The Yemeni Conflicts: A Mismatch Theory Interpretation

Focusing on separatists, we can apply a formal model drawn from Morelli and Rohner (2015) to outline how the location of resources and the concentration of the group involved predicts the secessionist direction of the fight that broke out between the STC and the Hadi government.

Southern separatists, who are geographically concentrated by definition, are on the sidelines of Yemeni politics and have complained heavily about being plundered of their natural resources. Although some oil reserves also exist in the north (for example in the Ma’rib region), 80% of Yemeni oil is located in the southern governorates of Hadhramaut and Shabwah. Moreover, the only gas terminal in the country is in the southern city of Balhaf. In the Yemeni case, before the intervention of the UAE made secessionists at least as strong as the government, their relative military strength did not mirror the fact that they were richly endowed in natural resources. In other words, the elite in power, rather than southerners, controlled southern natural resources by deciding how to allocate their revenues, as the government was stronger militarily. Hence, by seceding, southern Yemen would have had the opportunity to autonomously enjoy the revenues that its natural resources generate, as it would have had complete control over them.

We now introduce the model. Assume the Yemeni state comprises only two regions, north (1) and south (2), inhabited respectively by two major groups, a pro-government one (A) and a separatist one (B), with population sizes NA1, NB1, NA2 and NB2 (N indicates total population). Assume that A is in power and that each group is particularly concentrated in one of the two regions. In particular, as we already know, the pro-government group is concentrated in the north or region 1, while secessionists are southerners and can thus be found mainly in region 2. R₁ and R₂ are the values of the extractable resources in region 1 and 2.

Group A controls the ensuing revenues and decides which fraction β of the total surplus R = R₁ + R₂ can be distributed to the group of separatists B. In turn, group B can choose between accepting that arrangement peacefully (action p), start a centrist war (action c), or request a secession (action s), implying a separation between the two regions. In the latter case, group B would attempt to make South Yemen or region 2 an independent state governed by southern secessionists themselves.

Group A can also trigger a centrist war (in other words, a preventive repression of the minority group), or wait for B’s choice and react to it. Assuming that a faction’s military power is roughly proportional to its population size, in case of a centrist conflict, separatists win with probability Pc=NBNA+NB+γ, where γ is the advantage accumulated by the group in power A and that may also include relative wealth. If B asks for a secession, group A can accept it or fight against it. In the latter case, B wins with probability Ps=NB2NA2+NB2+γ′, where γ′ is A’s accumulated advantage only in region 2.

The separatist group has a higher probability of winning a secessionist conflict than a centrist conflict if NB2N2>NBN, meaning that it is particularly concentrated in region 2 as by construction the case of southern secessionists is, and/or if γ>γ′. In the Yemeni case, one possible mechanism that decreased the advantage accumulated by the state in the south was the fact that secessionists were allied with the UAE.

We now turn to the computation of expected utilities to understand when the incentive for southern separatists to trigger a secessionist conflict materializes. The peaceful status quo gives them a payoff β (R₁ + R₂). Secessionists’ expected utility from a centrist conflict is NBN+γR1+R2−dcB, where dc_B is the destruction cost of a centrist war for group B. Instead, secessionists’ expected utility from a secession is NB2N2+γ′R2−dsB with ds_B < dc_B (ds_B is the destruction cost of a secession for group B).

Assume that dsA<NA2+γγN2+γ′R2, meaning that the government never accepts a secession peacefully. This assumption is credible in the Yemeni scenario: the south is vital for the survival of the state given its strategic position for oil shipments and the fact that a significant part of its revenues come from natural resources, which in turn are mainly concentrated in Hadhramaut and Shabwah, as already stressed. Hence, it seems likely that the government in Sana’a would fight back any secession attempt by southern governorates. One concrete example that can be mentioned with respect to the fights that took place between supporters of Hadi and secessionists is the Aden battle of January 2018.

Both the government and the secessionists have a threshold for β beyond which peace is less convenient than fighting: there is a minimum share of resources β‾s southerners require from the government not to start attempting a secession, but simultaneously there is a maximum β‾ beyond which the government will prefer to fight rather than extending the share of resources it gives to group B. β‾s can be found by equating the payoff for group B from peace with its expected utility from a secession:

which can be rewritten as

Group B prefers peace over secession when β≥β‾s, that is when the share of resources the government concedes is higher than the minimum percentage separatists request. β‾ can be found by equating the payoff for A from peace with its expected utility from a centrist war:

which can be rewritten as

The government prefers the status quo as long as β≤β‾. Hence, the condition for peace is β‾s≤β‾. When NB2N2+γ′R2−NBN+γR>dcA+dsB, there is no value of β that leads both groups to choose peace. In other words, there is no bargaining space for secessionists and the government to find a compromise since β‾s>β‾. This was the deadlock the STC and the Yemeni government found themselves in.

The model just outlined is such that war becomes almost unavoidable when natural resources are mostly present in the region where the out-of-government group, or southern secessionists in our case, is more concentrated and/or when conflict is not too costly. In this respect, although we recognize that costs may be high in terms of destruction of natural resources and related infrastructures (even though they may be not in terms of opportunity costs since the country is very poor), we still believe that the concentration of resources and the history of marginalization made the incentive to secede outcompete these costs. With respect to resources, given that 80% of them are in South Yemen we can argue that they are present almost exclusively in region 2, hence R≈R2. In this scenario, even if we had assumed that γ≈γ′, meaning that the central government has no particular difficulties in fighting in the territory of secessionists, we would have had that the condition for war dcA+dsBR2<NB2N2+γ−NBN+γ would have been satisfied because R₂ is really high and the out-of-government secessionist group is concentrated in region 2; all of this outweighs the fact that the cost of destroying natural resources may be high.