Page 55 - ISC PROCEEDINGS 21.4

From a theoretical standpoint, modern Public Financial Management (PFM) is built
upon three core pillars: fiscal discipline, efficient resource allocation, and improved
spending performance. ST transparency and public investment oversight are considered
essential preconditions for reducing information asymmetry, curbing rent-seeking
behavior, and strengthening public trust. However, international research shows that
when the scale of public investment is large and project value chains extend across
multiple stages - from planning, appraisal, capital allocation, and procurement to
disbursement and final settlement - the risks of fraud, cost overruns, and delays increase,
particularly when data systems are fragmented and analytical tools remain limited.
In practice, PIE from the ST in Vietnam in recent years has been maintained at
approximately 26-32% of total state budget expenditure, equivalent to over 6-7% of GDP
annually. Notably, during the implementation of socio-economic recovery and
development programs, the scale of public investment has expanded significantly,
focusing on transport infrastructure, digital transformation, and energy. Nevertheless,
uneven disbursement across ministries, sectors, and localities; adjustments to total
investment levels; and delays in the settlement of completed projects indicate substantial
room for improvement in oversight mechanisms. Traditional inspection and audit
methods, which rely primarily on ex-post review and sampling techniques, make it
difficult to cover the entire project lifecycle or detect early signs of anomalies in millions
of expenditure transactions.
In this context, BDA and AI offer a new approach to monitoring PIE. BDA enables the
integration of data from the State Treasury system, public investment management
platforms, e-procurement systems, and local databases, while processing large volumes
of multidimensional information in real-time. Machine learning (ML) algorithms can
identify anomalous spending patterns, assign project risk scores, and forecast
disbursement trends, thereby supporting financial authorities in shifting from a “post-
violation response” model to a “data-driven preventive” approach.
Building upon the existing legal framework, the theoretical foundations of modern
Public financial governance, and the practical need to enhance public investment
efficiency, this study analyzes the potential application of BDA and AI in monitoring PIE in
Vietnam. It seeks to answer the central question: How can BDA and AI be effectively and
transparently integrated into public investment oversight in a manner consistent with
Vietnam’s institutional context? On that basis, the study proposes a framework aimed at
advancing smart, transparent, and sustainable public financial governance in the digital
era.
2. Theoretical framework, literature review and research methods
2.1. Theoretical framework
The theoretical framework of this study is built upon three pillars: (i) Big Data
Governance; (ii) Predictive Analytics; and (iii) AI for Public Accountability, integrated
within the Smart PFM approach.
(i) Big Data Governance
Big Data Governance refers to a system of principles, mechanisms, and processes
designed to ensure that data is collected, processed, stored, and shared in a lawful,
secure, accurate, and with utility value. In the field of PIE, data is not merely technical
input but a strategic asset for fiscal management and ST oversight.
From a theoretical perspective, BDA is characterized by three core dimensions:
volume, velocity, and variety. In PIE, data is generated from multiple sources, including

54