Comment on “What Can Students Gain from China's Higher Education?”

Li et al. (2023) use a unique dataset collected from a large-scale survey conducted by the authors to study college students in China. The data provide many interesting and new descriptive statistics about college students, such as their technical certificates received, their internship experiences, and so on. Their study examines how different mechanisms, including human capital, social network, and signaling, help explain the returns to a college education.

The human capital measures include grade point average (GPA), technical certificates, etc. Social networks are measured by Communist Party membership, participation in a student union, parental income and education. Students' skills possessed before enrolling in college such as the type of college enrolled in are assumed to be related to signaling.

Li et al.'s main findings are: (i) human capital is the least useful in raising the starting salary; (ii) social networks are a strong predictor; and (iii) signaling matters. Li et al. suggest the need to reform the higher education system in China.

A novel contribution of Li et al. (2023) is their investigation of the channels explaining the labor market effects of higher education. It complements similar studies that use data from other countries with more sophisticated causality analyses. For example, Dale and Krueger (2002) compare students who attended more selective colleges and those who had seemingly comparable ability but chose to attend less-selective ones and found that they obtained similar earnings in general. Black and Smith (2006) measure college quality with multiple proxies and find that existing studies understated the wage effect of college quality. Zhong (2011) finds that significant return gaps exist between high-ranked and low-ranked college in China.

Li et al. (2023) use students' wages for their first job to measure the effect of human capital. A potential issue with the wage for the first job is the quality of the job match. It is known that job turnover rates for new college graduates are very high. For example, Takeshita et al. (2022) find that the turnover rate for new college graduates is around 20% within a year in Japan. In China, 34% of college graduates leave their first job within 1 year.¹ Therefore, the first job may not reflect a person's true human capital due to the potential wage penalty of the mismatch (Sellami et al., 2017). This study is limited by the data, yet it would be helpful to discuss more of the implications regarding this issue.

As Li et al. also acknowledge, the three mechanisms that determine wages are represented by variables that may overlap for the different mechanisms. For example, the involvement in a student union may be related to their non-cognitive abilities, such as openness, a part of human capital. For the signaling mechanism, the type of school that a student enrolled in has a signaling effect but may also represent the quantity and quality of learning. Although there is some ad hoc flavor in classifying those variables into three channels, the findings about those individual variables still provide useful information.

Li et al. (2023) use standardized scores for college GPA. It is possible that the GPA in a top-ranked college is generally lower than in a low-ranked college. By pooling those scores together, the standardized scores may distort their relative magnitudes and their relationship with wages. Li et al. control for college fixed effects and the results still show that GPA is insignificant. Additional explanations could include that an incremental change of GPA does not influence wages, but a large change may do so, such as an increase from a low to a high GPA. Even within the same school, the comparability of GPA can still be problematic due to, for example, possible grade inflation in “soft” subjects.

Whether GPA has any effect on earnings should have very important implications and thus warrant further investigation, for example, applying additional techniques to make them comparable and better reflect the true learning.

In sum, Li et al. (2023) present new facts and findings about college students in China. They provide useful information for policy makers and will help further research.