Page 93 - 36-1

NTU Management Review Vol. 36 No. 1 Apr. 2026

The overarching goal is to demonstrate how management accounting can evolve
from a descriptive and historical system into a forward-looking, value-oriented, and AI-
enabled decision support mechanism. This mechanism offers both theoretical advancement
and practical relevance to firms engaged in channel distribution and sustainability

transformation.

2. Design / Methodology / Approach

The research employs a mixed-method approach combining conceptual model
development, case-based analysis, and empirical testing. The theoretical foundation
draws on Cooper and Kaplan (1988) and Kaplan and Anderson (2004) to extend the AVM
framework toward a dynamic, predictive value system.

The study empirically utilizes product-level and activity-level data collected from P
Company, a channel distributor in the retail sector. The dataset spans 2016 to 2021, with
2020 serving as the primary year for predictive modeling. To enhance analytical rigor, the
study incorporates AI-based algorithms, including Extreme Gradient Boosting (XGBoost)

and Neural Networks, to estimate PLCV and to forecast product profitability across
different life cycle stages.
The selection of AI-based predictive models is motivated by the increasing
complexity and nonlinearity inherent in product life cycle dynamics. Traditional linear

models often struggle to capture interaction effects among activities, customer structures,
and temporal factors. In contrast, machine learning techniques are capable of learning
complex patterns from high-dimensional transactional data. To mitigate concerns
regarding model opacity, the study explicitly incorporates explainability mechanisms that

allow managerial users to trace predictions back to underlying value drivers and activity
configurations, thereby balancing predictive performance with interpretability.
The proposed AI-integrated PLCV-AVM model consists of three main modules.
(1) Value Identification Module maps value drivers and activity attributes to product value

creation.
(2) Predictive Analytics Module employs machine learning to forecast PLCV and identify
high-value product segments.
(3) Strategic Alignment Module links AVM results to managerial decision variables, such

85