TRANSITION TO ENGINEERING ANALYSIS OF PUBLIC FINANCES: ALGORITHMIC AUDIT OF INTERBUDGETARY TRANSFERS OF TERRITORIAL COMMUNITIES

Abstract

The article substantiates the feasibility of applying Data Science methods to the analysis of interbudgetary transfers of territorial communities using open budget data. Using the open arrays of OpenBudget, an algorithmic audit of the dynamics of subventions and subsidies of 1,469 territorial communities of Ukraine (2022-2025) was conducted. The actual curtailment of financial decentralization due to the withdrawal of the “military” personal income tax was empirically recorded: the collapse of the reverse subsidy from UAH 16.73 billion in 2023 to UAH 14.37 million in 2024 transferred the system to centralized discretionary allocation through massive non-targeted subsidies. A two-level machine learning methodology was used to detect discretionary distribution of funds. Comparative analysis indicates that hierarchical agglomerative clustering helps to describe the underlying macrostructure, while time series analysis using the DTW (Dynamic Time Warping) metric allows for the detection of transactional anomalies. The DTW algorithm identified statistically significant deviations in the processes of distributing funds for reconstruction, automatically isolating microgroups of wealthy donor communities that absorbed disproportionately large capital injections. Given the lack of sufficiently complete and accurate data in the open access, the algorithmic model cannot guarantee absolute reliability. However, the identified anomalies are an important algorithmic indicator indicating the potential excessive concentration of budgetary resources, thereby confirming the conclusions of regulatory authorities. The proposed methodological framework can serve as a practical decision-making tool for state institutions, fiscal regulatory authorities and local governments, allowing for continuous monitoring of inter-budgetary transfers, early detection of anomalous financing models and objective assessment of the effectiveness of budget allocation. Its application helps to strengthen budget transparency, accountability and data-based management of public finances. This article also emphasizes that the artificial secrecy surrounding national databases under the pretext of security makes it impossible to conduct a comprehensive audit. In this context, ensuring procedural reproducibility of research (research reproducibility) becomes a key tool for solving the problem of lack of transparency in public finance management.