Posted by Francois Michel
If one wanted to summarize briefly how research on corruption has evolved in recent years, one could say that it has made progress in four different areas:
The search for determinants of corruption and its transmission mechanism to growth and the exploration of linkages between corruption and other economic—GDP per capita, capital flows, aid, income distribution, inflation, etc.—or political variables. This is often achieved through panel data analyses;- The improvement of transparency indexes—see Daniel Kaufmann and Aart Kraay’s recent article on "Governance Indicators: Where Are We, Where Should We Be Going?";
- Efforts to leverage insights from the corruption literature into sectoral, country-specific reform plans, and that have formed the core of the World Bank’s strategy in recent years. As the Bank’s recent flagship publication on corruption makes clear, public financial management reform are instrumental in tackling corruption;
- New microeconomic models explaining how corruption can originate in auctions of procurement contracts—e.g. on allowing ex-post collusion opportunities between the bureaucrat and one bidder.
The recent article “Bribery and public procurement: an experimental study” (Public Choice; Oct2008) from Susanne Büchner, Andreas Freytag, Luis González and Werner Güth belongs to the fourth category. The article’s model focuses on how bidders solve a tradeoff between price-oriented and bribe-oriented competition. It predicts actual pricing and bribing by (risk-neutral) competing bidders.
But this article is especially interesting in its account of a laboratory test of the model, a carefully designed game involving a total of 114 participants. While not fully confirming the model’s predictions of bidders’ strategies, the experience still provides some very insightful information on how individual pricing and bribing decisions are formed. Overall, the experience confirms that bidders engage actively—though with much variation in individual behavior—in bribery, whether or not bribes are framed as socially detrimental.
A similar approach could be usefully applied to testing specific features of PFM systems—in particular of internal and external control systems—designed to reduce corruption. For example, one could model and test how transparency and the modalities and strength of ex post controls—probability of contract audits, quality of these audits, etc.—impact bureaucrats’ inclinations to accept bribes.
