數位轉型對於我國產業的影響

Abstract

在數位技術快速進展的時代浪潮下，數位轉型已被視為推動產業升級與經濟成長的重要驅動力。然而，數位轉型是否能有效轉化為實質產出，仍需透過實證驗證。本研究以我國 225 個產業小類為分析單位，結合行政院主計總處公布的4期工業及服務業普查資料，建立涵蓋資訊設備、人力結構、營運管理等多面向之數位轉型指標，並透過平均整合與主成分分析（PCA）建立數位轉型程度的衡量指標，從而探討數位轉型對產業全年生產毛額之影響。 本研究採用加權迴歸固定效果模型進行整體估計，同時導入非線性項以檢視數位轉型是否具有邊際報酬遞減的特性。為進一步探討轉型效果之異質性，本研究設計納入產業廠商規模之交乘項估計，並針對不同產業類型（製造業、服務業與其他產業）進行分組估計。最後，為了更進一步控制差異，採用傾向分數配對法（PSM）與廣義傾向分數法（GPS），分析不同數位轉型強度下的處理效果。 實證結果發現數位轉型整體對產業的全年生產毛額具有顯著正向影響，本研究另以非線性模型及四分位數分組分析發現轉型程度與全年生產毛額之間並非單純的線性關係，而是呈現倒U型的非線性結構，因此中高強度的數位轉型可創造最大邊際效益，但在高強度轉型階段則因整合成本升高或技術瓶頸而出現效益遞減。此外，服務業與中小型產業的數位轉型效果較為明顯，其在導入數位工具時具備較高的即時效益，然而製造業可能會因結構與技術門檻較高，短期內成效不如預期。 數位轉型對於產業全年生產毛額是具有影響力的，但其效果是會受到轉型強度及產業條件的影響。數位轉型並非「越多越好」，而是需要視產業性質採取不同策略。最後，本研究建議在推動數位政策時，應強化階段性規劃與產業差異化，促使產業能在轉型過程中達到效果，為未來數位經濟之永續發展奠定基礎。

Digital transformation has become a key driver of industrial upgrading and economic growth. However, whether it translates into actual productivity gains remains an open empirical question. This study analyzes 225 sub-industries in Taiwan using data from four waves of industrial and service sector censuses. A composite digital transformation index---capturing IT infrastructure, workforce composition, and operational practices---is constructed through averaging and principal component analysis (PCA). To evaluate its impact, we employ weighted regression models that account for nonlinear effects and interaction terms related to industry size. Subgroup analyses are conducted across manufacturing, services, and other sectors. For robustness, we further apply propensity score matching (PSM) and generalized propensity score (GPS) methods to estimate treatment effects under varying levels of digital adoption. The results indicate that digital transformation significantly boosts industrial output, but the relationship is nonlinear—following an inverted U-shape. Moderate to high levels of digitalization produce the greatest gains, while excessive transformation may lead to diminishing returns due to integration costs or technical constraints. The benefits are more pronounced in the service sector and among small and medium-sized enterprises (SMEs), whereas structural barriers in manufacturing may hinder short-term gains. These findings highlight the need for sector-specific digital strategies. Rather than assuming “more is better”, policymakers should pursue phased and differentiated approaches to foster effective and sustainable digital development.