Meritocracy in Imperial China: a reflection on Mark Elliott’s Op-Ed in the New York Times

Mark Elliott, a Qing (1644-1911) historian at Harvard, achieved something incredible:  he published an Op-Ed piece in the New York Times devoted almost entirely to processes of elite recruitment in imperial Chinese history.  He shared his views on the question of whether the selection of officials in Qing China was meritocratic, and introduces readers to excellent relevant research by Ben Elman and others.  It wouldn’t surprise me if this were the first New York Times Op-Ed ever that went into such detail about empirical results on a specific feature of late imperial Chinese history, in this case, the procedures for selecting officials.

I hope that Mark’s piece sets a precedent that will allow me and my collaborators to publish Op-Ed pieces about Qing history.

In the meantime, I was inspired to write a response here on my blog.

I’ll begin by putting Mark’s piece in context.  There is actually a  longstanding debate in Chinese historical studies about the issues raised in Mark’s piece, and I’ll introduce some of the relevant work, some of it by me and my collaborators.  Mark’s piece is itself a contribution to a debate over meritocracy in contemporary China triggered by some rather controversial claims by Daniel Bell and Zhang Weiwei that the current system for appointing and promoting officials in China is meritocratic.  China Digital Times has a nice summary of the debate, with links to various pieces.  Bell, Zhang and others have invoked the examination system in imperial China as a precedent for the current system, and Mark is offering an important and needed corrective to some of the overheated claims made about the virtues of examination system.

I will offer my own take on the issue, which is that we need to think about the issues involved in a comparative fashion.  Rather than assessing whether China was a meritocracy by comparing it to what Weber referred to as an ideal type, that is a hypothetical society that might exist only in Plato’s world of forms or a Star Trek episode, assessment has to be made by comparison to other societies.  I’ll identify what I think the relevant dimensions are for comparison between China and other societies.  I’ll conclude with some comments about Chinese studies and stratification research.

I originally intended this to be a short piece inspired by Mark’s Op-Ed that would focus on my favorite subject: me.  Or rather, my published collaborative work.  As I began writing, though, this evolved into a larger meditation on what I think the appropriate approach is to addressing the issues raised in Mark’s Op-Ed, and the work he is responding to.  At some point I wondered if perhaps I should spend a few weeks to turn it into a formal academic essay, and publish it.  The problem with that is that I would spend a lot of time on it, it would take a year or two to appear, and then only five people would ever read it, most of them friends of mine who already agreed with me, or were afraid to disagree openly with me.  Hastily posting this core dump from my brain to the web is probably not going to do much for me professionally in the bean-counting world of modern academics, but given the wider attention to processes of elite recruitment in historical China generated by Mark’s piece, I thought it was an excellent opportunity to introduce a wider, non-specialist audience to some of the issues and debates in stratification research in historical China, and perhaps attract some of them to the field.

Mark’s Op-Ed piece on meritocracy is embedded in a longstanding debate in the study of Chinese history about whether the social composition of political elites was ‘open’ or ‘closed’.  This much broader debate about whether the system was open or closed, fluid or rigid, is more important than the narrower one about whether the political appointment system was meritocratic or not, and indeed subsumes that debate.  I would argue, and I think Mark and others would agree, that the technical details of the examination and appointment system are less important than their implications for broader patterns of access and participation, and for long-term patterns of turnover among elites.

Before I proceed, I want to note that I will limit the scope of my discussion to the openness of processes for recruiting political elites in past times, and largely ignore contemporary issues, as well as other important issues in historical stratification and inequality.  Thus I’m not even going to touch Zhang Weiwei’s controversial claims about ‘meritocracy’ in the process for political appointments and promotions in contemporary China.  I think that Zhang Weiwei’s claims are dubious at best, and I may opine on them at some future point in time, but for right now, I’ll stick to what I know best, which is historical China.  I’m also going to sidestep the issue of overall social fluidity in the past, since for the most part the sorts of data we would really like to have as a basis for comparison in largely rural historical societies are still rare.

I’m also going to skip the important issue of whether an examination system, or ‘meritocratic’ systems in general, are actually optimal from the perspective of recruiting a political elite that does the best possible job of governing the country.  I doubt there is a universal agreement on what the appropriate objective measure of ‘merit’ is when it comes to recruiting political elites.  It isn’t clear to me that mastery of Confucian classics was a reliable predictor of leadership ability in the past, any more than academic credentials predict leadership ability now.  Most people who hold a PhD, including myself, shouldn’t be trusted to manage a hot dog stand, let alone a country.

As Mark notes, imperial Chinese ideology was that the reliance on the examination system (keju) made for an ‘open’ system in which advancement was based on merit, not ancestry or personal connections, and in which everyone, or at least every male, had a chance to succeed.  At least in principle, the examination system selected candidates for office in an objective and meritocratic fashion, mainly based on their mastery of a set of classic texts as demonstrated in a rigorous exam.  In theory, any male who was not specifically disqualified from eligibility could sit for the exam.  Chinese history is accordingly replete with remarkable Horatio Alger stories of talented men from humble backgrounds who succeeded on the exams as a result of diligent study, and ended up attaining high office.  It is this tradition that forms the context for the contemporary emphasis on standardized exams for access to high school, college and civil service, as well as the more general concern with the accumulation of credentials such as degrees, prizes, licenses, certificates and so forth that are perceived to be awarded in an objective fashion.

One of the earliest systematic efforts to assess whether the examination system was indeed ‘open’ was Ping-ti Ho’s (何炳棣) classic The Ladder of Success in Imperial China, Aspects of Social Mobility (1962).  Inspired by his reading of studies of Western societies in the then-new study of social mobility and stratification, Ho carried out a remarkable and pioneering study of the family histories of successful exam candidates in successive dynasties.  He found that substantial proportions of successful exam takers in various dynasties were ‘new blood’ in the sense that neither their father nor their grandfather had held an exam degree.   Based on this finding, Ho argued that the openness of the system was not an illusion sustained by Horatio Alger stories of the occasional poor boy made good, but a reality, in the sense that the system was not dominated by a small number of elite families.  This is the work that was the basis of Mark’s possibly cryptic reference to “ladders of success.”

My own relevant work with James Lee (HKUST) on the transmission of status in northeast China during the Qing reached broadly similar conclusions (Campbell and Lee 2003, 2008; Lee and Campbell 1997).  Our focus was on the composition of a regional or even local political elite, not a national one.  We examined determinants of the attainment of salaried official positions in a largely rural population in part of what is now Liaoning province from the mid-eighteenth century to the beginning of the twentieth century.  These are relatively mundane positions in a local administrative hierarchy, not to be confused with the sorts of national-level positions that might be attained by the successful candidates in Ho’s study.

We found that individuals who held official posts had a relatively difficult time transmitting their status to their sons.  Men whose fathers held a position certainly did enjoy an advantage, and were themselves roughly 7-10 times more likely to themselves attain a position, but the baseline chances of obtaining a position were so low than even multiplying them by 7 to 10 yielded a probability that was still quite low.  As a result, the overwhelming majority of the male offspring of men who held position did not attain positions of their own.  Conversely, something like half of men in each generation who attained position were ‘new’ in the sense that they from families in which no one had held position in recent generations.  Again, while certain families clearly had an advantage, there is little evidence of the system being monopolized by a small set of elite families, and considerable indication of social fluidity.

As an aside, the basis of our analysis was a database we constructed from household registers, and which we have now publicly released as the China Multigenerational Panel Dataset-Liaoning (CMGPD-LN).  If you are interested enough in this topic to want to carry out your own analysis, you can download the data at ICPSR and access the most up to date news and documentation via entries at my blog.  It is especially important to check my blog for the latest editions of the User’s Guide and Training Guide since updates tend to take quite a while to appear at our ICPSR site.  Our public release of these data at ICPSR was supported by NICHD R01 HD057175-01A1 “Multi-Generation Family and Life History Panel Dataset” with funds from the American Recovery and Reinvestment Act.

As Mark notes, the national examination system had a variety of features that had the effect of limiting access to a relatively small segment of the late imperial Chinese population. Elman (1991), the source of one of Mark’s quotes, provides a comprehensive yet elegant review of the relevant literature while making the point that the system served the state well by creating the appearance of openness.  As in many historical societies, half the population was ineligible because of their gender.  The focus on demonstrating mastery of Confucian classics via a written essay in a standardized format further limited the pool of exam-takers to men who were lucky enough to grow up in a family or lineage that had the resources necessary to provide them with a classical education, or live somewhere where they had access to a charitable school.  Elman (2000) is a book-length study of the same topic, and commended to the attention of anyone seeking additional depth.

A specific critique of Ho’s (1962) suggestion that the large proportions of ‘new’ men in each generation were indicative of openness that has inspired some of our own recent work on this issue is the one by Robert Hymes (1986), who pointed out that many of the men who in Ho’s study appeared to be ‘new’ because neither their father nor grandfather held position, may have been from elite families, and may have had other kin who held position.  In other words, a sole focus on correspondence between son’s and father’s or grandfather’s attainment may overstate openness by obscuring the fact that disproportionate numbers of the men who made up the exam elite were from a small number of especially successful families.  Hymes’ empirical basis was a study of local elites in Fuzhou, Jiangxi during the Sung.

Our own recent work with the Liaoning household registers confirms Hymes’ suggestion that shifting the focus from the individual or father-son dyad to the larger kin network or lineage reveals a deep, persistent, kin-based structure that is invisible in an analysis of correlations between father’s and son’s outcomes (Campbell and Lee 2011).  We found evidence of subtle and systematic differences between descent groups in the chances that members would attain official position.  These patterns were distinct from the father-son correlations we reported in Campbell and Lee (2008).  That said, these differences were not so pronounced as to suggest that certain descent groups monopolized opportunities, and that others were shut out.  We also found that the relative standing of descent groups was remarkably stable, in the sense that rankings of descent groups according to their success in obtaining position were highly correlated from the eighteenth century to the beginning of the twentieth century.  Even more intriguingly, we reported evidence based on contemporary follow-up of a small subset of the historical descent groups that the relative status of the descent group in the eighteenth and nineteenth century was associated with descent group status in the late nineteenth century.

Another serious critique of Ho (1962) that Mark mentions is that success on the exams qualified a candidate for appointment to office, but did not by itself guarantee an appointment.  Since there were more successful candidates than there were offices, the process of deciding which of the successful candidates would be appointed to an office was much more vulnerable to the sordid or at least tawdry manipulations that complicate the selection of individuals to appoint to potentially lucrative or at least powerful positions in most societies, historical or modern.  The work by Lawrence Zhang that Mark cites sounds intriguing, and I look forward to reading it.  I don’t have expertise in this area so don’t have much to say about it.

My own take on the overall debate is that as is often the case in the humanities and social sciences, the underlying empirical facts are not in dispute, and what is contested is their interpretation.  The empirical findings of Ho, James Lee and myself, Elman, and others do not necessarily contradict each other because each one individually is one facet of a much larger and more complex process which is difficult to discern or comprehend in its entirety.  As James Lee and myself have shown in our analysis of northeast Chinese data, there is evidence both of weak father-son correlations suggestive of considerable openness consistent with Ho (1962), and strong intra-lineage correlations consistent with the suggestions in Hymes (1986).  Arguably, the debate is at heart a ‘glass half-full/glass half-empty’ debate or perhaps more charitably ‘tastes great/less filling’ debate in which all the participants have different interpretations of the implications of the same underlying body of facts.

The appropriate question, therefore, is how do we move forward, and avoid an endless back and forth that is little more than a repeated and contentious restatement of specific positions?

One problem with the current debate is that it occurs in a vacuum.  The participants, including me, justify interpretations of empirical results on China by explicit or more likely implicit reference to a society imagined from a Weberian ideal type, without comparison to actual societies.  Claims of openness by Ho and others, including myself, seem entirely reasonable if the reference for comparison is a hereditary aristocracy, or caste-based society with rigid, hereditary status distinctions.  There was clearly much more upward and downward mobility among the elites generated by the examination system than would be expected in a system where elite status was explicitly hereditary, or formal restrictions limited eligibility for office to only a tiny segment of the population that was defined by ancestry.

Conversely, claims of rigidity by Elman and others are compelling if the reference is the other extreme: a society in which the criteria for selection do not explicitly or implicitly limit the pool of eligible candidates based on their heredity or other characteristics, and where the distribution of wealth and parental education is sufficiently egalitarian that all families have the means to equip their children to compete.  While a few contemporary societies might come close to this ideal, few historical societies did.  Neither of these idealized frames of reference is entirely plausible as a basis for comparison or interpretation of results from historical societies, and the debate is unlikely to be settled if everyone involved continues to make use of them.

A more useful approach would be to anchor the interpretation of empirical results in detailed comparisons of quantitative or qualitative dimensions of recruitment into political elites across historical societies.  Instead of debating whether historical China conformed to one favorite ideal type or another, it would be useful to specify multiple meaningful and historically metrics of openness and access for different historical societies, and compare them.  While it is highly unlikely that China and other societies could be arrayed on a single, agreed-upon dimension of ‘openness’ or ‘meritocracy’ and then ranked to produce a conclusive result, it is more plausible that several relevant dimensions could be identified, and meaningful comparisons made.

I’ll try to get the ball rolling by identifying some basic dimensions for historical comparison of based on the criteria apparent in the work by Elman, myself and Lee, Hymes, and others.  In my somewhat open-ended and rambling specification of criteria, some possibly qualitative, I am departing from what seems to be the reigning orthodoxy in contemporary stratification research, according to which it sometimes seems that all meaningful variation in social openness can be reduced to parameters from a log-linear model, or coefficients from a regression of child’s attainment on parental characteristics.

The first would be the share of the population explicitly excluded from participation solely on the basis of what stratification researchers call ascribed characteristics: gender, race, ethnicity, caste, and other dimensions that individuals have little control over, but are the basis for labeling and categorization by others.  Almost every historical society was characterized by such formal restrictions based on heredity or other ascribed characteristics, though the size of the affected population varied.  One might imagine arraying societies on a spectrum ranging from monarchies governed by a hereditary aristocracy and/or nobility, to some contemporary developed societies in which there are no criteria for entry into the elite that are explicitly based on an inherited or other ascribed status.

The second would be the share of the population that was implicitly excluded from participation because the process by which political or other elites were recruited in each generation favored the offspring of families that had the resources necessary to invest in education or other activities that increase children’s chances of success.  Again, one might think of arraying societies on a spectrum that ranged from an imagined perfectly egalitarian society in which the resources that prepared candidates for an examination or other meritocratic selection process were equally distributed, to a perfectly unequal society where only one family had the resources needed to prepare children for the otherwise meritocratic selection process.

The third would be the extent to which parental status predicted child success in the population that remained after imposing the previous criteria.  This is essentially what most contemporary studies of inter-generational mobility focus on: statistical associations in parent and child outcomes as a measure of social openness.  Again, one could imagine arraying societies on a spectrum that ranged from one extreme in which all eligible adults had equal chances of being selected into a political elite, to another imagined extreme where only the children of elite were themselves able to enter the elite.

Right now we don’t have the quantitative data that would allow a rigorous comparison between China and other historical societies on these dimensions, but at least in my opinion, a casual comparison of qualitative features of the processes for elite recruitment in the past  suggest that China would come out looking reasonably well.

On the first criteria, the share of the population that was explicitly forbidden from participating, I speculate that China would come out of a comparison reasonably favorably.  As Mark notes there were periods when specific categories of people were excluded from eligibility for the exams, for example children from merchant were excluded during the Ming, and prostitutes, singers, entertainers and other “degraded” or “mean” occupations were excluded from eligibility during other periods.  For better or worse, Ronald Reagan, Arnold Schwarzenegger, and Al Franken all would have been forbidden from holding office in imperial China, at least in certain eras.  And of course each dynasty was ruled by an Emperor drawn from the imperial family.

The question is not whether privileged or excluded categories existed in China, but rather how their share of the population contrasted with the shares of the population accounted for by excluded groups in other societies.  I suspect it was overall much smaller, especially later in the Qing after the last remaining hereditary degraded statuses were largely abolished.   Most other historical societies were characterized by systems in which membership in the political elite was explicitly hereditary, and/or very large segments of the population were assigned to hereditary status categories that not only precluded participation in the competition to join the political elite, but also precluded participation of any sort.

On the second criteria, the implicit exclusion of individuals because their family circumstances didn’t allow for the investments in education necessary to make someone a viable candidate for recruitment into the political elite, I doubt China was much different from other societies.  Literacy and numeracy were formal or at least practical prerequisites for high office in most historical societies.  Before the advent of public education in the West in the nineteenth century, only a small proportion of families had the wherewithal to endow their children with the education necessary to prepare them for high government office, let alone fairly mundane office.  China may not have stood out in this regard, but it is unlikely that did especially poorly.

On the third criteria, the association between parent’s status and children’s outcomes in the population that was not explicitly excluded by virtue of membership in a particular category, or implicitly excluded by lack of resources, my own take based on Ho (1962) and my own work with James Lee on Liaoning is that China probably did reasonably well.  In Campbell and Lee (2003) we compared the associations we observed in Liaoning with ones we calculated from published results for Western populations, mainly urban, and found that the associations in Liaoning were much weaker.  Similarly, while Ho’s (1962) calculations are difficult to convert into a metric that would allow direct comparison with the West, I would speculate that the proportions of successful exam candidates from undistinguished families who went on to attain office were still higher than the proportions of children of modest origins who went on to hold important political office in the West.  Again, this is an empirical question.

What’s my conclusion?

While the recruitment of political elites in China may have had all of the problems that Mark identifies, it isn’t at all clear to me that it was any worse than any other society, and it certainly isn’t clear to me that empirical results justify Mark’s rather harsh judgment that ‘…among much of Chinese society before the 20th century the belief prevailed that “anyone could make it,” and the state connived at this; but literary sources make it clear that only the naïve clung to such a fantasy.’

First of all, it isn’t clear to me that non-literary sources confirm that the belief that “anyone could make it” was more of a fantasy in China than it was in the West, or anywhere else in the world.  It might have been more of a fantasy in China than in some imagined perfect society that exists only in Star Trek episodes in which everyone had equal chances of joining the political elite, but I’m not sure I understand the value of comparison with ideal types, or societies that only exist in Plato’s world of forms.  The same logic of comparisons to ideal types that is used to show that China was not a meritocracy could be used to show that the most Western countries are not democracies or even capitalist, and that the Soviet Union, Eastern Europe, and China were not communist, or even socialist.

Indeed, much of the non-literary and even literary evidence suggests the possibility that the system was at least as open in China than it was in the West, if not more so.  If we look at Western literary evidence, the assumption that social status depended heavily or almost entirely on ancestry is pretty clear in almost all of Western literature until the late nineteenth century.   I haven’t read Jane Austen for a while, but I don’t remember any characters not born into the gentry being described in particularly appealing terms.

In my own opinion, according to Rawlsian criteria, “behind the veil of ignorance”, an individual who had the choice of being born in China or the West before the nineteenth century but didn’t know what status he or she would have been born into, but who sought membership in the political elite, might have been indifferent between born in China or the West.

What are my conclusions?

The first is that to the extent the study of Chinese history wants to move beyond historiography and description, it has to be comparative.  It simply isn’t sufficient to assemble a collection of empirical facts about a society, and then based on those facts, make some general assessment of the properties of that society, for example, assess whether it was a meritocracy, or a democracy, or a theocracy, or a plutocracy, or some other -cracy.  To make such a statement in the absence of comparisons with other societies is an exercise in comparison with ideal types, and while intellectually stimulating, unlikely to resolve any debates.

The appropriate question isn’t whether China was a meritocracy overall, but whether the system for the recruitment of political elites was more or less meritocratic than other historical societies.  By that standard, China was probably comparable to other societies, and certainly not much worse.  The fact of the matter is that no society before the last half of the eighteen century was meritocratic by the high standards implied in Mark’s piece.  By the standards suggested in Mark’s piece, it isn’t even clear to me that most Western democracies would be considered meritocracies.  China probably deserves some credit for at least having articulated a ideology of meritocracy well before the West was even aware of the concept.

A related concern I have is that in this important discussion of the social origins of elites, sociologists who engage in quantitative studies of stratification are AWOL.  Stratification researchers appear to have become so fixated on applying log-linear models or estimation regressions on population-representative survey data in which substantively important but numerical few elites account for a small proportion of the sample that it seems to have abandoned interest in understanding the social origins of the people in the top tail of the distribution.   This is unfortunate.  Now, more than ever, it is urgent to understand the processes the lead to the formation of the elites whose decisions have a disproportionate impact on social organization, yet there are relatively few such studies.

I am grateful to Mark not only for bringing attention to the important work by Elman and others to the attention of a much wider audience, but also throwing some cold water on the extravagant claims about the system for the recruitment of political elites in historical China made by various parties.  I would suggest that in some ways access to entrance into the political elite in historical China was as open or perhaps more open than in the historical West, but that is a long way from concluding as some have that the system was an ideal or even an attractive one for selecting and promoting officials.  Whether the system actually worked as claimed and produced a talented and effective bureaucratic elite that governed effectively is an entirely different question from the one I am addressing here.

Campbell, Cameron and James Lee. 2003. “Social mobility from a kinship perspective: Rural Liaoning, 1789-1909.” International Review of Social History.  47:1-26. [LINK]  doi:10.1017/S0268416098003063

Campbell, Cameron and James Lee.  2008.  “Kinship, Employment and Marriage: The Importance of Kin Networks for Young Adult Males in Qing Liaoning.”  Social Science History.  32(2):175-214.  [LINK]

Campbell, Cameron and James Z. Lee.  2011.  “Kinship and the Long-Term Persistence of Inequality in Liaoning, China, 1749-2005.”  Chinese Sociological Review.  44(1):71-104. Pubmed[/a]

Elman,Benjamin A.  1991.  “Political, Social, and Cultural Reproduction via Civil Service Examinations in Late Imperial China.”  Journal of Asian Studies.  50(1): 7-28

Elman, Benjamin A.  2000.  A Cultural History of Civil Examinations in Late Imperial
China.  Berkeley: University of California Press.

Ho Ping-ti.  1962.  The Ladder of Success in Imperial China, Aspects of Social Mobility 1368-1911.

Hymes, Robert P. 1986.  Statesmen and gentlemen: The elite of Fu-chou, Chiang-hsi, in Northern and Southern Sung. Cambridge University Press.

Lee, James and Cameron Campbell. 1997.  Fate and Fortune in Rural China: Social Organization and Population Behavior in Liaoning, 1774-1873.  Cambridge University Press. [Link]

李中清 (Lee, James) and 康文林 (Campbell, Cameron).  2008. “中国农村传统社会的延续 – 辽宁(1749-2005)的阶层化对革命的挑战 (The Persistence and Challenges of Rural Social Stratification in Liaoning 1749-2005)”  清华大学学报哲学社会科学版  (Journal of Tsinghua University: Philosophy and Social Sciences).  23(4):26-34.