The Validity of Investor Sentiment Proxies

• Author: Robert B. Durand,Joyce Khuu,Lee A. Smales,Felix Chan
• Published date: 01 September 2017
• DOI: http://doi.org/10.1111/irfi.12102
• Date: 01 September 2017

The Validity of Investor Sentiment

Proxies*

FELIX CHAN,ROBERT B. DURAND,JOYCE KHUU AND LEE A. SMALES

School of Economics and Finance, Curtin University,Perth, Western Australia, Australia

ABSTRACT

Behavioral finance research relies on proxies for unobservable phenomena.

Different proxies for the same underlying phenomena should be correlated

(formal proof of this proposition is presented in this letter). This letter exam-

ines proxies for an unobservable variable, sentiment. We utilize a well-

known methodology to construct text-based sentiment proxies and compare

these with metrics from Baker and Wurgler. We find that they are not corre-

lated. At least one, but perhaps all, of these are not valid proxies of sentiment.

JEL Codes: G02; G12

I. INTRODUCTION

Sound proxies should have a strong positive correlation with the unobserved true

variable (Krasker and Pratt 1986). Where different proxies might represent the

same underlying phenomenon, these different proxies should be correlated with

each other (see Proposition 1).

Investor sentiment is important in empirical finance (Baker and Wurgler 2006,

2007; Tetlock 2007; Yu and Yuan 2011; Baker et al. 2012; García 2013; Smales

2014), yet sentiment cannot be measured directly. Therefore, we cannot consider

the soundness of these proxies by finding their correlation with unobservable in-

vestment sentiment. We can, however, consider their correlation with each

other.

We consider methodologies to proxy investor sentiment in empirical Finance.

We find that proxies for sentiment obtained using these methodologies are not

correlated. Therefore, at least one of these methodologies produces invalid prox-

ies for investor sentiment.

II. SENTIMENT PROXIES

Baker and Wurgler (2006, 2007) utilize a “top down”approach using market and

macroeconomic variables to proxy market sentiment. Their sentiment index,

* We are grateful for comments made by seminar participants at Curtin University and an anony-

mous referee.

© 2016 International Review of Finance Ltd. 2016

International Review of Finance, 17:3, 2017: pp. 473–477

DOI: 10.1111/irfi.12102