Anti-corruption practitioners regularly puzzle over the ‘whack-a-mole’ nature of their task: the problem with fighting corruption is that if you succeed in suppressing it in one place, it tends to pop up in another. By way of example, a recent study of healthcare reform in Uganda found that, although a drastic increase in oversight and penalties for bribe-taking achieved a short-term reduction in bribe requests, the positive effects proved short-lived because staff began instead to solicit ‘gifts’ or other ‘signs of appreciation’ to supplement their incomes. Law enforcement reports similar trends, i.e., when police clamp down on gangs operating in one area, criminals respond by moving to different geographical locations where law enforcement is weaker.
The fact that corruption is a constantly moving target strengthens the case for ensuring that interventions are flexible and adaptive. However, to make informed decisions about how to adapt programming, we need to observe a whole system and detailed data points within it, in order to detect any displacement effects and unintended consequences. That kind of detailed observation is rarely possible.
Our GI-ACE research, however, is based on observing patterns of corruption risk in big data on public procurement. Public procurement is critical to good governance. It typically accounts for around 30–50% of public spending, but it is also highly prone to corruption. Learning how to reduce vulnerability to corruption in public procurement thus promises to bring important benefits, in the form of making public service delivery more efficient, helping to foster competitive markets and boost economic development.
Our big data analysis is well suited to observing a whole system and monitoring how behaviour changes in different areas as the result of reform. In our most recent research article, we analysed data points from World Bank-funded development aid tenders over twelve years in over 100 developing countries. Having such a large procurement dataset that was carried out under the same regulatory regime – the rules imposed by the World Bank on its lending – meant that we could use the dataset to test what happened when a major reform was introduced.
We studied the effects of an anti-corruption reform which introduced limits on the discretion that public officials exercised over the process and increased opportunities for oversight and scrutiny, particularly by bidders. The reform also sought to open up competition by requiring that tenders were advertised more widely and by mandating greater use of e-procurement methods – changes which, by facilitating market entry, are expected to indirectly increase accountability.
The new rules applied to projects that commenced after November 2003, with many projects running for several years and hence awarded contracts long after the reform was introduced. This meant that we could match and compare contracts awarded under different rules in the same or very similar countries, buyers, markets, and financial years. We could focus on the impact of the rules change, controlling for these other explanations of corruption risks.
We found that the World Bank procurement reform was effective in the most direct sense: the share of tenders with a single bidder decreased from 22.4% to 18.7%, and the average number of bidders increased from 4.5 to 5.0. Thus the corruption risks associated with low competition were reduced. The share of repeat winners also fell from 71.8% to 65.4%, again suggesting an opening up to new entrants.
But, we also found evidence that the reform prompted some corrupt actors to switch their efforts to other parts of the procurement process – parts unaffected by the reform. Government buyers started to rely more heavily on non-competitive procedure types, which offer the easiest way to channel contracts to cronies; the use of these procedures increased from 7.3% to 9.6%. And they also appear to have exploited these procedures more intensively: the outcomes of procurement conducted through non-competitive procedure types became high-risk, with the share of tenders attracting only one bidder increased from 67% to 81%. In addition, foreign companies lost out – their market share dropped by 2 percentage points, suggesting that local companies with better connections might have fared better after the reform. Taken together, the net effect of the reform is at best ambiguous. The evasive tactics may have largely cancelled out the positive direct effects of the reform.
What is the conclusion from this rather bleak analysis? Does it mean that anti-corruption efforts are hopeless, because the corrupt will always find ways around new rules? We suggest a more nuanced learning. Anti-corruption reform must go beyond narrow technocratic solutions that only tweak the incentives facing corrupt actors, to address the underlying political and social conditions that drive corruption. Technical interventions are an important part of the toolbox, but they alone cannot tackle systemic practices that are deeply rooted in social and political structures and norms. We need strategies that also target the root causes of corruption and contribute to building new cultures of integrity.
As a professor in politics at University of Sussex, Liz David-Barrett leads the Centre for the Study of Corruption’s (CSC) activities in research, teaching, and policy impact. Her research focuses on corruption risks at the interface between politics and business, in public procurement, lobbying and bribery, as well as on private-sector action to prevent corruption. She engages widely with anti-corruption practitioners in governments, the private sector, and NGOs; has written reports on the UK Bribery Act, lobbying and the revolving door, and local government corruption; and has given evidence to parliamentary select committees. David-Barrett previously worked in Croatia and Hungary as a journalist, reporting for The Economist, the Financial Times, the BBC World Service and Business Central Europe. David-Barrett has a DPhil in Politics from Oxford, an MA in Slavonic and East European Studies from the University of London, and a BA in Philosophy, Politics and Economics (Oxford).
Mihály Fazekas is an assistant professor at the Central European University, School of Public Policy, with a focus on using Big Data methods to understand the quality of government globally. Fazekas also is the scientific director of the Government Transparency Institute, where he promotes the implementation of new measurement instruments of corruption and quality of government using ‘Big Data’. Fazekas’ research and policy interests revolve around corruption, favouritism, private sector collusion, and government spending efficiency. He regularly consults the European Commission, Council of Europe, EBRD, OECD, World Bank, and a range of national governments and NGOs across the globe. Fazekas received his PhD from the University of Cambridge and studied public policy at the Hertie School of Governance (Berlin), economics at the Corvinus University of Budapest, and teaching at the Corvinus University of Budapest.