz von Trotha (Portland: Hart Publishing, 2004), 109-143.
[10] Jamie Crawford, “U.S. suspends aid to Mali in wake of coup,” CNN, last modified March 27, 2012, http://www.cnn.com/2012/03/26/world/africa/mali-us-aid-suspended/.
[11] James C. Scott, The Moral Economy of the Peasant: Rebellion and Substinence in Southeast Asia (New Haven: Yale University Press, 1976).
[12] Rich Nielsen, Michael Findley, Daniel Nielsen, Zachary Davis, and Tara Candland, “Do Foreign Aid Shocks Cause Violent Armed Conflict?” American Journal for Political Science 55, no. 2 (2011): 219-232.