應用貝氏管制於永續盤存存貨系統之研究

Abstract

中美貿易戰、COVID-19疫情、俄烏戰爭、極端氣候等眾多外在環境變因，都嚴重衝擊供應鏈管理中的三大成本控管領域：製造、存貨、運輸；遂使得「供應鏈管理」在近五年來蔚為風潮，成為企業經營管理的核心要務。其中，在存貨管理領域，電子產業由於疫情前後的需求預測失準，存貨管理系統無法反應真實需求變化，分別造成了供不應求和存貨過剩的現象，致使許多企業蒙受重大損失；因此，本研究嘗試建立一套能夠有效因應外在環境變因的存貨管理系統，改善這一重大存貨管理問題。

過往研究中，許多學者將管制圖 (control chart) 的概念引入存貨管理中，試圖藉由管制圖監控需求變化，達成更好的存貨控管效果。吳致賢 (2017) 率先將貝氏統計的概念結合管制圖引入定期盤存制的存貨系統中，並模擬多種需求情境發現貝氏管制圖能更有效因應需求遞增情境，大幅降低此一需求情境中的存貨成本。

本研究在吳致賢的研究基礎上，將貝氏統計結合管制圖，改以永續盤存制為研究對象，同時在模擬情境中加入疫情期間電子業需求數據以及各情境中成本結構的敏感度分析。研究中以傳統永續盤存制以及本研究另外提出的動態調整永續盤存制作為比較對象，觀察貝氏管制永續盤存制是否更能有效因應外在環境變因。

研究結果發現，在需求穩定遞增和模擬疫情需求的情境中，貝氏管制永續盤存制在初始設定的成本結構下都能得到最低的存貨總成本。而在成本結構敏感度分析中，在需求穩定遞增時，當缺貨成本極高時動態調整法能獲得較低的存貨總成本；而在模擬疫情需求的情境中，除了缺貨成本低的情境外，貝氏管制永續盤存制都能獲得較低的存貨總成本。其中，在模擬疫情需求中，對比另外兩種永續盤存制，貝氏管制永續盤存制能將缺貨量壓低至50%以下，大幅降低缺貨成本，達到最好的績效表現。由此可見，對於外在環境急遽變動的情境，本研究所提貝氏管制永續盤存制將更能因應外在變化，降低總體存貨成本，提升企業經營績效。

During the past five years, supply chain management’s three major topics: manufacturing, inventory, and transportation have been severely impacted by external factors, such as the trade war, COVID-19 pandemic, Russia-Ukraine war, extreme weather…etc. As a result, supply chain management has become an important topic and a core concern in nowadays business management. Focusing on inventory management, the electronics industry has experienced shortages and inventory surplus during the pandemic, as inventory systems couldn’t respond to demand changes, causing huge losses for companies. Therefore, this study attempts to find an inventory system that can effectively cope with external factors and demand changes.

In previous studies, many scholars have introduced statistic control charts into inventory management, attempting to achieve better inventory control by monitoring demand changes. Wu Chih-Hsien (2017) first combined Bayesian statistics with control charts in a periodic inventory system; by simulating various demand scenarios, Wu discovered that Bayesian control charts were more effective when responding to increasing demand scenarios.

Following Wu's research, this study combines Bayesian statistics with control charts and focuses on a continuous review system; also simulates a demand data from the electronics industry during the pandemic and conducts sensitivity analysis for cost structures. A traditional and a dynamic adjusted continuous review system proposed in this study will be used to compare with the proposed Bayesian control continuous review system, in order to observe which can better respond to demand changes and cope with external factors.

The research findings reveal that in scenarios of steadily increasing demand and the simulated pandemic demand, Bayesian control system consistently achieves the lowest total inventory cost under the initial cost structure. In the sensitivity analysis, when demand steadily increases, the dynamic adjusted system obtains lower total inventory costs when the stockout cost is extremely high. And in the scenario of the simulated pandemic demand, except for situations with low stockout costs, Bayesian control system achieves lower total inventory costs, reduces stockout quantities up to 50%, significantly lower stockout costs and has the best performance.

In conclusion, during a steadily increasing demand environment, the proposed Bayesian control system performs the best under the initial cost structure, but when stockout cost is extremely high, the dynamic adjusted system will perform better. While in rapid changing environments, the Bayesian control system outperforms others, and therefore is proved to be more capable of responding to external changes, reducing overall inventory costs and enhance business profit.