A Comparison of New Factor Models in the Korean Stock Market

• Published date: 01 October 2019
• Author: Hankil Kang,Wooyeon Kim,Jangkoo Kang
• DOI: http://doi.org/10.1111/ajfs.12274
• Date: 01 October 2019

A Comparison of New Factor Models in the

Korean Stock Market

Hankil Kang

Securities & Derivatives R&D Center, Korea Exchange, Republic of Korea

Jangkoo Kang

College of Business, Korea Advanced Institute of Science and Technology, Republic of Korea

Wooyeon Kim*

College of Business, Korea Advanced Institute of Science and Technology, Republic of Korea

Received 17 January 2018; Accepted 13 May 2019

Abstract

We compare the empirical performance of the Fama and French (2015) five-factor model,

the Hou et al. (2015) q-factor model, and their variations in the Korean stock market.

Among the models considered, we demonstrate that the adjusted five-factor model, which

includes the quarterly- rather than the yearly-based profitability factor, best explains the size-,

value-, investment-, and profitability-sorted portfolio returns. We also document supporting

evidence that high-minus-low (HML) may not be a redundant factor in the existence of

q-factors. The adjusted five-factor model outperforms the other factor models in digesting

various anomalies in the Korean market.

Keywords Anomalies; Asset pricing model; Cross-section of stock returns; Factor model;

Time series

JEL Classification: G11, G12

1. Introduction

In recent years, researchers have introduced a number of linear asset pricing models

in the stock market. Specifically, Fama and French (2015, hereafter FF) extend their

1993 three-factor model with investment (Inv) and profitability factors to develop a

five-factor asset pricing model. Hou et al. (2015, hereafter HXZ) derive a four-

factor model from the q-theory of firm Inv. Although the two models appear to be

similar, they are different in their motivation, methods of factor construction, and

inclusion of the value factor (high-minus-low, HML). The relative performance of

*Corresponding author: College of Business, Korea Advanced Institute of Science and Tech-

nology, Seoul, Republic of Korea. Tel: +82-2-958-3693, Fax: +82-2-959-4645, email:

qdragon326@kaist.ac.kr.

Asia-Pacific Journal of Financial Studies (2019) 48, 593–614 doi:10.1111/ajfs.12274

©2019 Korean Securities Association 593

alternative asset pricing models to explain the various patterns in stock returns is

an ongoing debate.

1

It is important to ask whether well-known cross-sectional patterns in the US

stock market also exist, or whether the asset pricing models in the US mar-

ket also work well in international stock markets, since the US evidence on the

cross-section of stock returns may not appear to be the same as in international

markets.

2

For example, Fama and French (2017) report that the book-to-market

(B/M), Inv, and profitability effects are strong in North America, Europe, and

Asia–Pacific countries except for Japan, whereas only the B/M effect is strong in

Japan.

This paper compares the performance of the FF five-factor model, the HXZ q-

factor model, and their variations in the Korean stock market. First, we test whether

the size, book-to-market, profitability, or Inv effect exists in the Korean market.

Second, we discuss the performance of alternative asset pricing models in describing

the cross-section of portfolio returns sorted by size, B/M, profitability, and Inv.

Third, we consider a variety of anomaly portfolios in the US market and test

whether our candidate models effectively explain significant anomalies in the

Korean market.

Our empirical findings can be summarized as follows. First, in our sample from

July 2002 to June 2015, we observe value, profitability, and Inv effects in the Korean

market. We find little evidence of the size effect. Specifically, to exhibit a strong

cross-sectional pattern, we find that the profitability variable should be based on

quarterly earnings data, following Hou et al. (2015), instead of the annual prof-

itability measure of Fama and French (2015). Second, the adjusted FF five-factor

model with the quarterly profitability factor outperforms the FF three-factor model,

the FF five-factor model, and the HXZ q-factor model in explaining portfolio

returns sorted on size, book-to-market, profitability, or Inv. Third, the adjusted

five-factor model also performs better than the other candidate models in explain-

ing cross-sectional patterns in portfolio returns that exploit various anomalies such

as momentum.

This paper further contributes to the literature in the following ways. First, we

closely examine whether the size, value, profitability, or Inv effect exists in the Kor-

ean market. By investigating the empirical performance of the FF five-factor and

the HXZ q-factor models in the Korean market, we can test and compare these

models without being overly concerned about model mining or data sno oping

problems that can arise in tests on international market data. By analyzing the Kor-

ean local market, which is one of the most important local stock markets in the

Asia–Pacific region and has not been closely examined in an international study by

1

Recent examples include Hou et al. (2017) and Stambaugh and Yuan (2017).

2

See Fama and French (1998), Griffin (2002), and Fama and French (2012). In particular,

studies for emerging markets and the Asia–Pacific region include Kim et al. (2012), Cakici

et al. (2013), Hanauer and Linhart (2015), and Hahn and Yoon (2016).

H. Kang et al.

594 ©2019 Korean Securities Association