روابط متقابل در میان شاخص های بازار سهام صنعت پانل TFT-LCD ژاپن، کره و تایوان: استفاده از مدل سه جانبه FIEC-FIGARCH
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|16369||2012||10 صفحه PDF||سفارش دهید||7481 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Economic Modelling, Volume 29, Issue 6, November 2012, Pages 2724–2733
The purpose of this study is to analyze the interrelationships among the Taiwanese, Japanese and Korean TFT-LCD panel industry stock market indexes by applying a trivariate FIEC-FIGARCH model. The empirical results confirm that the FIEC-FIGARCH model can be used to capture long memory behavior and allow us to conclude that mean and volatility spillover, and long memory effects are found in these three markets. Furthermore, we found that deviations in the long-run equilibrium for Japanese TFT-LCD panel industry adjust back very slowly in comparison to the other two countries; and that, in terms of conditional covariance, dynamic interrelationships exist among the TFT-LCD panel industry stock market indices of these three countries.
In recent years the global economy has become more integrated due to the trends of internationalization and liberalization and the rapid progress of R&D, applied science, networking, etc. sensitizing local stocks to turbulences in foreign financial markets, for instance, the stock price vibrations derived from the Asian Financial Crisis in 1997, the acute stock price vibrations after the dreadful 911 terrorist attacks in United States, etc. Taiwan is also part of the internationalization trend; its entrance into the World Trade Organization (WTO) and its exportation oriented economy are characteristics that favor international influences in Taiwanese stock prices. Considering these circumstances and also that the volatility of equity markets might exhibit long memory effects, the international price transmission effects on the equity market of TFT-LCD panel industry are worth being analyzed by enterprises, government officials and scholars due to its importance for Taiwan economy. The TFT-LCD panel industry is characterized by (1) high capital requirement, entrance barriers and exit costs; (2) highly intensive technology along with complicated processes of production; (3) short-term product life cycle; (4) price sensitive to market demand, supply and economic cycles; (5) popular international division of labor; (6) high market concentration; (7) completeness of industry structure that helps to create competitive advantage; (8) material cost accounting for high percentages; (9) high brain circulation and (10) high concentration of shareholding of domestic TFT-LCD panel firms.1 The global development of the TFT-LCD panel industry is primarily concentrated in Taiwan, Korea and Japan. Japan was the largest producer of TFT-LCD panels until 1995, when Korea overtook the leadership by pursuing a low price strategy. The Won depreciation during Asian financial crisis (1997) sharpened the price differences, leading Japan to withdraw from the mass production of TFT-LCD panels, and to switch to research and development (R&D) in new technology and marketing channels. As for Taiwan, the production of small and medium sized TFT-LCD panels was established early and kept until Japan released the technology of large size TFT-LCD panels and transferred this knowledge to Taiwan. While Japanese firms gained from running brand channels, Taiwanese firms seize the technology transfer; it was a good deal for both countries. Taiwan began the mass production of large-size TFT-LCD panels in 1998 based on the competitive advantages of OEM and ODM for internationally known computer producers. Thus, the market share of Taiwan increased from only 2% in 1999 to more than 40% by 2004 and kept rising; in 2006 Taiwan TFT-LCD panel firms were given the award of global-best profit margin. Meanwhile the Korean TFT-LCD panel industry established a complete system of key parts of TFT-LCD panels obtaining large global market shares with self-owned brands. Between 2004 and 2005 the global production of large-scale TFT-LCD grew 25% from US$34 billion in 2004 to US$43 billion in 2005 with laptops, monitors and LCD TV screams as the principal components of a demand mainly supplied by Taiwan, Japan and Korea. This profitable industry attracts international securities investors that pay strong attention to the stock price volatility because their concern focuses on the portfolio diversification rather than in a single company performance. Their forecast analysis for risk hedging, arbitrage searching and correlations seizing technify local stock markets and increase the spillover effects of international events into the local economies. These circumstances make the investigation of the interrelationships among the TFT-LCD industry stock market indexes of Taiwan, Japan and Korea2 a very important topic for the future development of these industries as well as a model for other industries sharing similar characteristics. In order to make easier to understand this paper, some terms wide used in it will be explained as follows: TFT-LCD are liquid crystal displays (LCD) which use thin-film transistor (TFT) technology to improve image quality used in television sets, computer monitors, mobile phones, handheld video game systems, personal digital assistants, navigations systems, projectors, etc. Stock market index is a method of measuring a section of the stock market. In our study the indexes measure the TFT-LCD panel industry for Taiwan, Japan and Korea by weighting the stock value of each firm in the panel industry within the corresponding country to reflect the market capitalization of its components. Long memory in time series data means that past behavior and information affects the data for long time with dependency effects (Cheung, 1993 and Davidson, 2004). Various findings of long memory phenomenon in the volatility of financial instruments cannot be properly explained with GARCH models, as for example, slowly decaying autocorrelations and mean reversion phenomenon, which are exhibit for the sample data employed in this study as indicated later. In these cases, the conventional integer cointegration method can be used to establish stationary relationships between linear combination of different variables, but cannot be used to delineate the effects of long-term lag for autocorrelation series. Therefore, this study proposes employing the fractionally integrated error correction (FIEC) model and the fractionally integrated GARCH (FIGARCH) model to capture the long memory effects of financial time series data. Fractionally integrated general autoregressive conditional heteroskedacity (FIGARCH) model implies a finite persistence of volatility shocks (while there is no persistence in the GARCH framework), i.e., long memory behavior and a slow rate of decay after a volatility shock. An IGARCH model implies complete persistence of a shock and seemed to quickly fall out of favor. Interestingly, the FIGARCH (1,d,1) nests a GARCH(1,1) with d = 0 and the IGARCH model for d = 1 ( Beine et al., 1999). Stock prices volatility is unpredictable and heteroskedacity, therefore a precise heteroskedacity model is needed to simulate the volatility clustering phenomenon, i.e., large volatility usually followed by large volatility and small volatility usually followed by small volatility. The pioneer in these kinds of models was Engle (1982) who developed the ARCH model to analyze series that exhibit constant unconditional (or long-run) variance with periods in which the variance is relatively high. Bollerslev (1986) generalized the model allowing for both autoregressive and moving-average components in the heteroskedacity variance with the benefit that a more parsimonious GARCH representation is much easier to identify and estimate; however, GARCH models can only capture the persistence of short-term volatility while the FIGARCH model, derived by Baillie, et al. (1996), better capture the long memory dynamic dependencies in the conditional variance. A number of studies reveal that the FIGARCH model adequately simulates the long memory in volatility of financial data well, see, for example, Brunetti and Gilbert (2000), Beine et al. (2002), Tang and Shieh (2006) and Figuerola-Ferretti and Gilbert (2008). In particular, Mendes and Kolev (2008) investigate the interdependence in emerging markets, which can be driven by conditional short and long range dependence in volatility. Mendes and Kolev (2008) also fit copulas model to pairs of filtered returns, which using FIGARCH process to the joint excess residuals. Their empirical results indicate that long memory in volatility is responsible for chances in increasing external dependence. Fractionally integrated error correction (FIEC) is the name of the model established by Granger (1986) to analyze variables when fractional cointegration prevails. In general, two or more economic variables are fractionally cointegrated if both are of the same fractional order, whereas a linear combination of them has a smaller fractional order. By incorporating a FIGARCH model into a FIEC model, this study formulates a FIEC-FIGARCH model to analyze the interrelationships among the TFT-LCD panel industry stock market index (from now on TFT-LCD index) of Taiwan, Japan and Korea, created by the authors especially for this study and applicable for further analysis, by itself and important contribution of our study for the industry. Long term memory is a different approach that the widespread short-term methodology but the empirical results prove the validity of our proposal. Following our ideas, empirical results would provide useful information for investors looking for price discoveries, hedging or diversification benefits within the TFT-LCD panel industry, it would also help governmental agencies to make their policy-related decisions. The remaining parts of this study are organized as follows. Section 2 outlines the theoretical foundation of FIEC-FIGARCH Model. Section 3 summarizes the empirical methodology, followed by the empirical results reported in Section 4. Finally, Section 5 presents some concluding remarks.
نتیجه گیری انگلیسی
Long memory in time series data means that current behavior is affected by past long-run behavior, reflecting a phenomenon of long-range dependency effect. We conclude that the long memory model adequately simulates the volatility of financial data, and therefore our study formulates a trivariate FIEC-FIGARCH model, incorporating a FIGARCH model into a fractionally integrated error correction (FIEC) model to analyze the interrelationships among the Taiwanese, Japanese and Korean TFT-LCD panel industries. The sample period of this study extends from July 23, 2004 to September 22, 2008. With these data we constructed the TFT-LCD index and we calculated it for these three countries in order to perform an empirical analysis. The empirical results verify the reliability of the model, i.e., because it is correctly specified it can be used to capture long-term mean return and volatility behaviors, and provide some useful information for investors to make their stock investment decisions about the TFT-LCD panel industries of these three countries. Furthermore after big and long affectations of the mean, the error correction term (Z) can be used to adjust the model making it flexible and improving its usability. The results also demonstrate that there exists long-term memory in Taiwanese, Korean and Japanese TFT-LCD indexes. The three indexes have their own-market and cross-market means spillover effects so their rates of return can be predicted by using past information; furthermore the inter-industry reactions and feedback effects with long-term memory, mean and volatility spillovers effects and dynamic interactions generate a dependency effect between the three industries. All of the three indexes are autoregressive with long-memory, and therefore new information is not incorporated into the stock prices immediately. On the contrary it has a prolonged effect favoring the arbitrage, a topic of special interest for short-term investors who can apply the findings of this study to forecast the TFT-LCD index and obtain economic benefits until the TFT-LCD panel industry become more efficient in terms of information disclosing and spreading. Due to the cointegration of the indexes the stock price volatility of one country is reflected in the other two at least for a week before return to the mean of the industry, a topic of special inters for hedge investors (usually international trade firms) who can apply the findings of this study to design their budgets, financial plans and hedge strategies considering the industry trends exposed with our model. But the quite intense interrelationship between the TFT-LCD industry of Taiwan, Japan and Korea, spreading the investment between these countries does not bring significant benefits derived from diversification. Investors with this aim should rather buy stocks from different industries even within the same country than TFT-LCD index stocks from different countries. The model has demonstrated some market imperfections, and therefore governmental authorities should implement policies that improve the information disclosure and the investors' knowledge in order to incorporate new information faster into the stock prices and thus improve the business environment. Some strategies could include seminars, workshops, international trade barriers relaxation, cross countries relationships promotion, etc. The TFT-LCD index along with the model can be tools to monitor the effectiveness of these policies.