How Can One Get Ahead in the Contemporary Labour Market of China? — Examining the Changing Stratification Mechanisms through Job-Attainment Patterns

Using the 2003 China General Social Survey data, this study explores the changing stratification mechanisms in transitional China by examining the relative importance of three job-attainment channels, namely, state-led hiring, market-led hiring, and guanxi-led hiring. I argue that the labour market of post-socialist China is co-stratified by political power and the market. These two forces have achieved an equilibrium in which market strength is subject to political power. Firstly, the increasingly importance of education is not due to the emergence of the market, because state-led hiring has even stricter educational criteria. Guanxi is no longer a driving force of stratification, since guanxi-led hiring is more utilized by the disadvantaged. Second, an increasing tendency of market-led and guanxi-led hirings accompanied by a decline of state-led hiring is evident, over time and across space. Thus, new opportunity structures have indeed expanded, making it possible to move upwards outside of the state-controlled system. Third, Quantile Regression estimations show that the dominance of political power penetrates from the lowest to highest social stratum, so that being hired by the state-controlled channel is the most likely means of upward mobility. The market channel pays off only for the top social stratum.

Studies on the socioeconomic transformation of China have continuously drawn academic attention, as they are directly associated with the long-standing sociological interest in the institutional change of stratification mechanisms.Despite the influential market-transition debate (Bian & Logan, 1996;Cao & Nee, 2000;Nee, 1989;Xie & Hannum, 1996;Walder, 1996Walder, , 2003;;Zhou, 2000) that spanned the past two decades, the question "How, and to what extent, have stratification mechanisms been changed?"remains unresolved.As Fligstein (1996) said, the papers in the debate raised as many questions as they answered.
One school of thought, Nee's (1989Nee's ( , 1991Nee's ( , & 1996) ) market transition theory, argues that markets are (gradually and partially (Note 1)) substituted for the state socialist redistributive system as the new stratification mechanism of social resources, power, and privilege.According to Cao and Nee's summary, the transformation of state socialism will eventually lead to "(1) a decline of the advantage of redistributive power, and other forms of political capital, relative to non-state economic actors who possess market power; (2) higher returns to human capital than under a centrally planned economy; and (3) new opportunities cantered on market activities, for example, entrepreneurship."(Cao & Nee, 2000: 1175-1176).also been challenged.In their study on the regional variation in earnings inequality in urban China, Xie and Hannum (1996) found that the rates of return to education do not increase with economic growth.Instead, returns to education and work experience, are lower in cities with faster economic growth than in cities where growth has been slower.In a more recent study by Wu and Xie (2003), the argument that marketization does not necessarily lead to a higher return to human capital is confirmed.Using the 1996 survey of Life Histories and Social Change in Contemporary China, and comparing the later stage with the early period of the reform, the authors found that higher returns to education in the market sector simply result from an increasing ratio of well-educated individuals entering the market, versus staying in the state sector.They therefore pointed out that "the state and market institutions do not differ in how they reward human capital" (Wu & Xie, 2003:439).This study echoes Zhou et al.'s (1997) study on job-shift patterns in urban China.Zhou et al. started with a reasonable hypothesis that the market transition theory implies a shift of intellectual elites from the public to private sector.Using the 1993-94 survey on the life histories of a sample of 4,600 residents from 17 cities in China, however, they found no support for this hypothesis.Instead of the hypothesized job-shift pattern, they actually found that well-educated individuals are more likely to shift into government agencies and public organizations, rather than hybrid firms (Zhou et al., 1997).
Critique of the third prediction revolves around the persistent attractiveness of the public sector.Based on their findings regarding the effect of education on job-shifts to the governmental and public organizations, for example, Zhou et al. (1997) asserted: "Thus, state organizations apparently offer incentives to those with high human capital that are at least as attractive as those offered by hybrid firms."(Zhou et al., 1997:355) In another study (with panel data of 4,730 urban residents drawn from 20 cities in China used), Zhou (2000) confirmed that the dominance of the public sector is persistent, despite the reform.Zhou termed this phenomenon the "coevolution of politics and markets" model.As he concluded, "state-initiated reforms and emerging markets have been two major sources of economic transformation.Both interest politics and markets coevolve in interactions with each other…The central idea in the imagery is that neither one can be understood without a careful and substantive understanding of the other."(Zhou, 2000(Zhou, :1140) ) The thirty-year socioeconomic transformation of China has already shown that neither side of the debate provides a satisfactory answer.Obviously, Nee's market transition theory was undermined by pervasive critiques.On the other hand, Nee's critics have, unfortunately, not yet constructed a theoretical alternative that can capture the whole picture of the changing stratification mechanisms of China, as market transition theory did.Zhou's (2000) "coevolution of politics and markets" model could be seen as an attempt in this regard.Without specifying exactly how politics and markets have coevolved, however, this "model" is, at best, a contribution in terminology.
A more comprehensive insight probably comes from the analytical perspective of economic sociology.In their comparative study on post-Communist economies, King and Szelenyi (2005) abandoned the political power-market dichotomy.Instead, they focused on "the pattern of class conflict and intraclass alliance (or interelite struggle)" (King & Szelenyi, 2005:214) as determinants of paths to capitalism selected by state socialist economies.As for China, they argued that previous policy failures made by the Communist Party of China (CCP hereafter), like the Great Leap Forward and the Cultural Revolution, greatly undermined the legitimacy of the political bureaucracy, and led to the technocratic faction coming to power.On the other hand, "these disasters emboldened, if not empowered, the dominated class, who had no choice but to abandon central policies and experiment with reform on their own."(King & Szelenyi, 2005:220) Based on the state-led nature, therefore, the transformation of China can be viewed from two sides.
"On one hand, markets have flourished, and the private economy continues to expand, while the SOEs decline but behave increasingly like capitalist firms (see Nee 1992Nee , 2001)).On the other hand, the Chinese transition has been overseen by a strong state (see Nee 2000), and more important for Chinese growth has been the state's massive infrastructural investments, which provide support for new private enterprise formation and growth."(King & Szelenyi, 2005:222) This argument could be viewed as the connotation of Zhou's "coevolution of politics and markets" model.The question subsequently coming out is how this transition path has changed the stratification structure of the society.The failure of the market transition debate has proved that intuitively assuming the opposition of two driving forces of stratification, namely political power and market, would by no means help us understand the changing stratification mechanisms.

Job-Attainment Patterns Mirror the Changing Stratification Mechanisms
As aforementioned, Zhou et al. claimed that job-shift patterns are a more appropriate indicator than income disparities in reflecting the institutional change of stratification mechanisms.The reason, as argued by Zhou et al., is that income cannot capture the effect of nonmonetary benefits, and thus can only reflect a "relatively small aspect of social inequality in state socialist societies.(Zhou et al., 1997:340)" In contrast, "patterns of job shifts in China reflect not only individual preferences and the desirability of workplaces, but also, more importantly, the relative strength of economic institutions…Because the hierarchy of workplaces and the structure of political authority in China are so closely linked, job-shift patterns reflect China's economic institutions and political institutions.Therefore, changes in job-shift patterns signal fundamental changes in its institutional arrangement."(Zhou et al., 1997:342) Based on this assumption, Zhou et al. compared the job-shift patterns of three historical periods in China, and found that job shifts from the public to hybrid or private sectors did not significantly increase after the reform.They thus concluded that the institutional change in stratification mechanisms is much more modest than is shown in other studies (which focus on income differences).
Nevertheless, the authors do not infer an insignificant impact of the market economy.Rather, they admit that public organizations have indeed "become actively engaged in market activities, which help to provide incentives for their employees to remain in their current workplace.(Zhou et al., 1997:364)" In this sense, it may be problematic to use individual job shifts between the public and private sectors to indicate the changing stratification mechanisms, because "the attractiveness of jobs in some state organizations probably reflects both continuing redistributive benefits and benefits from the newly-developed market activities for such workplaces."(Zhou et al., 1997:364; Italic is added by the authors) If neither income differences, nor types of work organization, can be an appropriate indicator of the distinction between the state power and markets, what best indicates the changing stratification mechanisms, in the transformation context?I therefore propose to examine the changing job search patterns during the process, in which the previous state-controlled job assignment system had gradually been replaced by the free labour market.In the following section, I will articulate three job-attainment patterns, namely, "state-led hiring", "marked-led hiring", and "guanxi-led hiring".I will argue that the relative importance of those three employment channels reflects the newly emerged stratified social structure in post-socialist China.

A Historical Review of the Pre-Reform Labour Market
To understand the job-attainment patterns in the present labour market, a brief historical review is needed.As scholars acknowledge, in China, markets first emerged in rural settings, created initially by peasants themselves, and were later legitimatized by the CCP in 1978 (Bian, 2002a;Nee & Matthew, 1996;Wang & Zhou, 1994).Urban reforms were implemented after rural reforms, and thus the influence of urban reforms was not evident until the late 1980s (Bian, 2002b;Wang, 1996).A government program of job assignment was the only legal way by which urban citizens could acquire a job from the mid-1950s to the late 1980s."After graduation from school, youths were required to wait for state job assignments, and once assigned the youths were restricted from switching between places of employment."(Bian, 1997:370) Although the party-state monopolized the entire labour market, individuals had their strategies to achieve an optimal job-assignment outcome; that is, whether or not individuals could obtain desirable jobs depended on their application of the culturally deeply-rooted guanxi (personal relations) philosophy.Numerous traditional culture studies (Hwang, 1987;King, 1994;Yan, 1996;Yang, 1994) have depicted that part of the essence of Chinese culture is to interpret formal regulations flexibly, depending on the particular contexts and participants involved.As a result, informal institutions which use informal "guanxi" to manipulate the formal, written regulations emerged, and this structure became the fundamental system which dominated every socioeconomic aspect of the Chinese society (at least in the pre-reform era).Above all, it is clear that before the transformation occurred, political power determined the social hierarchy; the closer one was to the centre of political power, the higher the position one achieved in the political hierarchy.However, how to get close to the centre of political power was, to a great extent, up to individuals' abilities to manipulate informal "guanxi."Turning to job acquisition, as Bian (1997) commented that while individuals were waiting for assignments, "they mobilized their guanxi networks to contact job-assigning authorities", hoping to influence control agents so they would be assigned a good job (Bian, 1997:371)."In short, it is appropriate to assert that the stratification structure of the pre-reform society was co-constructed by two forces, namely, political power and informal guanxi.The former dominated the social structure in the foreground, with the latter pervasive in the background.

The Persistent Dominance of Political Power
Then, how has the stratification mechanism changed since the market formed?The socioeconomic reform of China is usually viewed as a transforming process from the command economy to market economy.Another significant change was almost completely ignored by most transition scholars, except for Andrew Walder.Walder and his collaborators' (Walder, 2003(Walder, , 2004;;Walder et al., 2000) studies resonate with King and Szelenyi's (2005) argument, by pointing out that the CCP would have spontaneously changed its policy agenda, even without the reform.Rather than by using backwards inference, it may be more appropriate to claim that the shift of the CCP's policy agenda to economic performance indicated the launch of the reform.Since the late 1970s, it has become vitally important for the party-state to recruit talented people from all social classes in its "pursuit of a technocratic economic growth policy" (Gabriel et al., 1996).Individuals who now are preferred by the party-state are those having "the skills desired by party leaders to accomplish their new policy agenda (Bruce, 2001)".As scholars observed (Martin, 1981;Townsend & Womack, 1986;Walder, 2004), after the Mao era (1955-77 (Note 2.)), in which party policies tended to discriminate against people in professional occupations with higher education, and favoured those from politically reliable backgrounds, the Party seemed to re-acknowledge the importance of education.As Townsend and Womack commented, "in general, post-Mao recruitment has reverted to the pattern of the mid-1950, except that recruitment now is less class-oriented and even more committed to including intellectuals." (1986: 299) As Walder et al. (2000) summarized, China's revolutionary regime has evolved in a meritocratic direction, in terms of careers that lead to professional positions.When it comes to positions that will exercise decision-making authority, however, education in itself is not sufficient; the ruling party has continued to operate as a political machine that allocates positions based on particularistic political standards.In this sense, the abolishment of the job assignment program in the late 1980s does never mean a delineation of political power in the labour market.Job positions in the public sector, especially cadre or managerial positions in the governmental and party institutions, or state-owned enterprises (even though they are marketized), are still overwhelmingly controlled by the CCP's bureaucratic system.Thus, one can term attaining a job through any channels controlled by the party-state as "state-led hiring."Therefore: Hypothesis 1: Political identity is still the crucial screening criterion of state-led hiring.That is, party members are more likely than non-party members to be hired through the bureaucratic system.
However, as mentioned above, the party has been evolving in a meritocratic-technocratic way, so education should be a second crucial criterion of state-led hiring.That is: Hypothesis 2: The higher one's education, the more likely one is hired through the bureaucratic system.

The Increasing Importance of the Markets
Although, contrary to Nee's (1989) prediction, the market transition did not orient Chinese economy in the direction of western market economies, the expanding market sector creates new opportunity structures where entrepreneurs, managers, and professionals can economically excel on the basis of their individual initiatives and human capital.As the government job assignment program was gradually abolished in the late 1980s, entry into the labour market has become an increasingly individual behaviour.The process of looking into hiring advertisements, attending job fairs, circulating CVs, interviewing, and finally obtaining offers has been a common way of attaining jobs in contemporary China.Hence, I term it "market-led hiring."Without any doubt, the meritocratic characteristic is the core of the market economy.As for China, however, one should bear in mind that the privatization of the Chinese economy has not taken place via the conversion of state assets to private ownership, but instead through new start-ups outside the state sector, and through foreign investment.Consequently, China's Party hierarchy has survived unchanged, and has solidified by absorbing technocrats into the bureaucratic system.One can therefore hypothesize: Hypothesis 3: The higher one's education, the more likely one is hired through the market channel.
However, market-led hiring is subject to state-led hiring.That is to say: Hypothesis 4: State-led hiring is more attractive than market-led hiring to highly-educated job seekers.Namely, one's high educational degree is more likely to channel one's job attainment into the bureaucratic channel than the market channel.

Guanxi-Led Hiring: The Buttress of the Disadvantaged
As aforesaid, guanxi (informal personal relation) played a critical role in the stratification mechanisms of the pre-reform society.However, it is reasonable to assume that the possibility of accessing political power by manipulating guanxi has declined, as the party started to construct its bureaucratic system in a meritocratic-technocratic direction.On the other hand, social networks and social capital scholars have convincingly demonstrated the importance of informal interpersonal relationships even in the western markets (De Graaf & Flap, 1988;Granovetter, 1973;Lin et al., 1981; see also Lin's review 1999;Marsden & Hurlbert, 1988;Wegener, 1991;Uzzi, 1996).As Lin (1999:31) reviewed, using informal channels plays a positive role in determining the flow of information, exerting influence on an organizational agent's decision, providing social credentials to an individual, and reinforcing the individual identifying and recognizing him/herself as a member of the work organization.Application of informal relations is viewed as relatively unique to the Chinese society, because it is embedded in a culturally-rooted guanxi philosophy, which emphasizes familiarity, trustworthiness, and reciprocal obligation in a firmly-bonded tie (see Bian's 1997 review).One can thus term the way of getting a job through informal relations "guanxi-led hiring." To what extent guanxi matters during an individual's job search is controversial.Based on the 1988 survey in a Chinese city, Tianjin, Bian (1997) found that a job seeker is inclined to seek help from a contact they have strong ties to.Note that, however, instead of using the entire survey sample, Bian based his argument only on 45% of the total sample, which consisted of individuals who reached the helpers either directly or through a third intermediary person.Thus, this study may not be sound evidence of advantages of guanxi over meritocratic job-attainment methods.Lin's 1999 review also pointed out that, despite the notion of informal relation use having positive effects in the labour market, scholars actually found that the disadvantaged (i.e.females, the less-educated or low-skilled), rather than the advantaged, are more likely to rely on contacts to get jobs (see Lin's 1999 review: 481-482).On the other hand, as aforementioned, studies on Chinese culture have suggested that a successful manipulation of guanxi is either based on previous, long-term cultivation, or is accompanied by "favour debts," which the beneficiary has to repay in the future.It is thus reasonable to hypothesize that individuals may refrain from using guanxi to find jobs, if there is another, less costly method available.That is, if one can get ahead through state-or market-led hiring mechanisms, one is less likely to rely on guanxi to attain a job.I thus claim that guanxi is no longer the way by which one can excel in the labour market; instead, it is a buttress to help the disadvantaged survive the increasingly competitive labour market.Therefore, Hypothesis 5: Contrary to state-and market-led hiring, guanxi-led hiring is more preferred by the labour-market disadvantaged.That is, (1) females are more likely to use guanxi; and (2) less educated individuals are more likely to use guanxi.

The Symbiosis of Three Hiring Mechanisms over Time and across Space
It is necessary to know how the relative strengths of the three hiring mechanisms have changed over time, in order to understand the changing stratification mechanisms.Needless to say, political control was indeed eliminated or reduced in many economic spheres.Numerous new opportunities, led by the market or guanxi, have been opened with the development of the market sector.At the early stage of the transformation, the increase of guanxi-led hiring was even more rapid than that of the market-led hiring.This is because the formation of a formal institution is gradual.The maturity of the market economy actually lagged behind the abolishment of the planned economy.As a result, informal guanxi use would spring up to fill up this institutional vacuum.On the other hand, instead of being an underground, illegal channel to get a job in the pre-reform society, guanxi use has been legitimized in the contemporary labour market.This would lead individuals who found jobs in the reforming era to report their guanxi use more voluntarily than their counterparts in the pre-reform era.As the market economy matures, however, market-led hiring should exceed guanxi-led hiring to dominate the labour market.Therefore, Hypothesis 6a: Over time, market-and guanxi-led hirings present ascending tendencies, while state-led hiring goes down.Hypothesis 6b: Guanxi-led hiring outweighs market-led hiring at the early stage of the reform, whereas the latter exceeds the former at the later stage of the reform.
Moreover, since the maturation of the market economy, or more specifically, the private sector, varies across geographical regions, the relative strengths of the hiring mechanisms should vary geographically, as well.King and Szelenyi (2005) termed the path of the economic transition of China "capitalism from below"; namely, "China's transition first started with the opening up of the large domestic markets, giving small firms an enormous space to grow in the production of consumer products, an area where SOEs were less concentrated (King & Szelenyi, 2005:221)."Only at the later stage of the reform, at which even SOEs have also marketized, has market-led hiring become dominating in the whole labour market.Thus: Hypothesis 7a: Market-led hiring is significant at lower jurisdictional levels of areas at an early stage of reform, and it later becomes dominating at all jurisdictional levels as the reform progresses.(Note 3.) Guanxi use is probably more directly related to the openness of opportunity structures, because more open opportunities increase the possibility of using guanxi to find a job.Since more opportunities are concentrated in areas of higher jurisdictional level, it can be hypothesized that: Hypothesis 7b: Guanxi-led hiring increases with the jurisdictional level of the area.

The Symbiosis of Three Hiring Mechanisms across Social Strata
The above section suggests an expansion of market-as well as guanxi-led hiring, and a decline in state-led hiring over time, and across space.Note, however, that it would be a severe misunderstanding to simply conclude that there is a shift of the driving forces of stratification from the state power to the market.This changing trend is certain to occur, whereby the market has expanded from near nothing to fill up the void created by the withdrawal of state power.Despite the rapid development of the market, one can argue that the market has not yet, and will not, replace the dominance of state power, at least in the long term.As long as the body of the economy, the state sector, remains un-privatized, the opportunity structures opened up by the market will always be limited, and individuals who can gain a market premium are thus relatively few.For those who are constrained by the limited market opportunities, state-led hiring will still be the optimal choice.In this section, I thus distinguish this study from all previous stratification studies.An intuitively-accepted, but not necessarily true, assumption in the existing stratification literature is that the driving force of stratification works evenly at every social stratum.This might not be a problem when only one force solely structures the society.However, when more than one driving force of stratification co-constructs the social structure, however, assuming the relative strengths of those driving forces are consistent across strata could severely mislead one's understanding of the stratification scenario.For example, transition scholars assumed the strength competition between political power and the market was consistent across social strata, and consequently developed their arguments in opposite directions.They either took market elites' high income as evidence of the growing market strength (Cao & Nee, 2000;Nee, 1989Nee, , 1992Nee, , & 1996)), or they argued for the persistence of political power by showing cadres' income gain (Bian & Logan, 1996) and incomparable nonmonetary benefits in the public sector (Zhou et al., 1997).
In this paper, I claim that the relative strengths of three hiring mechanisms vary with social strata.I use income as the indicator of the social stratum for three reasons.Firstly, as Zhou et al.'s (1997) study proved, nonmonetary benefits are almost completely monopolized by the state's redistributive system.A focus on the variation of nonmonetary benefits across strata would, at best, reflect the relative change of political power in the spectrum of social classes, rather than the relative change between the strengths of political power and the market.Second, at the later stage of the reform, the public sector has started to marketize as well, though the state control is never changed.Welfare systems have subsequently changed, for example, transferring direct housing supply to housing subsidies, which are now included in one's income.Finally, under the influence of propaganda: "economic performance is the centre," "everything should give way to economic benefits," income, rather than occupational prestige or other societal dimensions, is the basic gauge to distinguish one's class status.To support the argument that political power and the market co-stratify the society, but the market strength is subject to political power, the following two hypotheses are proposed: Hypothesis 8a: There is an equilibrium status between political power and the market strength at the upper social strata.That is, state-led and market-led hirings do not lead to a significant difference in one's income at the upper social strata.Hypothesis 8b: State-led hiring leads to a significant advantage over market-led hiring at the lower social strata.That is, individuals hired through the party-state channel earn significantly more than those hired by the market channel at the lower social strata.This is because only a few economic elites are able to access the relatively limited market opportunities and 'get ahead' through the market channel, as mentioned above.Under the state economy, the strength of the market is not to re-stratify the whole social structure, but just to create a second upward mobility channel for those social elites who are able to access and take advantage of the newly emerging opportunity structures.It does so without threatening the mainstream bureaucratic channel, so that the group of moving upward though the market is imaginably small.For the remaining body of social members, the best choice is still to be as close as possible to political power, since the state economy will be likely last for a long time.
Regarding guanxi use, this paper argues that it is no longer a driving force of stratification, if it ever was in the pre-reform society.In this sense, Guanxi users should always be worse off, compared to those who followed the other two hiring mechanisms, regardless of the social stratum.Therefore: Hypothesis 8c: Guanxi-led hiring leads to the lowest income attainment among the three hiring mechanisms, and it is consistent across social strata.
To test the hypotheses in the section, the Quantile Regression Model (QRM) will be adopted.This is because the general linear regression model (LRM) is based on the assumption that the effects of covariates do not vary with the respondent's location in the distribution, and is thus not suitable for a skewed distribution.A widely-adopted method is to transform the skewed distribution into a normal distribution, such as using the log transformation.However, a highly skewed distribution cannot be changed into a normal one, even after a transformation.A more serious problem is, as Hao and Naiman (2007) point out, that "monotone equivariance" of log transformation fails in the LRM, so that the logged income results cannot be used to make conclusions about the distribution of income (see more details in Hao & Naiman, 2007:39-40).By contrast, the QRM estimates conditional quantiles separately, and does not share the above problems.(Please refer to Hao & Naiman 2007 to see more details about the QRM.)

Data and Measurements
Data used in this study is taken from the 2003 China General Social Survey (CGSS), a four-stage stratified probability sampling survey on Chinese urban population.The sample consists of 5,894 adults aged from 18 to 69 years old, who were randomly selected from 30 provincial-level jurisdictions.
The first dependent variable is hiring mechanism, including hiring by the bureaucratic channel, the market channel, or the guanxi channel.Missing responses are kept as a separate category, identified as "missing".The second dependent variable, income is generated from the log-transformed annual income.
Hiring mechanism will also be used as an independent variable in the later part of the analysis.Other independent variables are as follows: Gender is a dummy variable, with male coded 0 and female coded 1. Age and age squared refer to the respondent's age, and its squared value, respectively, in the year the respondent entered their current workplace.Years of education refers to the educational level of the respondent when they entered their current workplace.Political identity refers to one's party membership.For the convenience of interpretation, CCP members are coded 0 and non-CCP members are coded 1. Jurisdictional rank of the area consists of towns, counties, prefecture-level cities, and metropolitan cities (coded 1, 2, 3, and 4, respectively).The urban hierarchy is ranked from the town (lowest), county, prefecture-level city to metropolitan city (highest).Era of job entry distinguishes three historical periods -the pre-reform era before 1979, the early stage of the reform from 1979 to 1992, and the later stage of the reform after 1992 (coded 0, 1, and 2, respectively).
Control variables include: type of sector, which consists of state-owned enterprises (SOE hereafter, coded 0), party, governmental, and public institutions (PGP hereafter, coded 1), collective enterprises (coded 2), and private/hybrid enterprises (coded 3); and family background, which is indicated by father's education and party membership.

Results
Descriptive statistics (table is not included, but it is available upon request) show that as of 2003, around 43%, 26%, and 19% of respondents reported being hired by the bureaucratic channel, the market channel, and the guanxi channel, respectively.Clearly, state-led hiring is the most pervasive channel.
Hypotheses 1 to 5 are verified by the multinomial regression baseline model shown in Table 1.With other characteristics being equal, every one-year increase in education: (1) significantly decreases the odds ratio of guanxi-led hiring versus state-led hiring by 20% = 1-e -0.217 , (2) decreases the odds ratio of market-led versus state-led hiring 13% (=1-e -0.138 ), and (3) decreases the odds ratio of guanxi-led versus market-led hiring by 8% (=1-e -0.079 ).That is, as one's education level increases, the likelihood of informal guanxi use decreases the most, followed by that of market-led hiring.State-led hiring is the most likely to be used by highly-educated job seekers.Regarding party membership, non-party members are significantly more likely to use either guanxi or the market channel, relative to state-led hiring, with other covariates controlled.Clearly, guanxi is the most likely channel by which females were hired, as coefficients of odds ratios of using guanxi versus the other two channels are both positively significant (0.247 and 0.163).In short, hypotheses 1 to 5 are substantiated.An improved model containing the interaction between time and space is used to testify hypotheses 6 and 7.With the interaction between time and space (and other covariates) controlled, the jurisdictional rank of the area has no significant effect on the likelihood of use of hiring channels.The effect of the era is significant to the likelihood of using guanxi-led hiring relative to state-led hiring, and of market-led hiring relative to state-led hiring.It is not, however, significant to the odds ratio of guanxi-led versus market-led hiring.By scrutinizing the interaction items, it can be easily seen that during the reform era, only metropolitan cities are significantly different from lower jurisdictional-level locations.At the metropolitan level, the likelihood of using guanxi versus the bureaucratic channel significantly increases, as does the likelihood of using guanxi versus the market channel, with other covariates controlled.The likelihood of market-led hiring does not significantly differ from state-led hiring, though.At the later stage of the reforms, the county level starts to be significantly different from the reference group -towns.Other covariates being equal, the odds ratios of guanxi use versus state-led hiring, and of guanxi use versus market-led hiring are both significantly positive at county and metropolitan levels.1a shows, before the reform, the state-led hiring dominates the other two hiring channels.At an early stage of the reforms, shown in Figure 1b, the likelihood of state-led hiring is still apparently the highest.However, market-led hiring increases significantly, especially at the town level.This should be attributed to the booming of Town and Village Enterprises in the 1980s.Guanxi-led hiring increases significantly at the metropolitan level in this era because, as aforementioned, newly emerged opportunity structures are more concentrated in metropolitan cities, than in small towns or counties.At the later stage of the reforms, it turns out that market-led hiring, rather than state-led hiring, is the dominating hiring mechanism, as shown in Figure 1c.State-led hiring decreases significantly, especially at the prefecture and metropolitan levels.On the contrary, guanxi-led hiring increases significantly at the county and metropolitan levels.This is in line with Cao, Qian, and Weingast's (1999) finding in their study on the accelerated marketization process of SOEs since 1995.They found that regions with larger private economies (Note 4.) are more likely to marketize their SOEs, because the newly unemployed will be more likely to find jobs.My finding shows that the marketization process of SOEs is much faster at prefecture and metropolitan cities than towns and counties.In short, hypotheses 6 and 7 are substantiated.3 shows estimations based on the Quantile Regression Model.Other covariates being equal, the effects of hiring mechanisms vary with one's social stratum: At the 25% quantile, market-led and guanxi-led hirings cause an annual income attainment of 18% (=1-e -0.196 ) and 29% (=1-e -0.196 ) lower than that through state-led hiring, respectively.At the median quantile, individuals who found jobs in the market earn 9% (=1-e -0.098 ) less than those hired through the bureaucratic channel, while those who used guanxi to get jobs earn 17% (=1-e -0.185 ) less than their counterparts who relied on state-led hiring.At the 75% quantile, market-led hiring no longer causes significantly lower income compared to state-led hiring, though the coefficient is still negative.Guanxi users are still worse off, earning 14% (=1-e -0.153 ) less than their state-preferred counterparts.Only when the top level, the 99% quantile, is considered, can one see no significant difference in the monetary outcomes of the three hiring mechanisms.Rather, market-led hiring actually presents a positive trend, which implies individuals may be better off than those using state-led hiring, though the positive coefficient is not significant.Other covariates are omitted.Absolute value of z statistics in parentheses + significant at 10%; * significant at 5%; ** significant at 1%; ***significant at 0.1% Figure 2 presents the relative change in strengths of three hiring mechanisms in a more intuitive manner.As one can see, the horizontal line represents the reference group, namely, income attainment through state-led hiring.The blue solid line, representing income attainment through market-led hiring, is always higher than the brown dashed line, which represents income attainment through guanxi-led hiring.The blue line goes above the horizontal line after the 75 th quantile, whereas the brown line is always below it.Without the confidence interval shown in the graph, however, one cannot assert that market-led hiring leads to higher income than bureaucratic-led hiring does from the 75% quantile onwards.A more reasonable assertion is that, for one whose income is located below the 75 th percentile of the income distribution of the whole society, getting a job through the bureaucratic channel apparently leads one to be on the top of the particular social stratum, other conditions being equal.For the 'upper class,' whose income belongs to the upper 25 th percentile, on the other hand, the channel through which one is hired seems to be unimportant.In short, hypotheses 8a to 8c are substantiated.

Conclusion and Discussion
This study offers a unique analytical perspective, focusing on the job-attainment patterns to discuss the changing stratification mechanisms in transforming China.The socioeconomic transformation of China started with the emergence of the free market.While in rural areas the free market is mainly represented by peasants' autonomy in production and consumption, to urban residents it mostly means changes in a series of job-related systems.
Studying changes of job attainment can thus perfectly reflect changes of the macro-level stratification institutions.
By doing so, I argue that political power and the market are the two driving forces of stratification in transforming China.However, they do not rival each other, as the previous market transition literature assumed.Instead, the market was initiated, and is always led, by political power.I prove this argument from three aspects.
First of all, in examining the criteria of three hiring mechanisms, it is shown that one of the new stratification mechanisms is the (re)emphasis on education.However, one should bear in mind that this change is not due to the emergence of the market.Market-led hiring is just another means by which the state mobilizes intellectual elite, particularly those outside of the party and bureaucratic hierarchy, for the sake of its new policy agenda.
Empirical evidence shows that state-led hiring has even a stricter requirement on education than market-led hiring.Moreover, state-led hiring also has similar, or stricter, requirements than does market-led hiring, regarding other screening criteria, such as political qualification, gender, and age (Note 5).Guanxi is no longer a driving force of stratification, since guanxi-led hiring is always more likely to be used by the disadvantaged in the labour market.It may be appropriate to claim guanxi use as a buttress that helps uncompetitive job seekers survive the labour market.
Second, the three hiring mechanisms have changed over time and across space.Empirical results show that a great number of opportunities have indeed been created outside of the state-controlled channels.Thus, an ascending tendency of market-led hiring, as well as guanxi-led hiring, accompanied with a decline of state-led hiring is evident.The changing trends of the former two hiring mechanisms are geographically different.For example, guanxi-led hiring is more likely to be used in metropolitan cities than in small towns or counties, while market-led hiring seems to increase faster at the town level than metropolitan cities.
Finally, the relative strengths of the three hiring mechanism vary not only in the dimensions of time and space, but more importantly, also in the dimension of social stratum.The QRM estimations show that for most individuals who are located in the lowest to middle-upper social strata (below the 75% quantile of the countrywide income distribution), the influence of state-led hiring overwhelms that of the other two mechanisms.
That is, for an ordinary (or below-ordinary) individual, the bureaucratic channel is always the best channel to move upward, followed by market-led hiring, and lastly by guanxi-led hiring, at least economically speaking.However, for one who is at the top stratum (the top 25% quantile), it is difficult to tell which channel brings the best economic return.It is understandable, though, as all sources of power are concentrated at the top stratum.Power from politics, economy, and social networks can easily be converted from one form to another, by which the top class maintains and reinforces its social status (see also Bourdieu 1986).Nevertheless, market-led hiring presents a strong ascending trend of at the top stratum.That is, an individual could be better off by market-led relative to state-led hiring, when in the top social stratum.
In conclusion, the dominance of political power penetrates the society, from the lowest to the highest social stratum so that being hired by the bureaucratic channel is always the most likely means of upward mobility.The market could be an alternative that is "beyond the control of the state" (Cao andNee 2000:1183), but only for the top social stratum.Objectively speaking, however, the expansion of new opportunities structures has indeed provided a great number of opportunities for individuals to improve their socioeconomic status in contemporary China.

Figure
Figure 1a.(Upper-Left) Predicted probability of each hiring channel, before the reform; 1b.(Upper-Right) At the early stage of the reform; and 1c.(Bottom) The predicted probability of each hiring channel, at the later stage of the reform

Figure 2 .
Figure 2. Quantile functions of annual income by hiring mechanisms Note: Red solid line indicates the reference group -state-led hiring.

Table 1 .
Multinominal logistic coefficient estimates for job-attainment channels of entry into the current workplace (without interaction effects), urban China 2003Guanxi-led vs. state-led Market-led vs. state-led Guanxi-led vs. market-led

Table 2 .
Multinominal logistic coefficient estimates for job-attainment channels of entry into the current workplace (with interaction effects), urban China 2003 Figures 1a to 1c predict the variation of the probability of using each hiring mechanism over time and across space, based on the model presented in Table2.As Figure

Table 3 .
Quantile regression estimates for logged annual income by hiring mechanisms, urban China 2003