Twitter-based Uncertainty Indices

Download Twitter-based Uncertainty Data

We are pleased to introduce several new Twitter-based Economic Uncertainty (TEU) indices that run from June 2011 to the present. It was developed by Thomas Renault (University Paris 1 Panthéon-Sorbonne) in conjunction with Scott R. Baker (Northwestern), Nicholas Bloom (Stanford) and Steve Davis (University of Chicago).

We first extract all messages (tweets) sent on Twitter in since June 2011 that contain keywords related to Uncertainty and keywords related to the Economy.

The Uncertainty terms are as follows: 'uncertain', 'uncertainly', 'uncertainties', 'uncertainty'. Keywords related to the economy are the following: 'economic', 'economical', 'economically', 'economics', 'economies', 'economist', 'economists', 'economy'.

Using this database of tweets, we construct four TEU indices. The first, TEU-ENG, consists of the total number of daily English-language tweets containing both Uncertainty terms as well as Economy terms. Our TEU-USA index works to isolate the number of these tweets that originates from users in the United States using a geo-tag-based classifier. US users make up about 50 percent of the English language Twitter population in this sample.

Our TEU-WGT index is a variant of our TEU-USA index that weights each tweet by (1+log(1+# of retweets)). Finally, to control for changes in Twitter usage intensity over time, our TEU-SCA index scales the number of tweets each day by the number of tweets on that day that contain the word 'have'.

More details about these indices and data can be found in the paper: Twitter-Derived Measures of Economic Uncertainty.