| Summary: | 碩士 === 實踐大學 === 企業管理研究所 === 92 === Statistics showed that in 2001 there were 40 products whose production were listed as the world top three. (Industrial Technology Intelligence Services, Ministry of Economics, June 2002) Among them, 14 out of the 18 items listed as world top one were information technological products. And LCD monitor entering the list for the first time was listed in the world top one items. Besides that, the yearly production growth rate of LCD monitor was the largest compared to all other products in Taiwan from 2001 to 2002, and should continue in the year 2003. Apparently, LCD industry has become the most important one in Taiwan information industries. Viewing that, we tried to find out the leading index of LCD industry from the correlation of the stock price return among those companies located as in the upper, middle, or lower stream in the LCD industry.
The sample companies were chosen from those LCD related ones listed in TSE and OTC, and separated as upper, middle and lower stream in LCD industrial chain. There were 5 companies in each stream in our sample. The sampling criterion was that the LCD related products should contribute more than 50% out of its total revenue. The study period started from October 28, 2002 to April 3, 2003. The sample database comprised with the 30-minute quotation data in these 105 transaction days. We firstly transformed these quotation records to return records, and then connected each other as upper, middle and lower return indexes with Dow-Jones average method. Secondly, we used ADF unit root test to test whether the upper, middle and lower stream index series is stationary or not, and used the Schwartz’s Bayesian Criterion to decide the leading legs. And then, we used Variance Decomposition to identify the influence from the changes of the upper, middle and lower stream indexes. Besides that, we applied Impulse Response Analysis to give a brief description of the effects among them. At last, we used Granger causality test to test the causality relationship among these three indexes. The results of the study were,
1. All the upper, middle and lower stream index returns were stationary series with order zero, so there shouldn’t be any Spurious Regression.
2. From the results of Variance Decomposition and Impulse Response Analysis, the influence of lower stream index to upper and middle stream indexes was bigger than the influence of middle and upper stream indexes to lower stream index. Hence, we could predict the trend of middle and upper stream indexes with the lower stream index. That is to say, the lower stream index could be a leading index of middle and upper stream index.
3. When the leading leg was 1 period, the Granger causality test showed significant results for lower stream index to middle stream index and to the upper stream index.
|
|---|
