Paper Title: A federated learning approach for predicting AI capability synergy effects in manufacturing mergers and acquisitions: engineering collaboration and economic value creation
Authors: Tengfei Fan, Minghao Huang
Corresponding Author: Minghao Huang (mhuang@assist.ac.kr)/ South Korea
Abstract
Manufacturing mergers and acquisitions increasingly target AI capabilities, yet predicting synergy effects remains constrained by cross-enterprise data privacy barriers that render centralized approaches impractical. This study proposes a novel horizontal federated learning framework for AI capability synergy effect prediction in manufacturing M&A, integrating federated histogram-aggregated gradient boosting trees with FedAvg-optimized deep neural networks through a two-stage decoupled training strategy, alongside a systematically constructed engineering collaboration indicator system encompassing R&D compatibility, production system interoperability, and AI talent overlap. Empirical validation across 286 authentic manufacturing M&A cases from 23 enterprises demonstrates that the proposed framework achieves superior predictive performance to centralized machine learning and traditional econometric baselines while preserving complete data confidentiality, with engineering collaboration indicators contributing more substantially to synergy prediction than financial variables, and AI-intensive acquirers generating pronounced post-merger economic value premiums following a time-lagged pattern. These findings establish a methodological bridge between privacy-preserving machine learning and strategic management research, providing manufacturing executives with a comprehensive decision support toolkit for target screening, due diligence, and post-merger integration planning.
