Are All Readers the Same? Predicting Heterogeneity in Investors’ Information Retrieval from Financial Narratives

Research output: Working paperOther research output

Abstract

We develop a measure for differences in investors' consumption and processing of information contained in financial narratives based on observed user behavior. A growing stream of research in accounting and finance examines linguistic features of financial narratives, such as readability and tone, and how these features influence users' information processing and decision-making. However, users' information processing and evaluation of financial narratives is a complex phenomenon involving not only the linguistic features of the text, but also an individual users' context or knowledge. Using a tool that tracks users' reading and marking behavior in a controlled environment, we first elicit how a group of users with varying financial literacy process and evaluate excerpts from financial narratives in firms' MD&As. Next, we use this data to train state-of-the-art machine-learning algorithms to create out-of-sample predictions for how different user groups vary in their evaluation of financial narratives based on the linguistic features of the text and test whether this heterogeneity in users' evaluation is incrementally associated with capital market outcomes.
LanguageEnglish
Place of PublicationTilburg
Publication statusPublished - 2018

Fingerprint

Investors
Information retrieval
Evaluation
Information processing
Finance
Out-of-sample forecasting
Capital markets
Learning algorithm
Financial literacy
Machine learning
Readability
Decision making
User behavior
Train

Cite this

@techreport{a220447a62d44906911b49ce1db3eaed,
title = "Are All Readers the Same? Predicting Heterogeneity in Investors’ Information Retrieval from Financial Narratives",
abstract = "We develop a measure for differences in investors' consumption and processing of information contained in financial narratives based on observed user behavior. A growing stream of research in accounting and finance examines linguistic features of financial narratives, such as readability and tone, and how these features influence users' information processing and decision-making. However, users' information processing and evaluation of financial narratives is a complex phenomenon involving not only the linguistic features of the text, but also an individual users' context or knowledge. Using a tool that tracks users' reading and marking behavior in a controlled environment, we first elicit how a group of users with varying financial literacy process and evaluate excerpts from financial narratives in firms' MD&As. Next, we use this data to train state-of-the-art machine-learning algorithms to create out-of-sample predictions for how different user groups vary in their evaluation of financial narratives based on the linguistic features of the text and test whether this heterogeneity in users' evaluation is incrementally associated with capital market outcomes.",
author = "{de Kok}, Ties and Christoph Sextroh and {van Pelt}, Victor",
year = "2018",
language = "English",
type = "WorkingPaper",

}

TY - UNPB

T1 - Are All Readers the Same? Predicting Heterogeneity in Investors’ Information Retrieval from Financial Narratives

AU - de Kok, Ties

AU - Sextroh, Christoph

AU - van Pelt, Victor

PY - 2018

Y1 - 2018

N2 - We develop a measure for differences in investors' consumption and processing of information contained in financial narratives based on observed user behavior. A growing stream of research in accounting and finance examines linguistic features of financial narratives, such as readability and tone, and how these features influence users' information processing and decision-making. However, users' information processing and evaluation of financial narratives is a complex phenomenon involving not only the linguistic features of the text, but also an individual users' context or knowledge. Using a tool that tracks users' reading and marking behavior in a controlled environment, we first elicit how a group of users with varying financial literacy process and evaluate excerpts from financial narratives in firms' MD&As. Next, we use this data to train state-of-the-art machine-learning algorithms to create out-of-sample predictions for how different user groups vary in their evaluation of financial narratives based on the linguistic features of the text and test whether this heterogeneity in users' evaluation is incrementally associated with capital market outcomes.

AB - We develop a measure for differences in investors' consumption and processing of information contained in financial narratives based on observed user behavior. A growing stream of research in accounting and finance examines linguistic features of financial narratives, such as readability and tone, and how these features influence users' information processing and decision-making. However, users' information processing and evaluation of financial narratives is a complex phenomenon involving not only the linguistic features of the text, but also an individual users' context or knowledge. Using a tool that tracks users' reading and marking behavior in a controlled environment, we first elicit how a group of users with varying financial literacy process and evaluate excerpts from financial narratives in firms' MD&As. Next, we use this data to train state-of-the-art machine-learning algorithms to create out-of-sample predictions for how different user groups vary in their evaluation of financial narratives based on the linguistic features of the text and test whether this heterogeneity in users' evaluation is incrementally associated with capital market outcomes.

M3 - Working paper

BT - Are All Readers the Same? Predicting Heterogeneity in Investors’ Information Retrieval from Financial Narratives

CY - Tilburg

ER -