eThesis - School of Business electronic theses

The purpose of the study is to assess whether the negative language, defined as fraction of negative words as described in Harvard IV-4 psychosocial dictionary, extracted from earnings conference call participants’ comments can be used to assess future company earnings level and in addition, whether stock market underreacts to the negative language of the participants. Both issues are studied using event study methodology and ordinary least square regressions. Both multi- and univariate analysis is employed. The sample covers a unique dataset of 765 earnings conference call transcripts of 17 and 26 Finnish and Swedish large cap stocks, respectively, from 2003 – 2009. The transcripts have been gathered from LexisNexis online news service. Share price data, OMX Helsinki and OMX Stockholm exchange benchmark stock returns and control variable data have been obtained from Thomson Datastream and Thomson OneBanker service. I find that after controlling for several firm characteristics, lagged earnings per share and earnings level as well as the actual earnings level relative to latest analyst consensus level, higher current level of negative words in managers’ comments indicates lower returns during next fiscal quarter. There is some evidence indicating that higher proportion of negative words in managers’ comments also indicates lower the mean earnings during subsequent four fiscal quarters.