企業庫存週轉與股價報酬之關聯性實證探討

Abstract

本研究針對 2010 年至 2024 年臺灣上市 14 家企業的庫存週轉天數與股價報酬之關聯性進行實證分析，研究動機來自實務上企業庫存週轉常被投資人視作未來景氣與企業基本面前景的重要參考指標，但各產業間庫存涵義可能不同，過去的文獻對於「庫存週轉天數」與「股價報酬」的關聯分析仍缺乏系統性的比較及驗證，本研究以塑化、鋼鐵、水泥、汽車、半導體與 PC/NB（Personal Computer/Notebook）六大產業的代表性上市龍頭公司為樣本，包括台塑、南亞、中鋼、燁輝、台泥、亞泥、裕隆、和泰車、台積電、聯發科、日月光、宏碁、華碩、廣達，蒐集每季財報庫存週轉天數與同期還原股價季報酬，計算皮爾森相關係數並且檢視不同滯後期（一季、兩季、四季）下之相關係數差異，觀察庫存週轉天數以及庫存週轉天數變動率是否具有顯著的關聯性。

研究結果顯示，傳統產業中庫存週轉天數與股價報酬大多呈現負相關，表示當庫存週轉率下滑時，股價報酬往往走向下降趨勢，其中以塑化產業關聯最大，台塑的當期相關係數為－0.56，滯後的相關性則逐漸下滑；鋼鐵與水泥產業代表公司則呈現弱負相關（－0.29、－0.23），發生在滯後一季及當期；汽車產業的庫存週轉天數與股價關聯性則相對特殊，整體的趨勢多呈正相關（當期＋0.02、＋0.18），顯示汽車庫存週轉天數居高對股價並非拖累。

相較於傳產，電子產業的情況有所不同，半導體產業的庫存週轉天數與股價報酬呈現弱負相關（當期約－0.02至－0.13）；PC/NB產業則有個股呈現明顯正相關且具有滯後反應，以宏碁為例，庫存週轉天數滯後 4 季與股價報酬呈現中等正相關（約＋0.42），廣達及華碩也都是從當期到滯後 4 季正相關性加大，意即 PC/NB 產業庫存週轉天數在景氣循環中之高低轉折，可能領先於股價將近一年的時間，具有前導的角色定位。

另外，除了以庫存週轉天數與股價報酬進行計算，也以庫存週轉天數變化率與股價報酬進行相關係數分析，前者是以庫存水位對未來報酬率做評估，後者探討短期變動與轉折對股價報酬的關聯性分析，數據結果部分，也是以 PC/NB 族群差異最大，因此進一步討論台廠提前備貨與銷售狀況。

本論文根據上述數據結果，再深入比較傳產與電子間的差異，解釋不同產業庫存週轉天數與股價連動性的成因，同時，以 PC/NB 龍頭公司宏碁為代表進行跨時期的回溯分析，觀察庫存股價關聯在網路泡沫之前與之後的變化並解釋之。研究結論提供投資人及企業財務管理者等在解讀庫存水位、庫存循環週期時作為參考：庫存訊號對還原股價的關聯性因產業而異，投資決策應考量產業特性，不宜套用單一的標準。此外，本研究也填補了國內跨產業比較庫存效率與股價報酬的學術缺口，對未來臺灣上市公司相關研究與實務應用提出建議。

This study conducts an empirical analysis on the relationship between inventory turnover days (Days of Inventory, DOI) and stock returns of fourteen listed companies in Taiwan from 2010 to 2024. Motivated by the practical observation that inventory turnover is often viewed by investors as a key signal of business conditions and macroeconomic outlook, this research addresses the lack of systematic comparison across industries in previous literature regarding the correlation between DOI and stock performance. The sample includes representative firms from six major sectors: petrochemical, steel, cement, automobile, semiconductor, and PC/notebook (PC/NB), namely Formosa Plastics, Nan Ya Plastics, China Steel, Yieh Phui, Taiwan Cement, Asia Cement, Yulon Motor, Hotai Motor, TSMC, MediaTek, ASE Technology, Acer, ASUS, and Quanta Computer.

Using quarterly financial reports and adjusted stock returns, the study calculates Pearson correlation coefficients and evaluates variations in correlation under different lag periods (contemporaneous, one-quarter lag, and four-quarter lag). It further examines whether DOI and its rate of change exhibit statistically significant associations with stock returns.

The results show that DOI and stock returns generally exhibit negative correlations in traditional industries. For example, the contemporaneous correlation for Formosa Plastics is −0.56, while steel and cement companies present weaker negative correlations (−0.29 and −0.23, respectively), mostly appearing in the contemporaneous or one-quarter lag periods. In contrast, the automobile industry displays a different pattern, with DOI showing positive correlations with stock returns (+0.02 to +0.18), suggesting that higher inventory levels may not necessarily weigh on stock prices.

In the electronics sector, the correlation between DOI and returns is weaker. Semiconductor firms show slight negative correlations (−0.02 to −0.13), whereas companies in the PC/NB sector exhibit more distinct and positive correlations with a notable lag. For instance, Acer’s DOI shows a moderate positive correlation (approximately +0.42) with returns lagged by four quarters. Similar patterns are observed for Quanta and ASUS, indicating that shifts in DOI within this sector may precede stock price movements by up to a year, suggesting a potential leading indicator role.

The study also compares the correlation between stock returns and both the level and the change rate of DOI. While the former reflects long-term inventory positioning, the latter captures short-term adjustments or turning points. The PC/NB group shows the most significant variation across both measures, prompting further discussion on early stocking strategies and sales conditions among Taiwanese electronics firms.

To deepen the analysis, this study compares the dynamics between traditional and electronics sectors and explores underlying factors behind such variations. A case study of Acer further examines the correlation pattern before and after the dot-com bubble, offering insights into historical shifts. The findings offer practical implications for investors and corporate financial managers when interpreting inventory cycles and signals. Overall, this study contributes to the domestic literature by filling the gap in cross-industry analysis of inventory efficiency and market performance, and offers recommendations for future research and practical application in the Taiwanese capital market.