Tham khảo tài liệu 'handbook of empirical economics and finance _6', tài chính - ngân hàng, tài chính doanh nghiệp phục vụ nhu cầu học tập, nghiên cứu và làm việc hiệu quả | 136 Handbook of Empirical Economics and Finance econometric techniques can be brought to bear or new techniques have been adopted that are better able to deal with the complexity and high-dimensional character of the data. Bhattacharya et al. 2008 manually read and classified 171 488 news stories written during the Internet IPO frenzy to test whether the media hyped Internet stocks by giving them more positive news relative to a group of non-Internet IPOs during the same period. They find that there was indeed significant media hype but after controlling for other return-related factors find that the aggregated news affect from a given day would last only two days and explained only a small portion of the difference between Internet and non-Internet IPO returns. Tetlock 2007 looks for interactions between the content of a popular Wall Street Journal column and the stock market over a 16-year period. After reducing the dimensionality of the textual data through dictionary classifications and principal component analysis he uses a vector autoregressive methodology to identify the relationship between media pessimism and market returns and volumes. Findings indicate that the WSJ column does impact returns and volume and that his textual variable depends on prior market activity. Like Antweiler and Frank 2004 Das and Chen 2007 study the impact of stock message boards on market characteristics. Rather than using a single classification algorithm to label messages as BUY SELL or HOLD they employ five different classification algorithms which get to vote for the final label. These classifiers use different data sets extracted using various grammatical parsing and statistical techniques to decide on a label. Within the forecasting literature combining forecasts has been found to improve forecasting performance Armstrong 2001 Bates and Granger 1969 and compared to the Bayes text classifier the combined algorithms have better classification accuracy. Financial economics is .
