The Skill Structure of Export Wage Premium: Evidence from Chinese Matched Employer–Employee Data

• Author: Jianwei Xu,Mi Dai
• Date: 01 May 2017
• DOI: http://doi.org/10.1111/twec.12397
• Published date: 01 May 2017

The Skill Structure of Export Wage

Premium: Evidence from Chinese Matched

Employer–Employee Data

Mi Dai and Jianwei Xu

1

1

Business School, Beijing Normal University, Beijing, China

1. INTRODUCTION

TRADE theory has long emphasised that international trade is associated with substantial

distributional consequences across skill groups. Although a large body of literature has

investigated the impact of trade on the wages across skill groups using industry-level data,

there is less evidence at the firm. On the other hand, recent trade studies based on heteroge-

neous firms have emphasised firm’s heterogeneity in trade participation as an important deter-

minant of labour market outcomes. It is now well documented that exporters pay higher

average wages than non-exporters. However, is this export wage premium equally shared

among different skill groups? In this paper, we provide micro-level evidence to this question

using a newly released matched employer–employee data from China.

There are two reasons why we use Chinese data to investigate the skill structure of export

wage premium. First, the wage inequality between skilled and unskilled workers has long been

regarded as one of the major sources of rising overall wage inequality in developing countries.

However, to what extent international trade has contributed to this rising inequality is still con-

troversial (Goldberg and Pavcnik, 2007). China is the world’s largest developing country and

also the largest exporting country, making it a perfect candidate to study the distributional

effects of exports in developing countries. Second, the wage gap between skilled and unskilled

workers in China has been rising rapidly since China’s accession into the WTO in 2001 (Sheng

and Yang, 2012), suggesting that export expansion may have had a substantial contribution to

China’s recent rising inequality. However, detailed firm-level investigation is still lacking

because wage information by skill groups is rarely available in Chinese firm-level data sets.

The use of matched employer–employee data offers our study several advantages over the

firm-level data sets used by most existing studies on trade and wages. First, most firm-level

data only report average wages or at best distinguish blue-collar/ white-collar workers. There-

fore, they are not able to study how the export wage premium varies across detailed skill

groups defined by more detailed skill dimensions such as education. Second, the inability to

control for individual characteristics in firm-level data sets may also overestimate the export

wage premium if the workforce composition of exporters is more skewed to skilled workers.

We first investigate how export wage premium varies across skill groups, which are

defined by workers with different education. We find that export wage premium varies

We thank David Autor, Eric Verhoogen, Donald Davis, Sandra Poncet, Shi Li, Yang Yao, and workshop

participants in Kobe, IDE, CUFE, Nankai, Wuhan and Beijing Normal University for helpful comments

and suggestions. Financial support from the Key Project of the National Social Science Foundation

(Grant No.14ZDA082), National Science Foundation of China (Grant No. 71303021), the Fundamental

Research Funds for the Central Universities (Grant No. 2012WYB34) and Beijing Social Science

Foundation (No.14JGC100) is gratefully acknowledged. All errors are ours.

©2016 John Wiley & Sons Ltd 883

The World Economy (2017)

doi: 10.1111/twec.12397

The World Economy

systematically with education and is larger for more educated workers. This result is robust to

different control variables, wage measurement and subsamples. Quantitatively, the cross-skill

differences are sizable. In our benchmark specification, the premium for graduates, undergrad-

uates and college education workers is, respectively, 17.6, 13 and 6.5 percentage points larger

than the premium for workers with a junior high school education. We also find higher export

wage premium in occupations with a larger share of educated workers.

We next investigate the possible mechanisms of the larger export wage premium for more

educated workers. We focus on two non-exclusive mechanisms that are emphasised in the recent

trade and labour literature: the size-wage premium mechanism in Helpman et al. (2010) and the

quality upgrading mechanism in Verhoogen (2008). We find that the positive relation between

education and export wage premium is especially pronounced in sectors with higher scope for

quality differentiation. This is consistent with the predictions of the theory of Verhoogen (2008)

that exporters pay relatively higher wages to skilled workers because exporters produce higher

quality products, which require higher quality skilled labour. In contrast, we do not find the size

difference between exporter and non-exporters as emphasised in Helpman et al. (2010) to be

a major determinant of cross-skill differences in export wage premium.

Our paper is related to the large literature on exports and wages. Empirical studies using

firm-level data find that exporters pay higher average wages (Bernard et al., 1995; Frias et al.,

2009; Amiti and Davis, 2011; Krishna et al., 2011). Although most studies on this topic

investigate firm-level average wages, several studies based on matched employer–employee

data have started to uncover how the export wage premium varies across skill groups.

Munch and Skaksen (2008) find that the export wage premium only exists in firms with suffi-

ciently high skill intensity in Denmark. Schank et al. (2007) use data in Germany and find

that white-collar workers earn relatively more in exporting plants compared with blue-collar

workers. Also using German data, Klein et al. (2013) find an exporting wage premium in

high-skilled workers (defined by both education level and occupation) but a wage discount in

low-skilled workers. Our paper contributes to this literature in several aspects. First, most

existing studies on the skill structure of export wage premium are based on developed country

data. In contrast, our analysis is based on data in a developing country. Factor endowment

trade theory suggests that the impact of trade on the skill/unskilled wage gaps is opposite in

developed and developing countries; thus, it is worth investigating whether the skill structure

of export wage premium is different in developed and developing countries. Second, we

explore the possible mechanisms of the cross-skill differences in export wage premium. Our

results provide evidence on the validity of the alternative theories that are recently developed

to explain the relationship between firm’s exporting behaviour and wage inequality.

Our paper is also related to the literature on trade and skill premium, including earlier

works using industry-level data (Feenstra and Hanson, 1996, 1999) and more recent works

using more detailed micro-level data (Verhoogen, 2008; Bustos, 2011; Chen et al., 2013a,

2013b). Harrison et al. (2011) surveyed recent empirical works that use firm-level or matched

worker–firm data. Our paper contributes to this literature by extending the analysis to micro-

level data for China. More broadly, our paper fits into the literature on globalisation and

income inequality (as surveyed by Goldberg and Pavcnik, 2007).

The rest of the paper is organised as follows. Section 2 introduces the theoretical back-

ground for the skill structure of export wage premium. Section 3 describes the data and con-

duct preliminary analysis on the export wage premium. Section 4 examines the skill structure

of the export wage premium. Section 5 discusses the possible channels of the differential

export wage premium across skill groups. The last section concludes.

©2016 John Wiley & Sons Ltd

884 M. DAI AND J. XU