Online Appendix Appendix A
Note. Distribution of economic tweets per Reuters and Bloomberg account.
Appendix B
Appendix C
Coded Variables with Tweet Examples and Reliability Scores
Variables Examples Lotus Kappa Percent
Agreement
Retweets n/a .74 .887 88.9%
Favorites n/a .96 .893 89.6%
Expert Opinion
RT @freddiethekat: Buy oil, financial stocks on dips - analyst
http://t.co/IJzjmqnL6p @HavenTowerGroup @ReutersMoney“
.92 .328 89.8%
Hope “HSBC set to add 4,000 jobs in next 4 years in China's Pearl River Delta region”
.82 .596 88.2%
Fear “RT @business: Volkswagen shares
tumble after it admits cheating on
U.S. emission tests http://t.co/Cz3z8h9ELr http://t.co/fO7Pvbpl7B”
.71 .409 75.6%
No Emotions “Has Janet Yellen flip-flopped too many times? http://t.co/VYpk5uMgqA
http://t.co/u8J7SbBUWM”
.57 .393 70.1%
Focus of the News
#Volkswagen picks a company veteran to tackle its emissions crisis: http://t.co/Sv8mRAmfun
http://t.co/Mh483nIEVi
.69 .635 81.9%
Notes. Retweets and favorites were coded as indicated in the tweet when looked up online. The tweets were collected in September 2015; the coding took place between January 2016 and March 2016.
Appendix D
Ten Most Occurring Topics in Reuters and Bloomberg Tweets with Description and Examples Most Occurring
Topics Description Example for Tweet
Fed decision/ strategy
Tweets about the Federal Reserve System (i.e., Fed) and its decision/strategy regarding interest rates.
BREAKING: Federal Reserve leaves key policy rate unchanged. Volkswagen
scandal/ share price
Tweets that dealt with the Volkswagen emission scandal as well as information on the share price of Volkswagen.
#Volkswagen will dismiss CEO Martin Winterkorn over emissions-cheating scandal: German paper http://t.co/I9HyGwoTNe
http://t.co/usx82Hxjar Oil rally/
oil prices Tweets that dealt with the fallingoil price or information on the oil market/industry.
Oil is down again, erasing last week’s gains http://t.co/xsY5ii2e2J http://t.co/ZTFtEgxTHI"
Apple event Tweets that reported on the Apple event, where news on Apple products were released (iPhone 6S, Apple TV, new Watch, etc.).
RT @ReutersLive: iPhones, iPads and TV, oh my! The #AppleEvent will soon begin. Follow our LIVE coverage:
http://t.co/LHdbK5WPCN http://t.c ‰Û_
Brazil crisis/ weak real
Tweets about the economic/debt crisis in Brazil and news about the weak currency in Brazil (i.e., real) as a reaction.
RT @ReutersOpinion: Brazil is enduring an acute crisis -- but not all is lost. Here's why:
http://t.co/cMdoNvIBSm http://t.co/Gnbel99e2S US job report Tweets that dealt with the U.S.
jobs report for August 2015. Wall St. rally fades ahead of key jobs report: http://t.co/oZJSkt9moQ http://t.co/RWz0qKmGY4
SAB Miller/ Anheuser-Busch deal
Tweets that dealt with the merger between the beer companies SAB Miller and Anheuser Busch.
Anheuser-Busch said to be lining up banks for up to $70 billion in financing ahead of takeover proposal for SABMiller http://t.co/OQNeUEIIqq Wall Street down/
China data (01/09)
Tweets that dealt with reactions on Wall Street after the release of weak economic data in China.
RT @ReutersBiz: Wall Street opens sharply lower after weak data from #China: http://t.co/5MKdELc6ZE http://t.co/g4VELapNFg
China slowdown/ economic crisis
Tweets that dealt with the economic slowdown in China and its effects on other
economies and markets.
China could be about to cause global recession
http://t.co/9BBriWjn7u http://t.co/DoaTa58Ggt Glencore stocks
slump Tweets dealing with Glencore and the drop of its share price #Glencore shares fell almost 30 percent today, wiping off more than $5 billion in market value:
Appendix E
Descriptives of Variables for VAR Analyses
Notes. The high
standard deviations are due to the counting
characteristics of the variables (e.g., number of retweets, favorites). The VAR models were checked for robustness by means of the LM test, the Portmanteau (Q) statistics, and the Eigenvalue stability condition test.
Variables 1-Hour 20-Minutes 5-Minutes
Mean SD Mean SD Mean SD
All News |∆DJI Closing | ∆News Volume ∆Retweets ∆Favorites ∆Updates ∆Expert Opinion ∆Hope ∆Fear 53.641 0 -1.014 -.151 4.849 .027 .021 .007 54.493 8.406 348.92 5 152.28 5 544.11 8 3.89 3.915 3.728 24.549 -.019 -.692 -.246 -.017 0 -.005 -.014 21.113 3.392 142.81 0 72.341 221.60 3 2.055 1.851 1.918 13.057 0 -.076 -.023 0 0 0 -.001 17.724 1.358 74.058 38.006 74.923 .786 .906 .967 Market-Wide News |∆DJI Closing | ∆News Volume ∆Retweets ∆Favorites ∆Updates ∆Expert Opinion ∆Hope ∆Fear 53.641 -.048 -1.61 -.438 3.301 .007 -.007 -.048 54.493 7.283 273.55 6 112.48 3 539.27 6 3.581 3.034 2.990 24.549 -.014 -.578 -.179 -.017 0 0 -.0143 21.113 2.871 116.33 58.697 216.50 2 1.913 1.414 1.632 Firm-Specific News |∆DJI Closing | ∆News Volume ∆Retweets ∆Favorites ∆Updates ∆Expert Opinion ∆Hope ∆Fear 53.641 -.014 -.329 -.192 0 0 -.021 0 54.493 3.937 213.18 5 100.21 9 84.691 1.215 2.211 2.094
Appendix F
Appendix G