The Logic of State Failure: Learning from Late-Century Africa

Abstract

In the last decades of the twentieth century, the rate of civil war and state failure in Africa rose precipitously. In an effort to comprehend the reasons for the rapid rise of political disorder, the author advances a model that isolates the conditions under which rulers will serve as guardians and civilians disarm, arguing that these conditions specify a region within which political order and the state become possible. Using narrative materials and statistical data, the author then tests this argument, and in so doing helps to account for political chaos in late–century Africa.

The paper was written with financial support from the National Science Foundation (Grant SES 9905568), the Carnegie Corporation, and the Center for International Development and the Weatherhead Center for International Affairs of Harvard University. Special thanks go to Steven Block for his criticisms and corrections. As ever, Karen Ferree and Smita Singh deserve much of the credit for this work. I also wish to thank Macartan Humphreys, James Habyarimana, Matthew Hindeman, and Marcus Alexander for their technical assistance. The author alone is to be blamed for its shortcomings. The article draws heavily on Bates (2008).

Notes

That is, they have solved the collective action problem.

Including by the specialist in violence, who can seize their assets only in probability and at a cost that is certain.

As is well known, in a repeated game, multiple equilibria can exist; indeed, there can be infinitely many. In this article, however, it is not the game itself that is of interest; it is a particular equilibrium of that game—one that I have labeled “the state.” The reason why governments and private agents assembled in ways that sustain this equilibrium lie outside of the scope of this paper; they have been addressed by those studying the legacies of imperialism, such as Coleman (1958). Young (1994); or the impact of the nationalist period, as in Hodgkin (1956).

The central concern of this paper is with the breakdown of that equilibrium.

The analysis of a similar equilibrium in which the government punishes an agent who raided or failed to pay tax without reverting to the all defect equilibrium does not change the analysis.

As is standard, I treat the discount rate as exogenously determined.

Which is endogenously determined.

In the arguments that follow, I treat the level of temptation as exogenous in the short run.

Our phrase, not theirs.

Phrased differently, Bates, Greif, and Singh (2002) shift the emphasis from the impact of taxation on the tradeoff between labor and leisure to the impact on participation and credibility constraints.

Put another way, we depart from McGuire and Olson (1998) and Findlay (1991) by using sub-game perfection rather than Nash as our equilibrium concept.

The best account is offered in Ndulu et al. (2007).

See, for example, World Bank (1989; 1992).

See Blumenthal, E. (1982) and Huband, M. (2001).

See also Sambanis and Hegre (2006).

The data are available from http://africa.gov.harvard.edu//.

See, for example, the appendix to Ellis (1999), as well as the critiques of body count data—something of great concern to the International War Crimes Tribunals—which are reported at www.microconflict.eu.

In models not reported here, I also note the impact of the shift from military to civilian regimes. The results are consistent with the results presented in Table 4, Equation 2.

The value of diamond exports, mineral exports, resource rents, and so on. See Bates (2008).

As does Fearon (2005).

That is, missing at random, as opposed to missing completely at random.

I note and stress that in imputing missing values I employ a large number of variables other than those that I enter into the estimation models. Their inclusion helps to correct for sources of missingness in addition to political conflict.

Dropping “privileged region” and “president from a nonprivileged region” does not alter this finding. It therefore does not appear to result from multicolinearity.

One possible interpretation is that, as argued by Azam and Mesnard (2003) and Acemoglu and Robinson (2001), placing power over wealth in the hands of the relatively poor increases the credibility of policies of economic redistribution, thus reducing the incentives to use military force to extract that wealth.