- 學位論文
彭博資訊系統市場每股盈餘預測的資訊內涵–台灣實證
The Information Content of Bloomberg’s Consensus Forecast of Adjusted EPS – Evidence in Taiwan
摘要
彭博是一個追蹤賣方分析師發布預測數字的即時資訊系統，也是市場參與者廣泛查閱的資訊平台。本論文檢視摩根史丹利資本國際台灣指數(MSCI Taiwan Index) 的成份股於2005至2012年在彭博資訊系統上市場預測的調整後每股盈餘的資訊內涵。相對於過去股票評等變動的研究，我們發現市場預測的調整後每股盈餘沒有避免向下修正的跡象，說明市場預測的調整後每股盈餘是較少偏誤也可能是較好且中性的分析師預測修正指標。 我們發現市場預測的調整後每股盈餘的預測修正在事件日及事件日隔日皆存在統計上顯著的資訊內涵，然而該預測修正似乎僅捕捉到資訊內涵的末端因為事件日之前的交易日總是產生較高的累積平均異常報酬。此外我們發現預測修正通常緊接著市場反轉的訊號，該訊號在預測上修時於事件日後三日內出現，預測下修時則於後六日內才出現，暗示市場反應好消息的速度快過壞消息面。就產業而言，非科技類股的市場反轉訊號又較科技類股至少晚一日出現。 我們也利用預測修正建立簡單的交易策略以顯示可能潛在的獲利，但是發現這些獲利很容易在計入合理的交易成本之後消逝。儘管如此，我們的結果有較高的信心在買進股票的策略上當預測修正幅度較大時與金融類股小幅度向上預測修正時(假設我們有足夠的交易成本折扣)。
並列摘要
A financial terminal that market participants refer to extensively, Bloomberg is a timely information system that tracks all forecasts issued by sell-side analysts. This paper examines the information content in the consensus forecast of adjusted EPS changes on the Bloomberg for constituents of MSCI Taiwan Index starting from 2005 to 2012. We document that there is no sign of downward revisions avoidance on consensus adjusted EPS as opposed to stock recommendations changes in past research. Our finding implies that consensus forecast of adjusted EPS are less biased and may be a better and neutral indicator of analysts’ forecast revision. We also find statistically significant information content to exist in ex post consensus adjusted EPS revisions on the event day and the next day. However, such revision changes seems to simply capture the end of information content as trading days prior to the event day always yield higher cumulative average abnormal returns. Separately, we discover that the revisions are followed by market reverse signals. The reverse signals show up within 3 days for upward revisions and 6 days for downward revisions, thereby suggesting that the market exploits good news faster than bad news. In terms of industry level, the reverse signals in non-technology sector are at least one day later as compared to technology sector. We also demonstrate potential profits from using a simple trading strategy based on the revision changes, but find that these profits could vanish easily after accounting for reasonable trading costs. Nonetheless, our findings note that there is higher confidence in buying share strategies for larger level of revision changes and small level of upward revision changes in financial sector, given that we had sufficient discount for transaction cost.