Risks From Introducing the Anticartel Government IT System

Abstract

The paper provides an ex ante assessment of risks incurred by developing and implementing the Anticartel state information system (GIS), which will conduct automated screening of Russia’s state procurement in order to detect bid rigging (in other words, software analysis of procurement data using algorithms that identify collusion based on certain formal characteristics). The system will also apply artificial intelligence. Determining the risks of implementing this system is difficult because there is no publicly available information on the specific criteria that will be used in this screening process. This forces specialists to make do with partial data. This paper attempts to reconstruct the operating principles of the Anticartel GIS from external sources including public speeches by representatives of the Federal Antimonopoly Service of Russia together with the technical specifications provided to developers entering the competition to build and implement the system. Key questions prompted by examining Anticartel’s reconstructed features are put forward in order to facilitate analysis during an “expedited” regulatory impact assessment based on the assumption that the decision to employ such a digital screening system is irreversible. The questions and analysis will structure discussion of the risks, associated benefits, and costs of different options for implementation. The principal risks result from probable Type I errors in law enforcement and from the extent of their consequences, which will be amplified as the automated screening system rolls out to include the entire body of regulated procurement, as well as by the use of artificial intelligence. The main consequences of this type of error and how it would arise are explained in the article. In addition, certain risks from using digital tools for identifying collusion especially in commodity markets are identified.