Prevent content from getting lost
The new issue of "Cultural Vertical" will be released in June 2024, click on the picture above or the lower left corner at the end of the article to read the original article, check the submission email: [email protected], "Cultural Vertical", postal code: 80-942
✪ Wu Xiaogang | NYU Shanghai ✪ Li Xiaoguang | Xi'an Jiaotong University
Recently, a notice of "post-90s post-retirement at the age of 65" was widely circulated on the Internet, causing an uproar in public opinion. Later, after verification by relevant departments, there was no official certification of the official account of the notice, and on the basis of quoting the "Decision of the Central Committee of the Communist Party of China on Further Comprehensively Deepening Reform and Promoting Chinese-style Modernization" issued by Xinhua News Agency on July 21, on the basis of "steadily and orderly promoting the reform of gradually delaying the statutory retirement age", a large number of content other than the original text of the "Decision" was mixed in, which was suspected of over-interpretation and over-exertion. While "delayed retirement" is widely discussed, there is also an argument that "the employment situation is grim and it is difficult to retire in a timely manner". Nowadays, young people often feel that the market is so "volume" that the value of the diploma in their hands seems to be getting lower and lower. So is that really the case?
This paper analyzes more than 20,000 sample data in the form of statistical research, and concludes that due to the need for further industrial upgrading in the mainland, more jobs in need of higher education will be developed, and overall, "over-education" is actually on a downward trend. However, because the higher the overall education level of the generation born later, the relative position of young people in the peer competition has a greater impact on their job search, so they will feel that the competition is becoming more intense, and the probability of a "mismatch" between career and education will occur. And if a "mismatch" occurs, it is likely to affect the individual's life.
Of course, this article also points out that for society as a whole, there is a long way to go to improve the quality of the people, and education is far from excessive. Moreover, "over-education" is not only a negative phenomenon, some "mismatches" can make people with a higher level of education change the connotation and work content of some positions. At the same time, education is not all about labor and employment, it is more about creating a better society. In the future, strengthening the connection between higher education and the job market, promoting further industrial upgrading, and allowing young people with high-tech vocational skills to be matched with suitable positions are the key to the coordination and promotion of industrial and education policies in the next step.
This article was originally published in Social Sciences in China, Issue 2, 2021, and only represents the author's point of view for your consideration.
in China's urban labor market
Changing trends in education matching
- Dynamic analysis based on age, period, and generational effects
▍Introduction
Education is a foundation that has a bearing on the development of the country and the future of the nation. Since the beginning of reform and opening up, along with the rapid economic growth, the mainland has also made remarkable achievements in the development of education. In 2019, the gross enrolment rate of the first three years of primary school was 83.4%, the enrolment rate of the primary school-age population was 99.94%, and the gross enrolment rate of junior high school, senior high school and tertiary education was 102.6%, 89.5% and 51.6% respectively. Higher education has also entered the stage of popularization, and the national college entrance examination acceptance rate has risen from 7% in 1978 to 80% in 2019; The number of college graduates rose from 165,000 in 1978 to 8.74 million in 2020. The average number of years of education for the new labor force is more than 13.7 years, and the level of education coverage in the mainland ranks among the world's upper-income countries.
The development of education has promoted the improvement of the quality of the people on the mainland and provided valuable human resources for socialist modernization. However, the "difficulty of finding employment" in the labor market has become an increasingly real and urgent problem. Many college students graduate without finding suitable jobs, and some have to accept jobs with lower skill requirements and earnings, resulting in declining returns to education. China also seems to be experiencing a mismatch or mismatch in the labor market of developed countries after the expansion of education, especially over-education, which has become the main manifestation of education mismatch or mismatch.
Generally speaking, the initial allocation of occupational positions in the labor market in modern societies is mainly based on the schooling and skills training that people receive. According to the degree of matching between education level and vocational position, we can distinguish three types of education matching states: under-education, education match, and over-education. Many studies have shown that the income and job satisfaction of the undereducated are generally as high as those of the education match, i.e., the undereducated tend not to be at a disadvantage in the labor market. Over-education has received special attention in the study of education matching, because it can reduce the return to education and affect job satisfaction for individuals. For families and societies, investments in human capital are wasted. In this sense, this paper focuses on a specific situation of educational mismatch in China's labor market, namely the problem of over-education.
Unlike developed countries in the West, China's educational expansion, industrial structure transformation and vocational structure upgrading are basically synchronized. Especially since the 90s of the 20th century, with urbanization, industrialization, especially the development of the tertiary industry, a higher demand has been put forward for talent development. In line with this trend, higher education began to gradually change from elite education to mass education in the late 90s, providing valuable human resources for socialist modernization.
In the process, there is also a certain degree of development incongruity. On the one hand, the expansion of higher education that began in 1999 began before secondary education was fully developed, when the economic structure dominated by manufacturing and low-end services could not create enough suitable jobs to absorb the influx of university graduates into the job market. At the same time, there are many positions (such as skilled tradespeople) that cannot be recruited. On the other hand, due to the rapid expansion of higher education in a short period of time, the quality and content of training have not been improved with the increase in quantity, resulting in many graduates with university diplomas not having the skills required for certain professional or managerial positions.
This paper focuses on the trend of education matching in China's urban labor market, especially the problem of over-education, and incorporates the trend of over-education into the time frame composed of macro social processes and micro life courses, and strictly distinguishes and estimates the age, period, and generation effects of over-education change at the methodological level. Specifically, this article aims to answer the following three questions.
First, is over-education a short-term or long-term phenomenon as individuals grow older and live?
Second, with the process of marketization and modernization in the mainland, will excessive education increase or decline over time?
Third, with the change of the education system in mainland China, does the role of education diplomas obtained by different generation groups during their growth period increase or decline in the role of education matching in the labor market?
In order to answer these questions, our analysis is mainly based on CGSS 2003-2017 data, and constructs an age-period-generation model of overeducation, so as to outline the evolution trajectory of overeducation with macro social processes and micro life courses.
▍Theoretical analysis
The problem of education matching, especially how over-education evolves over time, is one of the important issues in the field of education matching in recent years. At present, the field of social sciences has identified three main trends of change: the age effect, the period effect, and the generation effect. The so-called age effect mainly reflects the age-related changes that accompany the life course and the change of social roles, such as the probability of over-education, and there may be obvious differences in different life course stages such as individuals getting a first job, reaching the peak of their career, and retiring. The so-called period effect mainly refers to the similar impact of macro-socio-economic conditions in a specific historical period of investigation on people of all ages. Changes in the structure of the labor market at different times may affect the probability of over-education among all populations, i.e., the period effect of over-education. The so-called cohort effect is essentially a social change, which mainly reflects the impact of early living conditions, social factors or social experiences on a specific birth generation. There may be significant differences in the opportunities for higher education among different birth generations, which in turn affects the probability of over-education, which is the generational effect of over-education. These three effects have their own rich sociological implications, and researchers need to distinguish between them when analyzing overeducation, so as to help accurately estimate the trend of change and rationally explain the mechanism of occurrence.
(1) The age effect of over-education
Is over-education a short-term or long-term phenomenon in an individual's career? The academic debate continues to this day. Short-term scholars believe that over-education is a temporary phenomenon. Long-term scholars argue that over-education is a persistent phenomenon. This paper argues that the matching status of education and career is closely related to the stage of an individual's life course. Life course theory emphasizes that the social roles of individuals will change at different life cycle stages, and the role relationships at specific stages will profoundly affect the choices and actions of individuals. In fact, there are differences in individual job search behavior and career mobility at different stages of the life course, such as "standing at 30, not confused at 40, and knowing the destiny of heaven at 50", which reflects the age expectation of traditional society for individual career achievements and life status. Therefore, we need to discuss the transition of over-education with the individual's career into the life course stages of the age composition.
The short-term theory of over-education is based on the theory of occupational mobility. The theory argues that the form of return for education in the labor market is not only reflected in direct income gains, but also implies more opportunities for upward career mobility. When given an initial occupational position, those workers with higher education will have more opportunities to achieve upward career mobility, in other words, those who are over-educated in the same occupational position, and their "excessive" part of their education will bring more opportunities for upward mobility, thus helping them to get rid of over-education. Therefore, over-education is short-term for the individual.
From the empirical research in China, although some scholars have found that over-education shows a downward trend with age, they do not strictly distinguish between the age effect and the generational effect of over-education. In addition, there is a significant shortcoming in the occupational mobility theory, that is, it ignores the occupational mobility situation at different age stages. In fact, career mobility declines with age, and individual education qualifications may depreciate with age. Therefore, the explanatory logic of the career mobility theory may only apply to the transition of over-education in youth or middle age, but it is difficult to apply to the long-term trend of over-education with the whole career.
The long-term theory of over-education is mainly derived from signal theory. This theory argues that the recruitment process in the labor market is an investment behavior with high uncertainty, because there is an information asymmetry between employers and job seekers, which makes it difficult for employers to judge the productivity of job seekers in advance. For this reason, during the job search process, job seekers will enhance and release signals of their job ability (e.g., educational background), which in turn will become the main basis for employers to screen and screen job applicants. However, along the way, some employment status or work experience may send negative signals. For example, in situations where an employer cannot determine a candidate's productivity, seeing an experience of over-education, regular unemployment, or unstable employment on their resume may create a negative impression of the candidate's potential productivity or personal abilities. At the same time, mismatch in education can lead to a scarring effect, which, if it occurs, can leave a scar on workers' resumes that can seriously hinder their future job mobility. As a result, the experience of over-education will limit people's long-term career mobility, and even if they change jobs or accumulate experience, they may not be able to completely escape the predicament of over-education, but may only move from one over-educated work state to another. According to China's empirical research, some scholars believe that over-education is not a short-term phenomenon by analyzing the changes of age and working years. Therefore, this paper infers that over-education is more likely to be a long-term phenomenon in an individual's career.
(2) The period effect of over-education
Will over-education show an upward trend over time? This paper argues that the trend of over-education with the period mainly depends on the characteristics of occupational structure in the labor market corresponding to the industrial structure in different periods. With the economic development and the adjustment of the industrial structure, if the occupational structure shows a trend of upgrading over time, the stronger the ability to absorb higher education qualifications, the lower the probability of over-education. Conversely, the probability of over-education is higher. Empirical studies in Europe and the United States have found that over-education has shown an overall upward trend over time, but it should be noted that the occupational structure in these countries is moving towards a "polarization" trend, that is, high-skilled occupations dominated by management and technology are expanding, and lower-skilled occupations centered on interpersonal services and interactions are also growing, but the growth of middle-skilled occupations is gradually stagnating or even declining. This hollowing out of middle-skilled occupations may reduce the ability of the labor market to absorb a workforce with a higher education diploma. To analyze the changing trend of over-education in China over time, we need to accurately grasp the overall trend of occupational structure.
Over the past 40 years of reform and opening up, the mainland's occupational structure has been rapidly upgraded. Lu Xueyi (2004) analyzed a large number of empirical data and found that from 1978 to 2002, the occupational structure of mainland China as a whole "tended to be advanced", which mainly reflected the change trend in three aspects: first, the number of absolute jobs continued to grow, second, the increase of occupational types, and third, the proportion of middle- and higher-level occupations increased, while the proportion of lower-level occupations gradually decreased. There are two main driving forces for this elevation of the occupational structure: modernization and marketization. Since China's accession to the WTO in 2002, the mainland's modernization process and marketization process have developed rapidly, which has effectively promoted the upgrading of the mainland's occupational structure.
The process of upgrading the occupational structure will improve the ability of the labor market to absorb and absorb workers with higher educational qualifications, thereby reducing the probability of over-education. This is mainly reflected in three processes: First, the process of upgrading the overall occupational structure will promote the upward trend of education demand within the vast majority of occupations, thereby helping to absorb workers with higher academic qualifications. Second, the proliferation and rapid development of some occupations have strongly promoted the demand for education in the labor market. Driven by marketization and modernization, some occupations such as professional and technical personnel, commercial service personnel, and production and transportation personnel have developed at a faster pace, and the demand for education has grown rapidly, which will help improve the ability to digest workers with higher education qualifications. Thirdly, the process of boundary penetration between different occupations is helpful to expand the employment path of higher education and reduce the probability of over-education. The process of marketization and modernization has created a large number of occupations such as commercial service workers, and the level of education required for them is also rising, providing opportunities for those who are potential over-educated to find modest educational jobs. This process of searching for matching across occupational boundaries helps to absorb the labor force with higher education qualifications in the mainland labor market. Based on the above analysis, we can infer that over-education will show a downward trend over time.
(3) The generational effect of over-education
What is the trend of over-education with different birth generations? This paper argues that the generational effect of over-education is rooted in the trajectory of change in the education system itself. The vast majority of individuals receive schooling during their formative years, so changes in a macro education system will only affect access to education for specific generational groups, but not for all generations. As a result, the changes in the macro education system will be reflected in the distribution of education among different birth generations. At the same time, the labor market value of an individual's education diploma depends on the relative position of the individual's educational distribution in the generation at which he was born: the higher the relative position, the stronger the signal value of the education diploma, and the lower the probability of over-education. A recent empirical study based on 12 Western industrial countries, including the United Kingdom, France and the Netherlands, shows that with the expansion of education, the risk of over-education among the younger generation has increased significantly, and higher education diplomas have become a necessary condition for the younger generation to compete and survive in the workplace, but the actual economic returns are declining.
The education system in mainland China has gone through four stages of development: foundation, reform, enrollment expansion and connotation improvement. The changes in the education system in these four stages have led to generational differences in the composition of education in the labor market, and have affected the relative educational position of workers in the market. Overall, the later generations are born, the more opportunities they have for compulsory and tertiary education, and the higher their overall educational attainment. However, since educational attainment rises with the generation at birth as a whole, the relative educational position of the same educational attainment will decline. As a result, the same degree is less valuable to younger generations.
Specifically, due to the information asymmetry between employers and job seekers in the labor market, employers mainly judge people's productivity by their relative educational position. Because individual job competition and other behaviors often occur in the birth generation group of their own age, for this reason, those job seekers with a higher relative education position in their birth generation are more likely to be selected by employers and more likely to find a job that matches their education level. Therefore, the younger generation is born later, the higher the chance of over-education.
▍Data, variables, and models
(1) Data sources
The data used in this paper are mainly from the China General Social Survey (CGSS) from 2003-2017. For this study, the CGSS data has two distinct advantages: it has sufficient time span and the consistency of the measurement questioner. First, we established the APC model based on the data from 2003 to 2008 and 2010 to 2017, respectively, and the estimated results were basically consistent with the pooled samples. Second, after we incorporate the weights into the APC model, the estimated age, period, and generation effects are robust. We first combined the 9-phase data of CGSS (see Table 1 for details), limited the sample to those who lived in the city and were aged 18-60 years, and then excluded the samples with missing values in the core variables such as occupation and education, and finally entered the analysis sample into the model with 26,820 samples.
(2) Measurement of excessive education
The measurement of educational matching (over-education) is based on realized matches. The so-called reality matching method is based on the observation data, the researcher first uses the mode or standard deviation to define the standard education level required for each occupation, and then compares the actual education level of the individual with the education level required by the occupation to determine whether over-education occurs. This is the most widely used and recognized method in the field of educational matching research.
However, there are two drawbacks to this approach. First, it ignores the changing nature of the system. The reality matching method is suitable for countries with relatively stable economic systems and education systems, while the mainland's economic and education systems have undergone major changes in recent decades. In order to overcome this shortcoming, we introduce the "period-generation" time grid, and use the reality matching method to define over-education within each grid. Second, it ignores the heterogeneity of education within the profession. Many professions require a co-composition of individuals with different levels of education, and the reality matching method statistically classifies some moderate educators as over-educated. To overcome this bias, we introduce the Intra-Occupational Educational Heterogeneity Index (IQV) and incorporate it into the prediction equation for over-education, thereby reducing the interference caused by measurement bias.
Based on the improved reality matching method, this paper measures over-education through four steps. First, the "time grid" is divided according to the survey period and generation, in which the birth generation includes three groups, that is, the birth generation before 1960, 1960-1980, and after 1980. Second, according to the education status of the respondents, the education level and the number of years of education are divided, of which the education level includes six categories (primary school and below, junior high school, high school, junior college, undergraduate and graduate students and above). Thirdly, within each time grid, the mode of educational attainment and the standard deviation of years of education within each occupation are calculated according to the three-digit occupation code of the International Standard Classification of Occupations (ISCO-88). Fourth, over-education is defined using the majority method and the standard deviation method by comparing the individual's actual educational status with the level of education required for his or her occupation. Based on different methods, the results of over-education were 35.09% (CGSS mode) and 35.27% (CGSS standard deviation method), respectively.
It can be seen that the proportion of over-education that we have derived using different data and statistical criteria is consistent and robust (about 35%). In addition, although the proportion of undereducated individuals is about 17%, many studies have shown that undereducated people tend not to be at a disadvantage in the labor market, so to simplify the analysis, this paper combines both undereducated and educationally matched individuals into moderate education. In the sensitivity analysis, we excluded under-educated samples and found that the findings were also robust.
Table 1 presents additional variables defined based on the CGSS database. First of all, in terms of demographic characteristics, the average age is 41 years old, about 53% of men are men, 78% are urban hukou, about 6% are ethnic minorities, 32% are members of the Communist Party of China, and 83% are married. Secondly, in terms of job characteristics, the average work experience is 23 years, and the proportion of employment within the system is about 49%, and the proportion of employees in monopoly industries is 46%. Among the occupations, 10% are managers, 21% are professional and technical personnel, 13% are clerical workers, 20% are commercial service personnel, and 36% are manual workers. The value of the intra-vocational educational heterogeneity index ranges from 0 to 1, and the mean value is 0.86, which means that the degree of intra-vocational educational heterogeneity is relatively high. Finally, there are family background and regional variables. In this paper, the highest values of the father's and mother's education level are taken as the proxy variable of family background. Regions are divided by province, with 53% of the sample from the eastern region, 28% from the central region and 19% from the western region.
(3) Research methods: age-period-generation model
In order to study the trend of educational matching (over-education) over time, this paper adopts the Age-Period-Cohort Model (APC model). The APC model is able to distinguish and estimate trends in over-education with age, period, and generation. Drawing on previous studies, this paper expresses the APC model of over-education as follows:
In equation (1), pi represents the probability of overeducation, μ is the intercept term, β1 represents the age effect, β2 represents the period effect, β3 represents the generation effect, and Xp represents the control variable. However, in the actual estimation process, due to the collinear relationship between the three variables of age, period and generation, the APC model has the problem of unidentification. In this regard, scholars have proposed a variety of solutions, such as the local limitation method, the dummy variable grouping method, the endogenous factor method and the multi-layer cross-random effects method, but each method has its limitations. There is no consensus on this, but it is suggested that researchers should use at least two methods as robustness tests.
In this paper, a dummy variable grouping model and a multi-level cross-over random-effects model were selected. The dummy variable grouping model solves the problem of insufficient identification of the original equation by grouping the ages, periods, and generations of sample members into time intervals of different lengths, and this method allows researchers to flexibly group ages, periods, and generations according to actual needs.
Here, we use two dummy variable methods for estimation, one is to include age, period, and generation as dummy variables in the model, where the period is the actual survey year, and the age and generation are at a time interval of 5 years. The second is to include age as a continuous variable (including age, age square, and age cube at the same time), and period and generation as dummy variables into the model. Previous empirical studies have shown that these two dummy variable methods can break the collinear relationship of the original APC equation (the estimation results of the two methods in this paper are highly consistent). On the other hand, the multi-level cross-random effects model is essentially a multi-level model, that is, when estimating the changing trend of over-education, the researchers estimate the age effect through the fixed effect of low-level variables, and estimate the period effect and the generation effect through the high-level random intercept effect, which can break the collinear relationship of the original APC equation. Since periods and generations are not nested, but intersecting, this method is known as the multi-level cross-random effects method. In this paper, the estimates using the dummy variable model and the multi-level cross-random effects model are basically consistent.
▍Empirical analysis results
(1) Age effect
In order to test whether over-education is a short-term or long-term phenomenon, this paper uses the APC model to estimate the trend of over-education with age. Table 2 shows the APC model estimation results based on the dummy variable method (in which age, period, and generation are included as dummy variables), and the dependent variables are overeducation measured by the mode method and the standard deviation method, respectively.
Table 2 shows that over-education declines with age, but this downward trend mainly occurs in young adulthood, and does not decline significantly with age in middle age and beyond. Graphically graphing the coefficients in Table 2 (see Figure 1 for details), the results clearly show that over-education decreases with age, regardless of whether the mode or standard deviation method is used, but this decline occurs mainly in young people. As robustness estimates, we used a dummy variable model (age as a continuous variable) and a multi-level cross-over random-effects model (see Figure 2 for details), both of which showed that the decline in over-education with age occurred mainly before the age of 28 years, and the decline was not significant thereafter. It follows that it is possible to get rid of over-education, but this is mainly done in the early stages of career development; In middle age and beyond, it can be difficult to get rid of overeducation.
However, it is still not possible to directly determine whether over-education is a temporary or long-term phenomenon. This is because the age effect of the APC model only reflects a declining trend of over-education in youth and a continuing trend in middle age and beyond, but what percentage of the population that is over-educated in the early years of their careers is able to escape it? As a complementary analysis, we use CFPS tracking data from 2010 to 2016 and sequence analysis to characterize the trajectory of overeducation. The advantage of sequence analysis is that it can outline the trajectory of the same individual's educational matching at different time points, and the results of sequence analysis in Figure 3 show that, first, all individuals can be clustered into four educational matching trajectories, namely, continuous moderate education, continuous over-education, over-turning moderate, and moderately over-turning. Second, from 2010 to 2016, 51% of individuals continued to be moderately educated, 35% continued to be over-educated, and 8% of individuals shifted from moderate to excessive; However, only 5% of individuals went from excessive to moderate within 6 years, which is quite low, i.e., less than 1% of individuals per year went from excessive to moderate. This means that while some people can get rid of over-education, the vast majority of people have a hard time getting out of it. To further confirm this conclusion, we limited the sample to all over-educated individuals and found that less than 13% of over-educated people were free from over-education within 6 years, and by implication, 87% of over-educated people were unable to get rid of over-education. Based on the results of the APC model and sequence analysis, we found that over-education is a long-term phenomenon that accompanies the life course of individuals.
(2) Period effect
The results in Table 2 and Figure 1 clearly depict the period effect of over-education. Overall, over-education showed a downward trend with the course of the period.
First, from 2003 to 2006, over-education showed a slight upward trend, followed by a rapid decline. One background reason is the 1999 expansion of university enrolment, i.e. the rapid expansion of university diplomas in the labour market in a short period of time, when the labour market was unable to absorb this labour quickly, resulting in "inflation" of university diplomas, and the inability of many university students to quickly find a career that matches their education, which eventually led to an upward trend of over-education.
Second, over-education has shown a downward trend over time, which is diametrically opposed to the "rising theory of over-education" found in previous studies. At the same time, both the majority and standard deviation methods in Figure 1 show that this conclusion is robust. The analyses of both the dummy variable model (age as a continuous variable) and the multi-level cross-random effects model in Figure 2 also show that the conclusions of the study are robust.
This paper argues that the reason why previous empirical analyses have found that over-education increases over time is that it ignores the generational effect of over-education (i.e., the younger generation continues to enter the labor market and the younger generation has a higher probability of over-education), and that the overall over-education declines with the period when the age, period, and generation effects of over-education change are strictly distinguished and identified.
But how to understand the downward trend of over-education over time? In our theoretical analysis, we propose that the fundamental driving force of over-education change lies in the change of occupational structure. So, how has China's occupational structure changed over time? Some scholars have shown that from 1978 to 2001, the occupational structure of the mainland "tends to be advanced", and it is believed that the trend of upgrading the occupational structure will develop by leaps and bounds in the next 10 years. By further analyzing the urban data of China Labor Statistics Yearbook from 2002 to 2017, this paper finds that the proportion of occupations with high education and skill requirements, represented by professional and technical personnel, has increased rapidly, and has become the main type of occupation in the urban labor market. On the contrary, the proportion of occupations with low requirements for education and skills, represented by agriculture, forestry, animal husbandry and fishery production, has declined rapidly. Based on the distribution ratio of various occupations in 2006 and 2015, we calculated the net difference index of 17.4%, indicating that the occupational structure has achieved an upward shift of 17.4% in the past 10 years, indicating that the occupational structure in mainland China is indeed moving towards seniority.
So, how does the process of upgrading the career structure affect the change of educational needs within the profession? Based on the urban data of the China Labor Statistics Yearbook from 2003 to 2017, we depict the trend of internal education demand in different occupations (see Figure 4 for details). First, the average number of years of education within almost all occupations showed a steady growth trend, and only the education needs of managers decreased slightly; Second, among the demand for educational qualifications in different occupations, the demand for professional and technical personnel, clerical personnel and management personnel is relatively higher, while the demand for education level of agriculture, forestry, animal husbandry and fishery production personnel is the lowest. Why, then, is there an increase in the average number of years of education within a profession? We believe that there are two main reasons: first, the overall upgrading process of the occupational structure drives the growth of the demand for education within the profession; Second, the expansion of education since 1999 and the increase in investment in education in order to obtain better jobs will lead to a rapid increase in the number of highly educated people in the labor market, thus creating a "rising tide lifts all boats" effect, that is, an increase in the demand for education within the profession. On the whole, with the process of upgrading the occupational structure, the demand for education level in different occupations has increased, which is conducive to the labor market to absorb the growing number of education diplomas.
The above analysis shows that the transformation of occupational structure in the labor market has a profound impact on the changing trend of over-education. From 2003 to 2017, there was a significant transformation of the occupational structure in mainland China's cities, which showed a trend of upgrading the overall occupational structure, especially professional and technical personnel and commercial service personnel became the most important types of occupations in the urban labor market. This process has profoundly changed the demand for education within different occupations, which is manifested in the upward trend of education demand within almost all occupations, which not only raises the educational entry threshold of the corresponding occupations, but also effectively improves the ability of the labor market to absorb and digest educational diplomas, and finally promotes the overall downward trend of over-education with the period.
(3) Generational effect
We have shown that over-education has shown a downward trend over time. However, in real life, whether it is media reports or people's real perception, it shows that the problem of "difficult employment" is constantly highlighted. It follows that the level of over-education should rise. So, how to understand this phenomenon?
In fact, the perceived "difficulty in finding employment" is mainly based on the observation of the employment process of young people (especially recent college graduates entering the labor market), rather than on the observation of the entire age group. The generational effect of over-education provides a unique perspective for us to understand the over-education of different birth generations.
Table 2 and Figure 1 show the generational effect of over-educational change. First, the incidence of over-education increased among the generation born from 1943 to 1999. The use of either the mode or standard deviation method (Figure 1), the use of a dummy variable model or a multi-level cross-level random-effects model (Figure 2), indicates that the upward trend of over-education with birth generation is robust. Second, the generation born between 1953 and 1963 showed a clear downward trend in over-education, mainly because they had relatively few years of education (due to the influence of political movements such as the Cultural Revolution), so they were less prone to over-education. The over-education of the generation born after 1963 has increased rapidly, and the trend of over-education has been more pronounced in the generation born in 1980 and later. Overall, over-education shows a clear upward trend with the generation at birth.
So, how do we understand the upward trend of over-education with birth generations? We believe that the changes in China's macro education system have a profound impact on the educational access and relative educational position of different birth generations, and then affect the possibility of over-education of individuals. By analyzing the trajectory of education acquisition of different birth generations, this paper finds that: first, from the perspective of absolute years of education, the average number of years of education shows a significant increase trend with the birth generation: the average number of years of education of the generation from 1943 to 1953 is about 9.5 years, while that of the generation from 1983 to 1987 is about 13 years. Second, the composition of education has changed dramatically across generations. With the advancement of the birth generation, the proportion of junior high school education and below is decreasing; The proportion of college and undergraduate students increased slowly among the generations born before 1967 and increased more rapidly among the generations born after 1967. In particular, with the expansion of university enrollment in 1999, the proportion of those born after 1980 with a bachelor's degree has soared.
Relative changes in the composition of education within different generations will affect the occurrence of over-education. Because the composition of education changes relative to generations, the value of the same education will no longer depend only on the absolute number of years of education, but also on the relative educational position of individuals in the generation they were born in. In other words, the effectiveness of an individual's academic qualifications, or whether he or she can obtain a career that matches his or her education in the labor force, depends to a certain extent on the educational composition of the competitors in the generation in which he or she is born, and the higher the proportion of higher education in the educational composition, the greater the competitive pressure on the individual and the more likely it is to over-educate.
In order to test this hypothesis, we establish a multi-level random coefficient model of over-education based on the analytical model of relative educational position of previous scholars. Specifically, individual-level variables include age, period, years of education, and other control variables. The high-level variable is the generation of birth. At the same time, we calculate the proportion of college degree or above within different generations, use it as the explanatory variable of the second level variable, and interact this variable with the education level of the first level, and the interaction term coefficient can show whether the influence of individual education level on overeducation is affected by the internal education composition of the generation at birth. Figure 5 shows that the increase in the proportion of "college degree or above" in the birth generation will significantly increase the probability of over-education in individuals with higher education, but it will not affect individuals with lower education. Specifically, whether it is high school, technical secondary school and technical school, or college college, bachelor's degree or above, with the increase of the proportion of higher education within the generation of their birth, the probability of over-education gradually increases; However, for individuals with a junior high school education or less, the change in the probability of over-education is relatively small. In line with what we see in our daily lives, the later generations are more likely to be over-educated, mainly because of the rapid growth in the number of higher education recipients within their birth generation.
Based on the above analysis, due to the changes in the education system, there are differences in the access to education among different birth generations. From 1943 to 1997, the birth generation had more and more access to education, resulting in a clear upward trend in the overall level of education with the birth generation. Since individuals compete mainly with their peers of similar age in the labor market, the increase in overall educational attainment makes the later generations face greater competitive pressure in the labor market, which ultimately leads to an upward trend of over-education with the generation as a whole.
▍Conclusion and discussion
With the global expansion of education over the past half-century, the issue of educational matching, especially over-education, has attracted attention in the labor markets of different countries. Based on the data of the China Comprehensive Social Survey from 2003 to 2017, this study uses the age-period-generation model to evaluate the changing trend of over-education, distinguishes the forces and mechanisms driving the changing trend, and finally obtains three main conclusions.
First, over-education showed a downward trend with the course of the period, which was mainly driven by the upgrading of the mainland's industrial structure and the transformation of the vocational structure. With the process of urbanization, industrialization and modernization, the occupational structure of the mainland has generally shown a trend of seniorization, which has strongly promoted the labor market to absorb highly educated workers, so the phenomenon of over-education has actually declined over time.
Second, over-education continues to rise with the birth generation, which is mainly caused by the changes in the education system in the mainland, and with the popularization of compulsory education and the expansion of higher education in the mainland, the later the birth generation, the higher the overall education level. Since individuals compete primarily in the labor market with people of similar age, the later the generation is subjected to competitive pressures and the higher the probability of educational mismatches.
Third, over-education is a long-term phenomenon for individuals. Although the APC model suggests that over-education declines slightly with age, this decline occurs mainly in adolescence; Only a small percentage of individuals who are over-educated are able to escape from over-education, and the vast majority of individuals remain over-educated. It is important to note that over-education is not something that can be easily escaped, but can persist for a long time in an individual's career.
It should be pointed out that, firstly, this paper defines and discusses the phenomenon of "over-education" from the degree of education-job matching, mainly for individuals. As far as Chinese society as a whole is concerned, there is still a long way to go to improve the quality of the people, and education is far from excessive. Secondly, the function of formal education in schools is not only oriented to the job market, but also to provide the main basis for the allocation of jobs and income. Education plays a more important role in the transmission of cultural values and the improvement of the quality of the people. Even if there is an over-education in the labour market, the value of continuing to expand education for economic and social development is undeniable. Thirdly, the issue of educational mismatch also needs to be viewed dialectically. Appropriate mismatches may have their positive functions. For example, some positions, traditionally do not have high requirements for education, and the entry of higher education practitioners forms a mismatch in a certain sense, but on the other hand, the entry of these "over-educated" practitioners will also change the connotation and work content of the position to a certain extent, improve the quality and remuneration of services, and may attract more higher education practitioners to join and improve the degree of education matching. Finally, whether it is "educational matching" or "over-education", it is only a general description of the phenomenon in the labor market, and this article focuses on the macro trend of this phenomenon. As for what factors affect the degree of education matching, such as personal ability, personality and other unobserved characteristics may affect the employment of people with the same education level in the labor market, it is an endogenous issue that must be considered in future research on the labor market consequences of education matching.
Our description of the trend of educational matching relies on the age-period-generation model (APC model), which has become increasingly popular in social science research in recent years. Although the academic innovation of the estimation principle of the APC model is still a frontier topic in the academic community and will continue for a long time to come, the APC model still has its irreplaceable value as an important method to understand the trend of social change. By introducing the APC model into the study of over-education, we can more accurately identify the different temporal trends of over-education than previous studies, which is conducive to a more comprehensive understanding of the change trend of education matching. Of course, different estimation methods for APC statistical models have their own limitations. For example, in order to break the collinear relationship in the APC model, the early common method is to add constraint constraints to the APC model, such as limiting the equality of any two coefficients in the age (or period and generation) effect (also known as local limiting), but researchers often lack prior information and cannot know which constraint is consistent with reality, so this method is sensitive to the selection of constraint constraints, which may lead to inconsistencies in age, period, or cohort estimates. Later, scholars proposed the intrinsic factor method (intrinsic estimator), which breaks the collinear relationship of age, period and generation through the parametric spatial decomposition of linear algebra, but this method has also been challenged, and critics believe that the implicit assumptions of the endogenous factor method are too strong and difficult to establish in reality. The dummy variable grouping model and the multi-level cross-random effects model used in this paper do not need to rely on prior information and premise assumptions, and their estimation principles are easier to understand and accept by researchers, and the two methods are consistent in estimating the age, period, and generation effects of overeducation.
Despite the above limitations and topics to be expanded, this study generally enriches our description of the changing trend of educational matching (over-education) and its theoretical explanation. Western scholars' research on the changing trend of over-education mainly focuses on the changes of individuals with age and career. Following the previous research context, this paper incorporates the two temporal dimensions of period and generation into the trend of over-education, focuses on the analysis of the impact of macro institutional structure changes on over-education, and expands the research framework of the trend of over-education. At the same time, based on the trajectory of the change of vocational structure and education system, we propose a new theoretical explanation for the trend of over-education, that is, the theory of occupational structure upgrading is used to explain the period change of over-education, and the relative educational position theory is used to explain the generational change of over-education. The empirical results also support these two theoretical explanations, that is, the changing trend of over-education has been deeply branded with the imprint of industrial structure transformation, vocational structure upgrading and education system change. These findings help us understand how the phenomenon of over-education has changed over time and across generations.
On the basis of the conclusions presented in this paper, the following aspects need to be considered in the future macroeconomic adjustment of the labor market:
First of all, for the younger generation, they are not only the main beneficiaries of the shift from elite education to mass education in higher education, but also the main bearers of "over-education" caused by the imbalance between education and economic and social development. How to create vocational jobs that match education requires systematic coordination and cooperation between industrial policy and education policy.
Second, further reform and development of the education system requires special attention to the design of institutional linkages between education and the labour market. Both vocational and higher education play an irreplaceable role in the employment of the labor force. In China's future education reform, it is necessary to strengthen the development scale, training quality and social recognition of vocational education, so as to strengthen the connection between the education system and the labor market. At the same time, the development of higher education should improve the quality of training and employment, and promote the education-career matching of highly educated workers by strengthening the cultivation of basic skills and comprehensive qualities of individuals.
Thirdly, the development of higher education in mainland China needs to be leading. The mainland's future economic growth will pay more and more attention to connotation and quality, and put forward higher requirements for scientific and technological innovation. The further upgrading of the industrial structure and the development of high-end manufacturing and service industries not only provide opportunities for reducing the phenomenon of over-education, but also pose new challenges to the construction of a high-quality education system and talent training. Education matching can be included in the measurement of employment quality, so as to play a feedback role in college admissions, major settings, and talent training.
Finally, education not only has employment and economic goals, but also has far-reaching significance for improving the quality of the people, promoting social civilization and national rejuvenation. The Fifth Plenary Session of the 19th Central Committee of the Communist Party of China put forward the long-term goal of "building a cultural and educational power" by 2035. Strengthening the dynamic research and policy design of the transition from school education to the labor market, striving to unleash the potential of each individual, and creating an institutional environment in which "people can make the best use of their talents" is the proper meaning of achieving this long-term goal.
This article was originally published in Social Sciences in China under the title "The Changing Trend of Education Matching in China's Urban Labor Market: A Dynamic Analysis Based on Age, Period, and Generation Effects". Welcome to share personally, please contact the copyright owner for media reprinting.
There is no upper limit on tipping, which supports cultural reconstruction
Subscription service hotline: 010-8559710713167577398 (WeChat) Working hours: Monday to Friday 8 a.m. to 8 p.m