探討半導體循環：總體經濟指標、股價指數、次產業之分析

Abstract

半導體產業所製造之晶片廣泛運用於現代社會中，包含手機、筆電等等，推動著世界的發展，半導體產業存在明顯循環性，稱為「半導體循環」，其中包含兩個一小循環，「應用週期」：新殺手級應用推動長期半導體銷售的增長；與「需求循環」：市場對於產品需求的強弱有關，帶動半導體銷售的上下行循環。

本研究希望能提供質化量化分析，過往研究僅專注於半導體循環定義與預測，本文將針對半導體循環有更系統性、整體性研究。期許提供投資人調整投資權重、半導體廠商能及早調整策略，避免錯誤之財務、投資策略或是面臨營運風險之目的。

研究使用世界半導體貿易統計組織所統計之每月度之全球半導體銷售額資料，以及美、中、臺之總體經濟指標、大盤指數績效、次產業產值、庫存與資本支出之資料，以時間序列迴歸方式來評估半導體循環相關之指標、股市反應、次產業變化。

研究發現，指標端以臺灣製造業PMI、新訂單減庫存指數、半導體外銷訂單及美國新訂單減庫存指數為領先指標，美國實質經濟成長率與臺灣外銷訂單為同時指標。股市端以美國S&P 500指數、費城半導體指數、臺灣加權指數分別領先半導體循環2、3、4季度，若將指數拆分，其中的每股盈餘代表之基本面為領先指標帶動股價，本益比代表之評價面則為同時指標，於半導體循環轉向時變化。次產業端以半導體設備與材料、IC設計產業領先2季度受影響；晶圓代工、封裝測試產業隨著半導體循環同時變動，並以封裝測試產業表現最吻合半導體循環，IP產業則與半導體循環較無關，且廠商之資本支出無法幫助評估半導體循環。

The semiconductor industry manufactures chips that are widely used in modern society, including mobile phones, laptops, and more. Semiconductor industry drives the development of the world. The semiconductor industry exhibits a clear cyclical pattern known as the "semiconductor cycle," which consists of two main cycles: the "application cycle" and the "demand cycle." The application cycle is driven by new applications that promote long-term semiconductor sales growth, while the demand cycle is influenced by the strength of market demand for semiconductor products, leading to cyclical fluctuations in semiconductor sales.

This research aims to provide qualitative and quantitative analysis. Previous research has focused only on defining and predicting the semiconductor cycle. This research seeks to conduct a more systematic and comprehensive analysis of the semiconductor cycle. The objective is to provide investors with insights to adjust their investment weights and enable semiconductor companies to proactively adjust their strategies, avoiding financial and investment strategy or operational risks.

The research utilizes monthly global semiconductor sales data from the World Semiconductor Trade Statistics organization, as well as macroeconomic indicators, stock indexes, sub-industries output, inventory and capital expenditures data from the United States, China, and Taiwan. Time series regression is employed to assess indicators related to the semiconductor cycle, stock market reactions, and sub-industries changes.

The research findings indicate that leading indicators include Taiwan's manufacturing PMI, the new orders-to-inventory ratio, semiconductor export orders, and the US new orders-to-inventory ratio. The real economic growth rate in the US and Taiwan's export orders are simultaneous indicators. In terms of the stock market, the US S&P 500 index, the Philadelphia Semiconductor Index, and Taiwan's Weighted Index lead the semiconductor cycle by 2, 3, and 4 quarters, respectively. When examining these indexes individually, earnings per share represents the leading fundamental indicator driving stock prices, while the price-to-earnings ratio serves as a simultaneous indicator, changing during the turning point of the semiconductor cycle. In the sub-industry aspect, the semiconductor equipment and materials industry, as well as the IC design industry, receive the earliest impact. The foundry industry and packaging testing industry follow the fluctuations of the semiconductor cycle, with the packaging testing industry showing the closest alignment. However, capital expenditures by companies are not helpful in evaluating the semiconductor cycle.