Page 76 - ISC PROCEEDINGS 21.4

Recent studies have increasingly applied Grey Relational Analysis (GRA) to evaluate
factors influencing domains associated with the digital economy. Wang Q. et al. (2024)
integrated GRA with Interpretive Structural Modeling (ISM) to delineate the influence
structure among elements of the digital economy in China. Guo F.H. (2022) employed
GRA to rank the relative importance of determinants impacting digital finance, whereas
Sun X. et al. (2024) combined GRA with a Vector Autoregression (VAR) model to analyze
the degree of integration between the digital economy and traditional economic sectors.
A synthesis of the aforementioned literature reveals that each methodological paradigm
possesses distinct advantages. Regression techniques and panel data approaches are
highly suitable for investigating heterogeneities across multiple spatial units; structural
models are optimal for deciphering complex causal relationships; conversely, GRA-based
methodologies prove exceptionally efficacious when the primary objective is to evaluate
the relative magnitude of influence among variables under conditions of limited or
incomplete information.
Originating from the characteristics of the dataset and the overarching research
objectives, this study selects Grey Relational Analysis (GRA) as its foundational analytical
methodology. GRA is a fundamental component of grey system theory, originally
proposed by Deng Julong (1989), which aims to mathematically measure the degree of
proximity between data sequences through the conceptual framework of grey
relationships. This specific methodological approach permits the precise determination of
the relative magnitude of influence exerted by individual independent variables upon a
dependent variable, notably without necessitating strict a priori assumptions regarding
probability distributions or linear relationships. The principal econometric advantage of
GRA lies in its robust capacity to process small-sample data, incomplete datasets, and
variables characterized by heterogeneous measurement scales; thus, it is exceptionally
well-suited for analyzing aggregated national-level macroeconomic data over a
constrained temporal horizon. The quantitative component of this research utilizes GRA
to rank the independent variables based on their relational grade with the dependent
variable-specifically, GDP-thereby facilitating the empirical identification of critical factors
that warrant targeted policy prioritization. This analytical technique is particularly
pertinent within the context of Vietnamese empirical data, which frequently exhibits
inherent limitations in time-series length and underlying uncertainty (Liu & Lin, 2025).
Consequently, the empirical analysis is executed utilizing GRA in strict adherence to
established standard procedural protocols (Deng, 1982; Liu & Lin, 2025):
Step 1. Determine the reference sequence and comparability sequences:
Reference sequence: = { (1), (2), …, ( )}, representing GDP growth.
0
0
0
0
Comparability sequences: = { (1), (2), …, ( )} , where = 1,2, …, ,

represent the independent variables (Technical Infrastructure, Human Resource
Infrastructure, IT Application, e-Governance, Productivity, Innovation, E-commerce).
Step 2. Data normalization
To eliminate unit discrepancies, the data are normalized according to the following
formulas:
If the variable follows a "larger-the-better" beneficial orientation:
( ) − ( )

'
( ) =

( ) − ( )

If the variable follows a "smaller-the-better" beneficial orientation:
75