Behavioral heterogeneity in return expectations across equity style portfolios

• Published date: 01 December 2021
• Author: Philip A. Stork,Milan Vidojevic,Remco C. J. Zwinkels
• Date: 01 December 2021
• DOI: http://doi.org/10.1111/irfi.12323

ORIGINAL ARTICLE

Behavioral heterogeneity in return expectations

across equity style portfolios

Philip A. Stork

1,2

| Milan Vidojevic

1,3

| Remco C. J. Zwinkels

1,2

1

Vrije Universiteit Amsterdam, The

Netherlands

2

Tinbergen Institute, The Netherlands

3

Robeco Institutional Asset Management, The

Netherlands

Correspondence

Remco C. J. Zwinkels, Vrije Universiteit

Amsterdam, The Netherlands.

Email: r.zwinkels@vu.nl

Abstract

We estimate a heterogeneous agent model on five prominent

equity investment styles—value, size, profitability, investment,

and momentum—and find evidence for behavioral heterogene-

ity in expected return formation. Our model features two

groups of boundedly rational investors, fundamentalists and

chartists, whose demand functions for the investment styles

depend on their respective expected style return forecasts. The

fundamentalists form return expectations using a model based

on time-varying stock-level characteristics and dynamic factor

premia, and the chartists do so based on heuristics commonly

employed by technical analysts, such as moving average rules.

Our results cast doubt on the theories that assume perfect ratio-

nality of the representative agent in financial markets, and give

support to the behavioral theories with heterogeneous agents.

KEYWORDS

asset pricing, behavioral finance, heterogeneous agent models,

style investing

JEL CLASSIFICATION

G11; G12; G14

1|INTRODUCTION

The notion that market prices exhibit dynamics consistent with the rational expectations hypothesis is repeatedly

challenged in the finance literature. The theory of rational expectations, originally proposed by Muth (1961),

Received: 24 October 2018 Revised: 18 May 2020 Accepted: 10 June 2020

DOI: 10.1111/irfi.12323

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and

reproduction in any medium, provided the original work is properly cited.

© 2020 The Authors. International Review of Finance published by John Wiley & Sons Australia, Ltd on behalf of International

Review of Finance Ltd. 2020

International Review of Finance. 2021;21:1225–1250. wileyonlinelibrary.com/journal/irfi 1225

postulates that the representative agent in the market is fully rational and sets the clearing prices and quantities of

assets. The last two decades have seen a proliferation in theoretical and empirical work that produced a considerable

amount of evidence for existence of agents with boundedly rational and heterogeneous beliefs about future asset

prices, who trade on these expectations, thereby potentially causing prices to deviate from their rational values.

Depending on which group of agents prevails in the market, different price dynamics are generated. Structural asset

pricing models that feature boundedly rational agents have shown promise in explaining empirically observed pat-

terns, such as volatility clustering, skewed return distributions, and persistent deviations from the fundamentals sub-

sequently followed by corrections, that remain a puzzle for the rational models (see Lux, 1998).

Brock and Hommes (1998) propose a discrete time,

1

structural asset pricing model that features two types of

agents in the market, who use different rules to form expectations about future prices. One group of agents believe

that the prices of assets revert toward their fundamental values and, as such, their actions have stabilizing effects on

market prices. This group of agents is referred to as the fundamentalists. On the other hand, there is a group of

investors who believe that patterns in past prices have predictive power over future asset returns, and thus form

expectations using heuristics based on past prices. These traders are often referred to as the technical traders or

chartists. In this framework, fundamentalists resemble the “rational”investors; however, they are only boundedly

rational as they do not take the existence of chartists into account when forming price expectations. A key question

is whether chartists can persist in the market equilibrium, or their actions on average cancel out, and the market ends

up in a steady state where the prices are set by the rational marginal investor (see Friedman, 1953). If both groups of

agents are identified in the market, the asset price dynamics are determined by their interactions; that is, which

group is present in a greater quantity determines whether prices converge toward or diverge away from their funda-

mentals. Agents in this type of models are assumed to choose which expectation formation rule they follow based

on the past performance of the rule they had used relative to the one used by the other group.

2

In addition, the

models feature an endogenous parameter—commonly referred to as the intensity of choice—that determines how

sensitive the agent groups are to the past performance when choosing which strategy to follow. A high value of this

parameter indicates that agents are sensitive to past performance, and therefore there is a lot of switching between

groups.

The choice of modeling expectation formation based on a combination of fundamentalists and chartists is based

on a substantial body of mainly experimental studies. For instance, Bloomfield and Hales (2002) find evidence for

the combination of trend following and mean reverting expectation in a laboratory experiment on MBA students,

who are forecasting a random walk. Furthermore, Hommes, Sonnemans, Tuinstra, and Velden (2005) show that

agents with the knowledge of the past returns, dividend yields, and interest rates utilize technical rules to make

return forecasts. Landier, Ma, and Thesmar (2020) run a large-scale experiment and find that a forward-looking

extrapolation best fits the forecasts. We refer to Assenza, Bao, and Cars Hommes (2014) for an overview of the

experimental literature on expectation formation. A number of papers provide evidence for fundamentalism and

chartism based on survey data. For example, ter Ellen, Verschoor, and Zwinkels (2013) and Goldbaum and

Zwinkels (2014) find fundamentalism, chartism, and switching between them in foreign exchange surveys. Green-

wood and Shleifer (2014) find, among a set of six data sources, that the survey expectations are characterized by

trend extrapolation, and display negative correlation with model-based forecasts. Finally, there is a set of studies that

finds evidence for the two types of expectation formation in the actions of fund managers. For example, Frijns, Gil-

bert, and Zwinkels (2016) find that mutual fund managers move their exposure toward styles that performed well in

the past. Schauten, Willemstein, and Zwinkels (2015) find similar results for hedge fund managers.

Heterogeneous agent models (HAMs) have been developed in many different specifications, and extensively

tested in different contexts. Most early studies in this area used simulations in order to validate predictions of their

respective models; however, empirical estimation using real world data has been more frequently conducted over

the last couple of decades. Boswijk, Hommes, and Manzan (2007) are the first to empirically estimate an HAM fea-

turing agents that switch between groups using the equity market portfolio as the test asset,

3

and find evidence for

significant behavioral heterogeneity in beliefs about the future index levels. Chiarella, He, and Zwinkels (2014)

1226 STORK ET AL.