Examining the Effectiveness of Forensic Accounting Techniques in Detecting Banking Fraud: A Systematic Review of Existing Evidence

Abstract

Purpose: Banks remain among the organisations most exposed to occupational and external fraud, and forensic accounting has been widely promoted as a remedy. This paper systematically reviews published evidence, up to and including 2018, on how effective forensic accounting techniques are in detecting fraud in banks and closely related financial institutions.

Design/methodology/approach: Following the evidence-based review logic of Tranfield, Denyer and Smart (2003) and the reporting principles of PRISMA (Moher et al., 2009), peer-reviewed articles, professional reports and authoritative texts were identified through academic databases and reference chaining, screened against explicit inclusion criteria, appraised for methodological quality and synthesised narratively. Twenty-nine core sources formed the final evidence base, supported by related conceptual work on audit, controls and compliance published up to 2018.

Findings: The evidence consistently associates forensic accounting with improved fraud detection and deterrence in banks, but the strength of that evidence varies sharply by technique. Computational techniques (data mining, machine learning and digital analysis such as Benford's Law) have the most rigorous, outcome-based support, largely from card and transaction fraud settings. Investigative and analytical techniques (financial statement and ratio analysis, document examination, interviewing, audit-trail reconstruction) are supported mainly by perception-based surveys, concentrated in Nigeria, that report statistically significant positive associations but rarely measure actual detection outcomes. Professional survey data show that tips remain the single most common means of initial detection, which positions forensic techniques as complements to, rather than replacements for, whistleblowing and internal controls.

Research limitations/implications: The evidence base is dominated by perception surveys from a single country, so conclusions about individual techniques remain provisional. Five testable propositions are offered to guide outcome-based research.

Practical implications: Banks should combine continuous analytics with trained forensic investigators, whistleblowing channels and strong internal controls, rather than relying on any single technique.

Originality/value: The review organises a fragmented literature into a technique-by-evidence matrix, proposes an integrated conceptual model of forensic fraud detection in banks, identifies the methodological weaknesses that limit current conclusions, and sets out a research and practice agenda for banks and regulators.