Does oil corrupt? Evidence from a natural experiment in West Africa

This paper explores an oil discovery natural experiment to assess the role of natural resources in determining corruption. We argue that an anticipated oil boom may increase corruption by boosting the value attributed by an elite to being in power when the actual oil exploration begins. We test this proposition by analyzing the impact of the oil discovery announcements that took place in 1997-99 in Sao Tome and Principe (West Africa). For this objective we conducted purposedly-designed household surveys on perceived corruption in the public services/sector. These were carried out in Sao Tome and Principe and in Cape Verde, a control West African country sharing strong cultural ties and important contemporary economic/political shocks. The unique survey instrument was retrospective and used personal histories to elicit memories from the respondents. Urban subjects, public officials, and respondents with higher reported experience with the services/issues at stake are used as internal treatment groups. Comparisons are also made with corresponding groups in Cape Verde. In addition, the regressions control for well-known 'good old times' bias: this is done by using data from direct questions on optimism and from the inclusion of a 'placebo' period (when no major occurrence had arisen). We conclude that a clear increase in perceived corruption has occurred in Sao Tome and Principe in recent years, ranging from 21 to 38% of the subjective scale. Consistently with our theoretical mechanism, which underlines the importance of being in power when the oil boom occurs, these effects are most robust in vote buying, education, and state jobs.