Many figures in traditional dress are gathered beneath a structure from which hang paper lists. In the foreground are horses. The colours are muted with age

Excerpt from the scroll Viewing the Pass Lists, traditionally attributed to Qiu Ying (1494-1552). National Palace Museum, Taipei/Wikipedia


The exam that broke society

Keju, China’s incredibly difficult civil service test, strengthened the state at the cost of freedom and creativity

by Yasheng Huang + BIO

Excerpt from the scroll Viewing the Pass Lists, traditionally attributed to Qiu Ying (1494-1552). National Palace Museum, Taipei/Wikipedia

On 7 and 8 June 2023, close to 13 million high-school students in China sat for the world’s most gruelling college entrance exam. ‘Imagine,’ wrote a Singapore journalist, ‘the SAT, ACT, and all of your AP tests rolled into two days. That’s Gao Kao, or “higher education exam”.’ In 2023, almost 2.6 million applied to sit China’s civil service exam to compete for only 37,100 slots.

Gao Kao and China’s civil service exam trace their origin to, and are modelled on, an ancient Chinese institution, Keju, the imperial civil service exam established by the Sui Dynasty (581-618). It can be translated as ‘subject recommendation’. Toward the end of its reign, the Qing dynasty (1644-1911) abolished it in 1905 as part of its effort to reform and modernise the Chinese system. Until then, Keju had been the principal recruitment route for imperial bureaucracy. Keju reached its apex during the Ming dynasty (1368-1644). All the prime ministers but one came through the Keju route and many of them were ranked at the very top in their exam cohort.

Keju was sheer memorisation. Testing was based primarily on the Confucian classics. And there was a lot to memorise. There were some 400,000 characters and phrases in the Confucian classics, according to Benjamin Elman’s book A Cultural History of Civil Examinations in Late Imperial China (2000). Preparation for the Keju began early. Boys aged as young as three to five began to practise their memorisation drills. After the immediate environs of their families, Keju was their first exposure to the world. Keju, which was open only to the male gender, was fiercely competitive. Using figures provided by Elman, during the Ming dynasty, 1 million regularly took the qualifying tests and, of these, eventually about 400 would make it to the final Jinshi round. Passing the first tier of Keju, known as the provincial exam, was a lot easier – working out to be 4 per cent on average during the Ming. Still, this was more cut-throat than getting into Harvard in most years.

The prestige of Keju was such that even an emperor coveted its bona fides. According to a legend, an emperor in the late Tang dynasty (618-907) hung on the wall of an imperial palace a wooden tablet proudly displaying his Keju degree – only it was fake. The emperor had it made for himself. This credentialism pervades officialdom today. Many Chinese government officials claim PhD degrees – earned or otherwise – on their résumés.

Much of the academic literature focuses on the meritocracy of Keju. The path-breaking book in this genre is Ping-ti Ho’s The Ladder of Success in Imperial China (1962). One of his observations is eye catching: more than half of those who obtained the Juren degree were first generation: ie, none of their ancestors had ever attained a Juren status. (Juren was, at the time, the first degree granted in the three-tiered hierarchy of Keju.) More recent literature demonstrates the political effects of Keju. In 1905, the Qing dynasty abolished Keju, dashing the aspirations of millions and sparking regional rebellions that eventually toppled China’s last imperial regime in 1911.

Keju cultivated and imposed the values of deference to authority and collectivism

The political dimension of Keju goes far beyond its meritocracy and its connection to the 1911 republican revolution. For an institution that had such deep penetration, both cross-sectionally in society and across time in history, Keju was all encompassing, laying claims to the time, effort and cognitive investment of a significant swathe of the male Chinese population. It was a state institution designed to augment the state’s own power and capabilities. Directly, the state monopolised the very best human capital; indirectly, the state deprived society of access to talent and pre-empted organised religion, commerce and the intelligentsia. Keju anchored Chinese autocracy.

Many young Chinese people are pictured in a long line outside an examination centre in a commercial district

Candidates queue for the national civil service examination on 27 March 2021 in Taiyuan, Shanxi province, China. Photo by Wu Junjie/China News Service via Getty

The impact of Keju is still felt today, not only in the form and practice of Gao Kao and the civil service exam but also because Keju incubated values and work ethics. Today, Chinese minds still bear its imprint. For one, Keju elevated the value of education and we see this effect today. A 2020 study shows that, for every doubling of successful Keju candidates per 10,000 of the population in the Ming-Qing period, there was a 6.9 per cent increase in years of schooling in 2010. The Keju exams loom as part of China’s human capital formation today, but they also cultivated and imposed the values of deference to authority and collectivism that the Chinese Communist Party has reaped richly for its rule and legitimacy.

But isn’t it the case that the West – Prussia, then the United Kingdom and the United States – all had their own civil service exams? How is it possible that a strong bureaucracy complemented rather than supplanted political and religious pluralisms in the West?

China and the West bureaucratised under an entirely different sequential order and under different contextual conditions, and these differences entail substantial implications for the subsequent political development. The civil service in the West was not a single-platform institution in the way that Keju was. There was a military civil service, a civil service for foreign affairs, for forestry, etc, etc. Multiple platforms of bureaucratic recruitment competed with one another and, collectively, they competed with other channels of mobility, such as the political parties and commerce. In the US, the Pendleton Act of 1883 removed the power of Congress and the political parties to control civil service appointments. Before the 1883 Act, federal appointees returned a portion of their salaries to the party that had appointed them. Civil service never replaced Congress or political parties in toto, as witnessed by the fact that Congress today wields enormous power over the bureaucracy, including the power of the purse that funds its operation.

Another difference – and this is a big one – is timing. In the 19th century, the US introduced bureaucracy when ‘[t]he two institutions of constraint, the rule of law and accountability, were the most highly developed,’ as Francis Fukuyama writes in Political Order and Political Decay (2014). The state in the US and the UK was already ‘a Shackled Leviathan’, to use the words of Daron Acemoglu and James A Robinson in their influential book, The Narrow Corridor (2020). The sequential order ran from politics to bureaucracy, not as in China from bureaucracy to politics. In the West, society was vibrant long before the state ramped up its administrative capacity. The rule of law, the principle of accountability, and the powers of the legislature and the political parties were already firmly entrenched. Yes, the Leviathan was shackled by society, but different parts of the Leviathan shackled each other. Bureaucracy in the US formed and gained power only under a myriad of constraints and contending forces, rather than the socioeconomic tabula rasa that greeted the arrival of Chinese bureaucracy.

Vladimir Putin’s autocracy pales in comparison with that of China’s president Xi Jinping

The civil service in the UK and the US was ensconced in pluralistic societies that enjoyed a degree of religious freedom and a modicum of emergent electoral democracy. A world of competing forces and constraints attended the arrival of bureaucracy, even helped to create it. Government bureaucracy competed in some situations or complemented in others with church, universities, commerce and other social groups for human capital, legitimacy and resources. For political development, birth order really matters.

In his book Strong Societies and Weak States (1988), Joel S Migdal identifies a common problem in the developing world – the struggle of the state to acquire autonomy and capabilities. China, through history and today, is exactly the opposite. The state dominates society. Vladimir Putin’s Russia is autocratic but his autocracy pales in comparison with that of China’s president Xi Jinping. Harassed and targeted by the state, opposition parties are still legal and tenuously legitimate in Russia and some of Putin’s critics command a sizeable following. Even the power to commit violence – war fighting – was outsourced to a private force, the mercenaries led by Yevgeny Prigozhin, an arrangement not even remotely conceivable in China.

An anxious looking young woman and a young man are reading notebooks or textbooks outside an examination hall

Last-minute revision before the 2010 civil service examination in Hefei, Anhui province, China. Photo AFP/Getty

Since 2013, against the increasingly dictatorial Xi, there have been two prominent critics of the president and both were dispensed with summarily. Unlike Putin who has to rely on extra-legal means to silence his critics, suggesting some formal constraints on him, Xi directed the full apparatus of the Chinese state after his critics. The Chinese court sentenced the businessman Ren Zhiqiang to 18 years in prison, and Tsinghua University promptly fired Xu Zhangrun, a law professor who wrote an open letter criticising Xi. Standing forlornly by themselves, neither Ren nor Xu commanded any formal political organisations behind them. In 2022, the Chinese regime put almost 400 million people under some sort of COVID-19 lockdown, a feat that is unimaginable in any other country.

An ultimate autocracy is one that reigns without society. Society shackles the state in many ways. One is ex ante: it checks and balances the actions of the state. The other is ex post. A strong society provides an outside option to those inside the state. Sometimes, this is derisively described as ‘a revolving door’, but it may also have the positive function of checking the power of the state. State functionaries can object to state actions by voting with their feet, as many US civil servants did during the Donald Trump administration, and thereby drain the state of the valuable human capital it needs to function and operate. A strong society raises the opportunity costs for the state to recruit human capital but such a receptor function of society has never existed at scale in imperial China nor today, thanks – in large part, I would argue – to Keju.

Keju was so precocious that it pre-empted and displaced an emergent society. Meritocracy empowered the Chinese state at a time when society was still at an embryonic stage. Massive resources and administrative manpower were poured into Keju such that it completely eclipsed all other channels of upward mobility that could have emerged. In that sense, the celebration by many of Keju’s meritocracy misses the bigger picture of Chinese history. It is a view of a tree rather than of a forest. The crowding-out effect of Keju is captured succinctly in a book from the late 19th century:

Since the introduction of the examination system … scholars have forsaken their studies, peasants their ploughs, artisans their crafts, and merchants their trades; all have turned their attention to but one thing – government office. This is because the official has all the combined advantages of the four without requiring their necessary toil …

This is the larger impact of Keju. Its impressive bureaucratic mobility demolished all other mobility channels and possibilities. Keju was an anti-mobility mobility channel. It packed all the upward mobility within one channel – that of the state. Society was crowded out, and over time, due to its deficient access to quality human capital, it atrophied. This is the root of the power of Chinese autocracy and, I would argue, it is a historical development that is unique to China and explains the awesome power of Chinese autocracy.

China has legions of intellectuals, but it is bereft of an intelligentsia

Take intellectuals as an example. Keju inculcated literacy and helped create a vibrant book readership. Book ownership was widespread as early as the Ming dynasty. ‘More books were available,’ writes Timothy Brook in The Troubled Empire (2010), ‘and more people read and owned more books, in the late Ming than at any earlier time in history, anywhere in the world.’ Brook sums up the impressions of Jesuits visiting China: ‘More surprising, perhaps, is that complete illiterates may well have been a minority in the late Ming.’

But a striking fact is that no organised intelligentsia of any significant size and visibility ever emerged in imperial China. There were no Chinese equivalents of the Royal Society in Britain or the many learned societies in France. One that left a mark is the Donglin Academy, a private discussion forum founded in 1111 by intellectuals of the Song dynasty (960-1279). The academy lasted as long as its founders’ lifespan and vanished into obscurity after their expiry. It was revived in 1604 during the reign of the Wanli emperor (1573-1620), but it operated as a political rather than an intellectual force. The scholar-officials formed a Donglin Faction, later brutally put down by the powerful eunuchs of the Ming court. The grand total of the second life of the Donglin Academy is 21 years, from 1604 to 1625.

The term ‘scholar official’ is of Chinese coinage and it is evocative of China’s lacuna of intellectuals as an institutionalised establishment. Compare that situation with Tsarist Russia, another autocracy. Russians coined the term ‘intelligentsia’ – intellectuals as a class – and Russian intellectuals have a long tradition of standing apart from and defining their identity as separate to the state. China has legions of intellectuals, but it is bereft of an intelligentsia.

Prior to Keju and even during the early centuries of Keju, China had a plurality of upward mobility. Within bureaucracy, officials were appointed through nepotism, family ties, heredity and recommendations. Commerce, while always curtailed, was a nascent force, promising to burst forward. The Song dynasty experienced a vibrant development of commerce and a market economy. Although Confucianism was always the first among equals, other ideologies, such as Legalism, Daoism and Buddhism, cohabitated with Confucianism and vied with one another for the Chinese population’s attention and adherence.

But these societal forces were too nascent and too embryonic by the time Keju arrived and matured. They had yet to acquire their own unique identity, significant organisation and autonomous agency. In imperial China, there never was a level playing field between state and society, and over nearly 1,500 years, Keju further deprived the congenitally deficient society of its oxygen – human capital. Fukuyama is right to assert that the Chinese state was precocious, but it was precocious in a particular fashion: its precocity contrasted sharply with the immaturity of Chinese society.

The most direct way Keju decimated Chinese society is through talent monopoly but there were others. Keju also monopolised the time and mental energy of its candidates. Keju was not a one-shot deal. A candidate could take the test multiple times. In a dataset that has information on the 11,706 Keju candidates during the Ming dynasty, the average age passing the final stage of Keju was 32, approaching middle age at a time when average life expectancy was much lower than today. The oldest in the dataset was was probably Gui Youguang (1506-1571). Before passing the provincial examination in 1540 at the youngish age of 34, Gui had already failed it on six occasions. He then proceeded to toil for more than 24 years of his life and finally attained his Jinshi degree in 1565, although ranking near the bottom of his class and at the ripe age of 59. Unfortunately, he did not bask in his exalted status for long, as he died aged 65. For him, and many others, Keju was a life-long endeavour.

The Keju curriculum was formidable and required memorising close to 400,000 characters. Is there spare residual energy, capacity and curiosity left to pursue other mentally taxing activities, such as ideation of new thoughts, new politics, and discoveries of natural phenomena? In my book The Rise and Fall of the EAST (2023), I show that Chinese technology began to stagnate as Keju gained dominance. The brain power that ended up in the state did not flow to Chinese society, the economy or human creativity.

Mental energy aside, the values drilled deeply into Keju candidates were pro-autocracy and authoritarian. Keju legitimates statism. Boys as young as three or four began to practise writing characters that were meant to instil admiration of, and devotion to, the ideas and teachings of the master – Confucius – which would eventually be tested on Keju. By the Ming dynasty, the initial plurality of the Keju subjects gave way to one subject only, Confucianism – ‘knowledge of classics, stereotyped theories of administration, and literary attainments’.

Autocracy and Keju became ever more intimately intertwined

Imagine repeated exposures to the statist values at that tender age, producing what psychologists call ‘an imprinting effect’. The autocratic values were incubated in substance but also by the format of Keju; this was standardised testing par excellence. When Keju was first established, candidates were tested on a wide range of subject matters but, after the Song dynasty, the Keju curriculum became progressively stratified and exceedingly narrow. Candidates were required to fill in the blanks with missing words or phrases in excerpted texts from the Confucian classics. The Yuan dynasty (1271-1368) narrowed the Keju curriculum further. Only a streamlined version of annotations of Confucian classics was allowed, the so-called Neo-Confucianism, which was the brainchild of the great Confucian scholar Zhu Xi (1130-1200) of the Song dynasty.

Neo-Confucianism is a pared-down version of classical Confucianism, and it strips away some of the moral veneer of its classical predecessor. Summarising a common view among historians, Peter K Bol observes in Neo-Confucianism in History (2010) that this version of Confucianism ‘provided a justification for seeking external authority in the ruler’ and stipulated the responsibility for transforming the world as that of the emperor alone. The Neo-Confucianist Keju curriculum was rigid, narrow and absolutist, and was single-minded in its advocacy of a hierarchical order – subordination to the ruler, to the elderly, and to the male gender. No scope for scepticism and ambiguity was allowed. Autocracy and Keju thus became ever more intimately intertwined.

There was, however, a massive operational advantage to the Neo-Confucianist curriculum: it standardised everything. Standardisation abhors nuance and the evaluations became more straightforward as the baseline comparison was more clearly delineated. There was objectivity, even if the objectivity was a manufactured artefact. The Chinese invented the modern state and meritocracy, but above all the Chinese invented specialised standardised testing – the memorisation, cognitive inclination and frame of references of an exceedingly narrow ideology.

Ming standardised Keju further: it enforced a highly scripted essay format, known as the ‘eight-legged essay’, or baguwen in Chinese (八股文), to which every Keju candidate had to adhere. A ‘leg’ here refers to each section of an essay, with a Keju essay requiring eight sections: 1) breaking open the topic; 2) receiving the topic; 3) beginning the discussion; 4) the initial leg; 5) the transition leg; 6) the middle leg; 7) the later leg; and 8) conclusion. The eight-legged essay fixed more than the aggregate structure of exposition. The specifications were granular and detailed. For example, the number of phrases was specified in each of the sections and the entire essay required expressions in paired sentences – a minimum of six paired sentences, up to a maximum of 12. The key contribution of the eight-legged essay is that it packed information into a pre-set presentational format.

Standardisation was designed to scale the Keju system and it succeeded brilliantly in that regard, but it had a devastating effect on expositional freedom and human creativity. All elements of subjectivity and judgment were taken out. In his book Traditional Government in Imperial China (1982), the historian Ch’ien Mu describes the ‘eight-legged essay’ as ‘the greatest destroyer of human talent’.

A bane to human creativity was a boon to autocracy. Standardised testing was conducive to authoritarianism. In his book Who’s Afraid of the Big Bad Dragon? (2014), Yong Zhao, professor at the School of Education of the University of Kansas, notes a natural compatibility between authoritarianism and standardised testing. Authoritarianism, he writes, ‘sees education as a way to instil in all students the same knowledge and skills deemed valuable by the authority.’ The standardised tests appeal to an authoritative body for correct answers; as Zhao said in an interview for the US National Education Policy Center, the tests ‘force students to comply with the answers or the way of thinking that the authority wants.’ The direction of deference is automatically established: ‘Then you hold the students, the teachers and, to a lesser extent, the parents accountable for being able to get the answers that the authority wants and to show that they have mastered the skills and the knowledge and possibly even the beliefs that the authority wants.’

Confucianism, thus, functioned as an equivalent of the abstruse and arcane vocabulary of the SAT

In his book The WEIRDest People in the World (2020), Joseph Henrich posited that the West prospered because of its early lead in literacy. Yet the substantial Keju literacy produced none of the liberalising effects on Chinese ideas, economy or society. The literacy that Henrich had in mind was a particular kind of literacy – Protestant literacy – and the contrast with Keju literacy could not have been sharper. Keju literacy was drilled and practised in classical and highly stratified Chinese, the language of the imperial court rather than the language of the masses, in sharp contrast to Protestant literacy. Protestant literacy empowered personal agency by embracing and spreading vernaculars of the masses. Henrich’s liberalising ‘WEIRD’ effect – Western, educated, industrialised, rich and democratic – was a byproduct of Protestant literacy. It is no accident that Keju literacy produced an opposite effect.

Why was there such a close affinity between Keju and Confucianism? The answer is not obvious. Ancient China boasted other great ideologies and traditions, such as Daoism, Mohism and Legalism, but they were completely absent in the Keju curriculum. This ideological single-mindedness of Keju is puzzling and it is puzzling still considering the following: in my book, I document that several emperors who played an instrumental role in inventing and developing Keju were not Confucianists themselves.

The answer may lie in an operational imperative of Keju. Standardised testing is necessary when you want to scale the evaluation. Subjective evaluations, such as relying on reputation, recommendations and interviews, are feasible when the number of candidates under evaluation is small. For example, the Big Three colleges in the US – Harvard, Yale and Princeton – began to embrace the SAT (the standardised test for college admissions) when they started recruiting beyond their traditional, narrow socioeconomic group – the white Anglo-Saxon Protestants (WASPs) in the elite private schools of the east coast. The Chinese emperors made the same decision when they expanded bureaucratic recruitment beyond the nobility and wealthy elites. Standardising and constricting the Keju curriculum were not an optional luxury; it was a necessity to scale Keju.

Confucianism offered an operational advantage. It is textually rich; the verbiage is massive, and the pontifications are incredibly involved, not unlike the verbal portion of the SAT. As noted before, there are approximately 400,000 characters and phrases in the Confucian classics. Using a website, Chinese Text Project, ‘an online open-access digital library that makes pre-modern Chinese texts available to readers and researchers all around the world’, I found that among the classical texts created before the Han dynasty (206 BCE-220 CE) Confucianism is paragraphically the richest, with 11,184 paragraphs. No other ideologies come remotely close. Legalism has 1,783 paragraphs; Daoism has 1,161 paragraphs, and Mohism has 915 paragraphs. Confucianism, thus, functioned as an equivalent of the abstruse and arcane vocabulary of the SAT, and it was most suited for screening and selecting the desired human capital from a large pool of candidates.

Is it at all possible that Keju successfully anchored and shaped the nature of the Chinese autocracy because of this accidental feature of Confucianism and on account of an operational technicality? Let’s pause, savour and ponder for a moment the momentous implications of this proposition.

This essay is adapted from the book The Rise and Fall of the EAST: How Exam, Autocracy, Stability and Technology Brought China Success, and Why They Might Lead to its Decline (2023) by Yasheng Huang.