Study
|
Why should we make EU public procurement more transparent?
A study commissioned by the Greens/EFA Rights and Democracy Cluster
The author of the study Vitezslav Titl is an Assistant Professor of Law & Economics at Utrecht University School of Economics. He is also a member of Utrecht University Centre for Public Procurement, principal investigator of a Junior STAR grant at Charles University, and an affiliated researcher at KU Leuven and The Free University of Brussels. Vitezslav obtained his PhD in economics from KU Leuven and held visiting positions at Bocconi University, Princeton University, ZEW Mannheim. The research interest of Vitezslav Titl comprises of economics of public procurement markets, corruption and law & economics.
EXECUTIVE SUMMARY AND RECOMMENDATIONS
• The aim of this study is to describe and analyse benefits and costs of increased transparency, interoperability, and machine-readability of public procurement data in the European Union. A special focus is given to the subsequent usage of machine learning algorithms in investigating fraud and corruption in public procurement.
• Public procurement is a prominent way of spending public resources in the European Union. The public procurement market is worth about 31% of general government expenditure in the European Union (equivalent of about 2 trillion euros; European Commission, 2017).
• The academic evidence shows convincingly that public procurement markets suffer from a number of issues such as corruption, political connections, and collusions. These issues cause large inefficiencies on the market.
• Higher transparency and the provision of interoperable open public procurement data in machine-readable format are likely to help in eliminating these inefficiencies.
• The interoperability of public procurement registers and other registers such as a company register and registers of politically exposed persons is another important element in strengthening the efforts to eliminate the inefficiencies.
• The public procurement data currently published on the European level are of low quality with many missing and incorrect values and it suffers from a number of other limitations. First, contract notices and contract award notices are not automatically merged, which makes it difficult2 to find, for example, the final price of a public procurement contract that we observe among contract notices. Second, the current reporting system is not linked to other datasets (for example, firm-level datasets or data about procuring authorities). Last, the reporting system does not cover the whole procurement process, i.e. there is no information about the project implementation. All this information would be useful in monitoring and control of the public procurement market.
• Higher publicizing standards and transparency in public procurement are shown to lead to significant savings for the public sector. Coviello & Mariniello (2014) find that due to the higher publicity standard of procurement announcements in Italy, the final prices declined by 7% of the estimated costs and the number of bidders increased by 9.3%.
• A more extensive and machine readable procurement data publication can be used by law enforcement agencies, public and private auditors as well as by the civic society and other volunteers to improve the efficiency of public procurement.
• Machine learning algorithms are effective in detecting potentially corrupt public procurement contracts as well as in detecting situations with conflicts of interests (firms with political connections). Corruption and political connections are associated with a lower procurement efficiency – higher costs and no quality improvements.
• Extensive online monitoring by the public and non-governmental organizations (NGOs) investigating collusion and corruption have been found to be associated with a reduction in the chance of collusive behaviour of procurement suppliers and with a decline in prices (Baranek et al., 2020).
• From a policy perspective, this study concludes that the thresholds for publication in the European Union procurement system should be lowered, the quality of the data should be better controlled and enforced, the public procurement datasets should be linked to other administrative datasets such as company registries and datasets from statistical offices and Eurostat, the whole procurement process should be covered including the information about the project implementation.
