Financial Analysts Penalize Female CEOs for Female-Type Speech

Female representation in top management remains stubbornly low. Of CEOs in public companies in the US for example, only 6.6% are women. One piece of the puzzle relates to audience reaction – analysts, who are sophisticated and informed actors reduce their recommendations following the first earnings call with an incoming female CEO (on average, from “hold” to “sell”). We use a novel dataset of containing the voice recordings of 60,038 earnings calls, and a novel computational approach that extracts to examine these earnings calls and ask why do analyst reduce (increase) their recommendation when a new Female (Male) CEO is appointed. The scale of our data (3,465 publicly listed firms, 203,159 executives and 266,340 analysts sampled at 15-second intervals – yielding 11.5 million observations of individual participant speech, and intonation), and it’s panel structure (from 2011 to 2019) allows us exceptional granularity. We find that although it seems that analysts penalize female CEOs, they only do so when female CEOs use a distinctly female type speech (uptalk). Non-uptalk using incoming female CEOs, or uptalk using male CEOs are not penalized. To our knowledge, this is the first paper to use large-scale voice data of earnings calls, and we, therefore, end by reporting how voice data compares to text data in established measures (e.g. positive, negative, financial, uncertainty etc) that demonstrate the usefulness of voice data over transcribed text.