Detecting bid-rigging coalitions in different countries and auction formats
شناسایی ائتلاف های تقلب در مناقصه در کشورهای مختلف و فرمت های حراج-2020
We propose an original application of screening methods using machine learning to detect collusive groups of firms in procurement auctions. As a methodical innovation, we calculate coalition-based screens by forming coalitions of bidders in tenders to flag bid-rigging cartels. Using Swiss, Japanese and Italian procurement data, we investigate the effectiveness of our method in different countries and auction settings, in our cases first-price sealed-bid and mean-price sealed-bid auctions. We correctly classify 90% of the collusive and competitive coalitions when applying four machine learning algorithms: lasso, support vector machine, random forest, and super learner ensemble method. Finally, we find that coalition-based screens for the variance and the uniformity of bids are in all the cases the most important predictors according to the random forest.
Keywords: Cartel detection | Screening | Machine learning | Procurement data
Estimating cartel damages with model averaging approaches
برآورد خسارات کارتل با رویکردهای میانگینگیری مدل-2020
This research offers an easy-to-implement forecast combination procedure to deal with issues of model uncertainty when evaluating cartel damages. We combine the Mallows model averaging (MMA) method with both the dummy variable (DV) and forecasting approaches to investigate the famous citric acid cartel case during the 1990s. The path of but-for prices generated from the MMA method with DV specification lies in-between those generated from the forecasting and DV methods, supporting the theoretical properties of the MMA method that weights over different forecasts generated from various candidate models. The findings indicate that the but-for prices generated from the MMA method could serve as a useful robustness check for cartel damage estimations.
Keywords: Cartel | Antitrust | Damage estimation
Machine learning with screens for detecting bid-rigging cartels
یادگیری ماشین با صفحه نمایش برای شناسایی کارتل های تقلب در مزایدات-2019
We combine machine learning techniques with statistical screens computed from the distribution of bids in tenders within the Swiss construction sector to predict collusion through bid-rigging cartels. We assess the out of sample per- formance of this approach and find it to correctly classify more than 84% of the total of bidding processes as collusive or non- collusive. We also discuss tradeoffs in reducing false positive vs. false negative predictions and find that false negative pre- dictions increase much faster in reducing false positive predic- tions. Finally, we discuss policy implications of our method for competition agencies aiming at detecting bid-rigging cartels.
Keywords: Bid rigging detection | Screening methods | Machine learning | Lasso | Ensemble methods
Editors’ JIF-boosting stratagems – Which are appropriate and which not?
JIF تقویت تدابیر ویراستاران - که مناسب هستند و که نه؟-2016
Keywords:Journal Impact Factor Editorial manipulation EthicsMisconduct Inappropriate behaviour Online queueThis extended editorial explores the growing range of stratagems devised by journal editors to boost their Journal Impact Factor (JIF) and the consequences for the credibility of this indicator as well as for the academic community more broadly. Over recent years, JIF has become the most prominent indicator of a journal’s standing, bringing intense pressure on journal editors to do what they can to increase it. After explaining the curious way in which JIF is calculated and the technical limitations that beset it, we examine the approaches employed by journal editors to maximise it. Some approaches would seem completely acceptable, others (such as coercive citations and cross-citing journal cartels) are in clear breach of the conventions on academic behaviour, but a number fall somewhere in between. Over time, editors have devised ingenious ways of enhancing their JIF without apparently breaching any rules. In particular, the editorial describes the ‘online queue’ stratagem and asks whether this constitutes appropriate behaviour or not. The editorial draws three conclusions. First, in the light of ever more devious ruses of editors, the JIF indicator has now lost most of its credibility. Secondly, where the rules are unclear or absent, the only way of determining whether particular editorial behaviour is appropriate or not is to expose it to public scrutiny. Thirdly, editors who engage in dubious behaviour thereby risk forfeiting their authority to police misconduct among authors.© 2015 Elsevier B.V. All rights reserved.Oh, what a tangled web we weave When ﬁrst we practise to deceive!
Journal Impact Factor | Editorial manipulation | Ethics | Misconduct | Inappropriate behaviour | Online queue
Stigmergy at the edge: Adversarial stigmergy in the war on drugs
نشانه ورزی در لبه: نشانه ورزی خصمانه در جنگ با مواد مخدر-2016
As a consequence of the stigmergic coordination that occurs among criminal and government agents, resilience has been built into the system that supplies illegal drugs to American consumers. Criminal agents create technology responses that are simple and cost-eﬀective, and consistently defeat the actions of government agents. Those responses to stigmergic stimulus improve iteratively the resilience and sophistication of the clandestine supply chains. In what the author calls a ‘‘homeland security Chaos Monkey model” a constant but predictable governmental escalation in the war on drugs plays the role of a failure signal to build resilience in the narcotics system: Any success by governmental agents sends stigmergic signals to criminal agents. These signals communicate a failure in the supply chain that requires an adversarial innovation to defeat the updated shape of the interdiction. The result is a more resilient system. This cycle of adversarial stigmergy has encouraged the emergence of a well-coordinated system of clandestine innovation in the territory of the US–Mexico borderlands that takes advantage of the border switch to solve in an iterative form one particular problem: to identify and exploit the vulnerabilities of a complicated enforcement architecture to build resilient narcotics systems. For homeland security policies to be more eﬀective, interdiction policies and technologies should be built with a better understanding of the stigmergic forces that shape adaptation in the war on drugs system.© 2015 Elsevier B.V. All rights reserved.
Keywords: Stigmergy | Self-organization | Organized crime | Innovation | War on drugs | Drug cartels