1999: NATO: State Failure Task Force Report: Phase II Findings

* NATO: Environment and Security in an International Context: Executive Summary Report [PDF]; State Failure Task Force Report: Phase II Findings [PDF]. Review: Richard Matthew: Environment and Security in an International Context: Critiquing a Pilot Study from NATO’s Committee on the Challenges of Modern Society [PDF]

State Failure Task Force Report: Phase II Findings

Prepared by Daniel C. Esty, Jack A. Goldstone, Ted Robert Gurr, Barbara Harff, Marc Levy, Geoffrey D. Dabelko, Pamela T. Surko, and Alan N. Unger


Abstract: In response to a request from Vice President Al Gore in 1994, the CIA established “The State Failure Task Force,” a group of independent researchers to examine comprehensively the factors and forces that have affected the stability of the post-Cold War world. The Task Force’s goal was to identify the factors or combinations of factors that distinguish states that failed from those, which averted crises over the last 40 years. The study represents the first empirical effort to identify factors associated with state failure by examining a broad range of demographic, societal, economic, environmental, and political indicators influencing state stability. The Task Force found that three clusters of variables had significant correlation with subsequent state failures: (1) quality of life; (2) openness to international trade; and (3) the level of democracy. However, it is the interaction among these variables that provided the most important insights. Following are excepts from Phase II of the State Failure Task Force findings.


The initial report of the State Failure Task Force1 developed a global model of the factors that contributed to serious political crises over the last four decades. In this report, we describe the progress of the Task Force on four additional research issues:

• Confirmation and refinement of the global model. This work included testing the model on an updated problem set, varying the set of control cases, and testing new or refined variables. In particular, we refined the level-of-democracy variable to examine partial democracies—countries that combine democratic and autocratic features—and their risks of state failure.

• Fitting a model for Sub-Saharan Africa. We also examined how the global model might best be modified to apply to the countries of Sub-Saharan Africa. To improve the accuracy of prediction, the Task Force undertook a pilot study of event sequences in a limited number of Sub-Saharan African cases of state failure and state stability to identify factors that could
be precipitators or “accelerators” of crises.

• Transitions to democracy and autocracy. The initial study only examined cases of adverse or disruptive regime transitions. Because of the great interest in transitions to democracy, and the conditions that provide for stable or unstable democracy, the Task Force applied its methodology for analyzing risks of state failure to transitions toward and away from democracy. This report explores the preliminary findings of these analyses of the emergence and decay of democratic regimes.

• The role of environmental factors in state failure. It appeared from the Phase I results that environmental factors did not directly contribute to the risks of state failure. The Task Force believes that this finding was due, in part, to the paucity, poor quality, and lack of comparability of the national-level environmental data and, in part, to the impact of environmental factors on political conflicts being mediated by other economic, social, and political conditions. We, therefore, undertook special initiatives to assess the state of global environmental data and to develop a mediated, two-stage model of the role of environmental factors on the risks of state failure. In this model, it appears that environmental hazards—in states with underlying vulnerabilities and limited governmental or social capacity to respond to environmental deterioration—is associated with increased risk of state failure.


Updating the Problem Set and Revising the Control Cases

One problem frequently encountered in statistical analyses such as the one performed in the initial phase of the State Failure project is that specific results may be highly sensitive to a particular data set.2 If the results reflect statistical accidents, rather than underlying social and political forces, then slight changes in the data set may greatly shift the results. Adding or subtracting cases, or changing the particular control cases, could make some variables newly significant or remove some variables from the list of significant factors. Our first task in re-examining our results was to update the problem set to include state failure cases from 1994-96, and to select new control sets for testing this new data, to make certain that our initial results proved robust.

It was reassuring to find that despite significant revisions and updating of the problem set and analyses using two different sets of control cases and three distinct analytical techniques, the same three variables—infant mortality, trade openness, and level of democracy—emerged as the critical discriminators between stable states and state failures. Moreover, these analyses resulted in about the same two-thirds range of accuracy in discriminating failures and stable cases.

State Failure Cases3

The set of “state failure cases” in the initial State Failure Task Force Report was updated and revised by reexamining all of the cases and consulting area experts to identify recent events
(1994-96) for inclusion.4 A number of cases in the initial problem set were dropped as being of insufficient magnitude or not meeting the precise definitions for failure events. A considerable number of new cases from recent years were added. However, none of these changes affected the global model results.

Control Cases5

The two new sets of control cases were obtained, as before, by randomly selecting to match every country-year that preceded a state failure by two years, three countries that were stable
(experienced no crises for the succeeding five years). Changing the control sets made no difference to any of the global model results.

The three analytical techniques used were logistic regression, neural network analysis, and genetic algorithm modeling.6 Logistic regression and neural network analysis were
used to estimate the “predictive” accuracy of our models. Genetic algorithm modeling was used to help identify candidate sets of variables, as a check on the univariate regression methodology, and to validate the suggestions of Task Force social science and area experts. Although each method relies on different assumptions and methods of estimation, all techniques converged on identifying the same three-factor model as the most efficient discriminator between stable and failure cases and yielded models with accuracy of predicting case outcomes of about two-thirds.

Retesting With a Refined Level of Democracy Variable

The original global model, using infant mortality, trade openness, and level of democracy, measured democracy as a dichotomous variable, classifying countries as “more democratic” or “less democratic.” However, it became apparent that not all democracies were “equal” in their vulnerability to state failure. The rich and well-established democracies were extremely stable. In contrast, the more recently established and poorer democracies were at very high risk of failure. Given this result, and the interests of policymakers in democratic transitions, it was clearly important to better differentiate the democracy variable to examine the risks associated with “partial democracies.”

Using both the democracy and autocracy scales of the Polity III Global Data Set7, each country was classified as a full democracy, a partial democracy, or an autocracy, on the basis of its political institutions:8

• Full democracies have all the characteristics of liberal democracy—such as elections, competitive parties, rule of law, limits on the power of government officials, an independent judiciary—and few or none of the characteristics of autocracy.

• Partial democracies have some democratic characteristics—such as elections—but also have some autocratic characteristics, such as a chief executive with almost no constraints on his/her power, sharp limits on political competition, a staterestrained press, or a cowed or
dependent judiciary. Most are countries that have recently transitioned toward democracy but have not yet fully replaced autocratic practices and institutions; some resemble what Fareed Zakaria has referred to in a recent Foreign Affairs essay as “illiberal democracies.”9 They are countries that have adopted some democratic practices
but have not yet fully extinguished autocratic practices in their government.

• Autocracies have various characteristics of autocracy and few or none of the characteristics of democracies. Guarantees of political rights are essential to institutionalized democracies, and most such polities guarantee civil rights to all citizens. Therefore, while the democracy index is based on an analysis of political institutions, it correlates very closely (+.90) with Freedom House indices of political rights and civil liberties.


Using the trichotomized measure of democracy, we discovered that partial democracies are indeed far more vulnerable to state failure–type crises than are either full democracies or autocracies. To be precise, when using this measure of democracy in the global state failure model—along with infant mortality and trade openness—to discriminate between stable and crisis cases, we find that partial democracies, other things being equal, are on average three times more likely to fail.

This refined version of the global model also confirms and makes more precise our estimates of the impact of trade openness and infant mortality (or overall quality of material life) on failure risks. Using the updated problem set, revised data, and new control cases, we find that states with above-average trade openness, other things being equal, have one-half the failure risk of countries with below-average trade openness. In addition, countries with above-world median levels of infant mortality have, other things being equal, three times the risk of state failure as compared with countries with below-median levels of infant mortality.


In the initial work of the Task Force, there was some concern that grouping advanced democratic nations and poor autocracies in one global analysis was like comparing apples and oranges. We have, therefore, applied our analytic techniques to testing the model on those crisis events and a matched set of control cases, drawn solely from the countries of
Sub-Saharan Africa.10 In addition to testing all of the factors that emerged as significant in the initial report, we also examined a variety of additional factors that area experts suggested as specifically relevant to Africa, including a country’s colonial heritage, conditions of ethnic discrimination, and level of urbanization.

The model that most effectively discriminated between crisis cases and control cases in the Sub-Saharan Africa model had six significant elements.11

Level of Democracy

As with the general model, partial democracies were most vulnerable to state failure. This result again showed a high degree of statistical significance. However, while in the global model full democracies and autocracies were about equally stable, in Sub-Saharan Africa autocracies were slightly more stable than even full democracies—presumably because in Africa full democracies have greater problems managing ethnic conflicts and fluctuations in material living standards than do the full democracies of Europe and North America. In addition—and this is one of our most striking results—we found that the vulnerability of partial democracies to state failure was especially great in Sub-Saharan Africa and much higher than in the world at large. The precise results of this model were that in Sub-Saharan Africa, other things being equal, partial democracies were on average 11 times more likely to fail than autocracies. Full democracies were far less vulnerable; other things being equal, they were on average more than twice as likely to fail than autocracies.

Trade Openess

Trade openness is also confirmed as a highly statistically significant correlate of state failure. The greater a country’s trade openness, the less like that country is to experience a major state failure. As in the global model, other things being equal, countries in Sub-Saharan Africa that were above the median in trade openness were on average only about one-half as likely to fail as countries below the median.

Change in Material Living Standards

In the global model, which compared countries with a huge range of living standards, the level of material living standards – as measured by infant mortality (or by GDP per capita or a basket of health and welfare measures) – was found to be a powerful discriminator of risks of state failure. In the Sub Saharan African cases, where most countries are clustered at the low end of the scale of material living standards, recent changes in living standards emerged as a stronger indicator of failure risks than did absolute levels. In particular, other things being equal, countries that had experienced a negative annual change in GDP per capita were on average twice as likely to experience a serious political crisis two years later than countries that had a positive change in GDP per capita.

Colonial Heritage

The Task Forces – along with Sub Saharan Africa regional experts – discussed the possibility that differences in colonial heritage affect vulnerability to state failure. Although states of all varieties of colonial background did experience problems, the data showed that, holding other factors equal, former French colonies on average had only one-third the risk of failure of other African countries. This was a firmly statistically significant result. However, we note that until recently france has also maintained a higher level of engagement – political, financial, and military – with its former colonies that most other powers. As these levels of engagement decline, it may well be that French colonial heritage will become less significant as a moderating factor in regard to state crises.

Level of Urbanization

Although the absolute level of GDP per capita was not a significant predictor of state failure, when combined with the level of urbanization—as measured by the proportion of population living in urban areas—the impact was statistically significant. Having a high proportion of urban population increased the risk of political crisis only in countries whose GDP per capita was below the average for Sub-Saharan Africa. Among such low GDP per capita nations, the risk of failure was twice as high as for countries with higher levels of urban population.

Interestingly, the effect of the share of population in urban areas on failure risks becomes favorable in countries with higher levels of GDP per capita. Other things being equal, for countries that had—by Sub-Saharan African standards—above average GDP per capita, those that also were above average in their proportions of urban population were only one-fifth as likely to fail as those that had lower levels of urbanization. In sum, countries with either high GDP per capita and higher levels of urbanization—relative to other Sub-Saharan African countries—or low GDP per capita and low urbanization were more stable; it was only when relative levels of urbanization were “out of balance” with relative levels of economic
development that political risks increased.

This confirms the bimodal effect of urbanization on political risks described by Jack Goldstone in his work on early modern European states;13 namely, that if the economy is doing well, and urbanization takes place in the context of good employment opportunities, then migrants to cities are socialized into an urban context that they view as rewarding hard work and promising a better future. This is politically stabilizing. In contrast, if the economy is doing poorly and urban migrants find poor opportunities for employment, then migrants are socialized into an urban context that is frustrating and that they view as hostile and unresponsive. This situation greatly aggravates the risks of political crisis.

Ethnic Discrimination

The presence of communal groups that are subject to significant economic or political constraints appears to increase the risks of political failure, all other things equal, by almost a factor of two. However, this result was only weakly statistically significant and should be
viewed as suggestive rather than conclusively demonstrated.

The Sub-Saharan Africa model had roughly the same accuracy as the global model—about two-thirds—in discriminating between state failure and stable cases14 but resulted in substantially reduced “false positives” for Sub-Saharan African countries.15



Institutionalized democracies have increased significantly in number since the late 1980s. At the end of the Cold War, the number of full democracies in the world system exceeded the number of autocracies for the first time since World War II. As of 1991, full democracies numbered 57, compared with 55 autocracies. By 1996 the number of full democracies had increased to 71, whereas autocracies had declined to 49.

The post–Cold War transition—which Samuel Huntington calls “the third wave of democratization”16—also has seen the establishment of a large number of partial democracies. In 1996 there were 27 such polities, double their numbers in the 1980s.

The long-run trend by which democracies have come to outnumber autocracies has two sources. One is the significant number of new democracies established in the post-Communist states. The other, and more important factor, is that many countries that tried and failed to establish democratic polities tried again. South Korea, for example, shifted from autocracy to full democracy in 1960, but a year later lapsed back to autocracy. In 1963 it shifted again to partial democracy but returned to autocratic rule in 1980. South Korea’s most recent transition began in the mid-1980s and was completed in 1988 when it became, and has thus far remained, a full democracy.

In short, South Korea accounts for three transitions toward democracy and two cases of backsliding to autocracy. Pakistan, Turkey, Thailand, and Bangladesh—all full or partial democracies by 1997—also made three or more democratic transitions between 1955 and 1996.

Transitions are defined in terms of shifts among the three categories of regime type—full democracy, partial democracy, and autocracy. For the analysis of trends, the Task Force defined transitions to democracy as shifts from autocracy to either partial or full democracy as well as shifts from partial to full democracy.17

These transitions are said to be stable if the regime does not regress toward autocracy in the first five years after the initial transition.18 A regime is unstable if it regresses toward autocracy within five years. Thus, a country that changes from autocracy to partial democracy, then two years later transitions from partial to full democracy, is counted as having made one stable transition. A country changing from partial democracy to autocracy and remaining an autocracy for five years is counted as a stable downward transition; whereas a country that shifts from democracy to autocracy, then within five years returns to partial democracy, would be counted as an unstable downward transition.

Four major observations can be made about the evidence:

• Many democratic transitions do not endure. Between 1957 and 1991 there were 54 durable transitions— that persisted for at least five years—toward full or partial democracy in independent countries, including 16 democracies established during the period 1989-91 in the Soviet and Yugoslav successor states. Another 20 democratic transitions were attempted between 1957 and 1991 but reverted to autocracy during their first five years. An additional 33 democratic failures—durable democracies that shifted toward autocracy for at least five years—occurred.

• Post–Cold War democratic transitions may be more durable than earlier ones. Before 1986, 24 regimes made durable transitions toward democracy, more than offset by 44 failures—reversion to autocracy—of full or partial democracies.19 The 38 durable transitions toward democracy between 1986 and 1991, however, were offset by only nine failures. A more precise comparison looks only at the outcome of democratic transitions that were attempted between 1957 and 1991. Of the 36 transitions that occurred before 1986, 12 countries (33 percent) reverted to autocracy within five years; whereas, for the 38 transitions in 1986 or later, only eight (21 percent) failed to survive. The short-term survival of democratic transitions thus appears to have increased slightly in the post–Cold War period, although the difference is not quite statistically significant.

• World regions differ substantially in the success of democratic transitions. Before 1986, Africa south of the Sahara had only one durable democratic transition and the record in Asia was only slightly better. In Latin America and the Caribbean, half of the pre-1986 transitions endured to early 1997. The success rates of recent democratic transitions are highest in Asia—where Cambodia is the only recent democratizing regime to backslide (in 1997)—and in Latin America. Despite a great deal of concern about the durability of the post-Communist states, 14 of the 19 that became partial or full democracies during 1989-91 have maintained democratic regimes. The exceptions are Azerbaijan and Armenia—where democratic governance was undermined by civil war—and Belarus, Kazakhstan, and Albania where it was subverted by chief executives who dissolved or emasculated legislatures that constrained their power.

• Partial democracies are less durable than either autocracies or full democracies. There are inherent political contradictions in most partial democracies—a tension between demands for greater and more effective participation on the one hand, and the desire of political elites to maintain or enhance their control. Most partial democracies transition within a decade or so either to full democracies or revert to autocracy.


In developing statistical models of transitions, the Task Force used a narrower definition of transition than it did for the analysis of trends.20 Because crossing the autocracy-democracy divide was thought to be the more critical transition, and because the number of shifts between partial and full democracy was relatively small, the Task Force decided to limit its statistical analysis to transitions from autocracy to partial or full democracy and those from partial or full democracy to autocracy. In this analysis, models were developed that  attempted to answer two research questions:

• What social, economic, and political conditions differentiate countries that make durable democratic transitions from others?

• What conditions characterize countries in which democratic regimes fail to succeed? These questions are different from the issue of the conditions of “state failure” because the  democratic transitions are defined and measured differently from state failures. Moreover, few transitions from autocracy to democracy, and only about half of the transitions from democracy to autocracy, meet the criteria of adverse regime transitions.

Transitions from Autocracy to Democracy.21

A total of 39 transitions to democracy were available for analysis and were matched with 68 control cases—autocracies in the same region that did not shift to democracy during the matching years.22 Experts examined the state failure database to identify variables that they thought should contribute to democratic transitions, and statistical tests were used to determine which of them differentiated significantly between the transitions and the controls.

Then various combinations of these variables were analyzed to determine the most efficient set. From more than 60 models analyzed, the one with the highest accuracy included two variables: relatively low land burden—an index that is highest for countries with largely agricultural populations and scarce cropland—and low durability of the regime before the transition. This model correctly classified two-thirds of the cases in a set of 39 transitions and 68 controls. The best three-variable model correctly classified two-thirds of the cases and showed that durable democratic transitions were most likely when infant mortality was relatively stable, autocracy was already restricted, and land burden was low.

These models suggest some interesting substantive findings. The regimes most likely to undergo stable democratic transitions during the last 40 years:

• Already had shifted away from purely autocratic forms of government.

• Tended to have had less durable regimes; that is, they had attempted previous political experiments. Transitions were also more likely to occur in societies with greater economic capabilities (measured by low land burden) and less short-term variability in quality of life (measured by changes in infant mortality). Once a country has transitioned to democracy, the Task Force found that the likelihood that the transition will be stable depends on several factors:

• Countries whose democratic transitions are most likely to succeed have greater annual improvement in infant mortality, a lower level of infant mortality, greater trade openness, a higher proportion of the population in urban areas, and more years of experience as a democracy.

Transitions from Democracy to Autocracy.23

A total of 35 democratic failures—transitions from full or partial democracy toward autocracy—were available for analysis and were matched with 98 control cases;24 that is, democratic countries in the same region that did not fail during the matching years. The two-variable model with the highest accuracy—nearly three-quarters of cases correctly classified— included infant mortality normalized by world average and regime durability. High infant mortality and low regime durability characterized transitions to autocracy.

It is not surprising that newer democracies—those of low durability—are more likely to fail than long-lived ones, based on the evidence that many democracies fail during their first five years. The role of infant mortality—and by extension, other aspects of quality of life—in raising the prospects for democratic survival is consistent with the results of the general models of state failure.


Goals and Hypotheses

We set out to determine whether the proposition that there is a measurable connection between environmental degradation and state failure was true. Our goals were to:

• Test the argument with data drawn from all countries, over an appropriate time period. Although a number of scholars in recent years have claimed that there is a connection between environmental degradation and political violence, these claims have been largely based on individual case studies.25 These individual studies, albeit largely of high quality, fail to rigorously test the correlative claim.

• Determine whether it was possible to offer analytical guidance to decisionmakers as they face new security challenges. US policymakers—in the State Department, National Security Council, Defense Department, and other agencies—have increasingly framed environmental issues in security terms.26 No clear consensus exists, however, as to what kinds of environmental changes are most important, what factors make a given level of environmental change more or less dangerous, or what types of policy interventions are most promising.

• Construct a specific model, and test it with empirical data, to provide the foundation for monitoring and forecasting potential trouble spots, where environmental deterioration
could potentially enhance the likelihood of state failure.

Two primary expectations guided our analysis:

• We did not expect to find any direct, measurable correlation between environmental change and state failure. Although this expectation is at odds with some of the literature,27 we were guided by the following logic: models of environmentally induced political violence all include numerous intervening variables that are held to interact in a complex fashion.28 The large number of intervening variables makes it hard to find strong direct relationships between the environment and state failure. The complex interaction means that whatever relationships do exist are likely to be different from case to case. As a result, the linkages between environmental change and state failure are unlikely to be discovered by simply adding environmental variables to a state failure model.29

• We did expect that environmental change might have a significant, negative impact on one of the factors associated with state failure in the general model. In particular, we sought to explore whether environmental degradation would have an impact on quality of life
measures such as infant mortality. If so, then this would demonstrate an important, though indirect, connection between environmental degradation and state failure.

Analytically, we conceived of the factors interacting in the following manner: a given change in environmental conditions generates an impact on a society that varies according to the underlying environmental conditions—a society’s vulnerability—and which is mediated by a nation’s capacity to respond effectively. Where capacity is high, harm will be avoided.

To illustrate, consider crop yields as the impact and drought as the environmental change. Vulnerability is the degree to which crop yields might be expected to fall in the absence of effective intervention. It might be measured through extent of irrigation or sensitivity of crops to rainfall. Capacity is the degree to which the government and social actors are able to lower the actual impact, and might be measured as the size of the government budget, number of scientifically trained experts, or extent of communications infrastructure.

To be even more concrete, for the 1991-1992 growing season, El Niño–driven droughts were forecast for northeastern Brazil and for Zimbabwe, with more or less equivalent lead times given to decisionmakers and a comparable projected and actual change in environmental resources—rainfall. The vulnerability—the potential drop in agricultural production divided by loss in rainfall—was also about the same. However, the net social impact, or actual loss in output, was very small in Brazil but quite high in Zimbabwe, where 80 percent of the maize crop was lost. Many analysts attribute this difference to different levels of capacity in the two settings. Officials in Brazil acted on the knowledge early, implementing effective strategies, whereas in Zimbabwe the information was never used, and no responsive strategies were developed.30


Environmental change does not appear to be directly linked to state failure. To determine whether it was possible to find a statistical correlation between environmental change and state failure, we tested variables that measured deforestation and freshwater supply, but both failed to generate significant results. This was consistent with our hypothesis that the more direct effects of democratization, trade openness and quality of life—measured by infant mortality—had such a strong impact on state failure that they masked any impact of environmental deterioration.

This result is at odds with recent work by Hauge and Ellingsen,31 the only other study we are aware of that employs statistical tests to evaluate claims about the direct impact of environmental harm on political violence. Hauge and Ellingsen found a significant impact from deforestation, soil degradation, and freshwater access, results that we believe are due to
differences in how the dependent variables are operationalized and how the independent variables are used. Some of these differences are potentially large enough to account for the different results by themselves; taken together they make the two models essentially incomparable. Because the state failure model covers a greater time period and includes trade openness as an explanatory variable, we think its results have more validity. Nevertheless, the Hauge and Ellingsen model shows that there is more than one way to approach these questions, and we welcome the opportunity for scholarly debate.

Environmental change is significantly associated with changes in infant mortality. To investigate the merits of the mediated model, we assembled data on environmental change, vulnerability, and state capacity. Because of data limitations, we limited our scope to the period 1980-90; extending the time frame back further would have seriously reduced the number of countries and variables available for testing.

We chose infant mortality as the dependent variable because of the availability of data, the high significance of infant mortality as a factor associated with state failure, and the high correlation of infant mortality with a number of other measures of material well-being. We would have preferred to use a basket of indicators that captured the level of material well-being or quality of life, but the only well-being indices we located covered too few countries, spanned too few years, or included factors that were not relevant to our analysis.

Once the data were assembled, we screened potential capacity and vulnerability variables by computing their correlation with infant mortality. Those that were significantly correlated—telephones per capita, population in subsistence agriculture, and land burden—were then tested in combination with an environmental stress variable in a multiple linear regression model.32

As we expected, deforestation proved to be statistically significant only when tested in a model that included measures of vulnerability and capacity. For given levels of vulnerability, capacity, and baseline infant mortality rates, we found that the greater the loss of forest cover, the higher the increase in infant mortality rate.

The results for the model using soil degradation as the environmental stress were more complex, and no linear relationship could be measured. We obtained significant results, however, by multiplying the rate of degradation by its severity and including it as an interactive term. The results suggest that soil degradation has a negative impact when severe degradation occurs at a rapid rate; otherwise the impact is positive. One possible interpretation of this finding is that the same practices that induce soil degradation—such as agricultural production—might have a positive net impact, for example, by improving
nutrition or incomes, if the degradation does not proceed too rapidly.33


One major insight that emerges from the analysis is that available measures of environmental degradation do not currently serve as a direct signal of impending state failure. In part, this is a function of the long, complex chain of association between environmental change and state failure, with a number of factors intervening along the way. Those factors are strong enough to push some societies blessed with benign environmental conditions into failure and to prevent other societies suffering serious environmental damage from slipping into political instability. This finding is also a function of the seriously limited data at our disposal. On balance, we cannot say how large an impact environmental damage has on the risk of state failure.

Nevertheless, the results of our analysis provide evidence for an indirect connection between environmental change and state failure. Deforestation and soil degradation appear to diminish the quality of life, as measured by infant mortality rates, for low-capacity states that are socially vulnerable to disruptions in soil ecosystems; and infant mortality has been shown to have a direct impact on the likelihood of state failure.

Caveats on the Findings

While we believe that the results of the mediated environmental model are useful and significant, the model has several limitations:

• The process of converting analytic concepts into measurable variables has necessarily resulted in variables that are more narrow and arbitrary than the analytic constructs that they represent. This is most true for our core capacity variable—telephones per capita, which we recognize to be a very limited measure of governmental and societal response capability—but to a degree it is true for all the variables.

• The findings represent a general tendency that applies to the set of all countries for which data were available, over the ten-year period studied. That does not mean that this tendency will be true for each individual country at every point in time. Some countries might experience far more direct connections between environmental change and state failure than we observe; other countries might experience less connection between environmental change and infant mortality than our results suggest.

• Environmental data limitations mean that our conclusions are far from the last word. We simply did not have measures for some very important environmental changes—including water quality, with its impact on public health—that might prove more significant as precursors of state failure than those we tested. Data constraints also prevented us from testing whether state failure is associated with aggregate processes of environmental deterioration, encompassing the degradation of soil, air, and water systems.


The main result from retesting and refining the global model is a solid confirmation of the work undertaken in the first phase of the Task Force’s work. Even with an updated and expanded problem set, different control sets, and more refined measures of democracy, the basic global model continued to accurately classify roughly two-thirds of historical cases. Moreover, the same independent variables emerged as statistically significant in a variety of retests.

The major implication for forecasting is that as far as statistical data are concerned—given current limitations in accuracy and coverage for global data—using a large number of variables does not add to the effectiveness of forecasting models. In many cases, we found that the gaps in either the temporal or geographic range of particular variables were so great that any possible gains in prediction were offset by statistical uncertainties or missing data problems associated with measuring those additional variables. Thus, in all models and regional sub-models, a handful of variables emerged as providing significant power in discriminating between state failures and stable cases over the past 40 years. Although many additional variables—including those measuring nutrition, education, droughts, and civil rights—showed significant correlations with risks of state failure, they did not add statistical power to models based on our key variables. Those variables, which consistently emerged in a wide variety of models, are material living standards, trade openness, and democracy, and in more limited circumstances, youth bulge, regime duration, ethnic dominance or discrimination, and the urban proportion of the population.34 We shall have to wait until the accuracy and coverage of global data series improves before we can gain further accuracy by
building more complex models. In the meantime, there is a compelling need to improve global and regional data on these key dimensions, and on many other social, economic, political, and particularly, environmental conditions.

A secondary implication is that the accuracy of statistical models forecasting state failure risks two years in advance remains at a level that is useful, but insufficient for refined predictions. In order to bridge the gap between the two-thirds accuracy of our statistical model, and the better than 90-percent accuracy required for effective policy responses, the skills of individual country analysts and policymakers in assessing rapidly changing local conditions remain absolutely crucial.

The mathematical data analysis cannot prove causality, but the correlations are consistent with causal interpretations. Our findings also suggest policy implications that are interesting and complex, although the best focus and mix of policy responses will, of course, vary from case to case.

Involvement in international trade, as measured by trade openness, is associated with a lower risk of state failure in virtually all states and all contexts. This finding suggests that policies or measures—including internal factors such as dependable enforcement of contracts, modest or low corruption, and improved infrastructure, as well as bilateral or multilateral efforts to eliminate trade barriers—that help to foster higher levels of international trade could help prevent political crises. Interestingly, it appears that it is the involvement in international trade itself, and not the eventual prosperity that such trade provides, that is the key to this effect. The work of Etel Solingen has shown that free trade, if sustained, helps bring together coalitions of elite actors that support the rule of law and stable property relationships, as a condition for building wealth.35 Such coalitions may or may not be democratic, but in either case, they promote political stability.

Partial democracies—particularly in lower-income countries where the quality of life remains poor—are associated with elevated risks of failure. Although full democracies and autocracies are fairly stable, the in-between forms of government are at high risk of undergoing abrupt or violent change. This suggests that while a policy of promoting democracy may eventually lead to a world of stable liberal states, one cannot presume that the inevitable intermediate stages will also be stable. Policymakers need to be particularly attentive to the risks of failure in such states, and should seek and encourage progress toward full democracy. Moreover, if helping to increase the odds of stability in such states is a goal, then policymakers need to focus on developing policies that help foster international trade and on supplementing democratization programs with broad development programs that help improve the overall level of material living standards.

Material living standards have an undeniable effect on the risks of state failure. In some models, it is the overall level of material living standards that emerges as important; in other models, such as that for Sub-Saharan Africa, it is the direction of change that appears crucial. In either case, the policy implication is that efforts to improve material living standards are a significant way to reduce risks of state failure. In Sub-Saharan Africa, it turns out that high levels of urbanization reinforce this effect—for states with high levels of urbanization, states experiencing growth in GDP per capita have only a fraction of the risks of state failure of those states experiencing economic stagnation or decline.

Despite the prevalence of ethnic conflicts—especially in Sub-Saharan Africa—ethnic discrimination or domination is not the sole, or even the most important, correlate of state failure. Because ethnic factors do not emerge as the most powerful—or most statistically significant—factors associated with state failure, they bear monitoring, but other policy levers may be more readily available and more effective.

Environmental stress, vulnerability, and capacity form an interdependent triad that affects quality of life and, indirectly, the risk of state failure. Our findings imply that analysts concerned with the social impact of environmental change need to monitor not simply the environment, but also changes in a country’s vulnerability to environmental changes and its capacity to cope effectively with them. The increased appreciation of the need to develop indicators of environmental change and of sustainability should be complemented with equally vigorous efforts to develop useful indicators of vulnerability and capacity, where the recent track record has been less encouraging. At the broadest level, our findings also suggest that when it comes to minimizing declines in quality of life, increases in capacity and reductions in vulnerability are equally appropriate targets for policy intervention as increases in environmental protection.

Newer democracies, especially in countries where quality of life is relatively low, are more likely to fail than long-lived ones. The Task Force’s models and data can be used to inform
policymaking about the conditions under which democratic transitions are likely to succeed or fail. Most contemporary democracies in Latin America, Asia, and Africa established democratic institutions one or several times, then reverted to autocratic rule before making their most recent transitions to democracy. The problem-ridden history of democratic transitions in these regions raises questions about the future durability of newly established democracies there and in the post-Communist states. Analytic results suggest it is crucial that international support for democratic institutions be reinforced by policies that promote improvement in the quality of life.


Potentially fruitful future analytic directions that are suggested by the Phase II results include:

• Forming a better understanding of the conditions of successful democratic transitions. Initial results suggest that successful democratic transitions tend to be preceded by political experimentation–including previous unsuccessful attempts to establish democratic institutions–and to occur in countries where agricultural stress is low and material living standards are higher. On the other hand, backsliding to democracy tends to occur within a few years after democratic institutions are introduced, and in countries with relatively low quality of life and high agricultural stress. Analyses are needed of the extent to which successful democratic transitions depend on improvements in the quality of life, and economic performance generally, during the early years. Models of these relationships should also take account of factors such as elite ethnicity, urban growth, and youth bulge, which
have been shown to correlate with other kinds of state failure, especially revolutionary and ethnic wars.

• Further developing the concept that the impact of environmental degradation on state failure is mediated by vulnerability and capacity, and more thorough testing of the model. Additional steps would include:

→ Constructing additional indicators of environmental change—such as water and air quality—vulnerability, and capacity from currently available sources.

→  Building a set of “watch lists” for specific ongoing environmental threats that would focus attention on environmental deterioration in countries with high vulnerability and low capacity.

→  Developing a core set of environmental indicators—measured consistently across countries and over time—that could be used in future analyses. This effort would include using the next generation of remotesensing satellites to gather terrestrial and atmospheric
data and using intensive on-site monitoring to build an adequate database for other environmental problems such as water quality, air quality, and chemical hazards.

→  Developing models that capture regional variation—or localized “hot spots”—within a country that are masked by national level analysis. We know that the environmental impact on material quality of life will be stronger if there is a spatial correlation among the variables. For example, if a given unit of land has a high rate of deforestation, a high land burden, and
poor institutional capacity, we would expect a larger local impact on infant mortality, an hypothesis that could be tested using currently available high-spatialresolution data sets.

→  As additional data become available, continuing to test the hypothesis that environmental damage directly contributes to the likelihood of state failure.

• Developing a more detailed concept of “state capacity” to test as a mediating factor in general and regional models. Building on the results of the mediated environmental model, further examine and develop in more depth the concept of state capacity, develop quantitative measures that tap this dimension, and incorporate this concept as a mediating factor. We should also seek or develop data sets that are better able to capture
state capacity.

• Investigating the usefulness of pilot studies of event data for bridging the gap between model-based risk assessments and “early warnings.” The general models of state failure identify risk factors measured two years before the expected onset of failure. Even the best models identify a substantial number of false positives and fail to predict correctly some failures. The goal is to supplement general models with early warning models that track the immediate precursors of failure and provide more accurate and timely warnings than do risk assessments that are based on background conditions. Specifically, monitoring of events should concentrate on situations judged to be at high risk through expert- and model-based analysis, and statistical techniques should be applied to study the clustering of events before a crisis.

• Investigating the impact of international support on the risk of state failure. Many policymakers and analysts assume that bilateral and multilateral policies can forestall some state failures and minimize the severity of others. Previous Task Force analyses have assessed the impact of some kinds of international economic policies—such as IMF standby agreements—on the likelihood of state failure, but these analyses have not shown strong and consistent results. The impact of other kinds of international engagement, such as diplomatic and military support, development programs, and assistance with institution building remain to be studied. Appropriate data and indicators need to be gathered and tested in new models. Because the objectives and hence the likely outcomes of international policies have changed since the peak of the Cold War, such models should distinguish between preand post–Cold War patterns of international policy and their consequences.

Appendix A: Global Model and General Material


State Failure

State failure and state collapse are new labels for a type of severe political crisis exemplified by events of the early 1990s in Somalia, Bosnia and Herzegovina, Liberia, and Afghanistan. In these instances, the institutions of the central state were so weakened that they could no longer maintain authority or political order beyond the capital city, and sometimes not even there. Such state failures usually occur in circumstances of widespread and violent civil conflict, and are often accompanied by severe humanitarian crises. In a general sense, they are all part of a syndrome of serious political crisis which, in the extreme case, leads to the collapse of governance.

Only 18 complete collapses of state authority have occurred during the last 40 years, too few for meaningful statistical generalization. Therefore, the Task Force broadened its focus and sought to identify systematically all occurrences of partial as well as complete state failures that began between 1954 and 1996. We began from existing compilations of data on revolutionary and ethnic conflicts, regime crises, and massive human rights violations of the types categorized as genocides and politicides (political mass murders). An initial list—the basis for the Phase I analysis—was critically evaluated, updated, and refined for the present study. The four types of internal wars and failures of governance are:36

• Revolutionary wars. Episodes of violent conflict between governments and politically organized challengers that seek to overthrow the central government, to replace its leaders, or to seize power in one region. From the 1950s through the late 1980s, most revolutionary wars were fought by guerrilla armies organized by clandestine political movements. A few, like the Iranian revolution of 1978 and the student revolutionary movement in China in 1989, were mass movements that organized campaigns of demonstrations. These mass movements are included only if one or both parties used substantial violence.

• Ethnic wars. Episodes of violent conflict in which national, ethnic, religious, or other communal minorities challenge governments seeking major changes in their status. Most ethnic wars since 1955 have been guerrilla or civil wars in which the challengers sought independence or regional autonomy. A few, like those in South Africa’s black townships in 1976-77, involved large-scale, violent protests aimed at sweeping political reforms. Warfare between rival community groups is not considered ethnic warfare unless