An influential related research line that has also developed around the issue of colonization is based on the original contribution by Acemoglu et al. (2001). With the more general goal of establishing the importance of institutions for current development, they identify in colonial history an instrument for institutions which can address their potential endogeneity. Indeed reverse causation becomes a crucial issue when taking into account institutional variables, since the direction of causality with respect to income and growth is by no means obvious. They focus on a specific variable, the mortality rate of the initial settlers, about which they collect and compile information based on detailed historic sources. Settler mortality is exploited as an instrument for current institutions, by arguing that colonizers adopted very different policies in places with different mortality rates. In places where they faced high mortality, they could not settle and were therefore more likely to set up extractive institutions, which in turn persist in present times. Their two-stage least-squares estimates establish that institutions, and in particular the degree of property rights protection, do affect economic development. In the previously mentioned papers based on growth regressions, the potential for endogeneity is addressed indirectly by Bertocchi and Canova (2002), who employ predetermined explanatory variables dated at the beginning of the sample period, and directly by Bertocchi and Guerzoni (2011), who instrument with its lag the only variable for which they find evidence of endogeneity, i.e., government expenditure over GDP. For the sake of comparison between the two parallel streams of the literature on colonization, it should be also noticed that Acemoglu et al. (2001), because of the way their instrumental variable is generated, can only include former colonies in their sample. Moreover, they do not focus exclusively on Africa. Finally, they do not distinguish, as Bertocchi and Canova (2002) do, between colonies and dependencies, even though to some extent it could be argued that at least within Africa dependencies were established in lower mortality places.
Another related stream of the literature has gone even beyond the colonial period to discover the roots of current performances. Indeed one cannot exclude a role for long term factors that predate colonial domination. Examples are the following. Herbst (2000) points to the underdevelopment of pre-colonial polities. Bockstette et al. (2002) establish a link between state antiquity, a measure of the depth of experience with state level institutions, and institutional quality. Gennaioli and Rainer (2007) uncover for Africa a positive association between stronger pre-colonial political institutions and public goods provision. Michalopoulos and Papaioannou (2010) show that tribal pre-colonial political institutions and class stratification still exert an effect on local economic activity in Africa, which they measure using satellite data on light density at night following Henderson et al. (2010). Finally, Nunn (2008a) studies the impact of the slave trades, while Nunn and Puga (2010) focus on the historic interaction between terrain ruggedness and Africa’s slave trades, i.e., between geography and institutions. In Section 6 I return to the issue of the slave trades in more detail.
More recently, micro-level data have also been exploited, with an even more precise focus on single African countries or specific sub-regions. One example of this line of research is Huillery (2009), who employs household surveys data on French West Africa to produce evidence that early colonial investments had large and persistent effects on current outcomes.
While the literature described so far is empirical, a few efforts have been made to model the underlying causal relationships within a dynamic setting. Bertocchi (1994) explicitly introduces colonization into a standard growth model with overlapping generations, in order to determine the net effect of modernization and the drain of wealth on the colonial economy. Colonization is modeled in the form of restrictions on direct foreign investment and exploitative activities, which may induce permanent distortions to physical and human capital accumulation and thus lead to negative growth rates even after decolonization. Nunn (2007) develops a game-theoretic model with multiple equilibria, only one of which is associated with secure property rights and a high level of production. In this setting, external extraction may drive a society into a low production equilibrium. Since this equilibrium is stable, the society remains trapped into it even after external extraction ends. Both models thus provide an explanation for the lasting legacy of colonial rule.
Summing up, colonial history has been shown to matter for growth in Africa, in a number of dimensions, but its influence has not been captured by a single explanatory variable which can identify the impact of colonization. Moreover, its effect appears to have faded over time, while the region is still underperforming. This implies that even accounting for colonial history does not yet provide a complete and consistent answer to the question about the determinants of African growth.
5. State fragility
The concept of state fragility (from now on, fragility) is perhaps the most recent newcomer to the debate about growth in Africa. Rather than to specific economic, institutional, or historical characteristics, the condition of fragility has been associated with combinations of multiple dysfunctions, including a country’s inability to provide vital services, unstable and weak governance, persistent and extreme poverty, lack of territorial control, and high propensity to conflict and civil war. It is clear, however, that these dysfunctions can in turn be determined by repeated interactions across a number of factors over the long run. While it may be difficult to isolate the influence of each of these factors singularly, using a composite measure may facilitate the analysis.
Once again, Africa has played a central role in the analysis of fragility, since it is in this continent that fragility is especially widespread. Indeed the European Report on Development (2009) is entirely devoted to the problem of fragility in Africa. The potential growth impact of fragility and the consequent relevance of fragility for policy are also confirmed by the increasing attention of other international organizations. The 2011 World Development Report (World Bank, 2011) focuses on conflict countries. Development practitioners, such as the Government and Social Development Resource Centre (2010), also warn policymakers about the need to understand and respond to fragile situations.
One of the most widely used definitions of fragility is based on the Country Policy and Institutional Assessment (CPIA) which has been conducted by the World Bank since 1999. The ratings are intended to capture the quality of a country’s policies and institutional arrangements, with a focus on the key elements that are within the country’s control, rather than on outcomes (such as growth rates) that are also influenced by elements outside the country’s control. Since the CPIA ratings represent criteria for aid allocation, they carry huge practical implications for policy. On the basis of the CPIA, the World Bank defines as fragile those low-income countries scoring 3.2 and below (over a 1-6 range). From 1999 to 2005, the individual ratings have been kept confidential. However, the general rankings of countries have been made public. On the basis of the rankings, it is therefore possible to infer the distribution of the countries by quintile. On the basis of the resulting quintile distribution, the OECD defines as fragile those countries in the bottom two quintiles, as well as those which are not rated. There is a partial overlap between the CPIA-based definitions of fragility and other related indexes, such as the Failed State Index (published by the Fund for Peace), the Index of State Weakness (published by the Brookings Institution), the indicator of Failed & Fragile States (published by the Country Indicators for Foreign Policy project), and the Fragility States Index (published by Polity IV). While all these indicators record similar components, the choice of variables and their weighting schemes remain largely arbitrary.