| Abstract [eng] |
The aim of this thesis is to develop and empirically evaluate an internal audit method for trade receivables based on the Expected Payment Delay from Actual (EPDA) indicator and to compare two technological implementations of this method – a custom .NET application and a Power Automate Desktop (RPA) prototype. To achieve this aim, the thesis analyses literature on receivables auditing, traditional aging and payment behaviour models, and the use of RPA in audit; defines the research problem, object, aim and tasks; and designs a conceptual EPDA model using a Kaggle invoices dataset. The empirical research involves cleaning and preparing the dataset (48,580 invoices), calculating EPDA and traditional aging buckets, implementing two functionally equivalent EPDA prototypes in .NET and Power Automate Desktop, and experimentally comparing them on several dataset sizes (1k, 10k, 25k, 48k records). Performance metrics (execution time, CPU, memory and I/O usage) are measured with Windows Performance Monitor and normalised per one EPDA run per 1,000 invoices. The results show that EPDA provides additional behavioural insight compared to traditional aging, as it measures each invoice’s deviation from the customer’s typical payment delay and thus reveals behavioural changes and “hidden” risk in invoices that appear low-risk in standard aging buckets. Heavily overdue invoices remain high-risk under both methods, but EPDA additionally highlights a share of invoices in the 0–30 days past due bucket as behavioural outliers, which is important for audit risk assessment and sampling. The technological comparison confirms that both prototypes generate consistent EPDA results, while the .NET solution is several orders of magnitude faster and more resource-efficient, making it suitable as the core computation engine. Power Automate Desktop is more appropriate as a low-code orchestration and integration layer for automating data extraction and reporting. The thesis concludes that EPDA should be used as a complementary behavioural metric alongside traditional aging and DSO in receivables audit and that combining high-performance calculation platforms with RPA-based workflows is a promising direction for FinTech-oriented internal audit automation. The work is comprised of 69 pages, contains 4 tables and 7 figures. |
