Across the globe, the exposure of collusive behaviour of companies in procurement markets is predominantly based on qualitative information from firms or individuals involved in collusion. This makes detection rare and titled towards disintegrating bidding rings. Economic analysis, modelling and forecasting have a limited role in this field. However, the increasing availability of large administrative datasets on public procurement transactions and the development of new econometric methods make it possible to develop a wide variety of indicators signalling different forms of collusion.
Based on a synthesis of literature to-date, this paper provides a flexible indicator set deployable as a toolkit across many countries for detecting collusive bidding in public procurement. While no one-size-fits-all approach exists in detecting collusion, robust elementary indicators and analytical tools for adapting them to local contexts can be developed. The paper delivers a conceptual definition and theoretical discussion for each indicator as well as a complex empirical assessment using data on over 75,000 contract awards in Hungary between 2005 and 2012. Indicators are identified, selected, and tested based on relevant academic literature, current best practice policy among competition authorities, court material, and interview evidence from experts and key stakeholders.
Indicators include the relative price of goods and services, skew in the distribution of offer prices, repetition in the pattern of winning companies (i.e. cyclical winning), and co-bidding network constellations suggesting the recurrent submission of losing bids. The proposed toolkit embeds this wide set of collusion risk indicators in a framework which explores theoretically founded co-variation between them and generates benchmarks of ‘normal’ market behaviour using geographical and temporal variation.
The proposed approach differs from previous attempts at generating indicators of collusion in public procurement markets in that it develops a broad ex ante risk-based monitoring framework which is adaptable to a wide range of markets and countries. While it is only an initial synthesis of evidence to date, it serves as a suitable starting point for developing context-specific robust signalling systems.