Forecasting economic indicators using a consumer sentiment index: Survey‐based versus text‐based data

• DOI: http://doi.org/10.1002/for.2584
• Date: 01 September 2019
• Published date: 01 September 2019

RESEARCH ARTICLE

Forecasting economic indicators using a consumer

sentiment index: Survey‐based versus text‐based data

Minchae Song

1

| Kyung‐shik Shin

2

1

Big Data Analytics, Ewha Womans

University, Seoul, Republic of Korea

2

School of Business, Ewha Womans

University, Seoul, Republic of Korea

Correspondence

Kyung‐shik Shin, School of Business,

Ewha Womans University, 52

Ewhayeodae‐gil, Seodaemun‐Gu, Seoul,

120‐750, Republic of Korea

Email: ksshin@ewha.ac.kr

Abstract

Given the confirmed effectiveness of the survey‐based consumer sentiment

index (CSI) as a leading indicator of real economic conditions, the CSI is

actively used in making policy judgments and decisions in many countries.

However, although the CSI offers qualitative information for presenting

current conditions and predicting a household's future economic activity,

the survey‐based method has several limitations. In this context, we extracted

sentiment information from online economic news articles and demonstrated

that the Korean cases are a good illustration of applying a text mining

technique when generating a CSI using sentiment analysis. By applying a

simple sentiment analysis based on the lexicon approach, this paper con-

firmed that news articles can be an effective source for generating an eco-

nomic indicator in Korea. Even though cross‐national comparative research

results are suited better than national‐level data to generalize and verify

the method used in this study, international comparisons are quite challeng-

ing to draw due to the necessary linguistic preprocessing. We hope to

encourage further cross‐national comparative research to apply the approach

proposed in this study.

KEYWORDS

consumer sentiment index, economic indicator forecasting, sentiment analysis, text mining

1|INTRODUCTION

Keynes realized that employment and production deci-

sions are based on expected consumer demand and that

aggregate demand follows firms' production and employ-

ment decisions (Benhabib, Wang, & Wen, 2015). In

response to the widespread belief that consumers' opin-

ions and expectations influence the direction of the econ-

omy, a growing number of studies have set out to analyze

the relationship between consumer attitudes and eco-

nomic variables (Bram & Ludvigson, 1997; Gelper,

Lemmens, & Croux, 2007; Ludvigson, 2004). Several

papers argued that sentiment shocks can drive aggregate

business conditions (Angeletos & La'O, 2013; Benhabib

et al., 2015; Benhabib & Wen, 2004; Farmer & Guo,

1994). An economic interpretation of sentiment shocks

is simply that economic agents such as households, firms,

and governments sometimes become optimistic or pessi-

mistic about future consumption, future investment, or

future inflationary pressures, and those attitudes appear

in the form of sentiment shocks. Sentiment shocks are

therefore identified from the dynamic interactions

between observed expectations and realized macroeco-

nomic time series (Arias, 2016; Milani, 2017). Other

researchers have suggested that sentiment may reflect

information about future states of the economy held by

individuals but not (yet) observed in publicly released

data. Barsky and Sims (2012) found out that this dynamic

Received: 16 July 2018 Revised: 11 January 2019 Accepted: 12 February 2019

DOI: 10.1002/for.2584

504 © 2019 John Wiley & Sons, Ltd. Journal of Forecasting. 2019;38:504–518.wileyonlinelibrary.com/journal/for