Manipulation testing based on Benford’s Law for discrete scores | Marco Ventura (Sapienza University of Rome)
FBK-IRVAPP is pleased to invite you to the following seminar: Manipulation testing based on Benford’s Law for discrete scores.
With the participation of Marco Ventura (Sapienza University of Rome)
Abstract
This paper addresses the problem of running variable manipulation in the context of regression discontinuity designs. Leveraging the observation that manipulation can manifest as an asymmetry in the running variable’s density around the cutoff, we identify this asymmetry using Benford’s Law, a data regularity property widely used particularly in fraud detection. Our proposed test complements McCrary-type tests, offering the advantage of eliminating researcher-specified options and parameters that can affect the results. To do so, we first propose a new approach to determine a bandwidth consistent with Benford’s Law, and then we propose two distinct but complementary tests that make use of predetermined acceptance/rejection threshold values as proposed by Nigrini [2012], taking advantage of Benford’s Law’s main feature: its universality as a natural property of most empirical datasets. Finally, our approach overcomes a key limitation of the law itself by using probabilities and threshold values, rather than digits. Empirical examples and practical implementations are provided.
The seminar is held in English.
Speakers
-
Marco Ventura - Guest SpeakerSapienza University of RomeMarco Ventura is an Associate Professor of Econometrics at the Department of Economics and Law, Sapienza University of Rome. He holds a PhD in Economics from Sapienza University of Rome and an MSc in Finance from Birkbeck College, University of London. Prior to his current academic position, he served as a Senior Researcher at the Italian National Institute of Statistics (ISTAT) in the Methodological and Data Quality Division and at the Institute for Studies and Economic Analyses (ISAE). His primary research interests focus on the econometrics of causal effects, microeconometrics, counterfactual evaluation of public policy interventions, causal machine learning, and the economics of innovation and intellectual property. He has led and participated in numerous national and international research projects on firm performance, public subsidies, and policy evaluation, and teaches advanced econometrics and causal inference courses at both undergraduate and doctoral levels.
Registration
Registration to this event is mandatory.
Register