Can Cultural Norms Reduce Conflicts?

Confucianism
and Peasant Rebellions in Qing China
James Kai-sing Kung

Chicheng Ma

This version, August 2012

ABSTRACT: Based on a panel dataset spanning 260 years (1651-1910) uniquely constructed on 107 counties of a Chinese province strongly steeped in the traditions of
both Confucianism and peasant rebellions, we find that, while economic shocks as
measured by crop failures and suboptimal rainfall trigger peasant rebellions, Confucianismas proxied by schools, temples, and folk customsreduces them. These results are robust to the inclusion of a wide gamut of controls and to the use of the
number of ancient Confucian sages enumerated between 500 B.C. and 1643 A.D. as
an instrument for the strength of Confucianism during the Qing dynasty.
Keywords: Cultural norms, Confucianism, Economic shocks, Conflicts, Peasant rebellions, Imperial China
JEL Codes: H56, N45, O10, Z12

Corresponding author. Division of Social Science, The Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong. Email: [email protected].

School of Economics, Shandong University, 27 Shanda Nanlu, Jinan, Shandong Province, China.
Email: [email protected]

1. Introduction
There is now more or less a consensus that economic (climate) shocks tend to trigger social conflicts (Collier and Hoeffler, 1998, 2004; Miguel, Satyanath and Sergenti, 2004; Blattman and Miguel, 2010, Bruckner and Ciccone, 2010; Besley and Persson, 2011; among others).
But much less is known, at least empirically, about the potential attenuating effect of cultural
norms on conflicts, despite the theoretical claim that such an effect exists (Rasmusen, 1996;
Posner, 2000; Putnam, 2000; Funk, 2004).1 By employing a unique dataset, we examine
whether a set of cultural (Confucian) norms that have persisted for at least several millennia
have had the benign effect of reducing social conflicts triggered by economic shocks.
We predicate our analysis on a specific form of social conflictpeasant rebellionsand
examine whether the cultural norms associated with Confucianism served to attenuate the
effect of economic shocks in triggering peasant rebellions in the last 267 years of Chinas dynastic rule (the Qing dynasty, circa 1644-1911). We choose to study peasant rebellions because they are the principal form of social conflict in agrarian societies. In these societies,
peasants live so close to subsistence, that periodic food shortages caused by the vagaries of
weather easily give rise to what James Scott (1976) coins subsistence ethican ideology
that justifies the occasional robbing and plundering for the sake of survival. The ties with
which the peasant rebels have had with their village communities render them neither fulltime military nor criminals, only someone forced to switch from farm work to robbery in
times of economic hardship (Hobsbawn, 1972; Scott, 1976). This two-faced nature of peasants renders peasant rebellion a uniquely interesting form of conflict whose occurrence may
occasionally be sparked by economic shocks.
To suppress this subsistence ethic, Chinas emperors had tirelessly inculcated in the
populace the Confucian norms throughout the countrys long history (circa 206 B.C. to 1911
A.D.).2 By imposing shame and moral distress on those who fail to abide by the principles
taught, Confucianism stresses the importance of subordinationof a subject to the ruler,
of a son to his father, and of a wife to her husband; the intended result was conflict avoidance. To foster this ethos, the Chinese emperors developed an education and civil service examination system based almost exclusively on the memorization of Confucian classics and

A notable example is shaming punishments, which involve deliberate public humiliation and moral
distress of the offender, as alternative to the formal criminal justice system (Posner, 2000). Recent
endeavors of the United Nations and other international communities in implementing a peace
education program in those areas of Africa rife with civil conflicts are premised on this principle
(UNICEF, 2009; Blattman, Hartman and Blair, 2011).
2
It is easy to understand why the Chinese emperors were concerned about peasant uprisings.
Throughout the countrys long history, peasant rebellions had directly overthrown Qin (221-207 B.C.),
Han (202 B.C.220 A.D.), Sui (581-618), Yuan (1271-1368), and the Ming dynasties (1368-1643)
(Wakeman, 1975).
1

rote learning,3 built temples to honor Confucius and his followers, praised the so-called
chaste women (lie nv) for the sacrifice they allegedly made by remaining widows or even
committed suicide upon their husbands deaths, and so forth.4 While the Confucian ethos
was undoubtedly undermined during the tumultuous times of the Communist rule in the
twentieth century, it has been revived and persists even to this day.5 Indeed, few civilizations
have seen the persistence of cultural norms for such a long period of time, which presents an
ideal situation for testing the possible effect of culture on conflict reduction.
To do this, we construct a panel dataset of 107 counties in Shandong Province that covers
260 years (1651-1910), a period that overlaps basically with the Qing dynasty. We choose the
Qing dynasty for analysis because after several millennia of diffusion Confucianism had by
then been firmly established as the cornerstone of morality in the Chinese society, permeating thoroughly to even the bottom wrung of the social hierarchy (Yang, 1961; Ho, 1962). The
primary reasons for choosing Shandong Province are two-fold. Foremost is that Shandong
(Qufu County specifically) is the birthplace of Confucius, which explains why, throughout
Chinas long imperial history a good number of emperors had zealously promoted Qufu as
the holy land or Mecca of Confucianism. Choosing Shandong Province thus allows us to
examine the strongest possible effect of Confucian culture on reducing social conflicts. Shandong is also ideal for studying peasant rebellions because historically the province had been
prone to both floods (from the levee breach of the Yellow River) and droughts and had
turned out to be the heartland of peasant rebellions in China.
Empirically measuring the strength or more specifically the popularity of Confucianism,
however, is challenging. In light of the importance that the Chinese emperors accorded to
education, religion and folk customs more generally, we employ the numbers of schools,
Confucian temples, and chaste women to proxy for the strength of Confucianism. Given that
cultural norms tend to evolve very slowly over time, we consider our three measures of Confucianism to have been fairly stable throughout the Qing dynasty.
Our empirical analysis finds that, while economic shocks as measured by crop failures
and rainfall variations do have the expected positive effect on the number of peasant rebellions, the effect can be mitigated by the Confucian norms. Specifically, an additional school,
Confucian temple, or chaste woman (per 10,000 km2) invariably reduces peasant uprisings by
27.3 percent, 4.4 percent, or 0.2 percent, respectively.

Established in the tenth century, the civil service examination was designed to select qualified candidates to serve in the states bureaucracy.
4
The virtuous deed of steadfastly preserving purity and loyalty to their deceased husbands was
held in the highest honor. These women allegedly exemplified the important virtues of Confucianism,
namely subordination, loyalty and purity.
5
After being denounced during the Communist era, Confucianism was recently embraced, albeit cautiously by the Chinese Communist Party once again as the moral cornerstone on which to reestablish social morality and maintain a harmonious society. The Economist, April 28, 2011.
2

In examining the effect of culture, it is important that we disentangle its effects from those
other socioeconomic characteristics correlated with the Confucian norms. A notable case in
point is social mobility achieved via success in civil service examinations, whose syllabi drew
heavily on the memorization of Confucian classics. To rule out the possibility that Confucian
norms may affect peasant rebellions via the channel of social mobility (i.e., those who failed
the exam became a rebel), we control for the success in imperial exams in a county using the
number of provincial degree holders (juren) as proxy.6 The quality of local governance represents another concern (of omitted variable bias). We note that, while good governance is likely to reduce conflicts, it may simultaneously affect the diffusion of Confucian norms. To reduce this possibility, we employ the size of the local gentrywho in late imperial China
were typically responsible for providing the local public goodsto proxy for the quality of
local governance. Other control variables include famine relief, the effects of other cultural
norms (Taoism and Buddhism), the potential cost of rebellion (proxied by the number of imperial soldiers stationed in a county), population density, and economic prosperity (proxied
by a dummy of whether a county was a designated commercial center). We also control for
the possible effect of Western influence in the late Qing period using a restricted sample that
excludes the years after 1860, a time when China was forced to thoroughly open up to the
West. After controlling for all these factors, the result of Confucian norms mitigating the effect of crop failures on peasant rebellions remains unchanged.
Given that it is impossible to control for every conceivable factor that may bear upon both
Confucian norms and peasant rebellions, plus our proxies of Confucian norms may contain
measurement error, we adopt an instrumental variable approach in which we employ the
total number of ancient Confucian sages that ever lived prior to the Qing dynasty (from
500BC to 1643AD) to instrument for the strength of Confucian norms during Qing times. Our
choice is premised on the reasoning that cultural norms tend to evolve very slowly and have
a cumulative impact; hence the strength of Confucian norms as it had developed over the
past 2000 years would likely have continued into the Qing dynasty, but is otherwise uncorrelated with peasant rebellions at the same time other than through the channel of the Confucian norms. The instrumented results reaffirm the finding that counties with stronger Confucian norms experienced fewer peasant rebellions when confronted by economic shocks.
In addition to providing fresh empirical insights into the question of whether cultural
norms have a mitigating effect on social conflicts triggered by economic shocks, our study
also contributes to a small but growing literature that examines the direct effect of cultural
6

The civil service examination is made up of three levels of degrees. The lowest level, the shengyuan
(equivalent to a bachelors degree), was taken at the county and prefecture levels. The next level up,
the juren (equivalent to a masters degree), was taken at the provincial level. The highest degree, the
jinshi (equivalent to a doctoral degree), was taken at the metropolitan (palace) level (Chang, 1955; Ho,
1962). We choose the number of juren as the proxy for social mobility because this qualification
facilitated entry into the civil service.
3

norms on conflicts, violence and crime. For example, Voigtlander and Voth (2012) find that
medieval anti-Semitism, a norm that originated largely from blaming the Jews during the
Black Death (1348-50), can explain the violence against the Jews in the 1920s. Similarly, Fisman and Miguel (2007) find that diplomats from countries with severe corruption proclivities have a tendency to commit unpaid parking violations in the United Nations.7 Of course,
not all cultural norms have a negative impact on social behavior. We aim to demonstrate how,
in the context of Chinas long history of civilization, the stable cultural norms of Confucianism had had the effect of mitigating social conflicts triggered by economic shocks. In this respect, our work coincides with that of MacCulloch and Pezzini (2002), who find that Christian beliefs have the benign effect of reducing the taste for revolts, and is in line with Buonanno, Montolio and Vanins (2009) finding that a fear of social stigmatization and ostracism
can similarly deter criminal behaviors.
The remainder of this paper proceeds as follows. The next section, Section 2, provides a
historical background on peasant rebellions and Confucianism in late imperial China. Section
3 describes our data sources and the construction of our variables. The main effect of economic shocks on peasant rebellions is examined in Section 4, whereas Sections 5 tests the impact of Confucian norms on peasant rebellions. Section 6 deals with various estimation issues,
first by adding a wide array of controls to our estimations, followed by discussing the rationale and results of the instrumental variable estimations. Section 7 provides a brief conclusion to the study.

2. Historical Background
2.1. Economic Shocks and Peasant Rebellions in Historical China
By and large peasant rebellions had afflicted the entire imperial China (221BC-1911AD),
which for thousands of years had remained a predominantly agrarian economy subject to the
vicissitudes of weather.8 In this land of famine, to borrow Mallorys (1926) term, the repeated ravages of weather shocks, periodic food shortages (and famines in the more extreme
instances) and accordingly peasant rebellions were indeed a recurrent feature of rural Chi-

Other interesting examples of this endeavor include Miguel, Saiegh and Satyanaths (2008) study
showing that the history of civil conflicts in a soccer players home countrya proxy for a culture of
violenceis positively correlated with his propensity to behave violently on the soccer field, and
Beber and Blattmans (2011) study of the importance of rebel leaders ideological indoctrination in the
recruitment of child soldiers in Africa. Additionally, there are a number of studies examining the role
of religious (ethnic) diversity in triggering religious (ethnic) conflicts (Fearon and Laitin, 2003; Collier
and Hoeffler, 2004; Montalvo and Reynal-Querol, 2005).
8
In fact, China remained an agrarian economy until the early twentieth century, with roughly 95
percent of the population still classified as peasants then (Zhang, 1931).
4

nas economic and social reality.9

The periodic food shortages caused by the vagaries of weather easily gave rise to what
James Scott (1976) coins subsistence ethicone that justifies the occasional robbing and
plundering by peasants when their subsistence is threatened (see also Hobsbawn, 1972).
Frederic Wakemans (1975, p. 6) depiction of the Chinese peasant-rebel as someone who
constituted the economic foundation of traditional Chinese society (by devoting their lives
to laborious farm work), but also stepped out of this role and momentarily attached themselves just as firmly to ambitious bandits at the head of rebel armies when facing economic
scarcity, testifies to the typical characterization alluded to earlier by Scott (1976). Indeed, this
pattern of time allocation between farming and banditry augurs well with the records of
Qing Chinas Board of Punishments, which found that banditry was usually a seasonal activity closely tied to the agricultural calendar. In particular the prime time for bandits coincided with the slack farming seasons in the winter months of North China when there was
little if any work to be done, with banditry falling off remarkably after entering the busy
agricultural season of Spring (Esherick, 1987, pp. 22-3).10
Thus, notwithstanding the fact that some well-known peasant rebellions in the course of
Chinese history had resulted in dynastic decline, the vast majority of these revolts were of a
significantly smaller scale involving no more than hundreds of people and assuming the
form of merely local unrests, most notably banditry where the resource-rich and powerful
officials were plundered in times of economic hardship. This feature is well illustrated in C.
K. Yangs (1975) study of social unrests that occurred during the second half of Chinas Qing
dynasty (1796-1911), where he found that less than 4.7 percent of the peasant rebellions
enumerated could be classified as significant, i.e. involving over 10,000 participants.11
2.2. Confucianism

Evidence based on the past several thousand years shows that droughts had occurred in especially
the north and central provinces at the rate of roughly once every seven years, whereas floods about
once every four years (Chu, 1926; Mallory, 1926). Moreover, the lack of irrigation infrastructure in
most places suggests that crop yields were totally dependent on the vagaries of weather (Sun, 1957).
10
This feature of peasant rebellions was found especially pervasive on the North China plain during
the nineteenth century, where an aggressive survival stylearticulated in the forms of banditry,
organized feuds, and local militarizationsaw the peasants switch repeatedly between farming and
banditry conditional on the vagaries of weather (Perry, 1980).
11
More specifically, there were altogether a total of 6,643 incidents of social unrests in the whole
country in that period, or an average of about 57 incidents per year. The average number of
participants in each incident of unrest ranged from less than 100 to over 1,000,000, and the duration
from less than one month to several years. Ninety percent of these unrests were organized and carried
out by peasants and occurred in the rural areas. Wolf (1969), Chesneauxs (1973), Feuerwerker (1975),
Tong (1991), and Rowe (2009) similarly provide a rich description of peasant rebellions in historical
China.
5

Moral suasion, especially when applied via the subtle indoctrinating influence of cultural
norms, can be an effective means of governance. Throughout various dynasties, Chinas imperial authorities had indeed appealed to the virtues of subordination and pacifismthe
twin pillars of Confucianismin suppressing the subsistence ethic.12 The founding emperor of the Ming dynasty (1368-1643), Zhu Yuanzhang, was convinced that indoctrination is
the key to govern a country. As a matter of fact, most emperors in Chinese history had religiously adhered to the following creed from Confucius:
Lead the people with governmental measures and regulate them by law and punishment, and
they will avoid wrongdoing but will have no sense of honor and shame. Lead them with virtue and
regulate them by the rules of propriety, and they will have a sense of shame and, moreover, set
themselves right (The Analects).
Indeed, Confucianism had been actively promoted by various Chinese emperors as orthodox ideology (state religion) for over two thousand years (throughout almost the entire
dynastic regime) on grounds of the powerful conviction that its pacifist doctrines are capable of discouraging conflicts and violence. As a complex system of ethics, Confucianism
has persisted tenaciously for several millennia since sixth century B.C. when Confucius (551479 B.C.) first began teaching. The gist of Confucianism, in a nutshell, is to pursue harmonious human relations through the cultivation or (more appropriately) indoctrination of the
twin virtues of subordination and pacifism.13 Embodied in the Three Guiding Principles (san
gang), which had been actively promoted throughout various imperial regimes, the Confucian ethics advocate specifically the subordination of a subject to the ruler (with a distinct
emphasis on loyalty towards the emperorthe so-called Mandate of Heaven), of a son to
his father and of a wife to her husband. By distinctly emphasizing filial piety and subordination within the family, the latter two principles constitute the micro-basis justifying subordination to the rulersthe first and overriding principle (Yao, 2000).
While the Three Guiding Principles lay down a clear subordinating relationship between
actors within both the society and family, the Five Constant Regulations (wu chang) stress a
number of social norms (humaneness/compassion, righteousness, propriety, wisdom, and
faithfulness), the violation of which would lead to severe penalty in the form of stigmatiza-

12

According to historians of China, the use of moral suasion for social control was also a consequence
of: a) the sheer area of the country and its population, and b) the lack of a modern, sophisticated
bureaucratic system (Hsiao, 1960; Wakeman, 1975).
13
From the Confucian perspective, the causes of conflict lie in ones own heart; people will naturally
have peace when the heart is at peace; it is not riches or power, but virtuous manners that constitute
harmonious human relations. In contrast, an uncultivated character causes one to be unhappy and
prone to complaints, triggering conflicts and violence (Yao, 2000, pp. 179-80).
6

tion and ostracism within ones community and family. Fear of such punishments, it is believed, would help suppress violence and reduce conflicts (Yao, 2000).
2.3. The Diffusion of Confucianism among the People
From (as early as) the West Han dynasty (206209 B.C.) onwards, the imperial authorities
had consciously and persistently promoted the Confucian virtues through: 1) the formal education system, 2) the erection of temples for worshipping and honoring Confucius and
those adhering religiously to his cardinal principles, and 3) honoring behavior that emphasized these principles. In the realm of education, Confucian classics were almost exclusively
taught in schools (Weber, 1922; Elman, 2000; Yuchtman, 2010), mainly because social mobility could only be achieved via passing the imperial civil service examinations, which were
drawn heavily on the memorization of Confucian classics and rote learning.14 In addition,
Confucianism was favored for the moral impact the emperors believed it would have on the
public (Rawski, 1979). By around the late fourteenth century, an elaborate school system had
been established, and, after allocating resources to support the elementary schools, in 1652
the Qing court decreed that every township must set up a school and to select teachers with
honest and sincere character. The emphasis on a Confucian-based education/examination
system allegedly had the effect of allowing families from a commoner background
meaning families without educated literati or gentry scholars as membersto study Confucianism and possibly even achieve success in the imperial exams (Ho, 1962; Rawski, 1979).
Beyond education, the erection of Confucian temples within which the associated rituals
could be practiced represented another important means of promoting Confucianism (Yang,
1961; Ho, 1962; Chow, 1994). Specifically, local governments (at the county and township
levels) were required to build temples and to perform sacrificial rites to worship the Confucian sages like Confucius and Mencius and their followers (the Confucian scholars).15 To

14

Confucian classics refer to the Four Books of Great Learning, Doctrine of the Mean, Analects of Confucius,
and Mencius, and the Five Classics of Book of Changes, Classic of Poetry, Classic of Rites, Classic of History,
and Spring and Autumn Annals (see Nylan, 2001 for details). Students began their studies as young
children by learning basic Chinese characters and memorizing the thousands of characters composing
the Confucian classics. They then went on to master and memorize the Four Books and the Five
Classics, and to practice their composition skills of the eight-legged essays that were required in the
examinations. Small surprise, students at the time had little incentive to acquire knowledge beyond
what was required for exam success (Elman, 2000; Yuchtman, 2010).
15
Emperor Yongzheng (1722 to 1735) of the Qing dynasty decreed that counties should build temples
in which men (and women) who excelled in filial piety and loyalty should be duly honored. With
guidance provided by the imperial court, elaborate and complex sacrificial rites and ceremonies were
periodically held to honor those Confucian sages whose images/bodies/ideals were enshrined in the
Confucian temples. It is believed that the religious, social, and psychological dimensions of these
rituals are most effective in perfecting the characters of the people toward those of the
exemplary sages (Yao, 2000).
7

complement this endeavor, local officials and gentry,16 like their Western missionaries counterparts, similarly put a great deal of effort into popularizing Confucianism by persuading
peasants of the importance of the moral principles underpinning the Confucian values
(Chang, 1955; Wakeman, 1975; Fairbank and Reischauer, 1989).17
Last but not least, Confucian norms were also diffused through publicly praising local
examples of laudable behavior and conduct. A notable Confucian exemplar was chaste
women or lie nv, who were basically widows who had vowed not to remarry or, at the extreme, even committed suicide after their husbands deaths, in order to demonstrate their
unshaken determination to preserve fidelity and loyalty to their (deceased) husbands until
death.18 Reflecting the Confucian ethics of subordination and loyalty (which serves as the
foundation of subordination to the ruler), chaste women were held in the highest regard by
the imperial authorities and greatly honored in their own local communities from as early as
the Zhou dynasty (1046256 B.C.). During the Ming and Qing dynasties, the imperial authorities promoted this virtue to an unprecedented level. Not only were a set of sophisticated
laws established for the sake of canonizing the chaste women, rituals were also developed
for honoring them (Chow, 1994; Mann, 1997). For example, after getting nominated by the
local gentry, candidates were subjected to strict scrutiny by various levels of government authorities before their purity was certified and the coveted honor of a chaste woman was
eventually conferred on them by the court. After receiving the honor, memorial archways
would be constructed in their hometowns to broadcast these virtues.
Not unexpectedly, the number of chaste women increased by leaps and bounds in late
imperial times. Indeed, historical records indicate that, when measured on a decadal basis
only one chaste woman had been identified in every one million of the population in the Yuan dynasty (1271-1368), but that number increased sharply to 13 and 16 during the Ming and
Qing dynasties respectively (Figure 1).19 This surge suggests the growing importance that the
imperial authorities had placed upon the Confucian ethics and norms as the last Chinese
dynasty approached its eventual decline.
[Figure 1 about here]

16

According to Wakeman (1975), every gentry was deeply schooled in ethical Confucianism and
passed a set of imperial examinations drawing heavily Confucian classics, and insisted that the duty
of a gentleman was the moral improvement of society (p. 25).
17
To nurture peoples understanding and acceptance of the Confucian values, an imperial edict was
decreed in 1670 by Emperor Kangxi that the sixteen moral doctrines had to be taught twice a month.
18
The deeds of chaste women were recorded in detail in the local gazetteers during the Ming and
Qing dynasties. An example is Jing Shi, a native of Licheng County, Shandong Province, who was
conferred the honor of a chaste woman for her determination to continue looking after her son and
parents-in-law after her husband passed away when she was only 28 (Licheng Gazetteer, 1926).
19
The data on chaste women are obtained from the Gu jin tu shu ji cheng (Great Collection of Ancient and
Modern Books) (1726).
8

After two thousand years, Confucian ideology had arguably permeated the entire social
strata and established itself as the predominant social norms among the Chinese populace
(Yang, 1961; Ho, 1962; Fairbank and Reischauer, 1989). This may be evidenced, at least in
part, by the fact that every township in Qing China had a temple built to honor and venerate
Confucius and his followers, to the degree that veneration of the Confucian sages is likened
to the Christians worship of Jesus Christ and the Muslims reverence for their prophet Mohammed (Carus, 1918, p. 155).
The effective diffusion of Confucian values and norms can also be gleaned from the fact
that the overriding goal of suppressing the Taiping Rebellion (1850-1864) was intended not
so much to protect the Manchu regime as to uphold Confucianism against a distorted interpretation of Christianitythe Society of God Worshippers, or Bai Shangdi Hui (Fairbank and
Goldman, 1992).20 Likewise, the patriotic Boxers of the late nineteenth century who ended up
killing many Western missionaries and Chinese communicants did so out of the intentions to
preserve the Confucian values (Spence, 1990).21 Last but not least, that Confucian norms had
firmly established itself in the Chinese society was also borne out in the New Cultural
Movement (1915-1921)the Chinese Renaissance that occurred shortly after the collapse
of the Qing dynasty, wherein the Confucian culture was thoroughly condemned as being the
key contributing factor to Chinas failure to develop the economy and democracy (Schwartz,
1983).
The crux, for our purpose, is whether the diffusion of Confucian ethics played a role in
reducing social conflicts or specifically peasant rebellions. Anecdotal evidence suggests it did.
For example, the prevailing popular expression be a peaceful dog [rather than] a rebel
clearly opined that one should strive to become a courteous gentleman rather than a martial knight. Even the peasant rebel army head Li Zicheng (1606-1645) and many other social
bandits,22 who were forced (by adverse circumstances) into such roles, maintained a reverence for Confucius, to the extent that the Confucius family mansion and temples in the
hometown of Confucius were effectively spared from their plundering (Zeng, 1876). We thus

20

With the most profound demographic consequences in Chinese history (the number of casualties
was estimated to range between 50 (Perkins, 1969) and 73 million (Cao, 2000), the Taiping Rebellion
may be regarded as a holy war waged on Confucianism. It arose out of a disdainful denunciation of
Confucianism and the Qing rule, and was suppressed by two army generals who were Confucian
scholarsZeng Guofan was a jinshi and Zuo Zongtang was a juren (Fairbank and Goldman, 1992).
21
China was forced to open up to the West after its defeat in the First Opium War (1839-1842) and the
signing of a series of so-called unequal treaties, particularly the Tianjin Treaty of 1860, which resulted in a more thorough penetration by the foreigners. A group of spontaneously formed, patriotic antiimperialists in 1900 known as the Boxers attacked the foreign missionaries and communicants for
the alleged superiority the latter imposed upon the Chinese, which essentially was underpinned by a
clash of values between Confucianism and Christianity (Esherick, 1987).
22
Li Zicheng was a famous peasant rebel leader who overthrew the Ming dynasty.
9

expect that the diffusion of Confucian values and norms in China to have played a significantly positive role in reducing peasant rebellions.

3. The Sample
To examine the effect of Confucianism on peasant rebellions, we construct a panel dataset that covers the 107 counties of Shandong Province throughout the 260 years of Chinas
Qing dynasty (1644-1911). There are good reasons for choosing Shandong on the North China Plain (Figure 2) for examining the possible effect of Confucianism on peasant rebellions.
Foremost is that, being the birthplace of Confucius (Qufu County to be precise) and many
other great Confucian sages, Shandong Province was the origin of the orthodox Confucian
culture.23 Its sacrosanct importance was reinforced in late imperial (Qing) times as some emperors, most notably Kangxi and Qianlong, chose to perform sacrificial ceremonies and built
more temples there, and granted feudal nobility to descendants of the great Confucian sages
who continued to live there (Elliott, 2009). While the county of Qufu and its neighboring regions allow us to examine the effect of Confucianism on peasant rebellions, there were regions within the province (e.g., the northern and eastern parts) that were distinctly weak in
the Confucian tradition. This sharp contrast within Shandong Province thus provides us
with substantial variations across the counties (more on this below) for examining the effect
of Confucianism on peasant rebellions.
[Figure 2 about here]
Second, historically the western plain of Shandong Province was frequently struck by
natural calamities. While droughts were a common occurrence, the Yellow Riverthe second largest river in China that runs from western Shandong to the seafrequently
breached its levees to flood the western plain (Figure 3a). By contrast, the eastern part of
Shandong (the peninsula) enjoyed a reliably plentiful supply of rainfall and was far away
from the flooding areas of the Yellow River. By the mid-to-late nineteenth century the Yellow
River had shifted its course repeatedly (1852-1887) and resulted in series of floods, leading to
widespread crop failures. Even more unfortunate was that such mishaps were followed on
close heels by the great drought of 1896-1897, resulting in a large-scale famine in Shandong
and other parts of North China.

23

Both Confucius and the majority of his disciples came from Qufu County or its nearby regions. His
disciples, many of whom subsequently became great Confucian sages or scholars themselves,
continued to spread Confucius doctrines back in their hometowns or other places near Qufu (Esherick,
1987; Legge, 1991). That was why Qufu was promoted as the holy land of Confucianism by various
imperial authorities from the Han dynasty (202 B.C.-220 A.D.) onwards (Forsyth, 1912).
10

[Figure 3 about here]

Accordingly, certain pockets in west Shandong were also home of the social bandits and
popular unrests in historical China (Esherick, 1987). One of the four classic Chinese novels,
the Water Margin (also known as Outlaws of the Marsh), which tells a tale of how a group of
108 outlaws gathered at Mount Liang (or Liangshan Marsh) to form a sizable army to resist
the government, was set in southwest Shandong in the early twelfth century (the Song dynasty). By the late nineteenth and early twentieth centuries, Western missionaries saw southwestern Shandong as the place where the classic Eldorado of bandits were gathered (Stenz,
1987). Figure 2, which is constructed based on statistics compiled by C. K. Yang (1975), clearly shows that Shandong was indeed among the provinces with the most frequent outbreak of
peasant rebellions during the Qing dynastywith over 20 incidents per 10,000 km2 between
1796 and 1911.
It is also worth mentioning that, with a land area approximately 20 percent larger than
that of England, and with more than a hundred counties (107) with sufficient variations in
culture, weather conditions, and peasant rebellions, Shandong Province is uniquely suitable
for conducting a study of this nature. Not to mention that, as a single-province study this
study avoids the unobserved heterogeneities that typically afflict other studies involving
cross-country comparisons.
We choose the Qing dynasty (1644-1911) to be our sample period because, as we have
pointed out in subsection 2.3, Confucianism had by then become deeply ingrained in the
Chinese society after several millennia of promotion by many emperors. We begin our analysis from 1651 onwards to exclude those social unrests carried over from the Ming dynasty.
Similarly, we end our analysis in 1910 just to avoid the complicated political circumstances
that clouded the final year of the Qing dynasty (1911).

4. The Effect of Economic Shocks on Peasant Rebellions

Before examining the effect of Confucian norms on peasant rebellions, we first examine
whether economic shocks were indeed a primary source of peasant rebellions.
4.1. Variables
Peasant Rebellions. The dependent variable, peasant rebellions, is measured by the number
of uprisings that reportedly occurred in a county on a yearly basis, on which detailed records
on the place and time of each peasant rebellion were meticulously compiled by the Qing

11

court (Veritable Records of the Qing Emperors).24 According to Chinese historians, the Veritable
Records of the Qing Emperors are the most complete and systematic source of original information on social unrests that occurred during the Qing dynasty. Indeed, C. K. Yang (1975)
checked the accuracy of the entries on peasant rebellions by comparing those listed in the
Records against those enumerated in 40 county gazetteers of six provinces, and found that the
Records did adequately cover the details of virtually all the reported social unrests (specifically the location and time of occurrence).
Economic Shocks. Being a rain-fed agrarian economy, China was susceptible to economic
shocks, which oftentimes resulted in crop failures. To measure such shocks, we employ a
dummy variable to indicate whether a county had experienced crop failures in the previous
year. Our data source, namely the Veritable Records of the Qing Emperors, also contained detailed reports on counties suffering from crop failures that resulted from natural disasters on
an annual basis.25
To reflect more fully the magnitude of economic shocks, we construct an alternative variable by multiplying the number of crop failures in a given year by the varied levels of extreme rainfall (ranging from drought to waterlog) in the previous year (hereafter lagged extreme rainfall). The rainfall data are obtained from the Atlas on the Spatial Distribution of
Drought and Waterlog in China for the Recent Five Hundred Years, complied by the China Meteorological Administration (1981). The rainfall level is differentiated on a scale of 0 to 2, with 0
indicating normal weather conditions, 1 indicating slight drought or waterlogging, and 2 indicating severe drought or waterlogging.26 By multiplying this measure with the crop failure
dummy we have a continuous variable ranging from 0 to 2.
Figure 4, which maps the decadal incidence of both peasant rebellions and crop failures
in Shandong Province from 1651 to 1910, shows the close association between the two over
time. Where there were few crop failures, as was the case before 1800, there were correspondingly few rebellions. But as the number of crop failures climbed rapidly after 1800 and
peaked in 1850-1870s, the number of peasant rebellions surged in this period as well.
[Figure 4 about here]

24

Unfortunately, the Records contained no information on the size and causes of the rebellionsthey
were all uniformly referred to as fei luan (bandit unrests).
25
The natural disasters enumerated in the Veritable Records of the Qing Emperors included droughts,
waterlogged land, floods, locust plagues, and earthquakes.
26
The original values in the Atlas are given a range of 1 to 5, with 1 standing for severe drought, 2 for
slight drought, 3 for normal weather conditions, 4 for slight waterlogging, and 5 for severe waterlogging. To combine the two measures together, we convert 3 (normal weather conditions) to 0, 2 and 4 to
1 (slight drought or waterlogging), and 1 and 5 to 2 (severe drought or waterlogging).
12

Cost of Rebellion. Rebellion is not without cost. A direct cost of peasant rebellion was suppression by the imperial authorities. To proxy for this cost we employ not only the number
of imperial soldiers stationed in a county but also those in its (contiguous) neighboring counties in each year as the pertinent proxy, for the reason that the latter could be easily deployed
to help suppress an uprising.
Population Pressure. In agrarian societies, the occurrence of rebellions was usually a consequence of cumulative population pressure (Malthus, 1798; Bruckner, 2010). This was arguably the case in historical China, where the population grew rapidly in the Ming and Qing
dynasties (Cao, 2000; Chen and Kung, 2012). We therefore control for population pressure
using population density as the pertinent measure. Since population data are available at only the prefecture level (an administrative unit between the province and the county) for 1776,
1820, 1850, and 1910, we employ a linear interpolation of population density for the missing
years as an approximation, premised on the reasoning that demographic variables tend to
change slowly over time (see Jia, 2011).
Famine Relief. We expect that government relief measures could effectively reduce peasant
rebellions in times of economic hardship. In fact, in late imperial China the Qing administration, together with the local gentry, had striven to use granaries as a means to provide temporary relief to victims of natural calamities (Shiue, 2004). As we have no information on the
actual amount of grain stored in each county, we employ the capacity of the granaries in a
county, taken from the Provincial Gazetteer of Shandong (1890s), to proxy for the possible effect
of famine relief.
Economic Prosperity. Economic prosperity is an omitted variable that may be correlated
with economic shocks as well as peasant rebellions. To proxy for this effect, we employ
dummy variables to indicate whether a county was a designated a regional commercial center in Shandong Province during the early (1644-1730), mid (1731-1820), and late (1821-1911)
Qing periods (Provincial Gazetteer of Shandong, 1890s). For robustness we also control for an
additional set of covariates, which we will introduce later. Details on the sources of all the
variables and their descriptive statistics are provided in Appendix Table A1.
4.2. Results
We examine the effect of economic shocks on peasant rebellions using the following OLS
specification:

y it 1 shock it 1 1 X ' it county i yeart it

(1)

where yit denotes the number of peasant rebellions in county i in year t, shock it 1 denotes the
variations in economic shocks as measured by the crop failure dummy in one set of estima-

13

tions and extreme rainfall in another, X 'it denotes a vector of control variables including the
cost of rebellion, population density, and so forth. countyi refers to county fixed-effects that
capture the time-invariant and county-specific factors, whereas yeart indicates year fixedeffects, which capture the common trend faced by all the counties.27 it is an error term that
captures all other omitted factors.
We report the results in Table 1. Economic shocks are proxied by the crop failure dummy
in columns (1) and (2), and by extreme rainfall in columns (3) and (4). To provide a baseline,
we do not include the control variables initially. In all columns the county and year fixedeffects are fully controlled for. The result shows that economic shocks have a significant and
positive effect on peasant rebellions across all the estimations. In terms of magnitude, the coefficients in columns (1) and (2) indicate that the number of peasant rebellions in counties
with crop failures is about 0.01 times higher than in counties without crop failures. Evaluated
at the mean, this difference accounts for one-third of the total number of peasant rebellions
occurred (0.03). Consistently, the results in columns (3) and (4), when evaluated at the mean,
indicate that a doubling of the level of rainfall (for example, from normal conditions to slight
waterlogging) will increase the number of peasant rebellions by 0.00378 times
[(0.89+1)0.002] or 12.6 percent (0.00378/0.03). These results, in fact, are consistent with the
findings that weather shocks were a primary source of social unrests both in historical China
and in other parts of the world (Miguel et al., 2004; Yancheva et al., 2007; Zhang et al., 2007;
Bai and Kung, 2011; Jia, 2011).28
[Table 1 about here]

5. The Effect of Confucian Norms on Peasant Rebellions

5.1. Measures of Confucian Norms
Now we turn to examine whether Confucian norms are able to mitigate the effect of crop
failures on peasant rebellions. Our main challenge is constructing a valid measure to proxy
for the strength of Confucianism. Unlike other religions (such as Christianity), whose
strength can be measured by, for instance, the share of communicants in a population, Con-

27

A notable example of this common trend is changing state capacity over time, which historically was
stronger at the beginning (and in the middle) of a dynasty but declined toward the end (Goldstone,
1991; Sng, 2011).
28
It is possible that peasant rebellions may reflect the tenacious persistence of poverty accumulated
from shocks in the past. To capture this effect, we ran the regressions again by aggregating the number of crop failures that occurred in the past ten years, with values ranging from 0 to 10 to indicate the
degree of severity. The results show that crop failures accumulated over the past decade also
significantly impacted peasant rebellions (in the current year).
14

fucianism resembles more of a set of social norms, and therefore lacks a clear-cut standard
with which to identify the true believers. After careful deliberations, we came up with what
we think are reasonable proxies for the strength of Confucian norms.
Before introducing these proxies, we would like to point out that we are treating the
strength of Confucian norms as time-invariant, based on the reasoning that cultures are likely to remain stable over a long period of time (e.g., Boyd and Richerson, 1985; Putnam, 1993;
Tabellini, 2010; Nunn and Wantchekon, 2011). This is arguably the case for Confucianism,
which had become firmly established as the orthodox ideology in China since the thirteenth
century (Fairbank and Reischauer, 1989).
Our first measure of Confucian norms is the number of schools that a county had established, as Confucian classics made up the core of the curriculum. In this light, the strength of
the Confucian norms was likely correlated positively with the number of schools. The Provincial Gazetteer of Shandong (1890s) contains detailed information on the number of schools
(and their composition) in each county.29
The number of Confucian temples is our second proxy. Our choice is premised on the
reasoning that stronger Confucian norms would likely give rise to more Confucian temples
being built, which would in turn deepen these norms (Ho, 1962). According to the Qing
court, Confucian temples were differentiated based on whether they were specifically erected for honoring Confucius (wen miao), for honoring the notable officials and local virtuous
Confucians (ming huan xiang xian ci), or for honoring the local loyal and filial men (zhong xiao
jie yi ci) (Yao, 2000). As with the other important social and economic affairs of the time the
Qing government kept systematic records of these temples in local gazetteers (Provincial Gazetteer of Shandong, 1890s).30
Our third measure is the number of chaste women officially elected by the Qing court.
Due to data limitations, we can only cover the period 1644-1890about two decades before
the Qing dynasty came to an end. We consider chaste women a valid proxy for Confucian
norms because, as discussed in subsection 2.3, the deeds of the chaste women were considered an ultimate expression of the Confucian virtues pertaining to purity, loyalty, and subordination, and thus represented a code of conduct that the Qing government was eager to
promote. Moreover, unlike schools and temples, it is less likely for chaste women to be corre-

29

There were four kinds of schools at the county level. The official county school, xian xue, was financed and administered by the county government, whereas the academies, shu yuan, the
community schools, she xue, and private tutors, si shu, were run by the local literati and gentry.
30
One may be concerned that the numbers of schools and temples are based on the Provincial Gazetteer
of Shandong (1890s), and hence they may not adequately reflect the situation before the 1890s. This
should not be a serious problem, however, as more than 63 percent of the schools and 66 percent of
the Confucian temples had already been built before the Qing dynasty, with the remainder constructed during the 1660s-1790s (when the dynasty was in its heyday).
15

lated with economic prosperity.31 Officially conferred by the court, the honor of chastity was
also recorded in local gazetteers (Provincial Gazetteer of Shandong, 1890s).
To account for the possible effect of county size on our independent variables, we normalize them primarily by land area (per 10,000 km2) and, where possible, also by county
population.32 To reduce skewness in our three measures, we express them in terms of natural
logarithm.33
Figures 3b-d, which respectively show the numbers of schools, Confucian temples, and
chaste women (per 10,000 km2) at the county level in Shandong Province, indeed reveal a
significant and positive correlation among the three measures. Moreover, the geographic distribution of each of these measures apparently supports the idea that Confucian norms were
distinctly stronger in the holy land of Confucianism. For instance, the figures show that
there were indeed more schools, temples, and chaste women in Qufu and its surrounding
counties, as well as in the northwestern part of the province than elsewhere in the province.
Finally, in light of the strong correlation among Confucian temples, chaste women, and
schools, we construct a single index of Confucianism by taking their first principal components so as to capture the overall strength of Confucian norms in a county. In estimating the
effect of Confucian norms, we use the separate proxies individually and cross check their
robustness with the single, aggregated index.
5.2. Main Results
To examine whether peasant rebellions can be mitigated by Confucian norms, we employ
the specification in Equation (2):

y it 2 shock it 1 3Confucianismi shock it 1 2 X ' it county i yeart it

(2).

Confucianismi denotes the time-invariant strength of the Confucian norms as measured

by schools, Confucian temples, chaste women, and their aggregated index. The interaction
term Confucianismi shock it 1 is our key variable of interest, included to estimate the possible effect of Confucian norms. Specifically, we expect the effect of economic shocks to be

31

Indeed, Western historians condemn Confucian chastity as being completely lost (of) any
rationality and out of touch with reality (Theiss, 2004, p.1).
32
One shortcoming of the population data is that they are available only for the year of 1890 but not
beforehand. We thus rely more on the normalization using a countys land area.
33
As some counties had no schools (zero value), we use ln (1+number of schools) as the pertinent
measure.
16

smaller in counties with a stronger diffusion of Confucian norms. Should that be the case, the
sign of the coefficient 3 should be significantly negative.34
In Table 2, we examine whether Confucian norms are able to mitigate the effect of crop
failures on peasant rebellions using the various measures discussed above. In columns (1)-(4)
Confucian norms are normalized by a countys land size, and by county population in column (5). Throughout all estimations we control for the cost of rebellion, population density,
famine relief, commercial center, and county and year fixed-effects. The results show that,
while the effect of crop failures is significant and positive, interacting it with Confucian
norms results in a significant and negative impact on peasant rebellions across estimations.
The magnitudes of the effect of Confucian norms on peasant rebellions are far from trivial.
Our calculations show that, an additional school (per 10,000 km2) has the effect of reducing
the number of peasant rebellions triggered by crop failures by 0.0027 times or 27.3 percent,
an additional Confucian temple by 0.0004 times or 4.4 percent, and an additional chaste
woman by 0.000015 times or 0.2 percent.35
Table 3 is essentially the same as Table 2, except economic shocks are now measured by
extreme rainfall. Similar to our results in Table 2, extreme rainfall has a significantly positive
effect on peasant rebellions, but become considerably smaller where Confucian norms were
stronger.
[Tables 2 and 3 about here]

6. Robustness
6.1. Additional Controls
In examining the relationship between culture and conflict, a formidable task is to disentangle the effects of culture from those of other variables, most notably institutional environment and its attendant socioeconomic conditions (Fisman and Miguel, 2007; Miguel,
Saiegh and Satyanath, 2008). This is especially the case in the Chinese context, since Confucianism was intimately associated with institutions pertaining to governance, the imperial
examination system, and so forth. Against this background, one might wonder if the effect of

34

We also estimate the main effect of Confucianism on peasant rebellions without controlling for
county fixed-effects in Table A2. The OLS results show that Confucian norms are significantly
negative.
35
Using chaste women as an example, 0.000015 (times) is the product of the pertinent coefficient (0.016)
and the marginal effect of chaste women, obtained by adding 1 to the mean of chaste women,
ln[(1+1051.527)/1051.527]. We then divide the number of times (0.00015) by the coefficient of crop
failures (0.008), and then multiply with 100 to obtain the answer in percentage terms 0.2.

17

Confucian norms (in Table 2) came instead from other omitted factors correlated with both
Confucian norms and peasant rebellions. To address this concern, we add a vector of controls. Based on the pertinent literature and the specific historical context of China our choice
of additional covariates includes the following:
Quality of Governance. It is widely assumed that the quality of governance (or state capacity) plays an important role in determining conflict (Fearon and Laitin, 2003; Abadie, 2006;
Besley and Persson, 2011). In particular, to the extent that local authorities were required to
consciously promote Confucianism in imperial China as state orthodoxy, spatial variations in
the strength of Confucian norms were likely to be correlated positively with the quality of
local governance. In late imperial China, it was indeed the situation that county officialsin
conjunction with the gentry classwere made to administer a whole host of affairs ranging
from public goods provision, disputes resolution, to tax collection (Chang, 1955). In this context, the size of the local gentry, which is also available in the Provincial Gazetteer of Shandong
(1890s), is a reasonable proxy for the overall quality of local governance. In our regressions
we will employ the annual size of the local gentry (per 10,000 km2) at the county level (in
natural logarithm) to be the pertinent measure.
Social Mobility. Confucianism may also affect peasant rebellions via the channel of social
mobility, to the extent that the latter was intimately tied to the imperial examination, and
whose syllabus was drawn heavily on the memorization of the Confucian classics and rote
learning (Ho, 1962; Elman, 2000). It is thus necessary to control for social mobility using the
annual change in the number of provincial degree holders (juren) (per 10,000 km2, in natural
logarithm) in each county as proxy.36 To the extent that the juren qualifications facilitated entry into bureaucracy, we consider counties with more juren to enjoy greater social mobility.
In addition, we are concerned that some characteristics specific to a county may be correlated with both Confucian norms and peasant rebellions. We come up with five such possibilities (detailed below), which we endeavor to further control by interacting them with economic shocks in the regressions.
Taoism and Buddhism. Like Confucianism, Taoism and Buddhism similarly advocate harmony and condemn violence (Giles, 1989; Gethin, 1998). Also like Confucianism, the two had
a similarly long history and enjoyed immense popularity among the Chinese. This raises the
concern that omission of these two religions may be correlated with both Confucianism and
peasant rebellions. To deal with this we employ the numbers of Taoist temples and Buddhist
temples (natural logarithm per 10,000 km2) that existed in the Qing dynasty to proxy for the

36

The provincial degree holders were those who had passed the province-level imperial examination.
Given that the provincial examination was held once every three years, we fill out the missing years
using the number of provincial degree holders in the most recent examination. For example, if a
provincial examination was held in 1700, it would not be held again until 1703; we thus take the
number of degree holders in 1701 and 1702 to be the same as that in 1700.
18

respective strengths of Taoism and Buddhism in a county. The pertinent data are obtained
from the Provincial Gazetteer of Shandong (1890s).
Altitude. Geographic factors in general and terrain in particular have long been considered a potential co-determinant of conflict (Fearon and Laitin, 2003). In our historical context,
the rugged mountainous terrain in southern Shandong Province arguably provided an ideal
nest for the bandits and outlaws (Esherick, 1987) (Figure 2a). But rugged terrains are typically ill-served by the transport infrastructure and thus would impair the diffusion of Confucianism. To control for this effect we employ the average altitude of a county to proxy for the
effect of rugged geographic terrain on peasant uprisings.
Grand Canal. Counties strong in Confucian norms were also likely located along trade
routes and benefited accordingly from more commercial activities. A more thriving economic environment is likely to reduce social unrests. This is, for instance, the case for western
Shandong, where many of its counties benefited from their proximity to the Grand Canala
vital transport route linking the capital Beijing in North China and the prosperous lower
Yangzi Delta in the south (Figure 2a). To disentangle the effect of Confucianism from the
possible effect of the Grand Canal, we employ a dummy variable indicating whether a county is located along the Grand Canal.
Political Center. In China, the prefecture is an administrative (political) unit between a
province and a county. Typically, the Confucian culture was likely stronger in counties
where a prefectural government was seated. Indeed there were fewer incidences of social
unrests in these counties (where government centers were located). To disentangle these potentially confounding effects, we employ a dummy variable to indicate whether a county
was also the seat of a prefectural government (Figure 2a).
Western Influence. The other factor that may bear upon both Confucianism and peasant
rebellions was the increasing influence of the West after Chinas defeat by the Western powers in the First Opium War (1839-1842). Prominent among these changes were the rapid diffusion of Christianity (in particular Protestantism) and the eventual abolition of Chinas
longstanding imperial examination system,37 both of which had the likely effect of undermining the strength of Confucian norms. At the same time, this penetration of Western economic
and ideological forces had also bred popular social unrests (Wakeman, 1975; Spence, 1990);
the Taiping Rebellion and the Boxers Uprising alluded to earlier are but two salient examples
of the consequences of these influences.
If Western influence indeed had an effect on both Confucianism and peasant rebellions,
its omission would bias our estimates. While the year fixed-effects help to capture the com37

Although Christianity had long existed in China, its influence was significantly smaller before the
nineteenth century, thanks to a ban imposed by Emperor Kangxi in 1721 after a dispute occurred
between the Qing government and the Roman Catholic Church over whether it was acceptable to the
latter for the Chinese to continue worshipping their ancestors and other deities after converting to
Christianity (e.g., Hsu, 2000).
19

mon trend in social changes in the late Qing period for all the counties, we cannot control for
the differential impact of these changes across counties using these year fixed-effects. To isolate the potentially confounding effects of Western influence, we exclude the period when
Shandong Province was duly influenced by this penetration, which began after 1861 following the opening of Che-foo (Yantai) as a treaty port, simply by dropping the period after
1860.38
We report the regression results in Table 4, which includes all the additional controls
mentioned above. In column (1), we measure economic shocks using the crop failure dummy,
and in column (2) extreme rainfall. In column (3), we exclude the years after 1860 to control
for the possible effect of Western influence. To save space we report only the estimates based
on the indexed measure of Confucian norms (the results of the individual measures are similar). In addition to reporting the standard errors (in parentheses), we also report the robust
standard errors clustered at the county level (in square brackets) in order to control for any
intra-county correlations.39
The results of Table 4 continue to show that counties with stronger Confucian norms
have significantly fewer peasant rebellions in times of economic shocks. Even more encouraging is the result that the coefficient of the interaction term between Confucianism and economic shocks is nearly identical to those in Tables 2 and 3, suggesting that the effect of Confucian norms on peasant rebellions is unlikely to have come through the channels of the controlled variables.
[Table 4 about here]
6.2. Instrumented Evidence
Although we have already controlled for a number of covariates, there are likely still
omissions that may be simultaneously associated with both Confucianism and peasant rebellions. Naquin (1981), for example, observes that historically heterodox (evil) sects (such as
Popular Taoism and White Lotus) whose doctrines were incompatible with those of Confu-

38

We have also tried using 1890 as the cutoff year, based on the historical fact that Christianity and
capitalism had penetrated the economy and society of Shandong more deeply after 1890. The effect of
Confucian norms on peasant rebellions using this alternative sample period remains unchanged and
hence we do not report the results separately.
39
In addition, given that the clustered standard error may confront the problem of few clusters (107
counties) relative to large observations (of over 26,000 in most cases), we also employ the standard
error clustered at the county-period level, in which the period is coded by either decade or the reign of
emperors. To address the concern of possible spatial correlation, we employ the clustered standard
errors at the prefecture-period level. We obtain broadly similar results from these additional tests,
except the significant level of the coefficients has become smaller. These results are available upon
request.
20

cianism were potential sources of popular unrests. Likewise, corruption on the part of some
officials also tended to violate the Confucian norms that the imperial authorities attempted
to promote, and which may trigger rebellions.
We are more concerned, however, with the possible measurement error associated with
our proxies for Confucian norms, which, if they were really severe, could seriously bias our
estimates. One particular concern is this: the enumeration of only the numbers of schools,
Confucian temples, and chaste women in the Qing period may not be sufficient to reflect the
strength of Confucianism, given its long existence (of over 2,000 years). Measurement error
may also arise from the crudeness of our three proxies. For instance, we only know how
many schools and Confucian temples there were but not their sizes, let alone the number of
students in each school and pilgrims in each temple, the availability of which would allow us
to better capture the strength of Confucian norms. And while the chaste women measure
does not suffer from the same limitation, the fact that local governments were responsible for
nominating them to the court (who reserved the right of officially conferring the honorary
title on them) implies that the process could be subject to manipulationsdepending on the
(varied) ulterior motives of the local governments (Chow, 1994).
To deal with these concerns we thus employ an instrumental variable (IV) approach, by
exploiting the variation in the strength of Confucianism across the counties before the Qing
dynastyspecifically between 500 B.C. (when Confucius started his teaching) and 1643 (the
end of the Ming dynasty)as instrument of the strength of Confucianism in the Qing dynasty. The validity of this instrument relies on the premise that cultural norms evolve extremely
slowly. In the pertinent Chinese context, that means that the differential spatial strengths of
Confucian norms developed over 2000 years prior to the Qing dynasty are likely to correlate
significantly with those in the Qing dynasty, but not with peasant rebellions in the same period.
We proxy for the strength of Confucian norms in a county before the Qing dynasty using
the number of Confucian sages born in a county before 1644approximately between 500
B.C. and 1643. By Confucian sages we include, in addition to Confucius and Mencius, also
the famous Confucian scholars and notable dignitaries who, by virtue of their recognizably
virtuous deeds were conferred the official title of sages. This choice is premised upon the
reasoning that stronger Confucian norms would produce more Confucian sages, and that
more Confucian sages would in turn help to strengthen these norms. As we have done with
other similar variables, we normalize the number of Confucian sages by the land area of a
county (per 10,000 km2).40 Enumeration of the Confucian sages is made possible by the detailed biographical records contained in the Provincial Gazetteer of Shandong (1890s). As we
have estimated the interacting effect between Confucianism and economic shocks on peasant

40

To the extent that geographic proximity facilitates the spread of culture, we include in our
enumeration of Confucian sages also those born in the neighboring counties.
21

rebellions, we similarly interact Confucian sages with economic shocks as the pertinent instrumental variable.
We report, in columns (1) and (2) of Table 5, the correlation between Confucian sages and
the strength of Confucian norms. To save space we show only the results using the indexed
measure of Confucianism, which demonstrate that Confucian sages are positively and significantly correlated with Confucianism.
To ensure that Confucian sages have a significantly positive effect on Confucianism but
are orthogonal to peasant rebellions, we employ a reduced-form estimation in column (3)
and an over-identification test in column (4) of Table 5. The result in column (3) shows that
Confucian sages can mitigate the effect of crop failures on peasant rebellions. Once we include the interaction term between crop failures and Confucianism in the regression, however, the instrumental variable (i.e., the interaction term between crop failures and Confucian
sages) becomes insignificant (column (4)), reinforcing our reasoning that Confucian sages are
a sound instrumental variable.
[Table 5 about here]
The instrumental variable results are reported in Table 6. Columns (1) and (2) report the
results using crop failures as the proxy for economic shocks, whereas columns (3) and (4) report the results of using extreme rainfall as the pertinent proxy. As before, we first exclude
all the control variables before including them in the estimations. In all cases we control for
county and year fixed-effects, and use both standard errors and robust standard errors clustered at the county level.
Three important findings are observed from Table 6. First, the instrumented evidence
importantly confirms that the effect of economic shocks on peasant rebellions is significantly
reduced by Confucianism. The second important finding is that the coefficients of the instrumented estimations are nearly identical to those in Table 4 (for instance, the coefficient of
Extreme rainfall Confucianism is -0.003 in both OLS and IV-2SLS estimations). What is even
more encouraging, finally, is the finding that the significance of the interaction term (of Confucianism and economic shocks) does not change even after controlling for the additional
covariates. All of these reinforce the validity of our OLS findings.
[Table 6 about here]

7. Conclusion
The research question raised in this studynamely whether conflicts triggered by economic shocks can be mediated by strong cultural normswas motivated by a literature that

22

has consistently demonstrated the existence of a significant relationship between economic

shocks and social conflicts. Using historical China as the pertinent context, we constructed a
unique dataset to ascertain, first of all, whether there was indeed a significant relationship
between economic shocks and peasant rebellions, before we turned to examine, more importantly, whether the Confucian norms which emphasize conflict avoidance were effective
in alleviating this particular form of social conflict. Regardless of how economic shocks (be it
crop failures or variations in rainfall) and Confucian norms (education, religion, and folk
customs) were measured, and the extent to which we controlled for other covariates and endogeneity, we came up with the same robust results that economic shocks did impact significantly peasant rebellions, but what is of greater interest to us is the result that strong Confucian ethos had the consistent effect of reducing such conflicts. While in this study we use historical China and specifically the incidence of Confucianism and peasant rebellions as an example, the findings can easily be generalized to other contexts.

23

References
Abadie, Alberto. 2006. Poverty, Political Freedom, and the Roots of Terrorism. AEA Papers
and Proceedings, 96 (May): 50-6.
Bai, Ying, and James Kai-sing Kung. 2011. Climate Shocks and Sino-nomadic Conflict. Review of Economics and Statistics, 93(3): 970-81.
Beber, Bernd, and Christopher Blattman. 2011. The Logic of Child Soldiering and Coercion.
Working paper.
Besley, Timothy, and Torsten Persson. 2011. The Logic of Political Violence. Quarterly Journal of Economics, 126 (3):1411-45.
Blattman, Christopher, and Edward Miguel. 2010. Civil War. Journal of Economic Literature,
48 (March): 3-57.
Blattman, Chris, Alexandra Hartman, and Robert Blair. 2011. Can We Teach Peace and Conflict Resolution? Results from a randomized evaluation of the Community Empowerment Program (CEP) in Liberia: A Program to Build Peace, Human Rights, and Civic Participation. IPA Policy Report.
Boyd, Robert, and Peter J. Richerson. 1985. Culture and the Evolutionary Process. Chicago: University of Chicago Press.
Bruckner, Markus. 2010. Population Size and Civil Conflict Risk: Is there a Causal Link?
The Economic Journal, 120 (544): 535-50.
Bruckner, Markus, and Antonio Ciccone. 2010. International Commodity Prices, Growth,
and Civil War in Sub-Saharan Africa. The Economic Journal , 544 (5): 519-34.
Cao, Shuji. 2000. Zhongguo renkou shi: Qing shiqi (Population History of China: Qing Dynasty).
Shanghai: Fudan daxue chuban she.
Carus, Paul. 1918. Ceremony Celebrated under the Chinese Republic in Honor of Confucius. Open Court, A Quarterly Magazine, 32 (3): 155-72.
Chang, Chung-li. 1955. The Chinese Gentry: Studies on Their Role in Nineteenth-Century Chinese
Society. Seattle: University of Washington Press.
Chen, Menglei. 1985 [1726]. Gu jin tu shu ji cheng (Great Collection of Ancient and Modern Books).
Beijing: Zhong hua shu ju.
Chen, Shuo, and James Kai-sing Kung. 2012. A Malthusian Quagmire? Maize and Population Growth in China, 15501910. Working paper.
Chesneaux, J. (1973). Peasant Revolts in China, 1840-1949, Volume 2. Thames & Hudson Ltd:
London.
CHGIS, Version 4. Cambridge: Harvard Yenching Institute, January 2007.
China Meteorological Administration. 1981. Atlas on the Spatial Distribution of Drought and
Waterlog in China for the Recent Five Hundred Years (Zhong guo jin wu bai nian han lao fen bu
tu ji). Beijing: Di tu chu ban she.
Chow, Kai-wing. 1994. The Rise of Confucian Ritualism in Late Imperial China: Ethics, Classics,
and Lineage Discourse. Stanford, Calif.: Stanford University Press.

24

Collier, Paul and Anke Hoeffler. 1998. On Economic Causes of Civil War. Oxford Economic
Papers, (50).
Collier, Paul and Anke Hoeffler. 2004. Greed and Grievance in Civil War. Oxford Economic
Papers, 56 (4):563-595.
Confucius (translated by Chichung Huang). 1993. The Analects of Confucius (Lun Yu). New
York: Oxford University Press.
Elliott, Mark C. 2009. Emperor Qianlong: Son of Heaven, Man of the World. New York; Hong
Kong: Longman.
Elman, Benjamin A. 2000. A Cultural History of Civil Examinations in Late Imperial China. Berkeley, CA: University of California Press.
Esherick, Joseph W. 1987. The Origins of the Boxer Uprising. Berkeley: University of California
Press.
Fairbank, John K., and Merle Goldman. 1992. China: A New History. Cambridge, Mass.: Belknap Press of Harvard University Press.
Fairbank, John K., and Edwin O. Reischauer. 1989. China: Tradition and Transformation. Boston:
Houghton Mifflin Co.
Fearon, James, and David D. Laitin. 2003. Ethnicity, Insurgency, and Civil War. American
Political Science Review, 97 (March): 75-90.
Feuerwerker, A. (1975). Rebellion in Nineteenth-Century China. Center for Chinese Studies,
University of Michigan.
Fisman, Raymond, and Edward Miguel. 2007. Corruption, Norms, and Legal Enforcement:
Evidence from Diplomatic Parking Tickets. Journal of Political Economy, 115 (6): 1020-48.
Forsyth, Robert C. 1912. Shantung: The Sacred Province of China. Shanghai: Christian Literature
Society.
Funk, Patricia. 2004. On the Effective Use of Stigma as a Crime-Deterrent. European Economic Review, 48: 715-28.
Gethin, Rupert. 1998. The Foundations of Buddhism. New York: Oxford University Press.
Goldstone, Jack A. 1991. Revolution and Rebellion in the Early Modern World. Berkeley: University of California Press.
Giles, Herbert A. 1989. Religions of Ancient China. Singapore: Graham Brash (Pte) Ltd.
Ho, Ping-ti. 1962. The Ladder of Success in Imperial China. New York: Columbia University
Press.
Hobsbawm, Eric J. 1972. Bandits. Harmondsworth: Pelican.
Hsiao, Kung-chuan. 1960. Rural China: Imperial Control in the Nineteenth Century. Seattle: University of Washington Press.
Hsu, Immanuel C. Y. 2000. The Rise of Modern China. New York: Oxford University Press.
Jarvis, A., H.I. Reuter, A. Nelson, and E. Guevara. 2008. Hole-filled SRTM for the Globe Version 4, CGIAR-CSI SRTM 90m Database. http://srtm.csi.cgiar.org.
Jia, Ruixue. 2011. Weather Shocks, Sweet Potatoes and Peasant Revolts in Historical China.
Working paper.
Legge, James. 1991. The Chinese Classics. Taipei: SMC Pub.
25

Malthus, T. Robert. 1798. An Essay on the Principle of Population as It Affects the Future Improvement of Society. London: J. Johnson.
MacCulloch Robert, Silvia Pezzini. 2010. The Roles of Freedom, Growth, and Religion in the
Taste for Revolution. Journal of Law and Economics, 53 (2):329-58.
Maddison, Angus. 2007. Chinese Economic Performance in the Long Run. Paris: Development
Centre of the Organisation for Economic Co-operation and Development.
Mann, Susan. 1997. Precious Records: Women in Chinas Long Eighteenth Century. Stanford, Calif.:
Stanford University Press.
Miguel, Edward, Shanker Satyanath, and Ernest Sergenti. 2004. Economic Shocks and Civil
Conflict: An Instrumental Variables Approach. Journal of Political Economy, 112 (August):
725-53.
Miguel, Edward, Sebastin M. Saiegh, and Shanker Satyanath. 2008. National Cultures and
Soccer Violence NBER Working Paper No. 13968.
Montalvo, Jos G., and Marta Reynal-Querol. 2005. Ethnic Polarization, Potential Conflict,
and Civil Wars. American Economic Review, 95 (3): 796816.
Naquin, Susan. 1981. Shantung Rebellion: the Wang Lun Uprising of 1774. New Haven: Yale
University Press.
Nunn, Nathan, and Leonard Wantchekon. The Slave Trade and the Origins of Mistrust in
Africa. American Economic Review, 101 (7): 3221-52.
Nylan, Michael. 2001. The Five Confucian Classics. New Haven: Yale University Press.
UNICEF. 2009. Machel Study 10-Year Strategic Review: Children and Conflict in a Changing World.
UNICEF.
Perkins, H. Dwight. 1969. Agricultural Development in China 13681968. Edinburgh: Edinburgh University Press.
Perry, Elizabeth J. 1980. Rebels and Revolutionaries in North China, 1845-1945. Stanford, Calif.: Stanford University Press.
Posner, Eric A. 2000. Law and Social Norms, Cambridge, Mass.; London: Harvard University
Press.
Putnam, Robert D. 1993. Making Democracy Work: Civic Traditions in Modern Italy. Princeton,
NJ: Princeton University Press.
Putnam, Robert D. 2000. Bowling Alone: the Collapse and Revival of American Community. New
York: Simon & Schuster.
Rasmusen, Eric. 1996. Stigma and Self-Fulfilling Expectations of Criminality. Journal of Law
and Economics, 39 (2): 519-43.
Rawski, Evelyn S. 1979. Education and Popular Literacy in Ching China. Ann Arbor: The University of Michigan Press.
Rowe, William T. 2009. Chinas Last Empire: the Great Qing. Cambridge, Mass.: Belknap Press of
Harvard University Press.
Schwartz, Benjamin. 1983. Themes in Intellectual History: May Fourth and After. In John K.
Fairbank. The Cambridge History of China: Republican China 1912-1949. Cambridge: Cambridge University Press.
26

Scott, James C. 1976. The Moral Economy of the Peasant: Rebellion and Subsistence in Southeast
Asia. New Haven: Yale University Press.
Shiue, Carol H. 2004. Local Granaries and Central Government Disaster Relief: Moral Hazard and Intergovernmental Finance in 18th and 19th Century China. Journal of Economic
History, 64 (1): 101-125.
Sng, Tuan-Hwee. 2011. Size and Dynastic Decline: The Principal-Agent Problem in Late
Imperial China 1700-1850. Working paper.
Stenz, George M. 1897. Life of Father Richard Henle, S.V.D., Missionary in China, Assassinated
November 1, trans. Elizabeth Ruff. Techny, Illinois: Mission Press.
Sun, Jing-zhi. 1957. Hua bei jing ji di li (Economic Geography of North China). Beijing: Ku xue
chu ban she.
Tabellini, Guido. 2010. Culture and Institutions: Economic Development in the Regions of
Europe. Journal of the European Economic Association, 8 (4): 677716.
Theiss, Janet M. 2004. Disgraceful Matters: The Politics of Chastity in Eighteenth-Century China.
Berkeley, Calif.: University of California Press.
Tong, James W. 1991. Disorder under Heaven: Collective Violence in the Ming Dynasty. Stanford,
Calif.: Stanford University Press.
Voigtlnder, Nico, and Hans-Joachim Voth, 2012. Persecution Perpetuated: The Medieval
Origins of Anti-Semitic Violence in Nazi Germany. Quarterly Journal of Economics,
forthcoming.
Wakeman, Frederic E. 1975. The Fall of Imperial China. New York: Free Pres.
Wolf, Eric R. 1969. Peasant Wars of the Twentieth Century. London: Faber and Faber.
Xu, Tan. 1998. Ming Qing shi qi Shan dong shang pin jing ji de fa zhan (Development of Commercial Economy in Shandong during Ming-Qing Period). Beijing: Zhong guo she hui ke xue chu
ban she.
Yancheva, Gergana, Norbert R. Nowaczyk, Jens Mingram, Peter Dulski, Georg Schettler, Jrg
F. W. Negendank, Jiaqi Liu, Daniel M. Sigman, Larry C. Peterson, Gerald H. Haug. 2007.
Influence of the Intertropical Convergence Zone on the East Asian Monsoon. Nature,
445 (7123), 74-7.
Yang, C. K. 1961. Religion in Chinese Society: A Study of Contemporary Social Functions of Religion and Some of Their Historical Factors. Berkeley: University of California Press. 1961.
Yang, C. K. 1975. Some Preliminary Statistical Patterns of Mass Actions in NineteenthCentury China. In Conflict and Control in late Imperial China. Edited by Frederic E.
Wakeman Jr. and Carolyn Grant. 174-210.
Yao, Xinzhong. 2000. An Introduction to Confucianism. Cambridge: Cambridge University
Press.
Yuchtman, Noam. 2010. Teaching to the Tests: An Economic Analysis of Traditional and
Modern Education in Late Imperial and Republican China. Working paper.
Zeng, Guofan. 1876. Zeng wenzhen gong quanji (Collection of Zeng Guofan). Chuanzhong shuju,
1876.

27

Zhang, David D., Jane Zhang, Harry F. Lee, and Yuan-qing He. 2007. Climate Change and
War Frequency in Eastern China Over the Last Millennium. Human Ecology, 35 (4): 40314.
Zhang, Xin-yi. 1931. Shandong sheng nong ye gai kuang gu ji bao gao (Report on Agricultural Development in Shandong Province). Tong-ji yue-bao, 3 (1): 20-45.

28

18
16
14
12
10
8
6
4
2
0
Zhou (1046Qin, Han
Wei,Jin (220- Sui,Tang (581256BC)
(221BC-220AD)
589)
907)

Song (9601279)

Yuan (12711368)

Ming (13681644)

Qing (16441725)

Figure 1. Number of Chaste Women per Decade by Dynasty

Note: Normalized by the population of each dynasty (in one million).
Source: Chaste women are based on Gu jin tu shu ji cheng (Great Collection of Ancient
and Modern Books) (1726). Population is enumerated based on Maddison (2007).

29

Figure 2. Number of Peasant Rebellions in Qing China

(per 10,000 km2, 1796-1911)
Source: C. K. Yang (1975).

30

Figure 3b. Number of Schools (per 10,000 km2)

Figure 3a. Shandong Province

Figure 3c. Number of Confucian Temples (per 10,000 km2)

Figure 3d. Number of Chaste Women (per 10,000 km2)

31

800
700
600
500
400
300
200
100

Number of peasant rebellions

Number of crop failures

Figure 4. Numbers of Peasant Rebellions and Crop Failures

in Shandong Province (by decade)
Source: Veritable Records of the Qing Emperors.

32

1900

1890

1880

1870

1860

1850

1840

1830

1820

1810

1800

1790

1780

1770

1760

1750

1740

1730

1720

1710

1700

1690

1680

1670

1660

1650

Table 1. Economic Shocks and Peasant Rebellions

Dependent variable

Number of peasant rebellions

(1)

Crop failures

(2)

0.010
(0.003)***

(4)

0.009
(0.004)***

Extreme rainfall
Cost of rebellion
Population density
Famine relief
Commercial center
R-squared
No. of observations

(3)

No
No
No
No
0.18
27713

Yes
Yes
Yes
Yes
0.18
27713

0.002
(0.001)**

0.002
(0.001)**

No
No
No
No
0.18
27674

Yes
Yes
Yes
Yes
0.18
27674

Notes: County and year fixed-effects are fully controlled for. Standard errors are reported in parentheses. * Significant at 10%; ** significant at 5%; *** significant at 1%.

33

Table 2. The Effect of Confucianism on Peasant Rebellions

(Economic shocks are measured by crop failures)
Dependent variable

Crop failures
Crop failures schools

Number of peasant rebellions

(1)

(2)

(3)

(4)

(5)

0.010
(0.004)***
-0.012
(0.004)***

0.009
(0.004)***

0.008
(0.004)**

0.010
(0.004)***

0.009
(0.003)**

-0.007
(0.002)***
Yes
27713
107

-0.005
(0.002)***
Yes
27694
107

Crop failures Confucian temples

-0.008
(0.003)***

Crop failures chaste women

-0.016
(0.003)***

Crop failures Confucianism

Controls
Number of observations
Number of counties

Yes
27713
107

Yes
27713
107

Yes
27713
107

Notes: Confucianism is the first principal component of schools, Confucian temples, and chaste women, all
of which are normalized by a countys land area (in 10,000 km2) in columns (1)-(4) or county population
(in 10,000) in column (5). Controls include cost of rebellion, population density, famine relief, and commercial center. County and year fixed-effects are fully controlled for. Standard errors are reported in
parentheses. * Significant at 10%; ** significant at 5%; *** significant at 1%.

34

Table 3. The Effect of Confucianism on Peasant Rebellions

(Economic shocks are measured by extreme rainfall)
Dependent variable

Extreme rainfall
Extreme rainfall schools

Number of peasant rebellions

(1)

(2)

(3)

(4)

(5)

0.002
(0.001)**
-0.003
(0.001)***

0.002
(0.001)**

0.002
(0.001)**

0.002
(0.001)**

0.002
(0.001)*

Extreme rainfall Confucian temples

-0.003
(0.001)***

Extreme rainfall chaste women

-0.005
(0.001)***

Extreme rainfall Confucianism

Controls
Number of observations
Number of counties

Yes
27674
107

Yes
27674
107

Yes
27674
107

-0.002
-0.001
(0.001)*** (0.001)**
Yes
Yes
27674
27655
107
107

Notes: Confucianism is the first principal component of schools, Confucian temples, and chaste women, all of
which are normalized by a countys land area (in 10,000 km2) in columns (1)-(4) or county population (in 10,000)
in column (5). Controls include cost of rebellion, population density, famine relief, and commercial center.
County and year fixed-effects are fully controlled for. Standard errors are reported in parentheses. * Significant at
10%; ** significant at 5%; *** significant at 1%.

35

Table 4. The Effect of Confucianism on Peasant Rebellions: Additional Controls

Dependent variable

Number of peasant rebellions

(1)

Crop failures
Crop failures Confucianism

(3)

0.003
(0.001)**
-0.003
(0.001)***
[0.001]**

0.001
(0.001)
-0.002
(0.001)***
[0.001]**

0.012
(0.004)***
-0.009
(0.003)***
[0.005]*

Extreme rainfall
Extreme rainfall Confucianism

Baseline controls
Additional controls
Western influence
R-squared
Number of counties
Number of observations

(2)

Yes
Yes
No
0.18
107
27713

Yes
Yes
No
0.18
107
27674

Yes
Yes
Yes
0.09
107
22217

Notes: Confucianism is the first principal component of schools, Confucian temples,

and chaste women, all of which are normalized by a countys land area. Baseline controls include cost of rebellion, population density, famine relief, and commercial center. Additional controls include quality of governance, social mobility, the interaction
terms between crop failures (or extreme rainfall) on the one hand, and Taoism, Buddhism, altitude, Grand Canal, and prefectural capital, on the other. County and year
fixed-effects are fully controlled for. In column (3) we control for Western influence
by excluding the years after 1860. Standard errors are reported in parentheses, and robust standard errors clustered at the county level are reported in brackets. * Significant
at 10%; ** significant at 5%; *** significant at 1%.

36

Table 5. Testing the Validity of Confucian Sages as the Instrumental Variable

Dependent variable

Confucian sages

Confucianism

Confucianism

Peasant
rebellions

(1)

(2)

(3)

0.024
(0.004)***

-0.0002
(0.0001)**

Crop failures Confucianism

No
No
107
107
0.25

(4)

0.006
(0.002)***

Crop failures Confucian sages

Crop failures
Controls
Number of counties
Number of observations
R-squared

Peasant
rebellions

Yes
Yes
107
107
0.87

Yes
Yes
107
27713
0.14

-0.0001
(-0.0001)
-0.008
(0.004)**
Yes
Yes
107
27713
0.14

Notes: Columns (1) and (2) examine the OLS relationship between Confucian sages (IV) and Confucianism at the county level, in which all time-varying variables are organized at the county level by
taking the average. Column (3) examines the reduced-form relationship between the interaction term
of Crop failures Confucian sages and peasant rebellions. Column (4) tests the exogeneity of the interaction term of Crop failures Confucian sages as the instrumental variable, in which the interaction term of Crop failures Confucian sages and the interaction term of Crop failures Confucianism are both included in regressions. Confucianism is the first principal component of schools, Confucian temples, and chaste women, all of which are normalized by a countys land area. Controls refer to
all the control variables (both the baseline and the additional) as described in the text. In columns (3)
and (4) the year and county fixed-effects are fully controlled for. Standard errors are reported in parentheses. ** Significant at 5%; *** significant at 1%.

37

Table 6. The Effect of Confucianism on Peasant Rebellions: Instrumental Evidence

Dependent variable

Number of peasant rebellions

(2)
(3)

(1)
Crop failures
Crop failures Confucianism

0.011
(0.004)***
-0.006
(0.003)*
[0.004]

0.012
(0.004)***
-0.012
(0.006)**
[0.006]*

Extreme rainfall
Extreme rainfall Confucianism

Controls
Number of observations
Number of counties

No
27713
107

(4)

Yes
27713
107

0.003
(0.001)**
-0.002
(0.001)*
[0.001]*
No
27374
107

0.003
(0.001)**
-0.003
(0.002)*
[0.001]**
Yes
27674
107

Notes: The interaction term of Crop failures (or extreme rainfall) Confucianism is instrumented by the interaction term of Crop failures (or extreme rainfall) Confucian sages. Confucianism is the first principal component of schools, Confucian temples, and chaste women, all
of which are normalized by a countys land area. Controls refer to all the control variables as
described in the text. Standard errors are reported in parentheses, and robust standard errors
clustered at the county level are reported in square brackets. * Significant at 10%; ** significant
at 5%; *** significant at 1%.

38

Appendix
Table A1. Sources and Descriptive Statistics of the Variables
Variable
Peasant rebellions
Schools
Confucian temples
Chaste women
Crop failures
Extreme rainfall
Cost of rebellion
Population density
Commercial center
Famine relief
Quality of governance
Social mobility
Taoism
Buddhism
Prefectural seat

Altitude
Grand Canal
Confucian sages

Source
A
B
B
B
A
C
B
D
B
B
B
B
B
B
B
E, F
B
B

Mean
0.03
1.14
2.55
6.60
0.28
0.89
6.82
0.46
0.14
6.99
3.34
0.38
0.68
0.99
0.12
3.81
0.19
21.99

S.D.

Min.

Max.

0.21
0.63
0.90
0.85
0.45
1.48
1.21
0.46
0.34
4.57
0.86
0.56
0.59
0.98
0.33
0.99
0.39
30.55

0
0
0.19
4.17
0
0
0
-0.84
0
0
1.02
0
0
-1.39
0
1.10
0
0.22

7
3.07
4.91
8.90
1
2
9.60
1.50
1
12.58
5.46
3.28
2.46
2.95
1
6.09
1
241.89

Notes: Confucian temples, chaste women, schools, famine relief (capacity of

granaries), quality of governance (size of gentry), social mobility (number of juren), Taoism, and Buddhism are normalized by a countys land area (in 10,000
km2) and expressed in terms of natural logarithm.
A. The Veritable Records of the Qing Emperors.
B. Provincial Gazetteers of Shandong (1890s).
C. China Meteorological Administration. 1981. Atlas on the Spatial Distribution of Drought and Waterlog in China for the Recent Five Hundred Years. Beijing: ditu chu ban she.
D. Cao, Shuji. 2000. Population History of China: Qing Dynasty. Shanghai:
Fudan daxue chu ban she.
E. Hole-filled SRTM for the Globe, Version 4, CGIAR-CSI SRTM 90m Database. http://srtm.csi.cgiar.org.
F. CHGIS, Version 4, Harvard Yenching Institute, January 2007.

39

Table A2. The Main Effect of Confucianism on Peasant Rebellions

Dependent variable
Confucianism variables
are normalized by:

Schools

Number of peasant rebellions

area
(1)

area
(2)

area
(3)

population
(5)

-0.011
(0.002)***
Yes
27713
107

-0.010
(0.002)***
Yes
27694
107

area
(6)

population
(7)

-0.015
(0.004)***

Confucian temples

-0.015
(0.004)***

Chaste women

-0.016
(0.004)***

Confucianism
Controls
Number of observations
Number of counties

area
(4)

Yes
27713
107

Yes
27713
107

Yes
27713
107

-1.282
(0.703)*
Yes
107
107

-1.489
(0.621)**
Yes
107
107

Notes: In columns (1)-(5) the OLS random-effects regressions are used, and the controls include crop failures, cost
of rebellion, population density, famine relief, commercial center, and a full set of year dummies. Variables are
organized at the county level in columns (6) and (7). Confucianism is the first principal component of schools,
Confucian temples, and chaste women, all of which are normalized by either a countys land area (in 10,000 km2)
or county population (in 10,000). Standard errors are reported in parentheses. * Significant at 10%; ** significant at
5%; *** significant at 1%.

40