從證券分析師於選舉日發佈盈餘預測修正行為看產業對美國大選的敏感度

Abstract

本研究檢視證券分析師是否在美國大選選舉日對個股發布盈餘預測修正，藉以反映該股所屬產業之政治敏感度。本文利用美國選舉結果多在深夜時點揭曉之特性，以分析師加時工作反映(1)其對於總統及聯邦議會選舉事件衝擊大小之解讀與 (2) 分析師解析選舉事件衝擊之難度。本研究區分為總統與參議院選舉兩類別討論，測試分析師在選舉年開票時點是否發布較多的盈餘預測修正，及在選舉結果意外或相近之情況下，盈餘預測修正發布數是否有所差異。本文亦將樣本分成政治敏感與不敏感產業，探討分析師對敏感與不敏感產業之發布行為，是否有差異。實證結果顯示，分析師在總統選舉年與議院選舉年，選舉日所作發布數皆較非選舉年多；但在選舉結果意外或相近時，較非意外或相近時少。而在總統選舉結果意外或相近時，分析師較少於該日發布敏感性產業；在參議院選舉民主與共和兩黨當選人數意外或相近時，分析師針對敏感與不敏感產業皆較少發布預測修正。

This study investigates earnings forecasts for individual firms issued by security analysts on the U.S. election days, showing whether the industries the firms belonging to are sensitive to politics. The U.S. election results are typically announced at midnight, which serves to demonstrate analyst efforts and concerns over the difficulty in measuring the magnitude and direction of the impacts of Presidential and Senators/Congressmen elections on companies they follow. We use analysts’ working after hours as a proxy to interpret the extent of their perceived election’s effect as well as the time and efforts needed to measure the impacts of elections. We explore President and Senate/Congress elections separately to test if analysts issue more earnings forecasts revisions or not during the time interval immediately after the ballots are announced in the election years and especially when the election results appear to be surprising. Moreover, we partition the sample into politically sensitive industries and politically insensitive ones to investigate the difference between them when the election results surprise the general public. We find that both in Presidential election years and Senate/Congress election years, analysts issue more forecasts revisions than in the other years, but issue fewer revisions in the years when election results are surprising or close. Moreover, when the President elect announcement is unanticipated, analysts issue fewer revisions only for politically sensitive industries. However, when the senate ballots are close between the Republicans and the Democrats, analysts issue fewer ones for both.