A Case of Sticky Floors: Gender Wage Differentials in Thailand

• DOI: http://doi.org/10.1111/j.1467-8381.2011.02047.x
• Published date: 01 March 2011
• Date: 01 March 2011
• Author: Chris Sakellariou,Zheng Fang

A Case of Sticky Floors: Gender Wage

Differentials in Thailand*

Zheng Fang and Chris Sakellariou

Received 1 March 2010; accepted 28 September 2010

Using 2004 Household Socio-Economic Survey data for Thailand, weexplore the

gender wage differentials across the entire unconditional wagedistribution and find

a strong sticky floor effect. Glass ceilings, although observed along with sticky

floors in the raw data and pooled regressions, disappear in the counterfactual

decompositions, suggesting a strong sticky floor pattern in gender wage differen-

tials in Thailand.Whereas the evidence for most European countries points to glass

ceilings, our findings for Thailand are qualitatively similar to those from other

South Asian and Southeast Asian countries for which evidence exists; namely,

Singapore, the Philippines and Sri Lanka.

Keywords: sticky floors, gender wage differentials, quantile regression, counter-

factual decompositions.

JEL classifications codes: J7, J16, J31.

doi: 10.1111/j.1467-8381.2011.02047.x

I. Introduction

The mean wage gap between male and female workers has been extensively

investigated in the labor economics literature. For example, Altonji and Blank

(1999) use the traditional Oaxaca–Blinder decomposition methodology (Blinder,

1973; Oaxaca, 1973) and find that characteristics such as education, experience

and on-the-job training explain less than one-third of the male–female wage

differential in the USA during 1979–1995 (although the gap has decreased sig-

nificantly over time). The remaining unexplained component is usually regarded

in the published literature as a measure of discrimination against female workers

in the labor market.

Recently,research has shifted its focus from the mean gender differential to the

gender differential pattern along the whole earnings distribution. The quantile

regression methodology, developed by Koenker and Bassett (1978), when com-

bined with the Oaxaca–Blinder decomposition technique allows the characteriza-

tion of gender wage gaps at any point of the wage distribution; this potentially

enables more targeted policy. A growing published literature in this regard has

emerged in over the past decade. For instance, Albrecht et al. (2003) find an

*Zheng Fang: Humanities and Social Sciences, Nanyang Technological University, Singapore

639798. Email: fang0031@ntu.edu.sg. Chris Sakellariou (corresponding author): Humanities and

Social Sciences, Nanyang Technological University,Singapore 639798. Email: acsake@ntu.edu.sg.

Asian Economic Journal 2011, Vol. 25 No. 1, 35–54 35

© 2011 The Authors

Asian Economic Journal © 2011 East Asian Economic Association and BlackwellPublishing Pty Ltd

increasing gender gap along the wage distribution in Sweden, 1998.Arulampalam

et al. (2007) study 11 European countries and find the ‘glass ceiling’ effect

prevailing for most countries. A glass ceiling is an observed gender gap that is

wider at the top of the counterfactual wage distribution compared to the bottom.

Correspondingly, a wider gap at the bottom of the counterfactual wage distribu-

tion is referred to as a ‘sticky floor’.

The many studies that focus on developed economies, such as the USA,

Australia and countries in Europe, generally point to the existence of glass

ceilings; in contrast, emerging studies for South and South-East Asia point mainly

to sticky floors (Sakellariou 2004a; Gunawardena et al. 2008).The present paper

concentrates on Thailand. Using data from the 2004 Thai Household Socio-

Economic Survey, we explore the pattern of the gender wage differential by using

quantile regression and counterfactual decompositions.

The rest of the paper is organized as follows. Section II reviewsthe literature on

gender wage differentials using quantile regression. Section III introduces the

conceptual framework and details the decomposition methodology used.The data

is described in Section IV and results are presented and discussed in detail in

Section 5. Finally, Section 6 provides a summary of the findings.

II. Literature Review

The method of quantile regression was developed by Koenker and Basset (1978)

and has since been applied to many fields in economics.1Buchinsky (1994) was

the first to incorporate quantile regression into the study of the wage structure. An

extensive literature considering wage differentials, for example according to

sector, education or gender, using quantile regression analysis has followed. For

example, Mueller (1998) uses quantile regression to analyze the wage premium in

the public sector in Canada. Highest rent payments are found at the lower tail of

the distribution. Pereira and Martins (2000) study within-educational-level wage

differentials for 15 European countries for the 1980–1995 period and identify four

patterns of impact of education attainment on earnings. López-Nicolás et al.

(2001) examine gender wage discrimination in Spain and report a higher degree

of discrimination against women at the upper tail of wage distribution. Similarly,

Gardeazabal and Ugidos (2005) observe increasing gender discrimination in

absolute terms along the distribution of wages. However, in the Philippines and

Singapore, Sakellariou (2004a,b) determine that the gender wage discrimination

is more severe in the lower segment of wage distribution.

Albrecht et al. (2003) combine the quantile regression technique with the

counterfactual decomposition method proposed by Machado and Mata (2005)

and observe that the gender wage gap accelerates at the top of the wage distribu-

tion; that is, there is a glass ceiling effect. Research on glass ceilings and sticky

1. See Koenker and Hallock (2001) for an excellent survey on the applications of quantile

regression.

ASIAN ECONOMIC JOURNAL 36

© 2011 The Authors

Asian Economic Journal © 2011 East Asian Economic Association and BlackwellPublishing Pty Ltd