19 pages

Cartel Detection and Collusion Screening: An Empirical Analysis of the London Metal Exchange

of 19
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
"Title: Cartel Detection and Collusion Screening: An Empirical Analysis of the London Metal Exchange Author: Dr. Danilo Samà Abstract: In order to fight collusive behaviors, the best scenario for competition authorities would be the possibility
  Cartel Detection and Collusion Screening:An Empirical Analysis of the London Metal Exchange  ∗ Danilo Sam`a † LUISS “Guido Carli” University of Rome2014 Abstract In order to fight collusive behaviors, the best scenario for competition authorities wouldbe the possibility to analyze detailed information on firms’ costs and prices, being theprice-cost margin a robust indicator of market power. However, information on firms’costs is rarely available. In this context, a fascinating technique to detect data manipula-tion and rigged prices is offered by an odd phenomenon called Benford’s Law, otherwiseknown as First-Digit Law, which has been successfully employed to discover the “LiborScandal” much time before the opening of the cartel settlement procedure. Thus, themain objective of the present paper is to apply a such useful instrument to track theprice of the aluminium traded on the London Metal Exchange, following the allegationsaccording to which there would be an aluminium cartel behind. As a result, quick testssuch as Benford’s Law can only be helpful to inspect markets where price patterns showsigns of collusion. Given the budget constraints to which antitrust watchdogs are com-monly subject to, a such price screen could be set up, just exploiting the data available,as warning system to identify cases that require further investigations. Keywords:  Benford’s Law, Cartel Detection, Collusion Screening, Competition Author-ities, Data Manipulation, Monopolization, Oligopolistic Markets, Price Fixing, VarianceScreen. JEL Classification:  C10; D40; L13; L41 ∗ The present paper was prepared during a visiting period at the Toulouse School of Economics (France) andat LUISS “Guido Carli” University of Rome (Italy). The author, who remains the only responsible for the viewsexpressed, would like to thank Prof. Roberto Pardolesi and Dr. Giacomo Luchetta for the kind comments andsuggestions offered. The dataset built for the purposes of the current work is available upon request. † Ph.D. Candidate and Researcher in Economic Analysis of Competition Law and Law & Economics LABResearch Fellow at LUISS “Guido Carli” University of Rome, Faculty of Economics, Viale Romania 32, 00197Rome (Italy) (E-Mail: ds@danilosama.com - Web-Site: www.danilosama.com).   In Francia abbiamo seguito le vostre elezioni.Il capo del governo ha tre reti televisive?   S`ı   .  Perch´e in Francia non si potrebbe, c’`e una legge.Voi non avete la legge antitrust?   S`ı. S`ı e no. Pi`u no che s`ı   . Nanni Moretti 1 Libor Scandal In 2013, the European Commission imposed an administrative fine of 1.7 billioneuro to some of the world’s largest banking companies involved in what has beendescribed by the mass media as “Libor Scandal” 1 . The record sanction, being thehighest ever levied by the officials of Brussels for a cartel infringement, was issued to8 international financial institutions for participating in illegal agreements relatingto interest rate derivatives. As it is common knowledge, interest rate derivativesare financial products, such as futures, options, swaps, which are both employed asinsurance tools for managing the risk of interest rate fluctuations and traded world-wide as investment assets by financial intermediaries. The value of these financialderivatives comes from the level of a benchmark interest rate, such as the EuroInterbank Offered Rate (Euribor), which is used for the euro area, or the LondonInterbank Offered Rate (Libor), which is used for several currencies including theJapanese Yen. In turn, the value of these benchmarks reflects the averaged interestrate at which, respectively, a selected panel of Eurozone and London banks offer tolend funds in a given currency to other banks on the daily interbank market.In a nutshell, the cartel aimed at manipulating the pricing process of the Euriborand the Libor, distorting the competition in the underlying trading of interest ratederivatives. Since at least    800 trillion in derivatives, loans, securities and otherfinancial products are tied to the Euribor and the Libor, such was the dimension of the scandal, which  inter alia   has highlighted the urgency of a regulatory reform of the banking sector, the largest one to have been rigged so far. 1 Commission Decision of 4 December 2013,  Euro Interest Rate Derivatives  , Case AT.39914;Commission Decision of 4 December 2013,  Yen Interest Rate Derivatives  , Case AT.39861; EuropeanCommission,  Antitrust: Commission fines banks     1.71 billion for participating in cartels in the interest rate derivatives industry  , Press Release, IP/13/1208, 4 December 2013, Brussels, Belgium. 1  2 Benford’s Law A crucial expedient for revealing the “Libor Scandal” has been the leniency pro-gram, joined by a member of the cartel at issue providing an active cooperationin the investigation of the Commission in exchange of full immunity. Beyond thesuccess of the cartel settlement procedure and the relevant dimension of the marketinvolved, from a competition policy standpoint, the Libor case offers another inter-esting food for thought, being an excellent example of how antitrust authorities canemploy screening instruments to identify collusive behaviors.A fascinating technique to detect rigged prices is offered by an odd phenomenoncalled Benford’s Law, otherwise known as First-Digit Law. Although a primordialstatement must be attributed to Newcomb (1881) 2 , in a 1938 paper, the father of the law, a physicist working at General Electric, recognized the existence of a spe-cific pattern that often occurs in vast datasets 3 . In particular, the law consists in afrequency distribution which describes the probability according to which a numberpresent in a random dataset starts with a certain digit.Theoretically, if a set of numbers were truly random, each leading digit wouldappear about 11% of the time. On the contrary, Benford’s Law predicts a logarithmicweakly monotonic distribution, according to which the digit 1 occurs as leadingdigit about 30% of the time, while larger digits occur in that position less and lessfrequently (cf. Formula 1). In other terms, the leading digits are not distributedevenly, as it would be natural to expect, but following a distribution where 1 is themost frequent and 9 is the less common. The law, which has also been generalized todigits beyond the first, tends to be more precise in datasets which exhibit multipleorders of magnitude and for types of values which grow exponentially. Formula 1 -  Benford’s Law Logarithmic Probability Distribution Function  P  ( d ) =  log 10 ( d  + 1) -  log 10 ( d ) =  log 10  ( d +1 d  ) =  log 10  (1 +  1 d ) 2 Newcomb, S. (1881),  Note on the Frequency of Use of the Different Digits in Natural Numbers  ,American Journal of Mathematics, Vol. 4, No. 1, The Johns Hopkins University Press, Baltimore,United States, pp. 39-40. 3 Benford, F. (1938),  The Law of Anomalous Numbers  , Proceedings of the American Philosoph-ical Society, Vol. 78, No. 4, American Philosophical Society, Philadelphia, United States, pp.551-572. 2  A brief and intuitive explanation of why the law naturally occurs is that usuallywe start counting from the digit 1 until the digit 9. It is obvious that if we think tothe digits from 1 to 9, we have the same probability that a random number startswith any of these digits. But if we consider a range of numbers, for example from1 to 20, we count more numbers starting with the digit 1. The same happens if we consider the range of numbers from 1 to 30, where we count many numbersstarting with the digit 1, but also many others starting with the digit 2. In anycase, what matters is that, in order to have many numbers starting with the digit9, it is necessary to examine a large dataset. As a result, analyzing for instancedistributions of numbers related to populations or surfaces, the probability to havea number starting with the digit 1 will be higher than that to have a number with9 as leading digit. Accordingly, Benford showed that, for several types of distribu-tions, the probability that a number starts with a certain digit tends to be alwaysthe same (cf. Figure 1). Figure 1 -  Probability Distribution of Leading Digits according to Benford’s Law  3 Literature Review The predictive power of Benford’s Law has been ascertained valid in several situ-ations normally observable in the real world. Death rates, election votes, financialtransitions, government spendings, income distributions, physical and mathematicalconstants, population numbers and stock prices are just few examples over which3  the law applies. It is not a case that auditors have successfully employed it to detectfrauds and manipulations in accounting data since the 1970s. At that time alsoVarian (1972) 4 suggested the possibility to use the law to uncover falsifications insocio-economic data collected for public purposes, under the assumption that whoaims at rigging datasets tends to provide numbers distributed according to a uniformpattern. Nigrini (1999) 5 as well showed that the law can be exploited for taxationcontrols, after having tested it with success on real cases of fiscal scams.Thus, given its regularity, the law can be adopted to test economic data in severalcases. It’s application is rather straightforward: even though a dataset has been ar-tificially ordered in such a way to preserve randomness, the distribution of the digitswill definitely violate the pattern predicted by the law. Within the present frame-work, in a seminal paper by Abrantes-Metz  et al.  (2011) 6 , the authors consideredworthwhile to test the theory over Libor data, using the second digit distributionvariant of the law. The surprising result was that the benchmark interest rate atissue departed significantly from the Benford’s Law pattern over an extended periodof time, signaling the possibility of a rate manipulation. As a result, through a quickapplication of the test, the Libor cartel could have been discovered much time beforethe opening of the settlement procedure.In Br¨ahler  et al.  (2011) 7 , a Benford’s Law test was applied to investigate thequality of macroeconomic data reported by the EU member states to Eurostat inorder to comply with the Stability and Growth Pact criteria. Since governmentstatistics are comparable in nature to financial accounting, governments, like firmstowards auditors, might be tempted to adjust the national account balances, giventhe strict obligations to which are subject to. The authors of the study found thatthe official statistics submitted by Greece registred the greatest deviation from theexpected Benford’s Law distribution in comparison to all the other EU countries. 4 Varian, H.R. (1972),  Benford’s Law (Letters to the Editor) , The American Statistician, Vol.26, Issue 3, Taylor & Francis Journals, London, United Kingdom, pp. 62-65. 5 Nigrini, M.J. (1999),  I’ve Got Your Number: How a Mathematical Phenomenon Can HelpCPAs Uncover Fraud and Other Irregularities  , Journal of Accountancy, Vol. 187, Issue 5, AmericanInstitute of Certified Public Accountants, New York, United States, pp. 15-27. 6 Abrantes-Metz, R.M., Judge, G., Villas-Boas, S. (2011),  Tracking the Libor Rate  , AppliedEconomics Letters, Vol. 10, Issue 10, Taylor & Francis Journals, London, United Kingdom, pp.893-899. 7 Br¨ahler, G., Engel, S., G¨ottsche, M., Rauch, B. (2011),  Fact and Fiction in EU-Governmental Economic Data  , German Economic Review, Vol. 12, Issue 3, John Wiley & Sons, New York,United States, pp. 243-255. 4
Related Documents
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!