Learning sampling in financial statement audits using vector quantised variational autoencoder neural networks

Abstract

The audit of financial statements is designed to collect reasonable assurance that an issued statement is free from material misstatement ('true and fair presentation'). International audit standards require the assessment of a statements' underlying accounting relevant transactions referred to as 'journal entries' to detect potential misstatements. To efficiently audit the increasing quantities of such journal entries, auditors regularly conduct an 'audit sampling' i.e. a sample-based assessment of a subset of these journal entries. However, the task of audit sampling is often conducted early in the overall audit process, where the auditor might not be aware of all generative factors and their dynamics that resulted in the journal entries in-scope of the audit. To overcome this challenge, we propose the use of a Vector Quantised-Variational Autoencoder (VQ-VAE) neural networks to learn a representation of journal entries able to provide a comprehensive 'audit sampling' to the auditor. We demonstrate, based on two real-world city payment datasets, that such artificial neural networks are capable of learning a quantised representation of accounting data. We show that the learned quantisation uncovers (i) the latent factors of variation and (ii) can be utilised as a highly representative audit sample in financial statement audits.

Multi-view contrastive self-supervised learning of accounting data representations for downstream audit tasks

“…Audit Sampling: We evaluate the fine-tuned VQ-VAE models (VAESSL) and compare its audit sampling capability to different vector-quantized VAE models (VAE) proposed in [49] which we use as a baseline. The baseline models are specifically designed for the purpose audit sampling.…”

“…Such techniques encompass: (i) autoencoder neural networks [43,48,52], (ii) variational autoencoders [66], and (iii) adversarial autoencoders [50]. Lately, vector quantized variational autoencoders have been applied to learn representative audit sampling [49]. Concluding from the reviewed literature, most references either draw from human-engineered representations or learned representations highly optimized towards a particular audit task.…”

“…Driven by such efforts and the rapid technological advances of artificial intelligence, deep learning [39] techniques are applied in external and internal financial audits [42,54]. Nowadays, the majority of deep learning-inspired audit techniques rely on several models, each highly specialized towards a particular audit task, e.g., the detection of anomalous journal entries [48] or to learn a representative audit sample [49]. Caused by the characteristics of modern audit engagements, such as (i) challenging time-budget constraints, (ii) extensive documentation obligations, and (iii) strict model interpretability requirements, auditors often desire to train and apply only a single (or limited number of) model(s) to audit vast quantities of accounting records.…”

RESHAPE: Explaining Accounting Anomalies in Financial Statement Audits by enhancing SHapley Additive exPlanations

“…Furthermore, Zupan et al [54] used Long Short-Term Memory AAENs to detect temporal anomalies in journal entry data. In addition, Nonnenmacher et al [32] and Schreyer et al [40] demonstrated that AENNs could be used to improve audit sampling during an audit process. Furthermore, it was shown that AENNs can be trained in a self-supervised learning setup to detect accounting anomalies and complete additional down-stream audit tasks [38].…”

Federated and Privacy-Preserving Learning of Accounting Data in Financial Statement Audits

“…Lately, such models have been trained to accomplish a various downstream audit tasks, e.g. accounting anomaly detection [43,53,68], audit sampling [48,50] or notes analysis [45,54]. Figure 1 illustrates the exemplary application of a deep bottleneck Autoencoder Network (AEN) to detect anomalies in accounting data [49].…”