Large-sample Analysis of U.K. Annual Report Narratives: Methods and Evidence


Speaker:Martin Walker, University of Manchester
Date: Wednesday 18 March 2015
Time: 14:00
Location: Constantine Leventis, Building One

Further details

Preliminary and not for external distribution


This paper develops an automated procedure for analysing narrative content disclosed in annual reports supplied as PDF files and then applies the procedure to analyse the structure and content of over 10,000 annual reports published during the period 2003 through 2013 by London Stock Exchange-listed firms. Tests yield a number of insights and stylised features concerning the properties of annual report narratives.

First, results suggest a significant (albeit complex) interaction between financial statement characteristics and the properties of narrative commentaries.

Second, we show that narrative reporting properties are a function of unobservable firm-specific characteristics (e.g., management’s reporting style) and time. In contrast, industry factors appear to have less impact on narrative content.

Third, key narrative subcomponents such as chairman’s statements, performance commentary, governance statements and remuneration reports are characterized by different linguistic properties.

Fourth, narratives vary as a function of observable firm-specific financial characteristics such as firm size, stock exchange listing status, profitability, and the presence of intangible assets.

Fifth, the extent to which these firm-level factors determine narrative content varies across key subsections of the narrative commentary (e.g., chairman’s statements, performance commentary, governance statements, remuneration reports, etc.)