Complex networks in audit A data-driven modelling approach

In this thesis, we introduce data-driven audit methods using a network-based approach. Utilizing data from over 300 companies, it transforms transaction data into a network format, providing auditors with a clear overview of a company's financial structure.
Chapter 2 details the financial statements network, designed for straightforward interpretation by auditors. This network effectively represents the company's financial structure, aiding in developing universal data-driven audit methods.
Chapter 3's analysis reveals that the financial account nodes' degree distribution typically follows a heavy-tail distribution. Moreover, we found only minor variations in network statistics across industries. These findings help establish baseline expectations for network statistics, facilitating risk assessment.
Chapter 4 addresses the complexity of these networks, proposing a method to simplify them into a more understandable high-level structure for auditors.
Chapter 5 explores a similarity measure to compare financial structures, helping auditors identify deviations in a client's financial network compared to peers or historical data. Deviations could signal increased audit risks.
In summary, we pioneer data-driven audit methods using financial statement networks, providing new insights and tools for auditors and paving the way for more efficient and effective audit processes.