Is Somalia better off without a government?

Peter Leeson has a provocative new 2007 paper I have just seen:

A comprehensive view of the data that allow pre- and post-anarchy welfare comparisons suggest that anarchy has improved overall development. Contrary to the typical case, in Somalia social welfare has improved because of, rather than despite, the absence of a central state.

Somalia’s government was oppressive, exploitative, and brutal. The extent of this predation created a situation in which social welfare was depressed below the level it could achieve without any government at all.

The emergence of anarchy in 1991 opened up opportunities for advancement not possible before government’s collapse. In particular, economic progress and improved public goods provision in critical areas flourished in the absence of a monopolistic and corrupt state.

Recognition of this is not to deny that Somalia could be doing much better. It clearly could. Nor is this to say that Somalia is better off stateless than it would be under any government. A constitutionally-constrained state with limited powers to do harm but strong enough to support the private sector may very well do more for Somalia than statelessness.

…The relevant question for Somalia’s future is thus whether or not a government, were a stable one to emerge, would be more like the constrained variety we observe in the West, or more like the purely predatory variety that systematically exploited Somalis between 1969 and the emergence of anarchy in 1991.

I think it’s hard to disagree with the basic point: some governments are so bad that no government might be worse. Even so, I’m wary about the conclusion.

First, I have a skeptical little data monkey that sits on my shoulder. Right now, he’s saying: no way we can take Somali economic and health data seriously. I don’t know the details of what was collected in Somalia, but I see first hand that so many of numbers are just made up, or extrapolated from unlikely sources, and so a change over time could be a real change. Or it could be error. Or it could be changes in the paces and ways people get the data, even supposedly rigorous epidemiology.

But let’s say the data are basically right and living standards are better now than under the nasty Barre regime. This seems plausible.

Is the implication that statelessness is preferable to a tyrannical government, or even a relatively unconstrained one (like we see in most other parts of the continent)? I think a constitutionally-constrained state might be a step too far, at least right away.

But let’s assume Leeson’s right: an awful lot of autocratic state forms are worse for standards of living than the current decentralized approach. Would an unconstrained central state really be a worse move for Somalis? I’m not so sure.

I think it might be a mistake to focus on levels, or the short term. The question for me is: what trajectory gets the region to stability and growth and some kind of constrained state?

History and political science suggests that the path to a constitutionally-constrained state with limited powers often lies through hardship–long periods of violence, and probably, eventually, tyrannical petty rulers. At least for a while. (Here is a take by North, Wallis and Weingast, and here is a take from Bob Bates). I’d like to think that the technology of statebuilding has gotten better in the past 500 years, and that there are better and more peaceful paths, but some days I’m not so sure.

I’m not sure it matters in this case, though. I am no Somalia expert (to say the least) but I suspect that most of the bits of Somalia that are doing well actually do have a government. It’s just not a centralized one that fits within the international laws and rules about official states. So, I’m not sure that I’d call the average situation anarchy. The things that govern people are states in all but name (Somaliland especially).

For me, the tricky question for Somalis is what to do from here. International incentives or interventions that support a centralized government could lead to stability or violence–it’s hard to predict. The alternative–recognizing statelets–might mean empowering some nasty warlords, or giving incentives for takeovers and violence, and has possibly perilous precedents for separatist movements everywhere (the slippery slope argument that comes out every time independence is whispered anywhere).

“Leave ’em alone” is also an option, but I’m not sure it an ethical, secure, or realistic one. Inaction is still an action, especially with the international laws and norms that make holding the capital such a prize.

So, for me, the real question is not “government or anarchy”, or even “constrained or unconstrained government”, but “centralized state or statelets”.

Whether I agree or not, what I like most about Peter’s work is his short and sweet libertarian critiques of conventional wisdom in development. OK, maybe not sweet. But clever. Here is a nice point Peter makes on aid. And he blogs here. I am hopeful for a response to this response.

Hat tip to Suresh.

Update: An excellent comment from Peter Leeson

13 thoughts on “Is Somalia better off without a government?

  1. Hi Chris – is this a new paper? It seem to have published in the Journal of Comparative Economics in 2007 (

    In any case, I agree with your points wholly but would go further. Leeson largely dismisses Somaliland as an “ultra-minimal state, if it is a state at all.” Yet it is far more of a state than Afghanistan, the DRC or Sudan, for example, or many states in Africa which cannot extend state control to their peripheral areas. If there is a positive relationship between statelessness and public goods, then why isn’t southern Afghanistan or eastern DRC doing much better than they are?

  2. It seems like Leeson is using the wrong counterfactual. Present-day, failed-state Somalia shouldn’t be compared to 1980s Barre-controlled Somalia, but rather to the closest approximation of what present-day Somalia would look like if it had a state. Maybe Djibouti? Ogaden? Northeast Kenya? I’m guessing these regions are better off than Somalia.

  3. Chris,

    Leeson lays out his overall anarchy theory in his PC publication “Efficient Anarchy,” available here:

    Also check out Ben Powell’s similar paper “Somalia After State Collapse: Chaos or Improvement?” published in JEBO, but available here: it compares development measures of Somalia in anarchy to other countries around Somalia with governments.



  4. @Nomadiyaah reminds me that it is “Somalis” not “Somalians”. Correction made. In my (pathetic) defense, I did highlight my ignorance of most things Somali…

  5. “I think it’s hard to disagree with the basic point: some governments are so bad that no government might be worse. Even so, I’m wary about the conclusion.” I think you mean “…bad that no government might be *better*.”

  6. A really tough question. In the relative near term, statelets could be effective much sooner than a central government. And if a central government came afterward, at least it would be clearer who needs to be brought into the fold. But I agree that creating incentives for invasion could be a problem, especially since without international legitimacy, they wouldn’t really be seen as takeovers so much as just fighting.

  7. Whether or not the likeliest form of government for a reconstituted Somalia would actually be better for its citizens is a good question. Following your point about the trajectory it occurs to me that one way in which any government would be better would be that it allows Somalis to re-engage with the international community in a whole bunch of ways currently denied them. When you’re starving this probably isn’t much of a priority, but at some point you’re going to want to do this. My wider point, however, is that this, in particular, is what the international community wants to see, ref diatribes against the dangers of failed states. Thus, whether or not a functioning national government would be good for Somalis, they are likely to get pushed into having one, probably sooner rather than later. One might even suggest this is not such a bad quid pro quo for feeding them. (Although the notion of starving refugees signing forms consenting to a having a government is more than a little farcical and grossly unethical!)

  8. I take this as further evidence that the global community should abandon Mogadishu and support the nearly functioning societies in Somaliland and Puntland. This seems politically unpalatable, however, from the African Union’s perspective, since they have a strong interest in seeing a strategy that strengthens existing states even when they are unreasonable.

  9. Leeson’s Table 3 does indeed have some data issues — he draws from different household surveys for his estimates, which brings into question inter-survey consistency problems. If you look at single surveys that cover all of Somalia, Puntland, and Somaliland, you see a very different picture — as page 23 of the 2002 UNDP/World Bank Household Survey report points out, incomes in southern Somalia are dramatically lower than those in the regions that the UNDP considers to be a part of Somalia, but which are actually in the de facto independent state of Somaliland. ( link below) Other numbers I’ve seen on this suggests that many more of these indicators are heavily buoyed by Somaliland, which I don’t think is an “ultra-minimalist” state in any sense. Indeed, one need look no further than the current famine incidence ( to see how incredibly different southern Somalia is from Somaliland. I have many issues with how aid is currently administered, but I am very uncomfortable with lumping Somaliland together with the rest of Somalia (especially statistically) and making an argument about the region as a whole.

    UNDP/World Bank Survey:,,contentMDK:20220052~menuPK:462274~pagePK:141137~piPK:217854~theSitePK:367665,00.html

    Collection of Somaliland Stats on page 7 (disclosure — this is my own paper):

  10. I remember this same observation being mentioned in passing at a humanitarian conference on somalia a little while back, by someone from WFP I think, who showed ‘before” and ‘after’ statistics. As I remember it, the explanation given was that it was a result of humanitarian assistance in improving specific conditions (particularly for things like girls education and emergency nutrition programs). The conclusion that paying too much attention to the ‘progress’ in Somalia was immaterial since nutritional and education gains were dwarfed by the progress in just about every other country in Africa. I can imagine that the presence of a relief operation would also shift the counting effect (since there are more resources going into areas accessible for counting).

    At any rate, I was very surprised that there so little mention of the relief operation in the article or Blattman’s comments. The fact that the international community has spent hundreds of millions of dollars to specifically impact the indicators (and of course considering the impact on these indicators as a key metric in designing their programs) seems to feature nowhere in the article. Although development (and military) aid was given before the war, the onset of a major humanitarian operation seems hard to rule out.

    The assumption of a ‘purely predatorial regime’ that would show negative development over the course of 20 years as the counterfactual somali history seems to set a pretty low bar, too. I mean there were a lot of predatorial regimes in 1991; very few of them are still around. And what would be the non-finding of his argument, that somalia would be one of the only countries to de-develop over the last twenty years? I mean where has there been net negative growth over that period when measured in UN terms? Over 20 years, development indicators have been going up just about everywhere, including DRC, Afghanistan, Burundi, etc. Every single country that wasn’t majorly affected by the AIDS crisis went up: , and the only country that noticeably went down was Zimbabwe.

  11. I’m not able to get past the data quality issues. To me, that’s kind of the end of the story.

  12. And that’s a twofer: it’s not like the data from before the collapse of the state, or the data from the neighboring countries, are particularly reliable either.