In the literature of development economics, corruption is usually conceived as detrimental to economic growth. This conventional wisdom, however, may be called into question. Many countries witnessed growth despite corruption, e.g., commodity-dependent and high-growth East Asian countries. The paper argues, through a comparison of Sub-Saharan Africa and East Asia, that the relationships between corruption and economic growth are difficult to demonstrate. It highlights two crucial factors that explain the lack of robustness of this relationship. Firstly, this lack of robustness stems from the methods of measurement, which are usually based on the building of indices, modelling and econometric techniques. These methods are inappropriate for a concept such as ‘corruption’, which refers to complex and heterogeneous phenomena that are difficult to subsume in a single and stable definition. A second set of factors underlying the weakness of the relationship between corruption and growth is the dependence of causal processes on specific contexts. The effects of corrupt practices on an economy depend on its particular history, its economic structures, its political economy and types of institutions: for these reasons, they vary across countries and regions. Causal links between corruption and growth may exist, but they are non-linear and subject to threshold effects. Beyond certain thresholds, which are built by specific contexts (i.e., the combination of many contextual factors, political, economic, institutional), corruption phenomena can be detrimental to growth; before reaching these thresholds, the impact of corruption on growth may be limited. These thresholds can be assessed only ex post: they cannot be measured ex ante, as they precisely depend on contexts that vary across space, countries and history. In some contexts, economic and political factors may reinforce each other, e.g. corruption, political instability, economic distortions and vulnerability, such as commodity-based market structures. This results in ‘low equilibria’ that combine low growth and pervasive corruption, and thresholds, which, once low equilibria are stabilised, it is very difficult to get out from under (‘poverty traps’). In other contexts, these factors may all exist. They remain separated, however; corruption does not combine with other economic and political factors and is contained, which makes it possible for countries not to fall into ‘lower’ equilibria. The state is here the core entity able to prevent the reciprocal reinforcement of corruption and other economic or political structures - and hence the formation of poverty traps -, and to make corruption subservient to growth objectives. This state capacity that can confine and control corruption, which exists in some countries but not in others, is a key factor in the differences in impacts of corruption on growth.
