In our paper “Firm-Level Political Risk: Measurement and Effects,” as published in the Quarterly Journal of Economics , we construct measures of sentiment and risk for individual listed firms using textual analysis of their earnings conference calls. Our data is at the quarterly frequency and covers over 11,000 US-based and international firms that hold English-language conference calls, ranging from 2002 to 2019q2.
Reference & details
Measures of Risk
Our measures of risk use simple tools from computational linguistics to quantify the share of the earnings call devoted to discussing risk in general, risks associated with politics, and risks associated with particular political topics, such as health care and economic policy.
Measures of Sentiment
Our measures of sentiment count positive and negative tone words used during the earnings call to measure the overall tone of the conversation, as well as call participants' optimism or pessimism associated with political and non-political topics.