Returning to "Coup-Proof" with some stuff on coup-proofing

I have finally been poked out of blog-lethargy (blothargy?) by a very interesting discussion about the integrity of some African regimes in the face of rebellion. James Fearon was provoked by M23's seizure of Goma to muse:

Being the president in African countries (and many others besides) can be an incredibly lucrative deal.  Why don’t these rulers, in their own self-interest, take some of that money and use it to build crack units, presidential guards, or strong and loyal army divisions that would protect their hold on power against two dozen putchists, or a hundred or a couple thousand rebels armed with rifles and maybe some mortars?

Fearon's intuition is then the same as mine--that the best explanation is the tradeoff between coup risk and rebellion, as developed in the work of, inter alia, Philip Roessler: 

Keeping the military weak  may lower their coup risk somewhat, but effectively trades coup risk off against higher risks of rural rebellion, insurgency, and foreign depredations such as we are seeing in Eastern Congo. 

But then he notes that Congo is a pretty extreme case in Africa. There are lots of countries that don't seem to have these recurrent problems. It is unclear what, structurally, differentiates these two groups of cases, according to Fearon.

One bit of brush clearing about what we're talking about. Laura Seay, in comments on Fearon, points out that Congo's President Joseph Kabila has in fact done a bit of what Fearon suggests: 

like leaders in most African states, Kabila has a well-trained, equipped, and paid presidential guard that has saved him from at least one coup/assassination attempt as well as a military battle in Kinshasa in 2007.

Goma is pretty far from Kinshasa, as Seay points out, so for the moment Kabila is fine. But clearly he would rather that Rwandan-backed (that seems clear enough) rebels not control a major town in the east. So there is indeed a bit of a puzzle. 

I'm going to try to think through the strategic dimensions of this a little bit, as a first cut. If we want to understand why some countries face these tradeoffs and others don't, I think our best bet lies in predictors of coup risk, and in generating comparative statics through a better characterization of the strategic situation facing leaders.

I think Fearon captures the demand side of things pretty well: presidents do have a pretty compelling interest security services that are (a) effective and (b) loyal, if they can get them. Now, the threat of a successful coup is worse for the leader than the threat of a successful civil or international war, because at least in the latter case, the president probably has a better chance to flee before he/she is captured. Hence the president has to ensure against coups first, rebellions/international wars second. In practice, there are coup-proofing techniques, such as divide & rule, or running multiple internal security agencies all monitoring each other, or relying on loyalists from a communal in-group, that can help reduce the risk of a successful coup, but weaken the army in the face of wars (international and internal). So the president tends to reduce effectiveness for the sake of not having a coup.

I would point out an additional problem attendant on rebellions: they disrupt the calculations of coup-proofing. If, from sources that are in the first instance outside the military, the likelihood that the president is going to get ousted suddenly and dramatically increases, soldiers have much more of an interest in defecting. This was a problem Mobutu Sese Seko faced in 1996-97, or Gaddafi in 2011. The problem in rebellion is then not just effectiveness per se, but also, quite often, loyalty too, and what a president does to keep loyalty against a coup plot might not be enough to keep loyalty against a rebellion.

In any case, the President has a clear preference for a strong & loyal army, and has lots of money, but money doesn't seem to translate into that strong & loyal army. We observe in reality security services are often not effective, and if they are loyal, it's only in the sense that they are not actually launching a coup right this second, not that they are willing to fight particularly hard, or not desert in the face of a rebellion (see Jeffrey Gettleman's masterpiece of sarcasm: In New Tack, Congo’s Army Starts to Fight). There are then three non-mutually-exclusive possible short-run explanations, as I see them:

(1) The problem is on the supply side, not the demand side. That is, the population of individuals willing to supply their military services loyally is too small to meet the demand.

(2) Even if those individuals existed, identifying them is going to be pretty hard. It would probably require credible signals of some kind, but it is unclear what kind of credible signals can be offered.

(3) Direct cash payment is not what is needed to secure loyalty. If I am a colonel, I can hope for a payoff from the next president, not just from this one. In any case promises of cash can attract mercenary careerists over committed professionals. If it is hard to identify one from the other, it sets up something like Jeremy Weinstein's information problem in rebel recruitment, applied to governments. So the money that African presidents have just isn't all that useful in building effective, loyal armies.

So these are some areas where I think we can clarify just what the problems are. I happen to think the identification problem is a seriously tricky and important one. In my paper "Loyalty Strategies and Military Defection in Rebellion," I investigate some of its consequences. I see the reliance on a core group, such as an ethnic minority, as a way of constructing a way of identifying loyalists from the suspect endogenously. A president believes that members of group X are particularly loyal; so promotes members of group X; so members of group X cannot easily defect in the face of a rebellion and hope for good treatment by the other side, being too tightly associated with the old regime. It is a kind of self-fulfilling prophecy. This does, of course, alienate non-Xs, who will be more likely to defect, and as Roessler argues in a terrific paper, it can provoke rebellions in the first place. My paper shows how this played out in Syria in the Muslim Brotherhood uprising that culminated in the Hama Massacre of 1982, in a manner similar to the reliance of the Jordanian monarchy on certain East Bank groups in the Jordanian Civil War in 1970 (and also Saddam Hussein's Iraq), and in contrast to individualized reward-and-punishment approaches like the Shah's in Iran. I think some version of it has been playing out, horribly, in Syria again since 2011.

So, these are some strategic considerations. I think picking them apart will provide further insights into the short run.

But what makes one country more likely than another to have these tradeoffs in the first place? That is, what about Fearon's question about structural preconditions? In general, I suspect that if there are, indeed, structural predictors of the kind of coup-rebellion tradeoff that's at issue here--the specific problem of having a military too weak to deal with these kinds of challenges--that we'll find them in structural predictors of coups, not structural predictors of civil wars, if a coup is indeed the more important threat of the two.

Otherwise, I'm not really sure. Perhaps the mechanisms I suggest above can point us in the right direction on structural conditions. Perhaps the ethnic-identity piece depends on the president having close ties to a communal groups of a particular size (not too big, or else identity gets to be not very predictive of loyalties; not too small, or else it doesn't provide enough manpower), but even then, there are multiple ways to use group identities: Saddam relied on Sunnis specifically from Tikrit, for example. We could maybe look for structural conditions in which there is a large supply of individuals willing to fight loyally to overcome the supply problem. If we game this out, we can come up with some comparative statics. But this all remains a neat puzzle without an obvious answer.

How to handle uncertainty

Tantawi, Morsi, Enan. Image via Reuters.

The big news: Egypt's president, Mohamed Morsi, announced the retirement of Field Marshal Mohamed Hussein Tantawi, the head of the armed forces and minister of defense, as well as Sami Enan, the chief of staff. The BBC is also reporting that Morsi's cancelled a constitutional ruling issuing limits on presidential power. I don't know whether the ruling in question is the one that keeps the military independent of any real external authority, but one presumes so, given that replacing top military personnel would obviously be in violation of that declaration. (UPDATE: yep; h/t Rex Brynen)

This is a big deal: a sudden announcement that could reset civil-military relations in Egypt. If it carries through, then advantage civilian government.

A brief moment of navel-gazing. This puts punditry and commentary in a weird spot. I love trying to understand and explain things in politics, and I have a bad weakness for speculating when I don't have enough information. But in periods like Egypt is going through right now, when political rules are uncertain, I think it may be best to sit back and watch, rather than trying to explain. Morsi could well pull this off. Perhaps this was all negotiated; perhaps not. I find this doubtful, as I don't think SCAF would accept the public image of unilateral action to which they roll over. Or Morsi could spark a coup: according to Jay Ulfelder, Egypt has structural conditions that, at the outset of the year, placed it at 14th globally in coup risk, even if that risk was still, at the time, only about 3%.

Much of the outcome turns on hidden preferences and intentions, rumour and sketchy information: the ingredients of surprise. And there are no very clear rules at the moment; Egypt's politics, right now, are about setting those rules. What happens in the next few days could set Egypt down one path or another, and we can have a more or less educated guess about which path, but I think there's a very strong element of uncertainty, and I think it's important to remember it. For some, next week whatever happens will just seem inevitable. And who knows: at a certain point we might get good enough at explaining and predicting to be able to make strong judgments about what is likely in Egypt in the coming days, and to devise a compelling explanation about why. (I doubt it.) But in the meantime I think it's incumbent on analysts to remember these stochastic moments, and how they humble us.

COIN training: the pupil becomes the master

So, the US accuses Hizbullah [NYT] of training Syrian forces:

An American official, who spoke on condition of anonymity, said Hezbollah was using “its specialized skill set and understanding of insurgencies” to aid Syria. 

Recently, also, it turns out that Blackwater Xe Academi has trained Canadian Forces as far back as 2006 in some counterinsurgency-related skillz--like evasive driving and bodyguarding--without U.S. government permission, and they've had to pay $7.5 million in fines for breaking U.S. law in this and other matters.

I guess they've learned on the job.

The very existence of private military organizations like Hizbullah and Academi--oh, they'll love getting lumped together--means a limit on the state's monopoly of the use of violence. That's well known. But this phenomenon of non-state actors training state armies in warfare perhaps goes a touch beyond that. Usually in training missions we think about government-to-government (say, the current Canadian Forces mission in Afghanistan) or government-to-non-state-actor (say, the U.S., Soviet Union, and South Africa during the Cold War--or, frankly, Hizbullah itself). Government as tutor, given its specialized knowledge; non-state actor as pupil. But this is the reverse.

It's certainly not without precedent. Non-state actors have trained each other, and in some cases have become state armies through victory. For example, several leaders of the Rwandan Patriotic Front--the rebels in the 1990-94 civil war and genocide--honed their skills with the National Resistance Army in Uganda, fighting alongside them during the NRA's 1980-86 civil war. I wouldn't be surprised to see past instances of non-state actors training state armies in some military tasks; mercenary war in early modern Europe comes to mind. But this is definitely a role reversal in terms of conventional wisdom.

This sort of transaction also seems to go a bit beyond training armed forces in what are, in the first instance, civilian skills--like local knowledge, law, civil engineering, or logistics. Instead, it's a "specialized skill set", but a directly military one--not a non-military skill set applied to a military context. We sometimes refer to the state as a "specialist in violence" (see, e.g., here), and this is what I mean.

That Academi violated U.S. law in training CF personnel without permission suggests an effort on the part of the (American) state to control not just violence but knowledge and skill in its use. (It's still a state-centric logic in that the objectionable thing seems to be that the recipient was a foreign military, rather than a citizen; my friend Mike could teach you how to drive evasively if you really wanted to know.)

Now the logical extension of a non-state actor as a specialist in violence is that they can train others in it. In this vein it's worth noting that insurgency and counterinsurgency are pretty low-tech, at least in equipment terms. In fact, using vehicles in COIN is actually a poor idea, as Syria may be finding out. What really counts, it seems, is experience--that is, something Hizbullah and Blackwater have a fair amount of. There would be a steeper curve for an NSA to learn, for example, how to fly fighter planes and be able to teach someone else. (There is an additional political dimension for Hizbullah--if in fact they are supporting Syria, one assumes it is out of a perceived strategic and not strictly financial gain.)

The question is more on the other side, the state recipient of training. Why not train internally--why outsource your training? Can a government buy out an NSA like this or establish its own capacity? When would it try, and when would it keep with a market transaction? Firm theory from economics might have some insights here, but I'm also interested in the politics, especially in light of the apparent use of outsourced training in small wars. In particular, I wonder whether this is just another case of military resistance to being specialists in insurgency and counterinsurgency. If armies become good at COIN over the long term, then they may have to do more of it--which many don't want to. For example, the COIN debates in the U.S. military have very much revolved around training for COIN wars and how much they should bother--consider that setting up counterinsurgency school at Ft. Leavenworth and producing a new field manual on counterinsurgency were two of the key accomplishments of the pro-COIN crowd.

The key question, of course, is whether Academi and Hizbullah use godawful powerpoints.

Drug dealers, stationary bandits, and crack

Still from Cidade de Deus (Fernando Meirelles, 2002)

Here's a neat story on al-Jazeera: some drug dealers in Rio's favelas are starting to stop selling crack. Why? After all there is, as one dealer interviewed says, lots of profit to be had. The dealers themselves, and their surprisingly forthright attorney Flavia Pinheiro Flores, claim an interesting form of civic responsibility: crack is worse in its social effects than other drugs. The police claim this is a PR ploy, to get public authorities to lay off of sales of cocaine and marijuana, in a sort of live-and-let-live deal.

I don't know anything much about drug cartels or Rio, but I find these two interpretations interesting. I suspect that there's an element of truth in both. I wonder whether they are really speaking to the same thing: the logic of stationary bandits.

The great Mancur Olson argued (PDF) that bandits who stay in one place have incentives not to rob people blind. People will stop putting in efforts at making money, and so sources of theft will dry up. They also have good reason to provide public goods, like a degree of security and justice, for similar reasons. (I'd add that the focus is too much on a Laffer curve logic for my liking, and that there's an additional consideration for a bandit to keep in mind: the possibility of flight or defection, with civilians greatly preferring to live under some other sort of order.) In contrast, roving bandits have no such incentives. Olson then uses this "criminal metaphor" to understand the state: even if you think of the state as a predator on the income of the people, it still has some reasons for restraint and for providing something in return. Olson's far from the only scholar to compare the state to a criminal organization, but he spelled out an important part of the link by focusing on revenues.

Applied to the favelas, the logic might be something like this: while crack addicts are for obvious reasons a rather useful clientele, they come at a cost for drug cartels that influence many different facets of life in a district, and especially those who have revenue from multiple different sources. The victims of theft and physical violence have many more reasons to leave, so they're harder to tax. They might even prefer that another, anti-crack cartel take over the district.

So how do we assess the rival claims in light of this possibility? Like I said, Olson's hardly the first to discuss the criminal metaphor; that might be (I'm not sure, but it's a decent bet) Plato, who in Republic (I: 351c-e) used the analogy to suggest that because even a gang of pirates has to practice justice among themselves to be powerful, political society must likewise involve ethics. What is "actually" motivated by self-interest may therefore easily be understood as ethical behaviour. While it isn't a stretch to imagine that there actually can be some stronger ties of community than mere self-interest, it also doesn't do any harm to note how the two correspond.

As for the police, and the possibility of a PR move motivating a live-and-let-live arrangement: again, this can be understood in light of the gang's incentives not to make life absolutely horrible. If life gets bad enough--i.e. violent, chaotic, shot through with theft--more individuals would become inclined to cooperate with the police in a bid to oust cartels from favelas. Given that such efforts generally entail enormous violence and short-run suffering, especially when the cartels are quite powerful to begin with, it's reasonable to suppose that people's decisions about whether to support police action are actually affected by just how bad life is under a cartel.

All of this raises an interesting, really rather weird implication. The logic of the argument turns on the notion that the cartel loses revenue from chaos, much of the latter instigated by crack in ways that other drugs don't. The cartel's lots more likely to lose revenue from violence if its revenue streams are more diverse than just crack--if, for example, it's also involved in other, pricier drugs, or protection and extortion. I wonder if that means something really weird: the more activities a cartel runs, the less socially awful its rule is, generally.

But, to repeat, I don't really know much of anything about the Rio drug trade...

Sudan/South Sudan: Some Sovereignty Speculations

Things seem to be cooling down between Sudan and South Sudan. After pulling its troops out of the disputed, oil-rich region of Abyei, Sudan left 169 police officers. South Sudan claimed they were soldiers in police uniform. Whether this is true or not, apparently those are now leaving too. The two countries have apparently agreed on a "road map" for talks and they're set to begin Monday. This all seems like fairly straightforward Good News given the scary moments in the spring between the two countries.

But at the same time, the conflict between Khartoum and SPLM-North is heating up. Médecins Sans Frontières estimates that there are perhaps 100,000 Sudanese refugees in camps in South Sudan, fleeing the violence. There's a plausible connection between the two. Before South Sudan's independence, it seemed as though there were constant tradeoffs between Sudan's various conflicts. For example, it was after the peace agreements of 2003 that the conflict in Darfur reached its zenith of brutality. Whether because Darfur was left isolated by the logic of minimum winning coalitions (don't give up more than you have to to stay in power) or because peace with the SPLM left Khartoum a free hand to assault its other perceived enemies, it seemed like if it wasn't one war it was another. One wonders whether there's a repeat performance.

If there is, how much "real" difference does South Sudanese independence make?

First, by one way of thinking, inter-state conflicts are a problem for the international community in a way that intra-state conflicts aren't. Since the international community is first a community of sovereign states, the thinking goes, the members of that community have a much clearer interest in ensuring that states don't violate sovereignty by sending troops across borders than they do in ensuring that states don't kill their own citizens. It did take a lot of change for the United Nations Security Council, for example, to start dealing with civil wars. But of course internal wars rarely stay internal for very long: consider the path of death from Uganda through Rwanda to DR Congo, or the West African arc of conflict, or the Afghanistan/Pakistan/Tajikistan cluster. More strongly, many see human security as a vital value alongside or trumping international or national security. So if there is a tradeoff between international war between the Sudans and internal war in Sudan itself, the result is a neat barometer of how relatively important you think international and civil war are.

Second, sovereignty may imply taking on a set of new roles and new interests. Consider Somaliland: there is a case that it's because it has managed not to become independent that it's the most stable part of Somalia. It doesn't have access to rents like aid and international support, so it has to tax its population and thus develop some stability and accountability to its citizens. Now independent, South Sudan needs to get oil flowing again, and has to find a way to cooperate with Khartoum to do so. There are new incentives for it to behave with restraint towards its old enemy, to stay on its side of the border--incentives that were weaker when there was no border to mark the farthest its troops could go. And it is now cut off by this international convention from the SPLM-North. It could certainly outsource its conflict with Khartoum to the latter as it sees fit. But now the interests of the SPLM in Juba now differ from the SPLM-North in important respects. There will sometimes be a strong temptation for South Sudan to sell out its old comrades in arms in order to reach a pricing agreement with Khartoum.

Sovereignty is one of those maddening things: it's arbitrary in some respects, a line on a map. But we act like it is a real thing, and so it is.

(Photo: Associated Press)

In which I attempt to avoid a post title like "Bureaucracy and Defence Procurement: The Case of Canada"...

...which would probably be the Trifecta of Boring. I've already lost basically everyone foolish enough to read this. But seriously hold on, this matters.


As everyone knows, we were supposed to spend a Carl Sagan amount of money on the F-35 Lightning II, a plane that has the demerits of sucking and costing too much, but on the plus side comes armed with unintentionally hilarious music videos. So I wake up to find that the Conservative government is considering a new agency for military purchases at arm's length from National Defence. This has the right kind of image: a Responsible Government, well meaning but duped by clever pencil pushers in a broken administration that our heroes are now fixing. 


It is hard to know what to make of this. You don't do this sort of thing without signalling that somebody really screwed up. Even the most somnolent bureaucracy can bolt to its hind legs with the sudden manic, slavering energy of a coke-addled Rottweiler to defend its territory. Especially when the hound is a defense department, and the particular patch of turf in question is the one where it can dig up its tastiest chew toys. 


But we still may be buying the damn things. Maybe. Ah, who knows?


As for this new agency plan. The implication that it's all the bureaucrats' fault tempts me to language that would get me into serious trouble if I worked for a respectable publication. The government misled Canadians about the true cost of the fighter, and then ordered the RCMP to investigate the leaks on "national security" grounds that the Mounties could not bring themselves to buy. The Defence Minister claimed that it was standard practice not to include life-cycle costs, when in fact it was. Mr. MacKay told us, rather wonderfully, 

If you went out and bought a new minivan and it was going to cost you $20,000, you wouldn't calculate the gas, the washer fluid, the oil and give yourself a salary to drive it for the next 20 years.

 If nothing else, MacKay did buy himself a lot of attention from disreputable used-car salesmen.


Anyway, MacKay didn't improve his case by arbitrarily knocking 50 per cent off the cost of the Libya mission. Chump change compared to the Lightning-II, so why bother? I can only attribute that last bit of old Peter's bull to sheer force of habit.

Prosecutions and the LRA

The detention of Major General Caesar Acellam, apparently the fourth-senior commander in the Lord's Resistance Army, is major news. It's prompted Radkhika Coomaraswamy, the Special Representative of the Secretary General on Children and Armed Conflict, to urge a prosecution. This is understandable: the LRA is a pretty terrible organization, and Acellam is no doubt responsible for some pretty terrible crimes, despite the fact that he is apparently not among those indicted by the ICC. However, the government of Uganda has also announced that Acellam is eligible to apply for amnesty. This would make him the most senior LRA official to be granted amnesty.

How would amnesty affect the ending of the conflict, the capture of Joseph Kony, and the dismantling of the LRA? Coomaraswamy argues that "The arrest and subsequent prosecution of Acellam would send a strong message to the LRA leadership that they will be held accountable for their actions." This is probably true. But to what end? Is it likely, for example, to induce them to commit fewer crimes in the meantime, for example? I find this unlikely, if only because if they do face prosecution for the crimes they've already committed--28 counts of crimes against humanity and 54 counts of war crimes between the four senior LRA commanders under ICC indictment--they face little hope of lighter sentences down the road. Interestingly, the one group that I could see this signal affecting in the way that Coomaraswamy intends is precisely not the senior LRA leadership, but lower-ranking officers not presently under indictment. 

Defense Minister Crispus Kiyonga has signalled that whether Acellam was captured or gave himself up will be an important consideration. As they should. Acellam was apparently taken with his wife, daughter and a helper, a rifle and eight rounds, separated from the rest of the LRA. He may have split from the remainder over a dispute. In addition, he has long been on a list of potential defectors after falling out with Kony over making peace with Kampala--though Ugandan officials apparently deny that they'd been in talks with Acellam over his defection. But even in the scenario in which he had not intended to defect, but merely to hide, means that prosecuting Acellam sends a signal to remaining senior LRA personnel that they need to stay with the group, which has for so long successfully evaded capture. Deserting the LRA becomes less of an option. 

Here, we might be seeing an instance of a more general problem: the inability to make a credible commitment may keep wars running longer than they need to. It's not obvious that amnesty is an automatic solution to this difficulty. It's associated with greater postwar instability and a higher likelihood of resumption of civil war, for example; exile seems to work more effectively.


The very difficulty of tracking down Kony suggests another reason to consider amnesty: plea bargaining. Malcolm Webb argues that intelligence from defectors and captured LRA members has proven more effective than electronic surveillance or other high-tech methods, and that Acellam could prove to be an invaluable resource--if he talks. Amnesty may give him reason to do so.

Ultimately, then, Defense Minister Kiyonga's stated approach--"Let's wait. We assess each (amnesty) case on its merit. We are waiting for him to arrive and then we will take a cool-headed decision"--seems about right to me. Too much is unknown, and I find Coomaraswamy to be rushing to a recommendation.

One way or another, I don't think that my arguments, or Coomaraswamy's, or those of any other commentator deserve more consideration than the opinions of the people of northern Uganda. There are enough outside agendas at work in the LRA conflict that it's problematic for anyone to claim the final word. But there is a pretty strong case, here, for keeping amnesty on the table.