This thesis is an attempt to improve the understanding of the causes of corruption emergence and incidence around the world. It highlights an undocumented feature of corrupt transactions, that is, their contribution to informal risk-coping and risk-management mechanisms used by economic agents to protect against income fluctuations. We propose in a first introductory chapter a general state of art of researches on corruption definitions, measurements, typologies and determinants. In chapter two, we explain, apply and compare standard methods of computing instability indicators, using export revenue data from sample of developed and developing countries. In chapter three, we build a retrospective Economic Vulnerability Index - i.e. an index reflecting the risk for a country of seeing its development hampered by natural and trade shocks - for a sample of 128 developing countries over 1975-2008. In chapter 4, we analyse the effect of export instability on corruption in developed and developing countries. This effect is decomposed into an ex post effect, resulting from agents' experience of export instability, and an ex ante effect, resulting from their perception of export instability. We test empirically these effects using data on corruption perceptions and on firms' bribe payments. We find robust, significant and nonlinear ex post and ex ante effects of instability on corruption, and stress that their direction strongly depends on the frequency and size of export fluctuations. We show that the liquidity constraint is a credible channel for these nonlinear effects: when the liquidity constraint hardens, instability is found to foster corruption; while when it softens, instability is found to reduce it. Thus, corrupt strategies may act as a substitute for financial market imperfections and a low state capacity for mitigating the consequences of economic fluctuations on welfare.
