The Newcomb-Benford law for digit sequences has recently attracted interest in anti-fraud analysis. However, most of its applications rely either on diagnostic checks of the data, or on informal decision rules. We suggest a new way of testing the Newcomb-Benford law that turns out to be particularly attractive for the detection of frauds in customs data collected from international trade. Our approach has two major advantages. The first one is that we control the rate of false rejections at each stage of the procedure, as required in anti-fraud applications. The second improvement is that our testing procedure leads to exact significance levels and does not rely on large-sample approximations. Another contribution of our work is the derivation of a simple expression for the digit distribution when the Newcomb-Benford law is violated, and a bound for a chi-squared type of distance between the actual digit distribution and the Newcomb-Benford one.
Goodness-of-fit testing for the Newcomb-Benford law with application to the detection of customs fraud / Barabesi, Lucio; Cerasa, Andrea; Cerioli, Andrea; Perrotta, Domenico. - In: JOURNAL OF BUSINESS & ECONOMIC STATISTICS. - ISSN 0735-0015. - 36:2(2018), pp. 346-358. [10.1080/07350015.2016.1172014]
Goodness-of-fit testing for the Newcomb-Benford law with application to the detection of customs fraud
CERIOLI, Andrea
;
2018-01-01
Abstract
The Newcomb-Benford law for digit sequences has recently attracted interest in anti-fraud analysis. However, most of its applications rely either on diagnostic checks of the data, or on informal decision rules. We suggest a new way of testing the Newcomb-Benford law that turns out to be particularly attractive for the detection of frauds in customs data collected from international trade. Our approach has two major advantages. The first one is that we control the rate of false rejections at each stage of the procedure, as required in anti-fraud applications. The second improvement is that our testing procedure leads to exact significance levels and does not rely on large-sample approximations. Another contribution of our work is the derivation of a simple expression for the digit distribution when the Newcomb-Benford law is violated, and a bound for a chi-squared type of distance between the actual digit distribution and the Newcomb-Benford one.File | Dimensione | Formato | |
---|---|---|---|
FirstDigit_BCCP_JBES_Supplementary_Final.pdf
accesso aperto
Descrizione: Supplementary material
Tipologia:
Documento in Post-print
Licenza:
Creative commons
Dimensione
55.27 kB
Formato
Adobe PDF
|
55.27 kB | Adobe PDF | Visualizza/Apri |
BCCP_Benford_JBES17_online.pdf
solo utenti autorizzati
Descrizione: Published paper
Tipologia:
Versione (PDF) editoriale
Licenza:
NON PUBBLICO - Accesso privato/ristretto
Dimensione
595.22 kB
Formato
Adobe PDF
|
595.22 kB | Adobe PDF | Visualizza/Apri Richiedi una copia |
BCCP_JBES_Accepted.pdf
accesso aperto
Tipologia:
Documento in Post-print
Licenza:
Creative commons
Dimensione
397.16 kB
Formato
Adobe PDF
|
397.16 kB | Adobe PDF | Visualizza/Apri |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.