|
This study investigates the short-term and long-term effects of three major institutional investors chip factors on the stock returns of the IC design and IC manufacturing industries. The research object consists of the top three listed companies in terms of market capitalization in the semiconductor industry's IC design and IC manufacturing sectors at the end of 2021, include MediaTek (2454), Silergy (6415), Novatek (3034), TSMC (2330), UMC (2303), and VIS (5347). The study period spans from Jan. 5, 2015, to Dec. 30, 2021, comprising a total of 10,272 daily observations. In order to explore the short-term and long-term effects, a Transfer Function Noise Model (TFARMA) is used as the forecasting model, and the accuracy of the model is evaluated by the Mean Absolute Percentage Error (MAPE). The research results are as follows:
1.Foreign institutional investors' net buying/selling activity (NBS) has both short-term and long-term effects on the stock returns of most companies in the IC design and IC manufacturing industries. Additionally, the trading volume ratio (TVR) and shareholding ratio (SR) of foreign institutional investors have either long-term or short-term effects on the stock returns of the six companies in the IC design and IC manufacturing industries.
2.The NBS of trust institutional investors has both short-term and long-term effects on the stock returns of three companies, while it only has a short-term impact on the other three companies. Additionally, the TVR and SR of trust institutional investors have either long-term or short-term effects on the stock returns of all six companies.
3.The NBS of proprietary traders has both short-term and long-term effects on the stock returns of most companies. Additionally, the TVR and SR of proprietary traders have either long-term or short-term effects on the stock returns of the six companies.
4.The estimation results of the TFARMA model revealed that the MAPE for the six companies ranged from 0.6634% to 1.9974%. This indicates that the model's forecast errors were relatively small, demonstrating a high level of accuracy in the model's forecast.
|