Governments Against States
| Author | Neil A. Englehart |
| Date | 01 March 2007 |
| DOI | 10.1177/0192512107075403 |
| Published date | 01 March 2007 |
| Subject Matter | Articles |
Englehart: Governments Against States 133
International Political Science Review (2007), Vol. 28, No. 2, 133–153
DOI: 10.1177/0192512107075403 © 2007 International Political Science Association
Sage Publications (Los Angeles, London, New Delhi, and Singapore)
Governments Against States:
The Logic of Self-Destructive Despotism
Neil A. Englehart
Abstract. Although state failure has assumed considerable importance
in the post-cold-war world, attempts to predict its occurrence statistically
have not been very successful. Such attempts rely on off-the-shelf data
collected for other purposes. To predict state failure, we need data
more specifi c to the problem. A better body of theory is required to
identify causal patterns, and case studies are a promising way to proceed.
Case studies of paradigmatic state failures in Somalia and Afghanistan
suggest a pattern: rulers attack the state apparatus in order to prevent
opposition by the bureaucracy and military, precipitating the collapse
of the state.
Keywords: • Civil service • Military • Militias • State failure
The study of state failure and collapse has become a fi eld in its own right only
recently. Earlier research programs certainly touched on these issues, but focused
on different problems: the causes of revolution and civil war, the effects of cor-
ruption and black markets, the abuse of human rights, and the factors that lead
to, and interfere with, economic development. It is only since the end of the cold
war, and especially since 9/11, that state failure has become a fi eld of inquiry
with a literature of its own.
Like most new fi elds, the theory of state failure is sparse and underdeveloped.
Yet there is signifi cant policy pressure to generate accurate predictions of state
failure, and this has led to attempts to model the phenomenon using readily
available data not generated specifi cally for the purpose. Indeed, given the weakly
developed theoretical literature, it is diffi cult to determine a priori exactly what
data would be useful.
The statistical analysis of predictors of state collapse is still in its infancy. The
most thorough effort to date is that of the Political Instability Task Force (PITF,
originally the State Failure Task Force or SFTF), a body formed by the United States
134 International Political Science Review 28(2)
Central Intelligence Agency to develop a model of risk factors for state failure
that would inform US government decisions about foreign aid and intervention.
The PITF has engaged in an impressive data-collection effort, and mined the data
using a variety of algorithms.
Statistical models of state failure and collapse such as those developed by the
PITF are, however, handicapped by a lack of appropriate data and the poorly
developed state of the theoretical literature. They are limited to off-the-shelf data
which are not causally very closely related to the events they hope to explain.
Furthermore, given the lack of theoretical guidance, they are constrained to
perform data-mining operations rather than more theoretically guided probes
of hypotheses.
Of course, any statistical analysis of state failure and collapse will have its limits:
it will necessarily be probabilistic, rather than diagnostic. It will not signal imminent
collapse, but rather a heightened risk of collapse. Many countries with elevated
risk limp along indefi nitely, chronically weak or failing, but not quite collapsing.1
However, it is likely that the attempt to quantify the analysis of state failure was
made too soon, without an adequate theoretical basis. Even if correspondences
are found in the existing data, it would remain to explain them and to determine
what exactly the relevant variables are proxying. Theoretical and case-study work
can suggest new and more promising variables for cross-national data collection,
as well as suggesting how variables might fi t together in a causal pattern.
This article seeks to address these issues by outlining one set of risk factors
suggested by two cases of total state collapse: Afghanistan and Somalia. In both
cases, state collapse was precipitated in part by government attacks on the state
apparatus. This may seem odd, because studies of the developing world seldom
make the distinction between governments and the state apparatus. The distinction
is commonly made in studies of democratic politics in the industrially developed
countries, where it would be considered an amateurish mistake to assume that
the elected government and the professional bureaucracy are identical or share
the same interests and preferences. However, it assumes special importance in
failing states.
When failing states collapse, the agent of the state’s destruction is frequently
a government desperate to sustain its power in the short run. Typically, such
governments lack a strong popular base to sustain themselves, and sacrifi ce core
state institutions to purchase political loyalty from key constituencies. Sometimes
this strategy can be sustained indefi nitely, especially if resources such as mineral
wealth or foreign aid are available. However, in some cases such equilibria may
not be found, or may be later disturbed, leading to more serious attacks on the
state apparatus. At the extreme, such self-destructive despotisms may arm the
very enemies that eventually bring them down, hoping to play them off against
each other for short-term survival. The result in such cases can be a catastrophic
collapse of the state. While such behavior seems bizarre and counterintuitive at
fi rst, there is a logic to it.
Below, I fi rst discuss in greater detail the limits of statistical forecasts of state
failure. I develop case studies of Somalia and Afghanistan to provide a basis for
theorizing about government attacks on the state. I then discuss the logic of self-
destructive despotism in order to draw out some of the theoretical implications
of the cases studied. While two cases chosen on the dependent variable certainly
To continue reading
Request your trial