The changing faces and roles of communist party membership in China: an empirical analysis based on CHIPS 1988, 1995 and 2002

ABSTRACT This paper analyses the evolution of the Chinese Communist Party (CCP) using micro-data from three nationwide surveys. I examine how party membership affects people’s choice of employment, occupational attainment, and earnings, while including changes in the function of individual attributes and human capital. I endeavor to cover both urban and rural areas and gain a dynamic understanding of the topic. The proportion of people who joined the CCP as they grew older was relatively stable in the three surveys, while the overall education level of party members increased rapidly with time. Party membership and education had a statistically significant and positive effect on non-farm employment choice and earnings, but those effects decreased rapidly over time in rural China. In contrast, party membership and education exerted more and more of an effect on employment choice, occupational attainment, and earnings in urban China. Furthermore, in both rural and urban areas, the rate of return to education and the earnings premium for CCP membership are converging at the same level. This is evidence worthy of attention as it indicates that labor markets throughout the country are becoming integrated.


Introduction 1
Since the 1980s, the Chinese Communist Party (CCP) has promoted a systemic transition from a planned economy to a market-oriented economy and, despite the ongoing actualization of marketization, has been able to maintain a one-party political system. A system that draws capable individuals from various fields has been established, 2 and a rise in the regime's operating prowess has been noted as a background factor. 3 Party membership increased from 39.65 million in 1982 to 86.69 million by the year 2013. On the other hand, the proportion of party members among adults aged 18 and over (a requirement for joining the Party), increased only slightly from 6.6% to 7.8%. The rapid increase in party membership is a phenomenon that can be attributed to increases in the population and changes in the age structure. The actual supply of party memberships (approval of applications for party entrance) did not expand drastically. 4 In recent years, the CCP Central Committee has published the number of total party members, the number of new party members, and party composition by various personal attributes, such as education level, and occupation. By combining these statistics, we are now able to grasp a rough picture of the membership of the CCP. However, even when we make multifaceted observations regarding categories such as sex, age, ethnicity, educational background, affiliation, or place of residence (i.e. rural or urban areas, different provinces), there are many questions that remain difficult to answer. The determinants that stipulate acquisition of party membership, the degree to which membership effects employment, attainment of occupational status and income, and moreover, how these have changed in the process of economic transition cannot be well understood given just this published information.
In this paper, we will consider the profile of the individuals who become CCP members, as well as look at the determinants of membership acquisition via analysis of multiple large sample surveys that cover the entire country. Defining the various functions and trends of CCP membership will be our principal research objective. Specifically, to the degree that the data allows, we will econometrically analyze (1) the percentage of the adult population 18 and over which holds party membership, and the characteristics of this group, (2) the factors that stipulate acquisition of party membership, and (3) the degree to which party membership affects or directs occupation, occupational advancement, and income. Meanwhile, we will also perform a careful examination of the personal attributes, area of residence, characteristics of place of employment, and especially the levels of education (reflecting human capital). Through an examination of data from multiple nationwide surveys, including data from both rural areas and cities that were absent from prior studies, we will attempt to produce a profile of CCP members that goes beyond what can be deduced from published information, and gain a dynamic understanding of the changes that marketization has caused in the functions of CCP membership. This is different from a political science approach dealing with the organization and function of the CCP, but we believe that it holds great significance in forming a deeper, more versatile understanding of the Party.

Survey of prior research
Over the past 20 years, many nationwide social and economic surveys have been conducted, and academic research that utilizes micro-data is plentiful. 5 Research that 4 From the identity function (number of party members ÷ adult population) × (adult population ÷ general population) × general population, we can derive the relational expression (yearly percent increase in party members ＝ yearly average proportion of party members ＋ yearly number of adults as a percentage of the population ＋ yearly percent change in the general population). Based on that function we can estimate the average percent increase in party members and the average percent contribution of three factors. For example, in the 28 years from 1982 to 2010, the number of party members grew by 2.6% on average, but increases in the proportion of party members contributed to only 20% of that. Increases in the proportion of adults in the population and increases in the overall population contributed 40% each to the increase. Official statistics on the number of party members published by the CCP Central Committee were actually based on general population and adult population census released by the Bureau of Statistics. 5 Riskin, Zhao and Li, China's Retreat from Equality; Li, Sicular and Gustafsson, Zhongguo Jumin Shouru; Li, Sato and Sicular, Rising Inequality in China.
focuses on the effects of party membership on occupation, occupational attainment, and income has also accumulated. In China's transition from a planned economy to a market-oriented economy, the principle of competition has been introduced to various fields. In this context, it is thought that the value of political capital, which has a weak relationship to productivity, would decrease as the value of human capital, a factor that contributes positively to productivity, increased. 6 For example, a study utilizing micro-data from a farm survey performed an empirical analysis under the hypothesis that marketization weakens the earning premium for political capital, while raising the rate of return for human capital. 7 However, in related empirical studies, there are many documents that do not support Nee's hypothesis. Much of the literature supports the positive influence of both human and political capital on individual employment, promotion, and earning as marketization of the economy progresses. 8 Many studies utilizing micro-data from nationwide surveys such as CHIPS (The Chinese Household Income Project Survey) or CGSS (The Chinese General Social Survey) have been presented, 9 and many interesting insights have been gleaned from the results of the analysis of basic information on party members and the income premium associated with CCP membership (the party premium). 10 First, according to estimated values for party members as a proportion of the general population, terminal education record (or years of education), and ethnicity from the late 1980s to the early 2000's in China, (1) the average level of education of the population increased in both rural and urban areas, (2) the percentage of CCP members in the adult population 18 and over trended upward, (3) there was a large disparity in both the proportion of party members in the population and average years of schooling between rural and urban residents, (4) the proportion of CCP members was higher in higher income strata (and lower in lower strata), and (5) there was a large gap in the proportion of male and female party members.
Second, party members received higher incomes compared to the population at large, and this trend became stronger with time. The lower the social stratum, the greater the increase in the party premium. Furthermore, party membership had a significant effect on priority selection in schools and job hunting. However, there are some results that show a reduction in the party premium resulting from marketization of the economy. 11 6 Among questionnaire surveys and household surveys conducted in China, personal information questions regarding CCP membership are often seen from foreign/domestic joint research teams. This tendency can also be seen in surveys from Japanese studies of farming households in China. See Sato, "Keizaiteki Bunka"; Yan, Noson kara Toshi e; Chugoku Nominko. 7 Nee, "A Theory of Market Transition". 8 Yan and Wei,"Chugoku no Daitoshi". 10 More information on the surveys used in studies on social stratification in party membership and education and the effects thereof on social mobility can be found in Yan and Wei,"Chugoku no Daitoshi". 11 The party premium was largely the same between the urban and rural sectors. See Li, Lu and Sato, "Power as a Driving Force of Inequality".
Third, the rate of return to education as related to productivity gives a measured result that nearly aligns with the earnings function. That is to say, the rate of return to education increased with economic marketization, and that increase was greater in more competitive sectors (i.e. self-employed or foreign and joint ventures). 12 There are a number of excellent analyses that have been performed using high quality micro-data and a diversity of quantitative methods, but in order to establish a profile of CCP members and develop a multifaceted understanding of the function of party membership, the following three types of analysis should also be undertaken: (1) analysis that takes into account both urban and rural areas in order to form a complete picture of the Communist Party in China, (2) research that utilizes multiple surveys to gain a dynamic understanding of changes over time, and (3) analysis that takes full advantage of the properties of large sample surveys, creates a profile of CCP members from various angles, and inspects the features of party membership. This study includes all three types of analysis, and thus goes a long way towards filling in the gaps in previous research.

Data
The data used in this study will be micro-data from the CHIP Surveys (CHIPS) conducted in 1988, 1995, and 2002 by the Chinese Academy of Social Sciences (CASS). CHIPS is a nationwide survey conducted using the State Statistics Bureau's household survey system. The number of rural households included in the three surveys is 10,258 (covering 28 provinces), 7998 (19), and 9200 (19) respectively, and the number of urban households included are 9009 (10), 6931 (11), and 6835 (12) respectively. The staff involved in the survey were exclusive members of the Statistics Bureau at the local level, and this fact assures the quality of the statistical data. For this reason, it is believed that a full picture of the country can be inferred from the results of an analysis based on CHIPS data. 13 The year 1988, when the CHIP Survey was first performed, was a period from which can be seen the basic results of a shift in the priority of systematic reform from rural areas to urban areas. However, reform of state-owned enterprises had not yet begun at a meaningful scale, and 70% of employees worked in the state-owned sector (state agencies, non-profit institutions, or state-owned enterprises). In the rural sector, especially on the outskirts of cities and along the coast, township and village enterprises (TVEs) accomplished rapid growth. However, the number of individuals employed therein stalled at 95.45 million (23.8% of rural employees) and the number of migrant laborers (nongmingong) moving to the city for work is estimated to be approximately 27 million during this period. It can be said that marketization was in its initial stages.
Following the 1989 Tiananmen Square protests and Deng Xiaoping's Southern Tour in 1992, the reform of state-owned enterprises was accelerated and restrictions on internal labor migration were relaxed. By 1995, when the second CHIP Survey was conducted, the population employed by the state-owned sector had fallen to 59% in urban areas, the number of non-farm workers in rural areas had risen to 33.5%, and the total number of nongmingong had surpassed 56 million. Marketization reform was rapidly moving toward a transformative stage.
In the seven years leading up to the third CHIP Survey in 2002, sweeping reforms to state-owned enterprises had been implemented under the leadership of Prime Minister Zhu Rongji, and China had been inducted into the World Trade Organization (WTO). In 2002, the proportion of workers employed by the state-owned sector was just 28.5% in urban areas, and the number of nonmingong exceeded 100 million. It can perhaps be said that the Chinese economy as a whole had attained a high level of marketization by 2002.
As is apparent from the above description of the data analyzed in this paper and the characteristics of the periods in which it was collected, the distinctive feature of this paper is the attempt at a dynamic understanding of the changes in China's labor market via analysis of CHIPS micro-data, which shows the influence of marketization reform processes.

Profile of CCP members as seen within CHIP surveys
In this section, we will analyze micro-data covering both rural and urban areas from the three CHIP Surveys and develop a quantitative profile of CCP members that was not present in previous studies. More specifically, we will consider the proportion of party members in the total population and the composition, occupations and affiliation of party members, as well as the determinants for joining the Party both overall and in urban or rural areas respectively.

Characteristics of party members in terms of education background
According to the terms of the Chinese Communist Party, one must be 18 years of age or older in order to apply to join. For this reason, we will focus on the population aged 18 and over in each survey, calculate the number of adults that indicated that they were CCP members in rural and urban areas separately, and then find the percentage of the adult population that is made up of party members. Next, we will calculate the percentage of party members and their distribution by sex, ethnicity, and highest levels of education completed. As noted previously, the CHIP Survey is a large sample survey covering the entire country, and as such, the overall situation of the nation as a whole can be inferred from its results.
Let us give a description of the basic characteristics of the Communist Party members in China in Table 1. In the 14 years from 1988 to 2002, the number of CCP members as a percentage of the adult population increased from 5.9% to 7.7% and 22.7% to 26.1% in rural and urban areas respectively. Residents of urban areas were three to four times more likely to be CCP members than residents of rural areas, and moreover this difference shrunk only slightly during this period. In the same period in both rural and urban areas, the proportion of party members among adults 18 and over increased by only 2 or 3 points, despite an increase in the adult population of around 10 points. From this fact it can be deduced that the increase in supply of party memberships was being strictly controlled.
When looking at individual attributes and education we can see that there is a large gap in party membership acquisition between individuals in different categories. Men and individuals with college degrees and above were far more likely to have joined the Party than women or individuals with a senior high school education or below respectively. On the other hand, the difference is comparatively small between Han and minority groups. Moreover, the proportion of party members with higher educational backgrounds saw an overall decline with time in urban areas. This is likely due to the fact, in the context of the rapid expansion of access to higher education, many young people with high educational qualifications were not approved for membership in the Party.

Party membership and occupation
In China, Communist Party membership is both a sign of social status and a difficult qualification to acquire. As long as dues are paid in accordance with regulations and no major mistakes that would result in expulsion are made, one will always be a party member. Neither employment status, nor area of residence has any effect on maintaining membership in the CCP. Of course, if one does not have steady employment, party membership may not hold much significance. The distribution of party members by occupation is available in Central CCP statistics, but we cannot know the distribution or percentage of party members in each occupation. Here, we will take relevant data items from the three CHIP Surveys, tie them to party membership, and calculate that information.
First, let us look at the basic situation of party members by occupation. As shown in Table 2, in the 14 years between 1988 and 2002, as the percentage of party members aged 18 and over increased from 23.8% to 29.0%, an increase of 5.2 percentage points, we can see that the proportion of party members made up by executives in state agencies, party organizations, enterprises and public institutions decreased by 5.1 percentage points, the proportion of professionals, technicians and clerical staff remained roughly the same, and the proportion of workers increased by 5.7 percent points. However, if we look at the specialization coefficient of the distribution of party members which can be obtained by dividing the distribution of adults 18 and over by the number of CP members, the probability that an executive will be a member of the CCP is remarkably high. For example, the proportion of executives in the adult population was only 6.5%, but the proportion among party members was 22.2% in the 1988 CHIP survey. Supposing we ignore other factors, the probability of an executive becoming a party member was 3.44 times the average. In contrast, the probability of a worker being a party member was only 30% higher than the average. In the 1995 and 2002 surveys, the proportion of executives as a percentage of party members declined, and that of workers increased. Despite this, a large disparity between the two still existed. The opportunities for relatively high-level professionals/technicians and clerical staff to join the Party showed slight decreases in the 1995 and 2002 surveys. In China, it is said that the Communist Party championed the working class, but executives acquired greater opportunity from gaining party membership.

Party membership and affiliation (workplace)
In the different CHIP Surveys there is a question regarding the nature of individual affiliation, but the responses available in each vary subtly. In order to clarify how the nature of employee affiliation changes with time or if the affiliation of people who hold party membership trended in a meaningful way, we will consolidate similar items below and aggregate. Specifically, we will consider state agencies, non-profit institutions, and state-owned enterprises (including corporations where the state controls more than half of shares) as the state-owned sector and classify other types of enterprises into collective enterprises, private enterprises, foreign and joint ventures, and other ownership sector. Table 3 shows the number of party members and non-party members, the promotion of CP members, and the distribution of party members by affiliation.
According to this table, among individuals who work in the state-owned sector, the proportion of CCP members remained relatively unchanged at approximately 28% between 1988 and 1995, but rose by over 8 percentage points to 36.2% by 2002. In urban collective enterprises, the proportion of party members among employees rose continuously, with the rate of growth increasing after 1995. Of particular note is the rapid increase in the number of party members working in private enterprises in the seven years following 1995 (increasing by 9.1 percentage points). In contrast, the percentage of party members in foreign and joint ventures changed very little.
If we look at the distribution of party members by affiliation, we can see that the percentage of individuals in the state-owned sector dropped ten percentage points between 1988 and 2002. The implementation of extensive reforms in the second half of the 1990s that resulted in a broad reduction in state-owned enterprises and the overall number of state employees is a likely factor here. Relative to this, the comparative percentage and overall number of party members working in private, foreign and joint ventures, and other sectors increased. The position of party members changed greatly with the advancement of economic marketization. This was likely a background factor in the era in which the "Three Representations" appeared in the 16 th National Congress of the Chinese Communist Party, and led to the recognition of party members working in the non-state sector.

Determinants of party membership acquisition
Below is a quantitative analysis of the relationship between party membership and a number of variables, utilizing a multiple regression model that examines the adult population aged 18 and over. The model takes party membership as the explained variable (1 if a member of the CCP, 0 otherwise). The explanatory variables are gender (male = 1, female = 0), age (years), race (Han = 1, minority = 0), marital status (married = 1, single = 0), and education (dummy variables undergraduate, college graduate, senior high school, and primary school or below with junior high school graduates as a benchmark group). We run models using data sets from each individual CHIP Survey, as well as a model that pools the three surveys into one data set. Table 4 shows the results of a logistic model. Based on the estimates from each model (odds ratio Exp(B)), we can report the following findings regarding the determinants of party membership acquisition.
First, compared to women, the odds ratio that a man will acquire party membership (a value found by dividing the probability of successfully acquiring membership by the probability of failing to or not attempting to acquire membership) is approximately three times higher. However, the difference between the two tended to shrink as time passed, decreasing from a 4.3 times greater probability for men in 1988, to 2.6 times in 2002. It is not reflected in the table, but if only the urban sector is observed, those odds ratios decreased from 3.5 to 2.0 in the same period. It can be said that equality of opportunity in entrance to the Party increased in the period under consideration.
Second, while the probability of successfully acquiring party membership increased with age, the trend reversed past a certain age. This is based on an inverse-U relationship found in the above model.
Third, based on the pooled data from all three surveys, the difference in opportunity for party membership acquisition between minority groups and Han was shown to very slightly disadvantage Han people. No statistically significant difference between ethnicities was found in models using data from each survey. A statistically significant result that showed disadvantageous treatment of minority groups or prioritization of Han people in acquiring party membership was not found.
Fourth, the effect of education on party membership acquisition was significant and positive. The higher the educational level of an individual, the higher the chance of acquiring party membership. This effect increased with the passage of time. As Table 4 shows, the odds ratio of acquiring party membership for an undergraduate as compared to a junior high school graduate increased from 2.7 in 1988 to 3.5 in 1995, and increased even further to a factor of 8.2 in 2002. This rapid upward trend can be confirmed for college graduates as well. In contrast, it is becoming more and more difficult to obtain party membership for individuals with an elementary school education or lower.
Fifth, the odds ratio of acquiring party membership differed significantly based on whether the individual resided in urban or rural areas. However, this trend shrank with time (odds ratio of urban versus rural: 2.6 times in 1988, 2.3 times in 1995, 1.6 times in 2002). In the three largest regions, the Eastern, Central, and Western regions, there was a statistically significant difference in opportunity to join the Party, but the degree was only slight.

The functions of party membership and changes therein
As is made clear by much of the previous research, party membership provides similar advantages in the form of political capital to the human capital that reflects an individual's abilities, such as education. Below we will examine the function of party membership in both urban and rural areas. More specifically, we will empirically analyze the role that party membership plays in employment choice, attainment of social status, and earnings, and we will examine whether this role changes over time, while considering the effects of education.  Note: *** shows significance level at 1％,** at 5％, and * at 10％, respectively.
4.6. The relationship between party membership, occupation and earnings in rural china 4.6.1. Non-farm employment choice amongst the farming population We would like to consider whether party membership as political capital affects employment choice and earnings. First, we must outline the employment situation of the rural population. Since the 1980s, the non-farm sector has grown rapidly, and population movement from rural areas to cities has increased since the 1990s. 14 In this context, the proportion of individuals working in non-farm sectors has continued to rise amongst members of rural households. According to calculations based on CHIP Survey data, the percentage of individuals with non-farm earnings in rural populations aged 16 or over increased rapidly, growing from 6.9% in 1988 to 21.9% in 1995, and again to 34.5% in 2002.
What effect did party membership have on access to these various non-farm occupations, including access to posts in rural local government? Here, we examine the determinants of non-farm employment for individuals aged 16 or over at the time of survey. Specifically, we estimate a logistic regression model where the explained variable is 1 if an individual receives non-farm earnings and 0 otherwise. The explanatory variables are age, gender, ethnicity and other attributes, education, party membership, and more. Table 5 shows the results of models using data from each individual survey, as well as a model using pooled data from all three.
As can be seen in Table 5, the coefficients representing individual attributes were shown to be significant in affecting employment choice in each of the surveys. Specifically, we can point to the following items: (1) opportunities in non-farm occupations increased with age, but decreased again past a certain age; (2) men had a higher probability of choosing non-farm occupations than women, and that tendency further trended upward; (3) the odds ratio for a Han person choosing a non-farm occupation was high and trended upward; and (4) in the East, Central, and West regions, the opportunities for members of rural households to find non-farm work converged toward equality.
What we would like to pay particular attention to is whether the effect of education reflecting human capital, and CCP membership reflecting political capital, on employment choice has changed. As shown in Table 5, the odds ratio of non-farm employment choice increased by 1.169 with each additional year of education. The odds ratio was 1.067 in 1995 and 1.034 in 2002. The effect of education on non-farm employment choice weakened while other factors held constant. On the other hand, the effect (in the form of an odds ratio) of party membership remained at a high level throughout the three surveys. In the 1988 and 2002 surveys, the odds ratio for an individual with party membership versus a member of the general public to choose non-farm employment was a factor of approximately two. In 1995, it was a factor of three on average.
When looking at employment choice for members of a rural household, the effect of education was meaningful and positive as ever, but the degree of this effect decreased greatly. On the other hand, however, the effect of party membership for this group continued to be a significant factor.
In order to emphasize the difference between common party members and executive party members, we utilized a new model which distinguishes between common and executive party members in place of the party member variable in each of the models represented in Table 5. Based on this model, we can see that the odds ratio for non-farm employment choice between common party members and members of the general public is a modest 1.22 times, 1.81 times, and 1.17 times in each survey. For executive party members, however, the odds ratio is 6.6 times, 23.3 times, and 67.6 times respectively. Executive party members had a clear advantage in gaining access to non-farm employment.

Effects of party membership and education on non-farm earnings in the rural household
Next, let us consider the effect of party membership on non-farm earnings. Here we employ an OLS regression model with non-farm earnings as the explained variable, but we sort party members into common and executive members. Table 6 shows the results of regressions utilizing data from each of the surveys, as well as a model that uses pooled data from all three.
First, in regard to the relationship between individual attributes and earnings, let us consider the regression coefficients in the table and their respective levels of significance. From the 1988 rural survey, we were able to confirm that factors such as gender, age, or ethnicity for rural household members had almost no statistically significant effect on the amount of non-farm earnings received. In the 1995 survey, gender and age again showed no meaningful effect on non-farm earnings, while Han people earned close to 38.7% more non-farm earnings than minority groups. Then in the 2002 survey, all three factors came to show statistical significance. We can observe the following findings: earnings increased with age in youth and then decreased again past a certain age, men had higher earnings than women, and the earnings of Han people were higher than those of minority groups.
Second, education has a significant positive effect of 1% on non-farm earnings, but that effect decreased over time. The rate of increase in earning that resulted from a 1 year increase in years of schooling, or the rate of return to education, declined from 10.3% in 1988 to 7.1% in 1995, and again to 4.8% in 2002. It is possible that as the average years of schooling for members of rural households continued to increase, a phenomenon known as academic inflation occurred. In a time when the average level of education was low, those with lower educational attainment could still find work relatively easily. However, rapid expansion in education perhaps meant that higher educational attainment became underutilized as the level of education in rural households increased with no proportionate creation of jobs. 15 Third, while there is a positive and statistically significant correlation between party membership and earnings, there is no doubt that our findings show a remarkable decline in the party premium. In the 1988 survey, non-farm earnings for party members as compared to members of the general public were 74.1% higher. In contrast, this fell to 20.4% and 26.6% for the 1995 and 2002 surveys respectively.
Amongst party members, however, there was a difference in the party premium between those who held no managerial duties and officers in government, non-profit institutions, and other various organizations. Actually, as is shown in the last row of Table 6, there was only a slight increase in earnings of 8.9% between the general public and common party members, in contrast to a greater increase of 59.7% for executive party members.

The relationship between occupational status and earnings in urban china
Let us continue to look at the effects of party membership and education on employment choice, attainment of occupational status, and earnings in urban China based on CHIP Survey data. We will divide the analysis into three steps. First, we will separate valid respondents into two categories based on the place of work of urban residents. In this step, state agencies, non-profit institutions, state-owned enterprises and corporations where the state owns a majority of stock will be termed the state-owned sector. Meanwhile, privately-owned enterprises, foreign and joint ventures, and the self- employed will be termed other. Using these categories, we will clarify the determinants of employment in the state-owned sector at the time of each survey. Second, we will consider the determinants of employment in executive positions in each type of organization at the time of survey. We will analyze this by separating respondents to questions regarding occupation into two groups: Executives in various organizations, and individuals engaged in other occupations, such as professionals/technicians or clerical staff. Third, we will estimate an earnings function that expresses the relationship between individual attributes, education, and party membership. We will then examine the effects of each of these factors on earnings, the degree to which each factor influences earnings, and any trends that may exist therein.

The determinants of formal employment in the state-owned sector
When looking at employment choice in the state-owned sector, we will utilize a logistic regression model that takes state-owned sector formal employment as 1 and other employment as 0. As explanatory variables, we will take gender, age and age squared, race, level of education (with 5 strata), and an interaction of education level and party membership. Table  7 shows the results from models using data from each survey. Below, we shall focus on the regression of interaction between education level and party membership. Let us analyze and consider the relationship between employment choice, education, and party membership.
To begin with, whether or not an urban resident was able to find employment in the state-owned sector was strongly influenced by education at the time of the 1988 survey. If we take junior high school graduates as a benchmark, the odds ratio of an elementary graduate working in the state-owned sector was no more than half. For a senior high school graduate, college graduate, and undergraduate, the odds ratios versus a junior high school graduate were 1.96, 3.94, and 7.33 times higher respectively. Note: *** shows significance level at 1％,** at 5％, and * at 10％, respectively.
When looking at the relationship between state-owned sector formal employment and CCP membership, the following trends can be inferred from the coefficients. Compared to the general public, CCP members had a higher probability in general of working in the stateowned sector, but it was not necessarily the case that higher education levels amongst party members meant a higher chance of working for the state. Rather, it was the opposite. At the time of the 1988 survey, the odds ratio of employment in the state-owned sector decreased linearly from 2.6 to 1.5 between elementary graduates and those with undergraduate degrees or higher. A similar phenomenon can be observed in the results obtained from the 2002 survey data. The odds ratio fell even more as education level rose. The difference between college graduates and members of the general public disappeared. Those with undergraduate degrees or higher came to avoid the state-owned sector more than the general public.
At the time of the 1995 survey, CCP members in all educational strata had a higher probability of working in the state-owned sector than others, and the trend is particularly conspicuous for those with an elementary education or below. In 1997, the reform of stateowned enterprises began in earnest. It is conceivable that over the following five years, many capable individuals in state agencies moved to other sectors such as foreign and joint ventures, as privatization advanced and state-run businesses began to disappear.

The conditions for becoming an executive
Next, let us think about the requirements for becoming a high-ranking member of an organization. Here, we run a logistic regression model that takes respondents who indicated that they were an executive in state agencies, non-profit institutions, or other various organizations as 1, and respondents in other occupations as 0. The targets of analysis are adult respondents aged 18 and over, and CCP members within the population. For the former group, variables that express individual attributes, years of schooling, and party membership were used. For the latter group, ages were stratified and dummy variables for age of joining the Party (under 29, 30s, and 40 and up) were utilized. Table 8 shows the results of models using data from the three surveys (due to limitations in the data, a similar analysis on the party members of the 1988 survey cannot be performed).
Over the three CHIP Surveys, the proportion of executives changed somewhat, but only by a few percentage points. What kind of qualities do the executives in various organizations possess in terms of individual attributes, political capital, and human capital? Let us describe our findings as shown in Table 8, expressing the conditions necessary for one to assume office as an executive in an organization. Education in particular has a positive effect on acquiring the position of an executive. Furthermore, the tendency strengthens with time. The odds ratio for this rose from 1.121 in 1988 to 1.344 in 2002. In contrast, the effect of party membership on employment in a high-ranking position weakened somewhat. This odds ratio decreased from 7.9 to 7.5, and again to 5.5 over the three surveys. Despite this finding, we can still say that the absolute influence of the party membership is very strong compared with the general public.
Being a party member did not uniformly guarantee a high-ranking position for all, however. As the results for CCP members show, men had a markedly stronger tendency to become executives, and this trend increased even more with time. The advantages for Han CCP members observed in the 1995 survey disappeared by the time of the 2002 survey. Finally, the effects of education for party members, as with the result for members of the entire adult population, grew more and more significant with time. Table 8. Determinants of being an executive in urban China (total population 18 years or above, and party members). Note: *** shows significance level at 1％,** at 5％, and * at 10％, respectively.
One interesting point was that age at the time of joining the Party had a significant correlation with employment as an executive. In the 1995 survey, those who joined the Party in their 30s or before had a remarkably higher probability of finding a highranking position as compared to those who finally joined the Party in their 40s or later. By the 2002 survey, however, the difference between respondents aged 40 and up and those in their 30s was no longer significant, and the comparative advantage for those under 30 had decreased. This means that the sooner in life one joined the Party, the greater the probability of finding success later. However, this effect decreased over time.

The earnings function for urban residents
Finally, let us consider the relationship between party membership and earnings. Table 9 shows the earnings function for employees in urban China. From these regression coefficients and the associated levels of significance, we can deduce the relationships between each explanatory variable and earnings. When compared to the earning function for members of rural households, the effects here of individuals' attributes, education and party membership varied greatly, and each trended in a different direction.
First, consider the relationship between age and earnings. In every survey, earnings show an inverse-U relationship, increasing with age before turning downward past a certain point.
Second, there was a disparity in earnings between men and women, and this gap widened over time. Compared to women, the earnings of men were 13.6% higher in 1988. This gap was 11.0% in 1995 and 15.1% in 2002. Third, a statistically significant difference in earnings between Han people and minorities could not be found, but the results do tentatively show higher earnings for minorities. This is the opposite of the situation for non-farm workers in rural China (Han people have an advantage).
Fourth, the rate of return to education was 2.1% in 1988, no more than a fifth of that of rural households. This grew to 4.1% in 1995, closer to the number for rural households, but overtook rural households in 2002, growing to 7.4% (compared to 4.8%). This continuing increase in the rate of return to education is the opposite of the earnings function found for non-farm workers in rural households.
Fifth, the party premium for urban residents was also very different than that of members of rural households. The earnings for urban residents who hold party membership were around 8% higher than those of members of the general public. This is only one ninth of the party premium for rural households in 1988. By the 1995 and 2002 surveys, although the earning premium of CCP membership for rural residents had fallen rapidly, the party premium for urban residents still failed by far to reach the rural residents' level. We can say that the earnings premium for CCP membership in urban China was relatively stable, and was less significant than the effects of the CCP premium in rural households.
Sixth, when we add two interactions (executive of organization × party, and employment of state-owned sector × party) as dummy variables to the model, the party premium is found to change greatly. Just working in the state-owned sector granted an earnings increase of around 20% over those who worked in other sectors, and employment in a high-ranking position meant higher earnings. One very interesting trend was that executives in various organizations saw a drastic increase in earnings premium as time passed (4.6% in 1988, 9.7% in 1995, and 22.5% in 2002). By holding party membership, it became easier to find work in the state-owned sector, and the Table 9. The earnings function in urban China (at age 16-64, OLS). Note: *** shows significance level at 1％,** at 5％, and * at 10％, respectively.
probability of advancement to high-ranking positions increased. Consequently, those who held party membership had comparatively higher earnings.

Summary of results
Here, we will gather and summarize the results of analysis of the effects of party membership on employment choice, occupational attainment, and earnings. First, party membership had a statistically significant and positive effect on the degree of non-farm employment choice and the amount of non-farm earnings found in rural areas. Further, the effect of party membership on non-farm employment choice was relatively stable over time, while the effect on non-farm earnings weakened gradually. In contrast, the effects of education on both non-farm employment choice and non-farm earnings weakened markedly with time.
Second, in urban areas, education and party membership were found to be determinants of formal state-owned sector employment; the tendency to work in the state-owned sector increased with educational attainment, but this trend weakened over time. Those who held party membership were more likely to work in the state-owned sector than the general public, but that relationship was not necessarily stronger for party members with more education. Furthermore, among party members, the probability of working as an executive in an organization increased as the age at which individuals joined the party decreased, and was higher for men than for women. Education became increasingly important for finding work as an executive of an organization. Han people were not at a significant advantage when other conditions remained constant.
Third, when considering the degree of non-farm employment choice in rural areas, the earnings premium for common party members was comparatively small. However, for executive party members working in high-level positions, the effect was very large and increased further with time. In urban areas, on the other hand, the party premium not only remained constant regardless of the organization's form of ownership, but party membership led to attainment of higher ranking positions in organizations and higher earnings.
Fourth, over the 14 years covered in the surveys, the rate of return to education and the earnings premiums for CCP members moved in opposite directions in rural and urban areas. If we look at overall levels, both the rate of return to education and earnings premiums for CCP members appear to be moving toward convergence. This hints that marketization is deepening, and integration of labor markets is proceeding at a nationwide level.

Conclusion
In this paper, we analyzed how the composition of the group of individuals who belonged to the Chinese Communist Party changed over time using micro-data from three nationwide surveys conducted at different periods between which the degree of marketization changed, and the social and economic structure in China was transformed. We examined how party membership affected people's choice of employment, occupational attainment, and earnings, while including changes in the function of individual attributes and human capital. Previous studies utilizing CHIPS data performed chronological analyses of just urban or rural areas, or handled both together, but most dealt with only one or two surveys. In this paper, we endeavored to cover both urban and rural areas over all three periods to gain a dynamic understanding of the transformative process of marketization as a whole, and we believe we have approached that goal. Based on the results of our empirical analysis, we offer the following major conclusions.
First, the proportion of people who joined the CCP as they grew older was relatively stable in the three surveys, while the overall level of education of party members increased rapidly with time. Between men and women and between urban and rural areas, the probability of joining the Party was noticeably different, but that gap trended downward. A noticeable difference in CCP entrance between ethnicities cannot be detected. The supply of party memberships (approval of applications for party entrance) was strictly controlled and increases in the number of CCP members largely stemmed from increases in population and changes in age demographics. Emphasis was placed on selection of party members from highly educated and younger groups, and highly educated people were brought into the Party from various strata of society. This phenomenon could be called a sharpening and streamlining of the Party.
Next, in both urban and rural areas, those who held party membership not only had higher earnings overall than those who did not, but the trend also increased with time. However, it is not entirely clear how we should interpret the data related to party premium. Does CCP membership hold value in the form of political capital in communist-controlled China, or is party membership itself a kind of signal of ability, with those who hold it tending to receive higher earnings than those who do not?
In rural China, party membership and education had a statistically significant and positive effect on non-farm employment choice and earnings, but those effects decreased rapidly over time. In addition, for those with party membership, the earnings premium was significantly reduced for those who did not occupy high-ranking positions in their organization.
In urban China, however, party membership and education exerted more and more of an effect on individuals' employment choice, occupational attainment, and earnings. For party members with higher levels of education, and among those who joined the Party at a younger age, the odds ratios for attaining high-ranking employment were markedly higher. These individuals garnered higher and higher earnings compared to others.
In other words, party membership was found to be a valuable form of capital in the same vein as education, and it was also a determinant for higher earnings for those who attained high-ranking positions in a given organization. Because the urban sector offered the opportunity for individuals to join the Party, find work in the state-owned sector, and attain a high rank in a government institution or the like, incentive to join the CCP was maintained over time. Individuals who followed this path had better occupational attainment and higher earnings. Due to this, it may be the case that the communist party has been able to pull in high-achieving individuals from all strata of the society, increase its administrative capabilities, and adapt to a rapidly changing environment. When viewed from this perspective, the view of CCP membership as a signal of ability cannot be denied.
In both rural and urban areas, the rate of return to education and the earnings premium for CCP membership were converging at the same level. This is evidence worthy of attention that labor markets throughout the country are becoming integrated.
To conclude, some limitations of this paper need to be pointed out. As previously stated, limitations in the data mean that the profiles of CCP members, and the various functions of party membership after the very beginning of the 2000s cannot be determined from our analysis. In research by Yang et al. (2010) (2008), the results show that the earnings premium had declined greatly or had disappeared by 2008. More research is necessary to observe the changes in the function of party membership over a longer time span based on empirical analysis. Furthermore, among the questionnaires utilized in the three CHIP Surveys, there are some questions and categories for answers that are not exactly comparable due to differing expressions. We took the utmost care to convert the results from all three surveys into a comparable form, but we cannot deny that some parts were not perfectly compatible. We believe that these are not significant enough to sway the results of the study, however.

Funding
This work was supported by The Japanese government-sponsored KAKENHI [15K03482], and The special support of Tianjin University of Technology [2019].