Language Ability or Personality Works?: The Return to Possessing a Global English Test Certificate for College Graduates in China

Abstract

Purpose

—English language skills have great influence on labors’ earnings from the global perspective. To reveal the economic returns to English language ability in Chinese labor market, we investigate how the global English test certificate affects college graduates’ wage.

Design/Approach/Methods

—We adopt the ordinary least squares (OLS) and propensity score matching (PSM) methods, using data from Chinese Education Panel Survey (CEPS).

Findings

—The results indicate that with English test scores controlled, possessing global English test certificates have an additional positive effect on wage premiums whereas domestic English test certificates do not. Therefore for college graduates in China, the act of chasing certificates represents proactivity and is rewarded at the initial employment stage.

Originality/Value

—Our findings imply that global English test has great comprehensive value in labor market: the certificate is not only the important signal to students’ English language, but the crucial indicator to one's productivity personality which is rewarded at the initial employment stage.

Keywords

English language ability College Entrance Examination score wage premium human capital proactivity

1. Introduction

Language ability is one of the most important parts of both human intelligence (Gardner, 1983) and human capital (Chiswick & Miller, 2003). The ability of speaking a foreign language, especially English, is a valuable skill for a person from non-English countries. Numerous studies have tested the positive effect of English language proficiency and the economic return in labor market for English-speaking countries’ immigrants (Chiswick & Miller, 1995; Dustmann & Fabbri, 2003; Yao & van Ours, 2015). Furthermore, English-speaking ability is significantly rewarded many non-English countries, including European (Ginsburgh & Prieto-Rodriguez, 2011) and Asian (Choi, 2015; Lang & Siniver, 2009) countries. Although English language proficiency is without a doubt closely related to earnings, nearly all evidence is based on workers’ self-evaluation of English in surveys, in addition to the testing scores of language certificates issued by authorities. The self-evaluation of language, typically measured as basic, regular, or advanced, causes ambiguity; therefore, we cannot clearly define either the capacity gap between the different evaluation levels or the equivalence relation between evaluations from different workers. Nevertheless, it is important to relate the objective English ability with earnings in measurable and comparable ways. However, few studies have focused on this point.

In this study, the objective measures of English language, such as English language testing scores and English certificates, from both domestic and global authorities are adopted to establish the language-wage relationship for college graduates in China's labor market. In this context, language test scores are considered a well-defined indicator for a worker's language ability. In addition, the pursuit of language certificates, namely, the act of choosing and participating in tests, may indicate one's other non-cognitive characteristics besides language ability. Thus, after considering and controlling workers’ language ability, are the certificates still worthy in the Chinese labor market?

China, the world's second largest economy with the highest total volume of global import and export, has the world's largest population of English learners, approximately 390 million (Wei & Su, 2012). In China's education system, students have to learn English from primary school to university. English language is one of the three national primary courses and tested in the national College Entrance Examination (CEE). At the college level, Chinese students still have to learn English; they are obliged to undertake the College English Test (CET) and pass it before obtaining the undergraduate degree. Furthermore, some college students voluntarily choose and obtain global English test certificates, such as TOFEL and IELTS, irrespective of whether they are interested in applying for further studies abroad. English test scores or grade certificates are usually stressed in the resume of graduates who are seeking jobs in the domestic labor market. The abovementioned English learning context in Chinese universities provides good conditions to evaluate the effect of English proficiency on earnings, including the economic effect of English ability and English qualification certificates and the comparison effect between different types of English tests in the domestic labor market.

Furthermore, the choice of taking a global English test and obtaining similar certification is self-selected. In this study, we employ several methods to solve the problem of endogeneity. The panel data of survey of college graduates is used. The main results indicate that English language ability measured by CET score has a positive effect on the starting salaries of graduates. In addition, having international test of English certificates has a steady strong marginal wage premium effect; however, the effect of CET excellence qualification certificate is unobvious. We also observe that the English certificate effect occurs only at the stage of starting salary and diminishes at the wage growth stage. We deduce that the effect of Global English Certificate stems from a proactive personality and works through the process of job search and application. These results enhance our understanding of the worth of language mastery in the labor market.

The paper is organized as follows: Section 2 provides the background of English learning and tests in Chinese universities; section 3 presents a review of the literature; section 4 discusses the study methodology; section 5 presents the data, while section 6 presents the empirical results; the robustness test and causal relationship are examined in section 7.

2. Background of English Learning in China

As English has become the lingua franca of international business, English language ability has become a valuable global commodity, particularly in non-English-speaking countries.

In China, educational policymaking has seen tremendous changes in the past half-century. Since 1978, with the implementation of the open-door policy and the drive for modernization and internationalization, there have been a number of official measures for promoting foreign language education, especially for English study (Gil, 2016; Hu, 2005).

For Chinese students, English language learning begins at the primary school level; most students learn English as one of the three compulsory subjects and their English ability is tested in the Chinese national CEE, regardless of whether a student has chosen the science or humanity track in high school. After entering college, students who do not major in English still have to study English and pass the English tests. The main types of English language tests for Chinese college students are called CET driven by the National College English Testing Committee, a public testing service organization (Zheng & Cheng, 2008). CET is divided into two different level tests, CET4 and CET6. The total score of CET4 and CET6 is 710. If students score above the cutoff line of 550, they have the option of taking the CET-Spoken English Test (CET-SET) and obtain the excellence of English ability certificate. The Chinese undergraduate students in domestic 4-year colleges have to undertake CET band 4 (CET4) test and score more than 425. The CET4 tests are held twice a year at the end of each semester. Currently, CET4 is the world's largest English as a Foreign Language (EFL) test (Jin & Yang, 2008).

Other types of tests are organized by foreign private test organizations, namely GRE, TOFEL, and IELTS. These tests are executed at a global scale in non-English-speaking countries and considered a means for proving that one's English proficiency can meet the requirements of universities in the U.S.A., U.K., Canada, and other English-speaking countries. The fee for attempting these tests is more than ten times of the previous one (CET). However, more and more students are voluntarily obtaining these global English test certificates, regardless of whether they intend to apply for further studies abroad.

Although there are differences in the English tests, the basic elements of language ability are emphasized in all of the tests. In this study, we considered that the CET4 scores represent the basic English language ability of students, since these scores are comparable. Furthermore, the optional participation in CET-SET and global English tests are treated as the effort of obtaining English language qualification certificates.

3. Literature Review: Wage Premium Effect of English as a Foreign Language and Its Explanations

Language ability is acquired through learning; it is considered an important aspect in human capital investment. The concept of language capital was proposed in the framework of human capital theory (Casey & Dustmann, 2008; Chiswick & Miller, 1995). Initial research on economic returns of English language ability has focused on immigrants in English-speaking countries. The results suggest that there is a positive link between English proficiency and individual's earning in the U.S.A. and U.K. (Chiswick, 1991; Chiswick & Miller, 2002; Kossoudji, 1988). English is also crucial for wage premiums in the labor markets of some non-English-speaking countries, such as Germany (Stöhr, 2015), Israel (Lang & Siniver, 2009), India (Azam, Chin, & Prakash, 2013), and Korea (Choi, 2015).

The relationship between higher English language proficiency and higher earnings can be explained in a direct and indirect way. Indirectly, English fluency could serve as a medium that helps workers transfer their academic knowledge or skills to the production process, thereby enhancing labor productivity (Park, 1999). Directly, English language represents a mechanism of achieving more prestigious occupations and meeting job requirements (Aldashev, Gernandt, & Thomsen, 2009; Chiswick & Miller, 2010); English ability can help workers establish global trade links with English countries for business jobs (Ku & Zussman, 2010; Melitz, 2008); English language competency is representative of workers’ general cognitive skills for employers (Leikin, 2013).

As a special group, college graduates are concerned with labor economics. Although the main stream literature has measured language ability through self-evaluation, some studies, aware of potential bias of overestimated or under-estimated language ability (Akresh & Frank, 2011), have used objective test scores or certificates as effective proxy variables for individuals’ English language ability, such as the ACT sub-test scores in the U.S.A. (Bettinger, Evans, & Pope, 2011) and CEE sub-test scores (Yang, 2014) or CET scores (Guo & Sun, 2014) in China. Certain studies have focused on college students’ English ability, achievement, and wage, revealing that in Korea, passing the English tests of ETS has substantial positive effects on getting a good job and earning higher wages (Choi, 2015). In China, it is significant that CET scores have positive relations with graduates’ starting salaries (Guo & Sun, 2014). Thus, the literature tends to indicate that English ability indicates the cognitive ability of workers, since students with higher English test scores have higher GPAs (Cho & Bridgeman, 2012; Wait & Gressel, 2009).

Although the estimation of English language ability using test scores has shown to be closely related with wages, little is known about the effects of English qualification certificates. In addition, explanations about the English certificate's effect are not mature. If we adopt the explanation framework of human capital theory, introduced by Mincer (Mincer, 1974), or signaling theory, introduced by Spence (Spence, 1973), we treat the English test scores only as the indicator of one's cognitive ability. This effect is famously known by another classical name: sheepskin effect (Hungerford & Solon, 1987; Jaeger & Page, 1994). Nevertheless, a student voluntarily participating in an exam and then earning scores represent two different stages. The proactive selection, which might has been ignored by both traditional human capital theory and signal theory, may become a key point in representing one's characteristics, including motivation or attitude, which affect job search and wage premiums.

The proactivity theory provides a new explanation clue. Proactive personality is first defined as “the relatively stable tendency to effect environmental change” (Bateman & Crant, 1993, p. 103). Individuals with proactive personalities tend to act in advance with foresight about future events before they occur and are explicitly eager to make a difference (Grant & Ashford, 2008). Abundant empirical research has indicated that proactivity is a very good indicator for predicting one's professional success (Crant & Bateman, 2000; Erdogan & Bauer, 2005; Seibert, Crant, & Kraimer, 1999; Seibert, Kraimer, & Crant, 2001). However, few studies have addressed proactivity and wage effects among graduate students; a study revealed that proactive personality significantly influences the success of college graduates’ job search and subsequently their wage premiums (Brown, Cober, Kane, Levy, & Shalhoop, 2006). To learn a second language, or take part in an optional language test, is motivated by some unobserved reasons. In this study, English test engagement can be regarded as a proactive action that reflects one's proactive personality. After controlling for the observable human capital, the learning of a foreign language has positive correlation with 2%–3% wage premiums for college graduates in the U.S.A. (Saiz & Zoido, 2005). The mechanism is attributed to motivation. However, is it suitable to treat the act of participating in global English tests as a proactive action? How does this proactive action function in acquiring higher wages? These issues are discussed in the following section.

In this study, we advance the previous studies in three different ways: first, we compare the effects of different subjective tests on graduates’ economic returns, which could be a global concern but has not been investigated; second, we not only testify the sheepskin effect of the international English certificates but also indicate a different mechanism of the effect, instead of adopting the view of the human capital theory and signaling theory; and third, we attempt to detect the causal relationship between global English tests and wages.

4. Empirical Strategy

4.1 Basic Regression Model

Based on previous studies about the economic returns of language ability and college graduates’ starting wages (Chiswick & Miller, 1995; Li, Meng, Shi, & Wu, 2012), this study controls for individual employees’ factors of gender and health, family factors of parents’ income, education factors of major type, and employers’ factors of ownership structure. The linear regression model adopted in our study is similar to that proposed in the aforementioned literature. We focus on the discussion of whether there are additional returns for having global English test certificates. We assume that the earning function of college students after graduation is determined by the following equation:

L n W_{i} = β_{0} + β_{1} c e r t i f i c a t e_{i} + β_{4} C E E t o t a l_{i} + X_{i} Y + ε_{i}

LnW_i is the logarithm of monthly income of starting salary for student i. The variable certificate_i, which stands for whether a student has global English test certificates, such as TOFEL, GRE, or IELTS, represents the student's English ability. CEEtotal_i, which stands for the total score of College Entrance Examination in China, serves as a proxy variable for the student's general ability. X_i represents multiple variables, including individual factors other than CEE scores, family factors, education factors, and employers’ factors, that are all controlled to reduce bias. In this study, the OLS¹ method is used to estimate the coefficients.

The study sample comprised students from different provinces in China with different CEE scores, which represented by ability may not be comparable. In order to solve this problem, two conversions were performed. First, we changed different scores into 750 system points, and Chinese language, mathematics, and foreign language subjects were divided into 150 original score systems. This deformation unified the students’ fractional range. Then, we calculated the Z-scores on students’ score, which was considered the standard score; it provided the relative rank of each student's achievement in the sample. For the College English Test scores, we simply calculated Z-scores of students with the same grade.

4.2 Endogeneity and Sample Selection

The OLS estimation using observational data could have selection bias problem (Rosenbaum & Rubin, 1983). In an observation study, the assignment to treatment or non-treatment group is not random. Since in this study we analyzed observational data, this could cause bias in the OLS estimation of the treatment effect. The inverse probability weighting method is often used to solve this problem (Wooldridge, 2002); this method first attains the probability of assignment into treatment group, which is calculated by a logistic regression on covariates affecting the assignment. Thereafter, it estimates separate outcome regression models for subjects in each treatment level with weighting. The weight is the inverse of the estimated probability that a subject receives a treatment. The treatment effect is the difference of estimated weighted outcomes at different treatment levels. Two specific methods using inverse probability weight are often used: augmented inverse probability weighted (AIPW) estimation and inverse probability weighted regression adjustment (IPWRA) estimation. Both estimation methods are doubly robust (Robins & Rotnitzky, 1995; Wooldridge, 2007), that is, only one of the treatment models and outcome models must be specified correctly to consistently estimate the treatment effect. In this study, we used both AIPW and IPWRA to estimate the average treatment effect (ATE) of having a global English test certificate in order to check the robustness of the result from OLS estimation.

We also addressed the sample selection bias resulting from wage missing values by applying Heckman's correction (Heckman, 1979). Some college graduates did not enter labor market after graduation; rather, they pursued advanced studies in China or abroad. Therefore, we estimated the Mills Lambda to correct the model.

5. Data

The study utilized longitudinal data from the Chinese Education Panel Survey (CEPS). The survey, which began in 2009, sampled all undergraduates in public universities in Beijing and applied three-order random sampling of universities, majors, and students. In total, 5,100 undergraduates with 255 different majors in 15 universities were selected. These undergraduates were from the 2006 and 2008 cohort. Between 2010 and 2013, this group of students completed four rounds of follow-up surveys. Regarding the 2006 cohort students, the data collection extended from the third grade in college to three years after graduation. For the 2008 cohort, the data collection extended from the first grade in college to one year after graduation. The data achieved high success rate of retention. The analysis of 2006 students suggests that the success rate of follow-up was 78.46% in the second year after graduation, and that there was no significant difference in the loss of students with different sexes, nationalities, specialties, and schools. In this study, the starting salary was the wages in the first year after graduation. In the database, 902 college graduates for cohort 2006 and 573 for cohort 2008 entered the labor market immediately after acquiring their bachelor's degree.

Table 1 provides a descriptive summary of the dependent and independent variables involved. The dependent variable, Ln wage, represents the logarithm of wages. The individual factors include variables describing students’ individual characteristics. Global English Certificate measures whether a student has obtained global English test certificates, such as TOFEL and IELTS. “CEE” is a sum of Chinese, Mathematics, and English scores in the CEE of China. Chinese college students still have to learn English for one or two years and pass the national CET before receiving a bachelor's degree. The CEE and CET scores are listed. CET4 excellence certificate measures whether a student's CET4 score is above the cutoff line of 550. Science track measures whether a student chooses to study liberal arts or science in high school. If the student chooses to learn science, this variable equals 1. For “health”, each student rates the condition of his health on a scale of 0 to 100 in the questionnaire. Male equals 1 in the gender variable. Family factors describe the family status of a student. “Urban” measures whether a student comes from city or countryside; if the student is from the city, this variable equals 1. Family income stands for the parents’ monthly expenditures for the student. Education factors include three variables. “211” college represents the colleges selected in the “211” project, a national project to establish world-class universities. Of the 3,000 colleges in China, 112 have been called “211” colleges. Therefore, “211” colleges can be seen as a designator of high quality education. Students’ academic ranking in college is measured by “academic ranking”, which is self-identified by students. “Major type” measures whether a student chooses to study liberal arts or science at the university level; if one chooses to learn science, technology, engineering, or mathematics (STEM), this variable equals 1. Many studies have shown that college students who major in STEM receive higher wages after graduation (Bostwick, 2016; Mann & DiPrete, 2013). Due to the labor market segmentation, employers’ factors are controlled for. The ownership type and industry type of a company could affect a worker's wage. In total, there are 10 types of social enterprise ownership structures: (1) government organizations, (2) schools, (3) research institutions, (4) other public institutions, (5) central enterprises, (6) state-owned enterprises, (7) collective enterprises, (8) private enterprises, (9) Sino-foreign joint ventures, and (10) self-employment. The following 19 types of employment industries are also controlled for: (1) farming, forestry, animal husbandry, and fishery, (2) mining and quarrying, (3) manufacturing, (4) electric power, hot power and water, (5) architecture, (6) transportation, (7) information technology, (8) retailing, (9) catering, (10) finance, (11) real estate, (12) business and service, (13) geologic exploration, (14) hydraulic engineering, (15) resident service, (16) education, (17) social security, (18) culture and recreation, (19) public administration.

Table 1.
Summary of Variables.

Variable Obs Mean Std. Dev Min Max

Independent variable Dependent variable Wage 1,504 4,541.221 3,799.139 0 60,000

Ln wage 1,496 8.249 562 0.632 285 0 11.002 1

Global English Certificate 1,466 0.180 081 9 0.384 386 7 0 1

English Test factors CET4 excellence certificate 1,466 0.261 937 2 0.439 838 7 0 1

CET4 score 1,355 481.575 6 67.540 04 256 697

CEE total score 1,464 553.680 8 71.103 02 91 750

Individual factors Science track 1,385 0.761 733 0.426 177 0 1

Male 1,504 0.496 011 0.500 15 0 1

Health 1,504 76.830 45 13.469 21 7 100

Family factors Family income 1,494 1,274.218 14,234.06 0 400,000

Urban 1,501 0.673 551 0.469 07 0 1

University factors “211” college 1,504 0.474 734 0.499 527 0 1

Academic ranking 1,502 0.480 034 0.216 318 0.02 1

Major type (Science) 1,504 0.493 351 0.500 122 0 1

Company factors Ownership type of enterprise 1,442 6.274 619 2.444 562 1 10

Employment industry 1,335 10.165 54 5.359 38 1 19

	Variable	Obs	Mean	Std. Dev	Min	Max
Independent variable	Dependent variable	Wage	1,504	4,541.221	3,799.139	0	60,000
Ln wage	1,496	8.249 562	0.632 285	0	11.002 1
	Global English Certificate	1,466	0.180 081 9	0.384 386 7	0	1
English Test factors	CET4 excellence certificate	1,466	0.261 937 2	0.439 838 7	0	1
CET4 score	1,355	481.575 6	67.540 04	256	697
CEE total score	1,464	553.680 8	71.103 02	91	750
Individual factors	Science track	1,385	0.761 733	0.426 177	0	1
Male	1,504	0.496 011	0.500 15	0	1
Health	1,504	76.830 45	13.469 21	7	100
Family factors	Family income	1,494	1,274.218	14,234.06	0	400,000
Urban	1,501	0.673 551	0.469 07	0	1
University factors	“211” college	1,504	0.474 734	0.499 527	0	1
Academic ranking	1,502	0.480 034	0.216 318	0.02	1
Major type (Science)	1,504	0.493 351	0.500 122	0	1
Company factors	Ownership type of enterprise	1,442	6.274 619	2.444 562	1	10
Employment industry	1,335	10.165 54	5.359 38	1	19

6. Results

6.1 Wage Premium of Global English Test Certificates

Table 2 displays the effects of the Global English test certificate on starting salaries. In this study, Model (1) was the basic regression model. Models (2)—(4) added the control factors of family, university, and enterprise separately. Model (5) added all the control variables. In Model (5), we observed that the Global English test certificate increases the starting wage by about 20.0%. The results of Model (6) were obtained after Heckman's correction, which indicates that the sample selection bias is not obvious. The results of the Heckman model were similar to that of Model (5), implying that the Global English test certificate increases the starting salary of graduates.

Table 2.
Effects of English Ability on Starting Salaries.

VARIABLES (1) Ln(wage) (2) Ln(wage) (3) Ln(wage) (4) Ln(wage) (5) Ln(wage) (6) Ln(wage)

Global English Certificate 0.238^* (0.041 7) 0.200^* (0.043 3) 0.228^* (0.041 7) 0.236^* (0.045 6) 0.200^* (0.047 2) 0.199^* (0.047 4)

CEE total score 0.001 99^* (0.000 293) 0.001 82^* (0.000 300) 0.001 98^* (0.000 315) 0.001 57^* (0.000 319) 0.001 63^* (0.000 349) 0.001 63^* (0.000 350)

Science track 0.030 9 (0.038 5) 0.047 5 (0.038 6) 0.092 1^ (0.044 9) 0.062 0 (0.043 3) 0.114^ (0.050 0) 0.110^ (0.050 6)

Period 0.273^* (0.032 0) 0.249^* (0.033 5) 0.280^* (0.031 9) 0.232^* (0.035 0) 0.222^* (0.036 9) 0.224^* (0.037 1)

Male 0.019 2 (0.032 6) 0.016 5 (0.032 7) 0.076 0^ (0.035 1) 0.038 8 (0.036 1) 0.085 2^ (0.039 0) 0.083 9^ (0.039 1)

Health 0.001 04 (0.001 18) 0.001 05 (0.001 19) 0.000 600 (0.001 18) 0.002 51* (0.001 28) 0.001 91 (0.001 31) 0.001 85 (0.001 31)

Urban -0.029 2 (0.041 0) -0.036 2 (0.044 2) -0.033 9 (0.044 4)

Ln(family income) 0.074 7^* (0.019 6) 0.059 9^* (0.021 9) 0.059 5^* (0.022 0)

“211” college 0.007 37 (0.043 0) -0.051 2 (0.048 2) -0.049 2 (0.048 3)

Academic ranking -0.260^* (0.079 7) -0.264^* (0.087 3) -0.264^* (0.087 4)

Major type (Science) -0.101^*** (0.038 7) -0.082 7^* (0.045 6) -0.082 9^* (0.045 8)

Ownership type of enterprise Yes Yes Yes

Employment industry Yes Yes Yes

Province Yes Yes Yes Yes Yes Yes

Constant 6.859^* (0.184) 7.080^* (0.320) 6.999^* (0.195) 6.466^* (0.232) 6.868^* (0.365) 6.876^* (0.365)

Lambda Yes −0.549 (0.805)

Observations 1,313 1,293 1,312 1,137 1,118 1,120

R-squared 0.148 0.167 0.160 0.218 0.235

VARIABLES	(1) Ln(wage)	(2) Ln(wage)	(3) Ln(wage)	(4) Ln(wage)	(5) Ln(wage)	(6) Ln(wage)
Global English Certificate	0.238^*** (0.041 7)	0.200^*** (0.043 3)	0.228^*** (0.041 7)	0.236^*** (0.045 6)	0.200^*** (0.047 2)	0.199^*** (0.047 4)
CEE total score	0.001 99^*** (0.000 293)	0.001 82^*** (0.000 300)	0.001 98^*** (0.000 315)	0.001 57^*** (0.000 319)	0.001 63^*** (0.000 349)	0.001 63^*** (0.000 350)
Science track	0.030 9 (0.038 5)	0.047 5 (0.038 6)	0.092 1^** (0.044 9)	0.062 0 (0.043 3)	0.114^** (0.050 0)	0.110^** (0.050 6)
Period	0.273^*** (0.032 0)	0.249^*** (0.033 5)	0.280^*** (0.031 9)	0.232^*** (0.035 0)	0.222^*** (0.036 9)	0.224^*** (0.037 1)
Male	0.019 2 (0.032 6)	0.016 5 (0.032 7)	0.076 0^** (0.035 1)	0.038 8 (0.036 1)	0.085 2^** (0.039 0)	0.083 9^** (0.039 1)
Health	0.001 04 (0.001 18)	0.001 05 (0.001 19)	0.000 600 (0.001 18)	0.002 51* (0.001 28)	0.001 91 (0.001 31)	0.001 85 (0.001 31)
Urban		-0.029 2 (0.041 0)			-0.036 2 (0.044 2)	-0.033 9 (0.044 4)
Ln(family income)		0.074 7^*** (0.019 6)			0.059 9^*** (0.021 9)	0.059 5^*** (0.022 0)
“211” college			0.007 37 (0.043 0)		-0.051 2 (0.048 2)	-0.049 2 (0.048 3)
Academic ranking			-0.260^*** (0.079 7)		-0.264^*** (0.087 3)	-0.264^*** (0.087 4)
Major type (Science)			-0.101^*** (0.038 7)		-0.082 7^* (0.045 6)	-0.082 9^* (0.045 8)
Ownership type of enterprise				Yes	Yes	Yes
Employment industry				Yes	Yes	Yes
Province	Yes	Yes	Yes	Yes	Yes	Yes
Constant	6.859^*** (0.184)	7.080^*** (0.320)	6.999^*** (0.195)	6.466^*** (0.232)	6.868^*** (0.365)	6.876^*** (0.365)
Lambda						Yes −0.549 (0.805)
Observations	1,313	1,293	1,312	1,137	1,118	1,120
R-squared	0.148	0.167	0.160	0.218	0.235

Standard errors in parentheses;

***

p<0.01

p<0.05

p<0.1.

Other variables also affected the starting salary. As a representative of personal ability, CEE total scores had a significant positive influence. From the perspective of cohort, the starting salary of cohort 2008 was generally higher than that of cohort 2006 by approximately 20% because of economic development, which shows changes in graduate wages in the labor market for two years. From the gender perspective, the males’ starting salary was higher than that of females by about 8%. Family income and academic ranking both had a positive influence; their coefficients were 0.036 and 0.264, respectively. Finally, from the perspective of majors, salaries of social sciences majors were higher than those of the natural sciences by 8%. This is probably because most graduates in the social sciences enter into lucrative fields like finance and management.

In order to further examine the effect of the Global English test certificate on starting salary, Table 3 presents the results of quartile regression and Table 4 shows the regression result by subsamples. Overall, the higher the wage tier, the stronger influence of the Global English test certificate, and the coefficients of the Global English test certificate were all significant at different wage tiers.

Table 3.

Quartile Regression on the Effects of English Certificate on College Graduates’ Starting Salaries.

VARIABLES	(1)	(2)	(3)	(4)	(5)	(6)
VARIABLES		10%	25%	50%	75%	90%
Global English Certificate	0.200^*** (0.047 2)	0.107^* (0.062 7)	0.105^** (0.044 0)	0.104^*** (0.035 0)	0.174^*** (0.052 2)	0.335^*** (0.064 2)
CEE total score	0.001 63^*** (0.000 349)	0.002 06^*** (0.000 464)	0.001 60^*** (0.000 325)	0.001 65^*** (0.000 259)	0.001 48^*** (0.000 386)	0.001 22^** (0.000 475)
Science track	0.114^** (0.050 0)	0.061 0 (0.066 3)	0.012 0 (0.046 6)	0.079 3^** (0.037 1)	0.095 5^* (0.055 3)	0.095 0 (0.068 0)
Period	0.222^*** (0.036 9)	0.253^*** (0.048 9)	0.219^*** (0.034 3)	0.262^*** (0.027 3)	0.225^*** (0.040 8)	0.227^*** (0.050 2)
Male	0.085 2^** (0.039 0)	0.119^** (0.051 8)	0.070 7^* (0.036 3)	0.111^*** (0.028 9)	0.091 5^** (0.043 2)	0.136^** (0.053 1)
Health	0.001 91 (0.001 31)	0.000 459 (0.001 74)	-0.000 192 (0.001 22)	-0.000 227 (0.000 971)	-0.000 875 (0.001 45)	0.001 54 (0.001 78)
Urban	-0.036 2 (0.044 2)	-0.079 3 (0.058 6)	-0.045 0 (0.041 1)	-0.040 5 (0.032 8)	-0.055 9 (0.048 9)	-0.042 0 (0.060 1)
Ln(family income)	0.059 9^*** (0.021 9)	0.065 7^** (0.029 1)	0.067 8^*** (0.020 4)	0.064 1^*** (0.016 2)	0.065 3^*** (0.024 2)	0.085 2^*** (0.029 8)
“211” college	-0.051 2 (0.048 2)	-0.083 1 (0.064 0)	-0.054 1 (0.044 9)	-0.005 91 (0.035 8)	-0.034 3 (0.053 3)	-0.014 3 (0.065 6)
Academic ranking	-0.264^*** (0.087 3)	-0.224^* (0.116)	-0.145^* (0.081 2)	-0.130^** (0.064 7)	-0.124 (0.096 5)	-0.251^** (0.119)
Major type (Science)	-0.082 7^* (0.045 6)	-0.035 9 (0.060 5)	-0.074 5^* (0.042 4)	-0.107^*** (0.033 8)	-0.093 4^* (0.050 4)	-0.090 0 (0.062 0)
Ownership type of enterprise	Yes	Yes	Yes	Yes	Yes	Yes
Employment industry	Yes	Yes	Yes	Yes	Yes	Yes
Province	Yes	Yes	Yes	Yes	Yes	Yes
Constant	6.868^*** (0.365)	6.764^*** (0.484)	6.984^*** (0.339)	6.943^*** (0.270)	6.841^*** (0.403)	7.358^*** (0.496)
Observations	1,118	1,118	1,118	1,118	1,118	1,118
R-squared	0.235

Standard errors in parentheses;

***

p<0.01

p<0.05

p<0.1.

Table 4.

Effects of English Certificates on College Graduates’ Starting Salaries (Subsamples).

VARIABLES	(1) Work in Beijing	(2) Work in other district	(3) Work in foreign enterprise	(4) Work in state-owned enterprise	(5) Work in government organs	(6) Work in financial sector	(7) Work in IT sector	(8) Male	(9) Female	(10) Major in Science and Engineering	(11) Major in Humanities and Social Science
Global English Certificate	0.123^** (0.051 2)	0.341^*** (0.108)	0.260^* (0.151)	0.189^** (0.087 5)	0.105 (0.107)	0.179 (0.1 76)	0.201^* (0.108)	0.257^*** (0.075 5)	0.110^* (0.060 0)	0.220^*** (0.062 4)	0.131^* (0.076 8)
CEE total score	0.001 08^*** (0.000 368)	0.002 82^*** (0.000 900)	0.004 42^*** (0.001 42)	0.001 71^*** (0.000 532)	0.001 33^* (0.000 774)	0.002 61 (0.001 59)	0.001 93^** (0.000 848)	0.002 04^*** (0.000 547)	0.001 23^*** (0.000 461)	0.001 78^*** (0.000 593)	0.001 67^*** (0.000 498)
Science track	0.095 9^* (0.052 0)	0.265^** (0.122)	0.161 (0.170)	-0.017 1 (0.100)	0.242^** (0.099 3)	0.051 6 (0.143)	0.099 9 (0.145)	0.061 8 (0.102)	0.136^** (0.054 3)	0.074 2 (0.384)	0.117^** (0.058 4)
Period	0.217^*** (0.038 9)	0.162^* (0.087 9)	0.496^*** (0.133)	0.222^*** (0.066 5)	0.210^** (0.082 2)	0.308^** (0.1 34)	0.274^*** (0.071 9)	0.197^*** (0.061 0)	0.243^*** (0.045 5)	0.166^*** (0.048 6)	0.276^*** (0.059 7)
Male	0.068 5^* (0.040 7)	0.103 (0.096 7)	0.133 (0.136)	-0.015 3 (0.072 2)	-0.006 92 (0.089 9)	-0.113 (0.136)	0.124 (0.084 3)			0.078 9 (0.052 1)	0.087 7 (0.063 3)
Health	-0.002 34^* (0.001 41)	0.010 4^*** (0.003 02)	0.016 9^*** (0.004 39)	-3.45e-05 (0.002 35)	0.002 99 (0.003 11)	-0.000 596 (0.004 82)	-0.000 338 (0.002 99)	0.005 27^*** (0.002 04)	-0.003 52^** (0.001 74)	0.000 779 (0.001 67)	0.002 38 (0.002 20)
Urban	-0.121^** (0.048 5)	0.131 (0.100)	0.274 (0.170)	-0.027 0 (0.079 5)	-0.160 (0.109)	-0.205 (0.166)	0.015 0 (0.095 6)	-0.063 4 (0.069 2)	-0.080 6 (0.058 9)	-0.076 2 (0.055 2)	0.004 38 (0.077 1)
Ln(family income)	0.062 2^** (0.024 6)	0.085 8^* (0.046 3)	-0.028 6 (0.073 6)	0.062 0 (0.042 0)	0.101^** (0.049 5)	0.060 1 (0.083 3)	0.041 2 (0.046 4)	0.069 5^* (0.035 5)	0.062 5^** (0.028 3)	0.080 0^*** (0.028 3)	0.019 6 (0.037 0)
“211” college	0.016 4 (0.054 4)	-0.058 6 (0.106)	-0.278^* (0.162)	-0.041 1 (0.077 3)	0.047 4 (0.108)	0.117 (0.205)	-0.096 5 (0.101)	-0.075 7 (0.078 9)	-0.012 3 (0.059 5)	-0.127^* (0.072 0)	-0.025 8 (0.075 0)
Academic ranking	-0.040 7 (0.094 1)	-0.621^*** (0.203)	-0.883^*** (0.313)	-0.051 4 (0.163)	-0.011 7 (0.231)	0.179 (0.347)	-0.528^*** (0.185)	-0.406^*** (0.136)	-0.102 (0.116)	-0.163 (0.116)	-0.375^*** (0.141)
Major type (Science)	-0.127^*** (0.047 8)	-0.115 (0.117)	0.174 (0.161)	-0.132 (0.082 7)	-0.192^* (0.105)	-0.068 8 (0.168)	0.019 5 (0.109)	-0.059 0 (0.076 3)	-0.095 5^* (0.056 1)
Ownership type of enterprise	Yes	Yes				Yes	Yes	Yes	Yes	Yes	Yes
Employment industry	Yes	Yes	Yes	Yes				Yes	Yes	Yes	Yes
Province	Yes	Yes				Yes	Yes	Yes	Yes	Yes	Yes
Constant	7.228^*** (0.31 3)	5.432^*** (0.760)	4.175^*** (1.034)	7.209^*** (0.658)	6.620^*** (0.588)	7.535^*** (1.184)	7.244^*** (0.773)	6.469^*** (0.539)	7.563^*** (0.562)	6.518^*** (0.593)	7.030^*** (0.453)
Observations	775	343	168	364	74	169	201	571	547	591	527
R-squared	0.272	0.406	0.479	0.150	0.607	0.312	0.480	0.274	0.330	0.295	0.278

Standard errors in parentheses;

***

p<0.01

p<0.05

p<0.1.

For graduates who work in Beijing, the Global English test certificate exerted a significant influence, while for graduates who leave Beijing, the Global English test certificate was still significant but higher than in Beijing. From the view of the ownership type of enterprise, the Global English test certificate had a significant effect for graduates who work in state-owned enterprises and foreign enterprises. Furthermore, the coefficients of the Global English test certificate were insignificant for graduates who work in government organizations. From the perspective of employment, the coefficient of the Global English test certificate was significant for graduates who work in the IT sector, and non-significant for those in the financial sector. As for gender, the coefficients of the Global English test certificate in the male subsample were larger than those in the female subsample. In terms of majors, the coefficients of the Global English test certificate were both significant and positive for students with different majors.

Therefore, we observed that the Global English test certificate does not play an important role but different roles in determining starting salaries.

We added the variables related to English language ability to the regression to further explore the mechanism of the Global English test certificate. The results are presented in Table 5. The estimates of Model (2) suggest that even for the English ability, which is represented by domestic English test CET4 score controlled, the effect of possessing a global English test certificate still had a positive effect. Furthermore, in Model (3), the CET excellence certificate did not affect either wages or the Global English test certificate, but the coefficient was much smaller than the Global English test of Model (1). However, does the Global English test certificate continue to affect wage premium after several years? To address this issue, Models (4) and (5) compared the third year wages and the wage growth in three years for the 2006 cohort. The results indicate that the Global English test certificate only exercises a positive impact on starting wages and does not influence the wage after employment.

Table 5.

Comparison between Global and Domestic Certificates.

VARIABLES	(1) Ln(wage)	(2) Ln(wage)	(3) Ln(wage)	(4) Ln(wage) in third year	(5) Ln(wage growth)
Global English Certificate	0.193^*** (0.047 5)	0.135^*** (0.045 5)		0.061 7 (0.064 5)	0.013 8 (0.152)
CET4 score		0.001 58^*** (0.000 303)		0.001 33^*** (0.000 501)	0.003 06^*** (0.001 05)
CET4 certificate			0.102^** (0.042 7)
Science track	0.124^** (0.050 2)	0.080 9^* (0.048 7)	0.125^** (0.050 5)	-0.035 4 (0.079 5)	-0.103 (0.165)
Period	0.208^*** (0.036 8)	0.192^*** (0.035 4)	0.194^*** (0.037 0)
Male	0.084 7^** (0.039 4)	0.132^*** (0.038 5)	0.099 7^** (0.039 6)	0.194^*** (0.056 9)	0.244^** (0.120)
Health	0.001 35 (0.001 32)	-0.000 327 (0.001 27)	0.001 25 (0.001 33)	-0.007 60^*** (0.002 43)	-0.004 63 (0.004 61)
Urban	-0.028 3 (0.045 0)	-0.084 2^** (0.042 6)	-0.031 8 (0.045 3)	-0.107 (0.065 4)	-0.156 (0.132)
Ln(family income)	0.080 8^*** (0.022 0)	0.082 5^*** (0.021 2)	0.087 5^*** (0.022 0)	0.088 0^*** (0.030 5)	0.142^** (0.062 8)
“211 college”	0.032 9 (0.045 4)	0.020 3 (0.044 1)	0.034 2 (0.045 6)	0.106 (0.066 8)	0.191 (0.149)
Academic ranking	-0.273^*** (0.087 4)	-0.111 (0.086 9)	-0.284^*** (0.087 8)	-0.109 (0.135)	-0.255 (0.267)
Major type (Science)	-0.102^** (0.045 8)	-0.087 5^** (0.044 0)	-0.099 2^** (0.046 1)	-0.088 4 (0.070 7)	-0.018 5 (0.142)
Ownership type of enterprise	Yes	Yes	Yes	Yes	Yes
Employment industry	Yes	Yes	Yes	Yes	Yes
Province	Yes	Yes	Yes	Yes	Yes
Change job				-0.064 0 (0.058 0)	0.113 (0.115)
Constant	7.693^*** (0.326)	7.131^*** (0.334)	7.735^*** (0.327)	8.049^*** (0.528)	5.358^*** (1.018)
Observations	1,150	1,045	1,150	669	330
R-squared	0.223	0.269	0.215	0.299	0.315

Standard errors in parentheses;

***

p<0.01

p<0.05

p<0.1.

6.2 Robustness Check

As previously mentioned, this study employed the IPWRA and AIPW methods to check the robustness of the OLS estimation results. First, we calculated the probability of students attempting global English tests by taking a logistic regression on their major, gender, household type, household income, university ranking, and GPA ranking. Figure 1 indicates that the propensity score overlapped between the treatment and control groups, which is a requirement for using IPWRA and AIPW. Then, we performed the covariate balance checking after inverse probability weighting. Table 6 suggests that the inverse probability weighting balanced the covariates between the treatment and control groups. In other words, the differences of covariates became extremely minimal after weighting.

Figure 1.

Propensity Score Overlap Checking.

Table 6.

Covariate Balance Checking after Inverse Probability Weighting.

	Standardized raw	Differences weighted
Science track	-0.104	0.030
Male	0.024	0.037
Health	-0.060	-0.046
Urban	0.306	-0.032
Ln(family income)	0.377	-0.090
“211” colleges	0.292	-0.066
Academic ranking	-0.145	0.062
Major type (Science)	-0.107	0.077

Table 7.

Results of Inverse Probability Weighting Regression Adjustment (IPWRA) and Augmented Inverse Probability (AIPW).

	IPWRA	AIPW
VARIABLES	ATE	ATE
ATE	0.198^***	0.197^***
	-0.037	-0.04
Observations	1,118	1,118

Robust standard errors in parentheses;

***

p<0.01

^**p<0.05

^*p<0.1.

Finally, in the outcome model, we regressed the outcome variable on covariates of entrance examination score, household type, household income, parental education, university ranking, GPA ranking, major, employer type, and work field. Table 6 shows the effect of possessing a global English test certificate using IPWRA and AIPW. The treatment effect was 0.198 or 0.197 for IPWRA and AIPW, respectively; both significant. The results are similar to the results of OLS estimate shown above. Therefore, the results of IPWRA and AIPW indicate that the OLS estimation of our study was robust.

6.3 Mechanism Analysis

Why can students with global English certificates enjoy a higher salary? This is because of the proactive personality. Proactive personality is first defined as “the relatively stable tendency to effect environmental change”. The generally accepted proactive personality scale (PPS) was composed by Bateman in 1993. A total of 17 items were tested among undergraduates and MBA students. In the CEPS survey questionnaire related to the psychological measurement, we found three similar items: “I am confident that I can address any unexpected events effectively”; “I am able to cope with the unexpected things myself”; and “When faced with a dilemma, I can usually find some solutions”. Table 8 shows the comparison of these three items. We found that the proactive personality of students with global English certificates was significantly higher than those without the certificates.

Table 8.
Proactive Personality Difference between Students Who have or do not have Global English Certificates.

Have Global English Certificate Do Not Have Global English Certificate P value

Proactive personality 1 3.176 585 3.107 870 0.000

Proactive personality 2 3.202 113 3.111 404 0.000

Proactive personality 3 3.249 472 3.133 158 0.000

	Have Global English Certificate	Do Not Have Global English Certificate
Proactive personality 1	3.176 585	3.107 870
Proactive personality 2	3.202 113	3.111 404
Proactive personality 3	3.249 472	3.133 158

The proactive personality is separately considered in the previous regression equation and in Table 9 from Column (2) to Column (4). We observed that the impact of Global English Certificate on wages is still significant, but increases slightly, compared to the coefficient of Global English Certificate, as shown in Column (2). Furthermore, the proactive personality significantly and positively affects the graduates’ salary from Column (2) to Column (4), indicating that a proactive personality may indeed be one of the mechanisms of the wage premium of Global English Certificate.

Table 9.

Effects of English Ability on Starting Salaries with Proactive Personality Variables Added.

VARIABLES	(1) Ln(wage)	(2) Ln(wage)	(3) Ln(wage)	(4) Ln(wage)
Global English Certificate	0.135^***	0.130^***	0.129^***	0.130^***
	(0.045 5)	(0.045 3)	(0.045 4)	(0.045 5)
Proactive personality		0.073 5^***	0.068 0^***	0.047 1^*
		(0.024 0)	(0.025 1)	(0.024 3)
CET4 score	0.001 58^***	0.001 58^***	0.001 58^***	0.001 57^***
	(0.000 303)	(0.000 302)	(0.000 302)	(0.000 303)
Science track	0.080 9^*	0.080 4^*	0.079 2	0.078 1
	(0.048 7)	(0.048 5)	(0.048 6)	(0.048 7)
Period	0.192^***	0.179^***	0.182^***	0.188^***
	(0.035 4)	(0.035 5)	(0.035 4)	(0.035 4)
Male	0.132^***	0.114^***	0.116^***	0.125^***
	(0.038 5)	(0.038 8)	(0.038 8)	(0.038 6)
Health	-0.000 327	-0.000 872	-0.000 549	-0.000 384
	(0.001 27)	(0.001 28)	(0.001 27)	(0.001 27)
Urban	-0.084 2^**	-0.084 5^**	-0.085 8^**	-0.081 8^*
	(0.042 6)	(0.042 4)	(0.042 4)	(0.042 5)
Ln(family income)	0.082 5^***	0.080 5^***	0.080 8^***	0.080 5^***
	(0.021 2)	(0.021 1)	(0.021 2)	(0.021 2)
“211 college”	0.020 3	0.025 6	0.023 9	0.026 8
	(0.044 1)	(0.044 0)	(0.044 0)	(0.044 2)
Academic ranking	-0.111	-0.097 9	-0.097 0	-0.101
	(0.086 9)	(0.086 6)	(0.086 8)	(0.086 9)
Major type (Science)	-0.087 5^**	-0.092 1^**	-0.090 8^**	-0.085 8^*
	(0.044 0)	(0.043 8)	(0.043 9)	(0.044 0)
Ownership type of enterprise	Yes	Yes	Yes	Yes
Employment industry	Yes	Yes	Yes	Yes
Province	Yes	Yes	Yes	Yes
Constant	7.131^***	7.009^***	6.978^***	6.986^***
	(0.334)	(0.335)	(0.338)	(0.342)
Observations	1,045	1,045	1,045	1,045
R-squared	0.269	0.276	0.275	0.272

Standard errors in parentheses;

***

p<0.01

p<0.05

p<0.1.

7. Conclusions, Discussion, and Implications

In this study, we discussed the effects of having global English test certificates in China's labor market for college graduates. After controlling for the English language score (CET4 score), processing global English test certificates had a steady strong wage premium effect; however, the effect of domestic English test certificates was not as significant as that of the global tests. In addition, the effect was evident only on the starting wage. The global English test certificate had no significant impact on the third year salary and salary growth.

The results of our study suggest that global English test certification has additional wage premium. The premium could be attributed to the matching of evaluation and labor market job requirements and also to non-cognitive factors. Therefore, how can wage premiums be understood? In our study, we observed that for individuals working in foreign firms and at senior positions, the effect of possessing a global English test certificate was greater. This highlights the job selection mechanism. Under the human capital theory, one can improve the ability though extra input on English learning, which subsequently improves their productivity. Under the signal theory, the certificate represents one's ability which is rewarded in labor market. However, is this true? If the extra input on English indeed improves the productivity, why is the domestic certificate not as effective as the global test? There is no doubt that the preparation of oral English test would also lead to the improvement of ability and also indicate the higher ability; however, the human capital theory and signal theory explanations are not convincing. A more plausible way of understanding the effect is the proactivity theory. The chasing certificate behavior, seen as a proactive action in labor market, is representative of one's motivation or attitude. The proactive action also makes people pursue more job applications and choose better jobs though proactive information collection. Another explanation would be that the labor market in China has a myth surrounding foreign certificates instead of domestic ones.

How to identify whether there is no effect of the global test certificate on wage growth? After a certain period of employment, the information asymmetry becomes weaker. The productivity of a person is more easily evident than at the stage of initial employment. Therefore, the productivity itself determines wage and not proactivity.

Traditional studies have focused on language ability but ignored the differences between language ability and some aspects of personality. The effect might be comprehensive. In this study, we separated the effect of language ability and personality. Therefore, we conclude that language ability works in all periods of employment but the certificate works only during the job search period. Therefore, proactivity among students must be emphasized. Moreover, it is crucial for employers to eliminate superstitions regarding certifications and focus on employees’ ability. Due to limitation of our database, the discussion about the mechanism of English test certificate is not sufficient and it is still necessary to further analyze the part of proactive personality.

Footnotes

1

In statistics, ordinary least squares (OLS) or linear least squares is a method for estimating the unknown parameters in a linear regression model.

Notes on Contributors

Sheng Cui, associate professor of School of Education at Renmin University of China and of Research Center for Educational Development and Public Policy at Ministry of Education. His research interests include educational economics, enrollment reform, student learning and development, local government investment in compulsory education. He has published quite a few academic articles and books in Peking University Education Review, Tsinghua University Education Review, China Higher Education Research and other CSSCI journals. He is the principal investigator of 1 grant project named “educational planning and strategy”. He won the Excellent Prize on 7th scientific research (social science) of higher education and on national conference of empirical research in education.

Kunfeng Pan, associate professor of School of Education at Renmin University of China, visiting scholar of University of Cambridge, and consultancy expert of Department of Development Planning in Ministry of Education of China. He has published more than 30 peer-reviewed articles in SCI, SSCI and CSSCI journals in poverty alleviation through education, student development, college enrollment and so on. He has been the PI of more than 10 projects funded by National Social Science Fund and other governmental funds.

Yangyong Ye, assistant professor of School of Education at Renmin University of China. His research areas are school finance, education law and teacher personnel policy.

References

Akresh

I. R.

, & Frank

(2011). At the intersection of self and other: English language ability and immigrant labor market outcomes. Social Science Research, 40(5), 1362–1370.

Aldashev

, Gernandt

, & Thomsen

S. L.

(2009). Language usage, participation, employment and earnings: Evidence for foreigners in West Germany with multiple sources of selection. Labour Economics, 16(3), 330–341.

Azam

, Chin

, & Prakash

(2013). The returns to English-language skills in India. Economic Development and Cultural Change, 61(2), 335–367.

Bateman

T. S.

, & Crant

J. M.

(1993). The proactive component of organizational behavior: A measure and correlates. Journal of Organizational Behavior, 14(2), 103–118.

Bettinger

E. P.

, Evans

B. J.

, & Pope

D. G.

(2011). Improving college performance and retention the easy way: Unpacking the ACT exam. National Bureau of Economic Research.

Bostwick

(2016). Signaling in higher education: The effect of access to elite colleges on choice of major. Economic Inquiry, 54(3), 1383–1401.

Brown

D. J.

, Cober

R. T.

, Kane

, Levy

P. E.

, & Shalhoop

(2006). Proactive personality and the successful job search: A field investigation with college graduates. Journal of Applied Psychology, 91(3), 717–726.

Casey

, & Dustmann

(2008). Intergenerational transmission of language capital and economic outcomes. Journal of Human Resources, 43(3), 660–687.

Chiswick

B. R.

(1991). Speaking, reading, and earnings among low-skilled immigrants. Journal of Labor Economics, 9(2), 149–170.

10.

Chiswick

B. R.

, & Miller

P. W.

(1995). The endogeneity between language and earnings: International analyses. Journal of Labor Economics, 13(2), 246–288.

11.

Chiswick

B. R.

, & Miller

P. W.

(2002). Immigrant earnings: Language skills, linguistic concentrations and the business cycle. Journal of Population Economics, 15(1), 31–57.

12.

Chiswick

B. R.

, & Miller

P. W.

(2003). The complementarity of language and other human capital: Immigrant earnings in Canada. Economics of Education Review, 22(5), 469–480.

13.

Chiswick

B. R.

, & Miller

P. W.

(2010). Occupational language requirements and the value of English in the US labor market. Journal of Population Economics, 23(1), 353–372.

14.

Cho

, & Bridgeman

(2012). Relationship of TOEFL iBT^® scores to academic performance: Some evidence from American universities. Language Testing, 29(3), 421–442.

15.

Choi

(2015). The effects of English training abroad on labor market outcomes in Korea. Research in Social Stratification and Mobility, 41, 11–24.

16.

Crant

J. M.

, & Bateman

T. S.

(2000). Charismatic leadership viewed from above: The impact of proactive personality. Journal of Organizational Behavior, 21(1), 63–75.

17.

Dustmann

, & Fabbri

(2003). Language proficiency and labour market performance of immigrants in the UK. The Economic Journal, 113(489), 695–717.

18.

Erdogan

, & Bauer

T. N.

(2005). Enhancing career benefits of employee proactive personality: The role of fit with jobs and organizations. Personnel Psychology, 58(4), 859–891.

19.

Gardner

(1983). Frames of mind: The theory of multiple intelligence. Journal of Policy Analysis & Management, 4(3), 19–35.

20.

Gil

(2016). English language education policies in the People's Republic of China. English Language Education Policy in Asia (pp. 49–90). Switzerland: Springer.

21.

Ginsburgh

V. A.

, & Prieto-Rodriguez

(2011). Returns to foreign languages of native workers in the European Union. ILR Review, 64(3), 599–618.

22.

Grant

A. M.

, & Ashford

S. J.

(2008). The dynamics of proactivity at work. Research in Organizational Behavior, 28, 3–34.

23.

Guo

, & Sun

(2014). Economic returns to English proficiency for college graduates in mainland China. China Economic Review, 30, 290–300.

24.

Heckman

J. J.

(1979). Sample selection bias as a specification error. Econometrica: Journal of the Econometric Society, 47(1), 153–161.

25.

(2005). English language education in China: Policies, progress, and problems. Language Policy, 4(1), 5–24.

26.

Hungerford

, & Solon

(1987). Sheepskin effects in returns to education. Review of Economics & Statistics, 69(1), 175–177.

27.

Jaeger

D. A.

, & Page

M. E.

(1994). Degrees matter: New evidence on sheepskin effects in the returns to education. Ann Arbor Michigan University of Michigan Population Studies Center Apr, 78(4), 733–740.

28.

Jin

, & Yang

(2008). The English proficiency of college and university students in China: As reflected in the CET. Language, Culture and Curriculum, 19(1), 21–36.

29.

Kossoudji

S. A.

(1988). English language ability and the labor market opportunities of Hispanic and East Asian immigrant men. Journal of Labor Economics, 6(2), 205–228.

30.

, & Zussman

(2010). Lingua franca: The role of English in international trade. Journal of Economic Behavior & Organization, 75(2), 250–260.

31.

Lang

, & Siniver

(2009). The return to English in a non-English speaking country: Russian immigrants and native Israelis in Israel. The BE Journal of Economic Analysis & Policy, 9(1), 50.

32.

Leikin

(2013). The effect of bilingualism on creativity: Developmental and educational perspectives. International Journal of Bilingualism, 17(4), 431–447.

33.

, Meng

, Shi

, & Wu

(2012). Does attending elite colleges pay in China? Journal of Comparative Economics, 40(1), 78–88.

34.

Mann

, & DiPrete

T. A.

(2013). Trends in gender segregation in the choice of science and engineering majors. Social Science Research, 42(6), 1519–1541.

35.

Melitz

(2008). Language and foreign trade. European Economic Review, 52(4), 667–699.

36.

Mincer

J. A.

(1974). Schooling and earnings. Schooling, Experience, and Earnings (pp. 41–63). NBER.

37.

Park

J. H.

(1999). The earnings of immigrants in the United States: The effect of English-speaking ability. American Journal of Economics and Sociology, 43–56.

38.

Robins

J. M.

, & Rotnitzky

(1995). Semiparametric efficiency in multivariate regression models with missing data. Source Journal of the American Statistical Association, 90(429), 122–129.

39.

Rosenbaum

P. R.

, & Rubin

D. B.

(1983). The central role of the propensity score in observational studies for causal effects. Biometrika, 70(1), 41–55.

40.

Saiz

, & Zoido

(2005). Listening to what the world says: Bilingualism and earnings in the United States. Review of Economics and Statistics, 87(3), 523–538.

41.

Seibert

S. E.

, Crant

J. M.

, & Kraimer

M. L.

(1999). Proactive personality and career success. Journal of Applied Psychology, 84(3), 416–427.

42.

Seibert

S. E.

, Kraimer

M. L.

, & Crant

J. M.

(2001). What do proactive people do? A longitudinal model linking proactive personality and career success. Personnel Psychology, 54(4), 845–874.

43.

Spence

(1973). Job market signaling. The Quarterly Journal of Economics, 87(3), 355–374.

44.

Stöhr

(2015). The returns to occupational foreign language use: Evidence from Germany. Labour Economics, 32, 86–98.

45.

Wait

I. W.

, & Gressel

J. W.

(2009). Relationship between TOEFL score and academic success for international engineering students. Journal of Engineering Education, 98(4), 389–398.

46.

Wei

, & Su

(2012). The statistics of English in China. English Today, 28(28), 10–14.

47.

Wooldridge

J. M.

(2002). Inverse probability weighted M-estimators for sample selection, attrition, and stratification. Portuguese Economic Journal, 1(2), 117–139.

48.

Wooldridge

J. M.

(2007). Inverse probability weighted estimation for general missing data problems. Journal of Econometrics, 141(2), 1281–1301.

49.

Yang

(2014). Are all admission sub-tests created equal?—Evidence from a national key university in China. China Economic Review, 30, 600–617.

50.

Yao

, & van Ours

J. C.

(2015). Language skills and labor market performance of immigrants in the Netherlands. Labour Economics, 34, 76–85.

51.

Zheng

, & Cheng

(2008). College English Test (CET) in China. Language Testing, 25(3), 408–417.