DIFFUSING KNOWLEDGE WHILE SPREADING GOD’S MESSAGE:
∗
PROTESTANTISM AND ECONOMIC PROSPERITY IN CHINA, 1840-1920
Ying BAI and James Kai-sing KUNG†
Hong Kong University of Science and Technology
This version, April 2012
Abstract
By using county-level data for 1920, and by exploiting the outcome of the retreat of missionaries due to the Boxer Uprising as instrument, we show how Protestantism promoted economic prosperity in China - a country Weber ruled out for the
development of Protestantism and capitalism. The result of decomposing the overall effect of Protestantism into its denomination-specific constituents shows that
while Calvinism - the denomination whose doctrine resembles Weber’s “Protestant ethic” thesis most closely - dominates the counties where Lutheranism and the
China Inland Mission had a strong presence, it had no distinct significant effect over
counties predominated by the Wesleyans and the Baptists. Moreover, any observed
differences among the denominations in economic outcome are fully accounted for
by the effect of knowledge diffusion activities in general, and by more advanced education (middle versus elementary schools), specific educational content (religious
versus human capital training), and healthcare (hospital) provisions in particular.
Key Words: Weber, Protestantism, Economic Prosperity, Knowledge Diffusion,
Denominations, China
JEL Code: N35, Z12, O18
∗
We thank Robert Barro, Sascha Becker, Davide Cantoni, Eric Chaney, Claudia Goldin, Li Han,
Avner Greif, Wenkai He, Philip Hoffman, Saumitra Jha, Yi-min Lin, Rachel McCleary, Nathan Nunn,
Albert Park, Nancy Qian, Satoru Shimokawa, Carol Shiue, Peter Temin, Robert Woodberry, Noam
Yuchtman, and participants at Harvard’s Economic History Workshop (2011), the Econometric Society’s
and Economic History Associations 2010 Annual Meetings, University of Hong Kong and Hong Kong
University of Science and Technology, for helpful comments and suggestions on earlier drafts of this
manuscript. All remaining errors are ours.
†
Direct all correspondence to James Kai-sing KUNG, Division of Social Science, Hong Kong University
of Science & Technology, Clear Water Bay, Hong Kong. Email: [email protected].
I. INTRODUCTION
There is little doubt that the best-known attempt to associate religion, or specifically
Christianity, with economic prosperity has been that of Max Weber’s.1 In his famous
work The Protestant Ethic and the Spirit of Capitalism, Weber (1930) hinted at the
“subtle link” between the teachings of Calvin and Puritan divines of encouraging savings, investment, and the relentless pursuit of profit, on the one hand, and the unintentional consequences of capitalist behavior and ultimately economic growth, on the other
(Hoffman, 2006, p. 82). While the results of the empirical vindication of Weber’s thesis
(of whether countries affected more by Protestantism tended to experience higher growth
rates) have been mixed,2 endeavors to examine the various alternative channels of growth
have considerably enriched our understanding of the mechanisms through which Protestantism may have contributed to economic outcome.3 These alternative channels include
moral codes (Lipset and Lenz 2000; Stulz and Williamson 2001),4 a country’s openness
(La Porta, et al 1998),5 “spiritual capital” (Arruada, 2010; Barro and McCleary 2003;
Barro 2004; McCleary and Barro 2006a, 2006b),6 social networks (Glaeser and Sacerdote
2008; Putnam 2000),7 and human capital (Becker and Woessmann, 2009, 2010; Gallego
1
Also known as the “most famous link between culture and economic development” (Acemoglu,
Johnson, and Robinson, 2005: 401).
2
For instance, while Glaeser and Glendon (1998) show that economic growth was faster in the Protestant than in the Catholic countries, Delacroix and Nielsen (2001) simply fail to identify any systematic
relationship between Protestantism and the spread of capitalism. And, in his review of the pertinent
literature, Iannaccone (1998) concludes that the available empirical evidence tends to reject rather than
support Weber’s thesis. More recently, Cantoni (2012) finds no evidence of any effect of Protestantism
on economic growth in Germany.
3
While Hoffman (2006, p. 82) suspects that the famous Weber thesis may have been the unwitting
result of his oversight of a parallel religion (post-Tridentine Catholicism) that shared a common ethos in
terms of “disciplined behavior associated with capitalism”, he does encourage endeavors to examining
“factors unrelated to Weber’s Protestant ethic” (p. 83).
4
Lipset and Lenz (2000) provide empirical evidence that Protestant countries are less corrupt, whereas
Stulz and Williamson (2001) contend that Protestant countries provide greater legal protection for the
creditors than do the Catholic countries.
5
La Porta, et al (1998) find that legal traditions that originated from France - a proxy for Catholicism - tend to provide less protection for outsiders than those originating from England - a proxy for
Protestantism.
6
“Spiritual capital” includes religious beliefs and such related characteristics or traits. The studies
undertaken by Barro and his collaborators show that economic growth correlates positively with religious
beliefs - especially beliefs in the concept of hell.
7
Both Glaeser and Sacerdote (2008) and Putnam (2000) regard the networks fostered by houses of
worship and churches are important elements of social capital. Given that social capital is productive,
church attendance should have a separate, positive effect on economic growth.
1
and Woodberry, 2010; Woodberry, 2006, 2007a, 2007b, 2010a, 2010b). In line with the
spirit of this literature, we examine if Protestantism had any distinct economic impact
in China during 1840-1920 - a period when it was forced to open up to the West - in
economic and other spheres. In addition, we also endeavor to identify the actual channel
through which Protestantism may have contributed to economic prosperity in China.
There are two unique reasons why it is important to examine the Protestant economic
history of China. To begin with, for Weber, China represents the counter-case of the development of Western religion; the failure for Protestantism to develop in China was seen
as the main reason behind the failed development of capitalism (Weber, 1930, 1964).8
The second is that China’s age-old tradition in Confucianism and Taoism means that the
Western religions and their underlying ethics - be they Protestantism or Catholicism were not embedded within the society and institutions of China. The external introduction of Western religion into the Chinese context thus allows us to exploit the exogenous
variations in the diffusion of Protestantism and accordingly the measureable effects of
their activities on economic development.
In addition, and this is especially important from Weber’s standpoint, is that the
period under analysis, 1840-1920, represents the genesis of capitalist development of a
more comprehensive nature in China, and thus is the setting when the Protestant ethic
should have been most effective. With the opening up of China following its defeat by the
Western powers since the opium war (circa 1839-1842), the missionary presence in China
increased enormously, resulting in a dramatic increase in the number of converts by the
early 1920s.9 These missionaries were now able to move around the country freely; to the
extent that eventually they were able to venture even beyond the treaty ports after 1860.
By drawing upon the variations in the diffusion of Protestantism across China’s then
8
According to Weber, the prevalence of, and protection afforded by, China’s kinship system not only
created enormous moral hazard problems, but also obstructed the development of factor markets (land
sales), technical innovations, a rational bureaucracy, and not the least a codified system of law, all
of which combined to undermine the development of capitalism (see, for example, Bendix, 1977, pp.
114-116).
9
For example, Bays (2012, p. 60) notes that there were only “a few hundred Chinese Protestant
converts before 1860, almost all in small congregations pastured by missionaries in one of the five treaty
ports”. But by the end of the century the number of Protestant communicants grew to approximately
100,000 (p. 67).
2
1175 counties, we show that Protestantism did indeed have a positive effect on economic
outcome in China in the 1920s.10
To exclude the possibility of a reverse causality in the positive association between
Protestantism and economic outcome, we exploit, as our instrumental variable, the consequences of the Boxer Uprising (circa 1900) on the subsequent spatial diffusion of Protestantism (circa 1920) - by which time the number of Protestant communicants had nearly
tripled (from 100,000 in 1900) to 289,874, and the overall size of the constituencies (including missionary personnel) was as large as 491, 952 (Stauffer, 1922).11 A group of
spontaneously formed patriotic, anti-imperialists, the Boxers’ hatred against the foreigners culminated in the killing of many missionaries in 26 prefectures in 1900;12 to avoid
being massacred, many missionaries fled for their lives.13 Our instrument is sound in
reasoning to the extent that where these missionaries eventually fled to and settled impacted the subsequent diffusion of Protestantism, but it is uncorrelated with economic
prosperity or specifically urbanization except through the effect of Protestantism. To
ascertain where the missionaries fled to, our first specific instrument entails the protocol
of the “Yangzi Compact”, according to which foreigners in the southern part of China
were distinctly better protected from the Boxers’ assault, in conjunction with the variation in the distance of the missionary retreat from the 26 prefectures in which the Boxer
Uprising occurred. Holding constant the distance of retreat, we predict that chances are
much greater that the missionaries would choose to flee to the safer region to the south
of the Yangzi boundary. Hence, we exploit the interaction between this policy difference
and the distance of the missionary retreat as the pertinent instrument; the resulting evidence robustly reaffirms the hypothesis that Protestantism did have a positive impact on
economic prosperity.
10
To the best of our knowledge, Becker and Woessmann (2009) and Cantoni (2012) are the only ones
to study the effect of the channels of Protestantism on economic outcome in a single country. Ours is
the first endeavor that studies China at the sub-national level.
11
While still small as a percentage of the total Chinese population - the communicants accounted for
0.088 percent and the constituents 0.148 percent, the speed and magnitude of growth were astounding.
12
The prefecture is the unit of administration just below that of a province.
13
For accounts of the Boxers incident see Cohen (1978), Esherick (1987), Hsu (2000), and Spence
(1990). For personal accounts of individual missionaries, see Glover (1918) and Saunders (1900), among
others.
3
Our endeavor does not stop there. In the light of the positive effect of Protestantism
on economic prosperity, we venture further to identify the actual channel of such an
effect. Specifically, we want to ascertain whether the positive and significant effect of
Protestantism on economic prosperity actually came from the “Protestant ethic” or some
other channels such as human capital (e.g., Becker and Woessmann, 2009). To do this we
break down the overall effect of Protestantism into its denomination-specific constituents
and compare their effects on economic prosperity. Based on the premise that Calvinism,
with its distinctly strong emphasis on “double predestination” or unconditional election to
both salvation and damnation, represents a doctrine that bears the closest resemblance
to Weber’s “Protestant ethic”, if the positive effect of Protestantism came from the
Protestant ethic, then the counties predominated by the Calvinist denomination should
have the largest effect on economic prosperity. While Calvinism did have a significantly
larger effect on economic prosperity than both Lutheranism and the China Inland Mission,
it enjoys no significant additional effect over those of the Baptists, the Wesleyans and the
rest of the denominations. These mixed results thus do not lend conclusive support to
Weber’s hypothesis.
In view of the fact that the Protestant missionaries had erected schools of various
levels and kinds and built and staffed hospitals both in China and throughout the world,
we test the alternative hypothesis that the positive effect of Protestantism and economic
prosperity may have come from the channel of knowledge diffusion activities. In fact,
after controlling for knowledge diffusion activities (the building and staffing of schools
and hospitals), there is no significant difference in the effect of Protestantism on economic prosperity between pairs of Protestant denominations, suggesting that knowledge
diffusion was the underlying channel through which Protestantism positively impacted
economic prosperity. Moreover, our conclusion that the main effect of Protestantism
on economic outcome comes from knowledge diffusion is further bolstered by the evidence that the overall effect of Protestantism is significantly positive in counties with a
strong presence of middle schools but not elementary schools, in counties predominated
by teacher-training schools rather than Bible schools, and not the least, in counties where
4
more hospitals had been built.
An estimation issue remains: the missionaries in China may have selected the level of
knowledge diffusion activities a priori, conditional upon a county’s socioeconomic characteristics. To ensure that this was not the case, we employ data on the kind and intensity
of knowledge diffusion activities committed by the same missionary societies in other
parts of the world to estimate their involvements in China, based on the assumption
that their “investment” behavior and strategy in one country or region are likely highly
correlated with those in another country. With very minor exceptions, we do find that
the knowledge diffusion activities of the various missionary societies in China mirrored
closely those in other parts of the world. Specifically, whereas the Bible (day) schools are
negatively correlated with the effect of Protestantism, the higher education institutions
have a significant and positive effect. Applying the same control (of knowledge diffusion
activities) to this cross country dataset yields the same consistent finding that Calvinismdominated counties has no significant additional effect over the counties predominated
by other denominations, thereby reaffirming the finding that knowledge diffusion is the
channel through which Protestantism’s positive effect on economic prosperity accrues.
While our account essentially complements the human capital story of Becker and
Woessmann (2009, 2010), our empirical examination of the Protestant history of China
helps identify additional channel(s) through which knowledge useful for economic development is diffused.14 This particular finding takes on additional importance when
placed against the context of the introduction of a qualitatively new educational curriculum especially in the middle schools in China, and the establishment of a new paradigm
for medical and healthcare through the erection of modern hospitals.15 An equally if
not more important and distinct contribution of our study is that we are able to show
14
Becker and Woessmann (2009) show that, while the counties in Prussia with a larger share of
Protestants, which includes Weber’s own hometown, fared better economically than those with a larger
share of Catholics, the observed difference was in fact not caused by the Protestant ethic but rather due
to the increase in literacy rate and accordingly human capital, as more people in the former had greater
opportunities to read the translated biblical texts. The term “useful knowledge” is from Simon Kuznets
(1965).
15
Mokyr (2002, p. 167), for example, shows how household knowledge of infectious diseases had
effectively resulted in the decline in mortality rates in the industrialized West even before effective cures
of such diseases were discovered.
5
that differences in the effect of individual Protestant denominations can in fact be fully
accounted for by the effect of knowledge diffusion activities in general, and by more
advanced education, educational content (religious versus human capital training), and
healthcare (hospital) provisions in particular.
The remainder of this article is organized as follows. In Section II, we provide a brief
overview of the development of Protestantism and the knowledge diffusion activities in
China - specifically those in the spheres of education (building of various levels and kinds
of schools and translation of Western knowledge) and healthcare (erection of hospitals).
We then introduce our data sources and define the variables that we employ in our
analysis in Section III. In Section IV, we test the hypothesis regarding the alleged effect
of Protestantism on economic outcome in the Chinese context. In Section V, we ascertain
whether the positive effect of Protestantism on economic prosperity actually came from
the “Protestant ethic” or knowledge diffusion by comparing the effects of individual
denominations of Protestantism on economic outcome, and then controlling for the effect
of knowledge diffusion activities on this outcome in Section VI. Section VII summarizes
and concludes the study.
II. THE DEVELOPMENT OF PROTESTANTISM AND THE DIFFUSION
OF WESTERN KNOWLEDGE IN CHINA
II.A. Development of Protestantism in China
Before the First Opium War (1839-1842), which marked the end of China’s long-term
isolation and the beginning of modern Chinese history, Christianity had been banned
entirely in China since 1721, after Emperor Kangxi (1661 - 1722) disagreed with Pope
Clement XI’s decree over the Chinese Rites controversy, which led to a complete severance
of ties with the Roman Catholic Church.16 But with the signing of various “Unequal
16
The Chinese Rites controversy was a dispute between the Chinese government and the Roman
Catholic Church from the 1630s to the early eighteenth century. It revolved around whether Chinese folk
religious rites and offerings to the emperor constituted idolatry. Pope Clement XI decided in favor of the
Dominicans (who argued that Chinese folk religion and offerings to the emperor were incompatible with
Catholicism), the result of which led to an enormous reduction in the presence of Catholic missionary
activities in China (Hsu, 2000, pp. 100-103; see also Bays, 2012, p. 28).
6
Treaties” with the Western powers between 1839 and 1842 (especially the Treaty of
Nanjing), the Qing government was forced to admit the Western missionaries into China,
initially to the five coastal cities or treaty ports, and with the signing of the second
round of treaties (especially the Treaty of Tianjin), to the entire country, after which
Protestantism spread quickly.17
According to the monumental survey conducted by Milton Stauffer and his associates
on the development of Protestant missionaries in China in 1919-1920 (Stauffer, 1922),18
by 1920 the Protestant missionaries had established a base in over 96 percent of the
counties (1131 out of 1175 counties) in 14 out of 18 Chinese provinces. Given that the
same survey also contained information on when missionary work began in 957 or more
than 80 percent of the counties (the pertinent information is missing for 218 counties),
we are able to deduce the diffusion of Protestantism in China since the Opium War
(Figure 1). Consistent with Bays’ (2012) evidence that only a handful of missionaries
had been assigned to China before 1839, Figure 1 does confirm that missionary presence
was indeed negligible in 1840. But as China gradually opened up to the Western powers,19
missionary presence grew rapidly and was recorded in approximately 12 percent of the
Chinese counties in the 1880s, and by 1920, the missionaries - representing a total of
some 93 different missionary societies - had penetrated nearly 95 percent of the counties
across the 14 Chinese provinces covered in Stauffer’s (1922) survey (refer to Appendix I
for details).20
Figure 1 about here
In the majority (725 or 61.7 percent) of counties where Protestantism was found
17
Before 1839 (the outbreak of the Opium War) only a total of 50 Protestant missionaries had been
assigned to China, and for reasons pertaining to hardship and frustrations, only a handful stayed (Bays,
2012, p. 46). Thanks to the increasingly efficient and professionalized mission societies back in the
homelands of North America and Europe, the number of Protestant missionaries in China “exploded
from barely 100 in 1860 to almost 3500 in 1906” (Bays, 2012, p. 68).
18
The publication was entitled “The Christian Occupation of China: A General Survey of the Numerical Strength and Geographical Distribution of the Christian Forces in China”.
19
Between 1842 and 1860, both Catholic and Protestant missionaries entered the treaty port in droves
under the Toleration Clauses of the treaties (Gregg, 1946, p. 8). As remarked earlier, however, it was not
until the signing of the second round of treaties (circa 1858-1860) that foreign missionaries were allowed
to venture beyond the treaty ports.
20
The penetration of the Protestant missionaries was so thorough that only a minority of counties (44
or 3.74 percent) had no record of their presence (Stauffer, 1922).
7
present, there was only one missionary society. In another 274 counties (or 23.3 percent)
there were two. Only slightly more than 10 percent of the counties - 132 counties or 11.2
percent to be precise - had two or more coexisting missionary societies (Panel A, Figure
2).21
Figure 2 about here
As we shall make clear later, the 93 missionary bodies identified in Stauffer’s (1922)
survey belonged to different denominations with their own distinct interpretations of
the Protestant doctrine. Before we dwell on that it is sufficient to enumerate at this
juncture the composition of these denominations. The first is the China Inland Mission
(CIM), which was the largest domestic missionary group founded in China in 1865 by
the Englishman Hudson Taylor. Joined by five other missions from Scandinavia, these
six churches covered almost one-third of the counties, or 30.8 percent (362 counties)
as surveyed by Stauffer.22 The denomination that had the most coverage was actually
the Calvinists, which included the reformed and Presbyterian churches. With a total
of 32 missionary bodies, the Calvinists left its footprint in 376 counties (32 percent).
The Anglicans were also well represented. Between their six missionary bodies and 33
non-sectarian missions the Anglicans had their presence identified in 30 percent of the
counties. With respectively nine and seven churches in operation, the Baptists and the
Wesleyans shared similar coverage - 17.4 percent (204 counties) in the case of the Baptists
and 17.8 percent (209 counties) for the Wesleyans. While the Lutherans operated with
more missions (18 altogether), their county coverage amounted to only 11.6 percent (or
136 counties). The summary relationship between the various denominations and their
corresponding coverage of counties are plotted in Panel B of Figure 2.
21
The county of Shanghai, with 25 missionary societies identified, clearly belonged to an outlying case.
The China Inland Mission was a “faith mission” from the start, meaning there were no guarantee of
finances and no real budget. Because of these limitations, the CIM avoided creating large operations in
cities; its strategy was to penetrate the countryside and unreached places instead (Bays, 2012, p. 68).
22
8
II.B. Introduction of Western Knowledge
It is important to recognize that the Protestant missions were not only interested in
spreading God’s message; the conviction that China needed also the Western culture led
the missionaries to “plunge into building, organizing, and regularizing not only schools
but an entire educational system ..” (Bays, 2012, p. 69; see also Hyatt, Jr., 1966). By the
same token, the missionaries were also “persevering providers and improvers of medical
care” (Bays, 2012, p. 69). Jonathan Spence (1990), a preeminent historian of China,
neatly sums up this quest of the Protestant missions to transforming China: “the impact
of the mission movement came through the spread of Christian texts, the publication of
general historical or scientific works, the development of schools, and the introduction of
new techniques of medicine” (p. 206). For our purpose we summarize the introduction of
Western knowledge by the missionaries into China as consisting of primarily two kinds of
activities: those related to education (building schools and translating texts) and those
related to health (erecting hospitals).
II.B.1. Education
An integral part of the institutional development that underpinned the endeavors of the
Protestant missionaries to transforming China lies in the translation and publication of
not only the Bible and Western science and technology but also law and a variety of
other subjects (Barnett and Fairbanks, 1985). The concerted efforts of many publishers
- chief amongst them Inkstone Press, who was renowned for being “the publishing Mecca
of missionary activities” (Elman, 2006, p. 102), had been instrumental in translating
texts from a whole host of Western scientific and engineering disciplines. In particular,
Alexander Wylie, who supervised Inkstone Press after 1847 and who worked with other
distinguished missionaries like Joseph Edkins and a number of outstanding Chinese scholars, had hugely influenced the subsequent diffusion of Western science in China. Under
his supervision, the Chinese terminologies of such important fields as chemistry, geography, animals and plants, astronomy, calculus, electricity, mechanics, fluid mechanics and
9
even optics and sound were introduced in a systematic manner on Chinese soil for the
very first time (Elman, 2006, p. 116).23 In addition, Spence (1990) notes that the missionaries, once again with Wylie playing a leading role, had translated works on Western
governments and history that were circulated widely in China beginning from the second
half of the nineteenth century.24 Likewise, the US Presbyterians, who expanded their
Shanghai printing press, had similarly been very active in this regard (Barnett and Fairbanks, 1985). In all, missionary journalism flourished in the larger treaty port cities since
the signing of the second round of treaties (circa 1860).
It is thus the systematic translation of a large corpus of literature on the modern sciences, and the concomitant introduction of a Western-based curriculum in schools erected
by the Protestant missionaries, that together became an important channel or channels
through which this body of new and useful knowledge was disseminated in China since
around the mid-nineteenth century.25 In the Chinese context, the Protestant missionaries
had been an integral part of the development of a national system of education (Gregg,
1946, p. 23).26 With the appointment of the School and Textbook Committee in 1877
and its eventual succession by the Educational Association of China, schools in general,
and mission schools in particular, grew rapidly in the midst of Empress Dowager’s reinstatement of the Imperial Examinations system after the failed Hundred Days’ Reform
in 1898.27 Whilst 1902-1905 saw the remarkable development of government education,
23
Woodberry (2010b: 10-11) disputes Habermas’ (1989) claim that it was printing and capitalism
combined that gave birth to civil society. Citing evidence from both Europe and Asia, he contends
that it was the conversionary Protestants’ preoccupation with the belief that “everyone needed access
to ‘God’s word’ ” that underpinned the mass expansion of texts and newspapers. In the case of Asia,
for instance, it was not until the arrival of Protestant missionaries in the nineteenth century did printing
begin to supplant manuscripts, and newspapers began to develop in earnest. More specifically, the first
newspapers in Chinese were actually printed by Protestant missionaries (p. 12).
24
One may note, in this context that while both Protestants and Catholics had actively engaged in
the translations and publications of journals on Western science, the former outnumbered the latter by
a considerable margin. For instance, in addition to publishing 4,000 books on Western sciences in 1835,
a year later the Protestants published twice as many journals as the Catholics did - 238 compared to
115 (Yang, 1968). This is consistent with the European evidence that, “from the 1600s on, Protestant
areas had and exported more printed material per capita than Catholic areas” (Eisenstein, 1979, cited
in Woodberry, 2010b: 11).
25
The diffusion of Western science was confined to the literati and intellectuals. However, it was also
linked to the development of a national curriculum via the organized effort of the School and Textbook
Committee in the second half of the nineteenth-century (Gregg, 1946, p. 19).
26
This is in fact consistent with the evidence stated elsewhere that “even when European governments
formed state school systems, they often merely nationalized religious schools” (Woodberry, 2010b: 14).
27
As E. T. Williams remarked in 1899, mission schools were “never more crowded” (cited in Gregg,
10
the period also witnessed the launching of a missionary system of education and, accordingly, a rapid increase in the number of schools and students. For example, the number
of boarding missionary institutions increased from just over 100 in 1902 to 456 in 1909,
with a fourfold increase in enrollments, from approximately 5,000 to 21,785 (including
919 college students). A similar trend was experienced by the day schools. By 1906, over
2,500 Protestant schools were founded, enrolling some 57,000 students (see also Bays,
2012, p. 69). One must note, in this context, that tremendous efforts had to be made to
prepare the Chinese associates for teaching, medical work, and evangelism.28
The Protestant missionaries brought not only the mission (day) schools, but, more
importantly, set up vast numbers of primary and middle schools and also teacher-training
and other professional schools. An authoritative source reveals the astounding development of school enrolment: by 1920s there were “150,000 pupils in lower primary schools,
33,000 in higher primary schools, 15,000 in middle schools, 600 in teacher-training schools
of various grades, and 2,000 in colleges and professional schools; a total of approximately
205,000 in Christian schools of all grades and types” (Committee of Reference and Counsel of the Foreign Missions Conference of North America, 1922, p. 21). And by 1922 the
Protestant missionaries had founded 956 senior (higher) primary schools and 5,637 junior
(lower) primary schools, enrolling 32,899 and 151,582 students, respectively (Lutz, 2001,
p. 186).
The Catholics’ efforts simply paled in comparison to the Protestants’ in this regard.
The same source cited above reveals that in 1921 a total of only 144,344 students were
enrolled in all the Roman Catholic schools (Committee of Reference and Counsel of the
Foreign Missions Conference of North America, 1922, p. 21), which was less than the
number of students in the lower primary schools established by the Protestant missionaries. Whilst the Catholics did step up their efforts in 1938, they only erected 362 senior
1946, p. 43).
28
The rapid expansion of education by the Protestant missionaries meant a great expansion of the
missionary force and of Chinese teachers. It is little wonder that the missionary body nearly doubled
between 1902 and 1909 (Gregg, 1946, p. 43). Again, the Chinese experience is consistent with that of
the European’s prior to the late nineteenth century, where religious groups were found to have devoted
a great deal of resources to teacher training, child-focused texts, and so forth in order to make mass
schooling possible (Woodberry, 2010b: 13-14).
11
primary schools and 3,135 junior primary schools, and enrolled 18,805 and 115,793 students respectively; clearly they were still far behind their Protestant counterparts after
playing catch-up for 16 years (Yang, 1968). An important reason why the Catholics were
less aggressive in the provision of education and health services had to do with their
greater emphasis on the cardinal virtue of preserving and nurturing faith among their
followers than on the direct involvement in social and community affairs (Wiest, 2001).
Based on the survey conducted by Stauffer (1922), we construct in Panel A of Figure
3 a graphical distribution of different kinds of schools (lower primary, higher primary,
middle, Bible, and teacher-training schools) founded by the Protestant missionaries across
the 1175 Chinese counties. Altogether lower primary schools were founded in 719 counties
(61.1 percent), higher primary schools in 314 counties (26.7 percent), middle schools in
107 counties (9.1 percent), Bible schools in 60 counties (5.1 percent), and teacher-training
schools in 20 counties (1.7 percent). Moreover, as Stauffer (1922) has listed also the
founding mission of the middle schools, we can sort them into the six denominations
present at the time in China and show the varying extent to which they had committed
resources to human capital training at this higher level. From Panel B of Figure 3 we can
see that the CIM and the Lutherans invested distinctly less than the other denominations.
While the CIM was the largest missionary group in China by 1905, as a “faith mission”
(meaning there was no guarantee of finances from its home institutions) and specifically
because of its non-urban strategy, the CIM had established only five middle schools
by 1920. Even the Lutherans - the second-most inactive group - managed to erect 14
middle schools. No doubt the Baptists and the Wesleyans - having invested in 25 and
26 middle schools, respectively - were more active in this regard. Nevertheless, they still
lagged behind the Calvinists, who had erected 60 middle schools. But by far the greatest
effort was made by the Anglicans and other non-sectarian missions, which altogether had
erected a total of 124 middle schools.
Figure 3 about here
12
II.B.2. Western Medical Knowledge and Hospitals
In line with their endeavours to diffuse the knowledge of Western science, the Protestant
missionaries similarly pioneered a series of medical and science translations under the
earlier efforts of Benjamin Hobson, an English medical missionary.29 More importantly,
these missionaries erected hospitals in earnest in the 1860s - initially in the treaty ports
but quickly spread to the interior - where such knowledge could be put to real practice
(Elman, 2006, p. 104). According to Bays (2012), “hundred of physicians were appointed
as medical missionaries (300 in China in 1905).” What is more, in order to “deliver care,
some large (and expensive) modern hospitals were built and staffed by highly qualified
physicians and researchers” (p. 69).
To be fair, by 1935 the Catholics had built 233 hospitals-cum-old people’s homes,
housing some 89,321 patients. Nonetheless, their efforts still paled in comparison with
the Protestants’, who by 1937 had built 271 hospitals housing over 224,258 patients,
while also having treated more than four million (4,150,342) out-patients. In addition,
the Protestants had set up 140 schools to train the nurses (Yang, 1968; see also Bays,
2012). In fact, long before these endeavors, the Protestants had already established a
Medical Missionary Society in China in 1838 and the China Medical Missionary Association in 1890, the purposes of which were to provide medical consultations, to contribute
financially to hospital constructions, and to publish medical journals and books, which
included the publication of the all-important bilingual (Chinese and English) medical
dictionary (Yang, 1968).
Indeed, Stauffer’s (1922) survey found a total of 220 hospitals belonging to various
denominations in the 14 sampled provinces. Panel C of Figure 3, which reports these
results, similarly finds that the CIM and the Lutherans paled in comparison with the
other denominations insofar as the provision of hospitals was concerned. Whereas the
29
Hobson’s Treatise on Physiology, which presented modern anatomy and reintroduced the centrality of
the brain and the nervous system, had remained standard in China until 1886, when it became superseded
by new translated texts of the Medical Missionary Association, which printed its own medical journal in
Chinese (Elman, 2006, p. 105). The translations of Hobson’s works into Chinese were important as they
led some literati to question the use of traditional medicine, which is something the Jesuits had not been
able to accomplish. These translations included the Treatise of Natural Philosophy, which was the first
work to introduce modern Western chemistry (and included in it also sections on physics, astronomy,
geography, and zoology) to China.
13
Baptists and the Wesleyans had erected 25 and 38 hospitals, respectively, the CIM had
established only 12 hospitals and the Lutherans only 17. As with the case of middle
schools, the Calvinists and the Anglicans, including other non-sectarian missions, had
each built 64 hospitals - the most of the six denominations.
III. DATA: URBANIZATION, PROTESTANTISM AND COUNTY CHARATERISTICS
In order to test the overall effect of Protestantism on economic prosperity and to identify
the channel(s) through which Protestantism impacted economic outcome we construct a
dataset that that allows us to exploit the variation in knowledge diffusion activities across
individual missionary denominations in China for the period 1840-1920. The dataset is
based primarily on digitizing Stauffer’s (1922) survey; the result is a dataset that covers
1175 counties in 14 provinces to the south of the Great Wall, whose locations are shown
on the map of Qing China (circa 1820, Figure 4).30 We choose 1840 as the starting point
of our analysis because before then missionary presence was negligible. We choose 1920
as the end point because soon after that China was caught in waves of civil conflicts
between the Communists and the Nationalists, not to mention the Sino-Japanese War.31
Before we conduct the empirical test we define the variables that we have employed in
our analysis in the remainder of this section. We begin with our dependent variable,
economic prosperity.
Figure 4 about here
30
These provinces are: Zhili, Shandong, Hebei, Henan, Shanxi, Shaanxi, Anhui, Jiangsu, Hubei, Hunan, Jiangxi, Zhejiang, Fujian and Guangdong. We exclude in our analysis the provinces of Yunan,
Guizhou, Gansu, Tibet, Tsinghai, Inner Mongolia and also those of Northeast China, for two reasons.
The first pertains to data availability, in addition, they were not settled agricultural communities.
31
Erupted initially in April of 1927, the Chinese Civil War was fought between the Nationalist Party
and the Communist Party which was intermittently interrupted by the Second Sino-Japanese War, which
ended in 1945, after which the two parties resumed full-scale confrontations for another four years.
14
III.A. Urbanization (Y ) measuring economic prosperity
To measure the consequence of Protestantism on economic prosperity, we need information on the regional distribution of income across China for the period 1840 to 1920.
Unfortunately, such data are not available. As a substitute, and following Acemoglu,
Johnson and Robinson (2002, 2005), we use the degree of urbanization to proxy for economic outcome, given its positive and significant correlation with per capita income.32
As with various other measures of Protestantism, data on urbanization in China in 1920
are obtained from Stauffer (1922), which contains population statistics on cities of sizes
larger than 25,000. This allows us to compute, at the county level, the share of urban
population in the overall population as our measure of urbanization - our dependent
variable. The overall mean of the share of urban population in a prefecture in 1920 was
roughly 4.30 percent (Table 1).
III.B. Protestantism
Level of Protestantism (P ). Evidence suggests that, by the 1920s, a total of 5,557 organized congregations had managed to spread across 1,175 counties in 14 Chinese provinces
(The Christian Occupation of China: A General Survey of the Numerical Strength and
Geographical Distribution of the Christian Forces in China, Stauffer, 1922). To measure
the effects of Protestantism (P ), we employ the number of total Christian communicants
normalized by the total population (communicants per 10,000 population) in each of these
1,175 counties as the pertinent proxy. The overall mean of this measure is 8.394 (Table
1). Moreover, we employ the total number of Christian constituencies, also normalized
by the total population (communicants per 10,000 population), as an alterative measure.
Denominational characteristics. Included in Stauffer’s (1922) survey is the presence
of individual denominations in each of the 1175 counties. This information allows us to
construct, for each county the share of a Protestant denomination to the total number of
missionary societies. Denote di as the share of the ith denomination in a county, specifically
32
It is worth pointing out that the foregoing relationship exists even after a country has undergone
industrialization.
15
we have d1 denoting the CIM, d2 the Lutherans, d3 the Calvinists, d4 the Baptists, and
d5 the Wesleyans, with the Anglicans and other non-sectarian missions serving as the
reference group (d6 ). In addition, as we are able to identify also the host country of each
missionary society we employ the host countries as controls, by dividing the missionary
societies into four groups (hj ) based on their home country. The first is the United
Kingdom, the second is the United States of America, the third is Europe, and the fourth
is the group of international denominations (mainly those missions without a specific host
country).
Knowledge Diffusion. We focus on two types of knowledge diffusion activities: educationrelated (schools) and health-related (hospitals) activities. Once again, Stauffer’s (1922)
survey contains information on whether a particular type of school had been erected in
each county. Specifically we employ five dummy variables to represent the lower primary schools (K1 ), the higher primary schools (K2 ), the middle schools (K3 ), the Bible
schools (K4 ) and the teacher-training schools (K5 ). Altogether 61.2 percent of the surveyed counties had lower primary schools, 26.7 percent had higher primary schools, 9.1
percent had middle schools, 5.1 percent had Bible schools, and 1.8 percent had teachertraining schools. Similarly, the same survey also enumerates the number of hospitals by
denomination in each county (K6 ); on average each county had 0.187 hospitals.
III.C. County Characteristics (W )
Catholicism. Although Roman Catholicism has had a long and convoluted history in
China since at least the eighth century A.D., the Chinese Rites controversy, which sparked
a dispute and resulted in a temporary break in ties between the Chinese government
and the Catholic Church from the 1630s to the early eighteenth century, substantially
reduced the presence of Catholic missionaries in China (Bays, 2012; Hsu, 2000). Similar
to the spread of Protestantism in China since the Opium War (of 1839-1842), Catholic
missionaries became active again. We employ the number of Catholic missionary stations
per one million population to proxy for the development of Catholicism in China. The
overall mean of this measure is 0.630 (Table 1).
16
Initial Economic Conditions. Given that urbanization is a gradual process, the level
of urbanization in 1920 would most likely be correlated with its lags. But since we do
not have the share of urban population in 1840, we can only control for this lag using the
three levels of urbanization generated by Rozman (1974) in the form of dummy variables.
According to Rozman’s classification, big cities were those with a population of 300,000
and up, the ones in the middle between 70,000 and 300,000, and small cities between
30,000 and 70,000. According to these criteria, 3.1 percent or 37 counties are classified
as small cities, 2.2 percent or 26 counties as mid-sized cities and 0.7 percent or 8 counties
as big cities (Table 1).
Western Economic Influence. In addition to religious influence, new and exogenous
economic forces began to penetrate China’s autarkic economy after the establishment of
the first five treaty ports in China in 1842.33 Since this process of economic penetration or
integration occurred simultaneously with the development of Christianity, it is necessary
to control for the effects of these Western economic influences on the Chinese economy.
We did so by means of two variables that were likely to be correlated positively with both
the development of Western religion and the Chinese economy. They are two dummy
variables for treaty ports and railway networks. We assign a value of zero to counties
with no treaty ports, and 1 to those with treaty ports. We did the same for the proxy
of railway networks. Table 1 shows that there are 17.9 percent of counties with treaty
ports, and 34.5 percent of counties with railway.
Political Center. To the extent that counties in which a prefectural government was
located were distinctly more prosperous, we employ an additional control of the initial
economic conditions by controlling for the location of the prefectural government seat.
Moreover, the share of government officials in a given urban population was higher in
the political centers. Table 1 shows that 15 percent or 176 counties in our sample were
situated in what may be regarded as political centers.
Geographical Factors. In development economics, geography is considered by some as
a key factor in accounting for the difference in economic performance. As Sachs (2000:
33
They are Shanghai, Guangzhou, Ningbo, Fuzhou, and Xiamen.
17
30) explains: “Geographical advantages might include access to key natural resources,
access to the coastline and sea-navigable rivers, proximity to other successful economies,
advantageous conditions for agriculture, advantageous conditions for human health”. To
control for the differences in geography among our prefectures, we use three dummy
variables; the first is whether a county is located along the Jinghang or the Grand Canal,
a vital route of transportation between north and south China, the second is whether a
county is located on the coastal seaboard (coast), whereas the third considers whether
a county is situated along the Changjiang (or Yangzi) River, the most navigable river
in China. Our inclusion of these three geographical variables is premised on the notion
that they were likely partial determinants of economic growth and were probably also
correlated with the development of Protestantism. Approximately 3.7 percent of the
sampled counties were located alongside the Grand Canal, 4.8 percent alongside the
Changjiang River and 10.8 percent on the coast. Moreover, the development of religion
may also be affected by population density and the size of a region, as foreign missionaries
supposedly chose to begin their work in populous or larger regions. As with treaty ports,
both population density and regional size should affect economic growth. Thus to control
for their effects, we include both the population density in 1850 and the size of a county
in our empirical estimations. In addition, we also control for the provincial dummies.
Details concerning the definition, summary statistics of the pertinent variables and their
sources are summarized in Table 1.
Table 1 about here
IV. PROTESTANTISM AND ECONOMIC PROSPERITY
IV.A. Baseline Results
We begin by regressing the economic outcome (y) of the 1,175 Chinese counties on the
proxy of Protestantism (P ) using the method of Ordinary Least Squares (OLS) based on
18
Equation (1):
y = ρP + W β + ε
(1)
in which ρ represents the effect of Protestantism on economic prosperity and β represents the coefficients of the control variables. In our estimations, we also control for a
number of independent variables (W ), including Catholicism, initial economic conditions
around1840, Western economic influences (treaty ports and railway networks), whether
a county was a political center, whether a county was located on the coast, along the
Changjiang River, and along the Grand Canal, the population density, the size of a
county, the provincial dummies, and the constant term.
We begin by reporting the baseline results, which show that a significant and positive
relationship exists between Protestantism and economic prosperity or specifically the degree of urbanization. For instance, column (1) of Table 2 shows that Protestantism has
the effect of significantly increasing economic prosperity, after controlling for only population density, county size and provincial dummies. In terms of magnitude, an increase of
one standard deviation (15.161) in the density of Christian communicants increases the
level of urbanization by 3.260 (0.215*15.161) percentage points. Column (1) also shows
that the estimated coefficient on Roman Catholicism is significant. However, the significance of Catholicism disappears after controlling for political centers, geographic factors
and provincial dummies (column (2)), but Protestantism remains significant even though
the pertinent magnitude decreases slightly from 0.215 (column (1)) to 0.195 (column (2)).
As for the other control variables, population density is found to be positively correlated
with economic prosperity, and counties where a prefectural government is seated also tend
to have a higher rate of urbanization (about 10 percentage points higher). Not surprisingly, some geographic factors also have a positive effect on urbanization; for example,
counties located in the coastal region or along the Grand Canal are also more prosperous.
Table 2 about here
Furthermore, it is necessary to control for the possible effect of the differences in initial
economic conditions prior to the massive entry of the Western missionaries around the
19
middle of the nineteenth century, on economic prosperity, for such differences may affect
both the diffusion of Protestantism and urbanization. Hence we include in our regressions
the initial level of urbanization in the 1840s. The result shows that this particular control
is significant and that it decreases the effect of Protestantism from 0.195 (column (1))
to 0.162 percent (column (3)). Lastly, since we cannot rule out the possibility that
other factors such as Western economic influences may also affect both the diffusion of
Protestantism and economic prosperity, we include two pertinent proxies; namely whether
a county belonged to a treaty port and whether it was served by the railway, respectively,
as pertinent proxies (column (4)). The result of including all these control variables is
that the effect of Protestantism or specifically the number of communicants per 10,000
population remains significant (column (5)). In terms of magnitude, a one-standarddeviation (15.161) increase in the density of Christian communicants increases the level
of urbanization by 2.411 (0.159*15.161) percentage points (column (6)), a magnitude
smaller than that of the baseline result (of 0.215, column (1), Table 2).
IV.B. Instrumental Evidence
In estimating the effect of Protestantism on economic development, it is necessary to
deal with the various endogeneity issues associated with the measure of Protestantism.
There is first of all the problem of measurement error. Our key explanatory variable,
Protestantism, is possibly a noisy proxy, so that the coefficient on the inexactly measured
Protestantism may be biased towards zero. Reverse causality presents another source of
endogeneity. Rather than religious development leading to economic growth, conceivably
it could be the other way round (e.g., Samuelsson, 1993; Barro and McCleary, 2003).34
Last, but not least, is the problem of omitted variable bias, despite our attempt to
control for a variety of factors that are likely correlated with both economic development
34
This is in fact what the secularization hypothesis has postulated (see, e.g., Samuelsson, 1993). This
thesis states that economic development has the effect of rendering individuals less religious as measured
by church attendance and religious beliefs. In a different twist, the same thesis argues that economic
development undermines the importance of organized religion both in the realms of political decisionmaking and in the social and legal processes (Barro and McCleary, 2003; Berger, 1967; Chaves, 1994;
Weber, 1930; Wilson, 1966).
20
and Protestantism (Iyigun, 2008).35 While the external introduction of Protestantism
in the Chinese context allows us to exploit the exogenous variations in the dispersion
of Protestant ethics, identification of the causal direction between Protestantism and
economic outcome requires an instrumental variable that is correlated only with the
former. In the pertinent historical context, the diffusion of Protestantism was affected
to a large extent by the consequences of an important historical incident known as the
Boxer Uprising.
Threatened by the assault of a group of spontaneously formed patriotic, anti-imperialists
whose hatred against the foreigners culminated in the killing of many foreign missionaries in 1900, many missionaries fled for their lives. Where they eventually settled, one
may reasonably expect, should have a distinct impact on the diffusion of Protestantism.
While we have no systematic knowledge of where the missionaries had fled to (and settled), scattered accounts of those who had experienced this trauma suggest that their
chief concern was safety, as can be gauged from their escape routes (e.g., Saunders, 1990;
Glover, 1918, more on this below). It was due to the assault of the Boxers on the missionaries that a policy emerged to render some parts of China a distinctly safer place for
the latter. As we will elaborate further, under the sponsorship by some political elites
who did not support Empress Dowager’s decision of declaring war against the Western
powers, a joint declaration known as the “Yangzi Compact” or literally “Joint Defense of
the Southeast” (Dongnan Hubao), was signed between some provincial leaders and the
foreign powers to provide explicit protection to the missionaries - foreign and domestic
alike. In what follows we provide a brief historical narrative of the Boxer Uprising and its
consequences in terms of the subsequent diffusion of Protestantism, based on the sharp
differences between the protection afforded by the “Yangzi Compact” and the resulting
safety implied by this agreement.
35
Christianity may be influenced by historical events, which may be correlated with economic growth.
For instance, Iyigun (2008) shows the Protestant Reformation was affected by the rise of the Ottoman
Empire.
21
IV.B.1 Boxer Uprising and the “Yangzi Compact”
In response to the seizure of the Taku Forts by the Western powers,36 Empress Dowager
Cixi changed her stance from originally suppressing the Boxers to eventually supporting
them. This resulted in the imperial courts issuing, in June of 1890, a proclamation
of war against the Western powers, followed immediately by the Boxers’ attack of the
foreign compounds in Tianjin and Beijing. Between June and August, when Beijing was
occupied by the Eight-Nation Alliance, the Boxers attacked the foreign missionaries in 26
prefectures across the country (Figure 5), resulting in the deaths of 53 foreign Catholic
missionaries and 188 foreign Protestant missionaries, with casualties among the Chinese
communicants - 23,000 Catholics and 5,000 Protestants, respectively - distinctly higher
(Yang, 1968).37
Figure 5 about here
An immediate consequence of the Boxer Uprising was that many missionaries attacked
by the Boxers abandoned their congregations and fled for safety elsewhere in the country
(Saunders, 1990; Glover, 1918; Song and Li, 1995; Austin, 2007).38 While no systematic
information is available on exactly where all the attacked missionaries had fled to, individual accounts of the surviving missionaries suggest that they had made tremendous
efforts to flee to where the protection of their lives would be enforced most rigorously.39
36
Located somewhere between Beijing and Tianjin, the Battle of the Dagu (Taku) Forts occurred during
the Boxer Uprising between China’s military and allied Western and Japanese naval forces, whereby the
allies captured the Forts after a brief but bloody battle.
37
There were a total of 26 prefectures where the Imperial Civil Examination had been temporarily
suspended as a result of the Qing government’s decree in June 1901. The Annex 8 of the Boxer Protocol
reaffirms this decree once again. We should point out though that the Boxer Uprising represented only
the pinnacle of the patriots’ hatred of foreigners - a sentiment that could be dated back to as early as
the 1860s. The Uprising was predated, for example, by the Tianjin “massacre”, or the vengeance of the
French against the killing of their Consul and several traders and their wives by a hostile Chinese crowd
four decades earlier. See, for example, Cohen (1978), Spence (1990).
38
In their account of the Canadian Protestant missionaries’ experience of the Boxer Uprising, for
instance, Song and Li (1995, p. 67) find that most of these missionaries had retreated from the north
of Henan Province when their congregations were attacked by the Boxers. Similarly, Austin (2007)
characterizes the evacuation of the missionaries from Shanxi Province to the coast as an “exodus” (p.
415). While many returned to the inland afterwards, others stayed in the safer areas; in both cases they
were joined by a new generation of missionaries “who did not know what had happened” (Austin, 2007,
p. 426; see also Graham, 1995, p. 10).
39
Many Canadian Protestant missionaries were found to have retreated from the north of Henan
22
Intuitively, if we hold constant the distance of retreat,40 the missionaries were more likely
to flee to the safer counties. Their hypothesized pattern of retreat can be represented by
the following expression:
E(R) = φ1 D + φ2 S + φ3 DS
(2)
in which D represents the distance of retreat from a county to the prefectures of the
Boxer Uprising, S represents safety, and DS their interaction term. We hypothesize that
φ3 > 0, which implies that the attacked missionaries would likely have fled to the safer
region - the region covered by the “Yangzi Compact”.
In this context, the “Yangzi Compact”, an agreement made explicitly between those
political elites who did not support the imperial edict of attacking the Western powers,
on the one hand, and the foreign consuls who were eager to protect their nationals on the
other, provides a useful clue. Just when the anti-foreign outbreaks threatened to spread
from the north into the Yangzi provinces (located on the southeastern coastal seaboard),
Sheng Xuanhuai, the then Ministry of Transportation, disregarded the imperial proclamation, and urged the provincial heads in the south - from whom he received tremendous
support - not to give it circulation. According to Feuerwerker (1958), Sheng was “instrumental in formulating a dual policy of neutrality toward affairs in the north and loyalty
to the Manchu court”, on the one hand, while “concluded with the foreign consuls in
Shanghai providing for the protection of life and property in the Yangzi area by the two
governor-generals” on the other (p. 72).41 Although the “Yangzi Compact” was applied
initially only to the Lower Yangzi region, it was quickly “followed by the authorities in
the other southern provinces with the result that the fighting and the destruction in the
north were effectively kept out of South China” (Feuerwerker, 1958, p. 73).
The “Yangzi Compact” and the policy difference it implies means that conceptually
Province to as far as Shanghai - a territory that afforded distinctly greater protection to the foreigners
by the “Yangzi Compact” (Song and Li, 1995: 67).
40
Throughout the paper the terms “distance of retreat” and “distance to the Boxer Uprising” are used
interchangeably.
41
The essence of the “Yangzi Compact” is that the Shanghai Enclave was to be jointly protected by
all the foreign nations with a presence therein, whereas Changjiang (the Lower Yangzi) and in particular
Suzhou and Hangzhou were to be protected by the local governor-general. This was how the property
rights and lives of both foreign and domestic merchants were protected.
23
we can divide China into two separate regions - bounded by whether or not a province
had formally committed to protecting the foreigners - between which there was significant
difference in safety. Figure 5 shows, with the thick, gray line physically demarcating the
two regions, that provinces in the south were protected by the “Yangzi Compact” (C = 1),
whereas those in the north were not (C = 0). Indeed, what Figure 5 also shows is that
the damage inflicted by the Boxers was confined largely to north China; the south, by and
large, was relatively unperturbed. This is most clearly articulated in the missionaries’
own account of their varied perceptions of safety during their journeys of retreat from
the Boxers’ assault.42 In other words, the demarcation line had the de facto effect of
sharply changing the protection afforded to the fleeing missionaries. That is, to the north
of the boundary the missionaries were likely subject to a greater threat of assault by the
Boxers (C = 1), but once they crossed the boundary to the south their safety increased
drastically, as is evident in Reverend Saunders’ remark about how well they were treated
by the officials in a province governed by one of the sponsors of the “Yangzi Compact”.43
To better interpret this discontinuity in safety at the boundary, we first define an
observed “assignment variable” - namely distance to the boundary (B) - as the shortest
distance between a county’s mass point and the boundary - calculated on the basis of
its latitudes and longitudes. Assuming that the boundary has a value of zero, we assign
positive values to counties lying to the south of the boundary, and negative values to
those in the north. For instance, suppose there are two counties both of which are 10
kilometers away from the boundary but located on opposite sides, the distance of the one
to the south of the boundary would be assigned a value of 10 and the one to the north
would be given a value of -10. The intuition that underlies this line of reasoning is that
safety will increase sharply where distance to the boundary (B) exceeds the cut-off value
of zero. More specifically, we denote S1 as representing safety on the south side of the
42
When Saunders and his family reached Hankou, the capital of Hubei Province and a “Compact”
province, Saunders (1900: 29) spoke with a great sense of relief: “Now we had come to the Hu-pen
(Hubei) Province, ruled over by Chang-chih Tung, and we were treated well by all the officials, and
instead of travelling as before, in cartsand the rest of our journey to Han-kow was accomplished in
comparative comfort. We arrived at our mission house on Tuesday morning, August 14th, in all 49 days
since we left Ping-iao” (refer to Appendix III for details).
43
See the previous footnote.
24
boundary (i.e. the “Compact” counties), and S0 as representing safety on the north side
of the boundary, and by assuming that S1 > S0 we can express the relationship between
distance to the boundary (B) and safety (S) as follows:
S=



S0 + λ0 B
C = 0 or B < 0


S1 + λ1 B
C = 1 or B > 0
(3)
where S1 > S0 implies the discrete or discontinuous increase in safety at the boundary,
and λ1 and λ0 each represent the differing effect of distance to the boundary on safety
between the two regions. We can now rewrite the relationship between S and B as:
S = S0 + (S1 − S0 )C + λ0 B + (λ1 − λ0 )CB
(4)
= ϕ0 + ϕ2 C + ϕ3 B + ϕ4 CB
A key assumption underlying our empirical strategy is that in the absence of the “Yangzi
Compact” there should be no difference in safety between counties on opposite sides of
the boundary. The sharp discontinuity in safety between the two sides of the boundary,
that is, ϕ1 > 0, is thus attributed to the effect of the “Yangzi Compact”.
Based on Equations (2) and (4) and in light of the asymmetrical effect on safety
induced by the “Yangzi Compact” we can derive our instrumental variable for Protestantism as follows:
E(R) = δ5 DC + δ0 + δ1 D + δ2 C + δ3 B + δ4 CB + δ6 DB + δ7 DCB
(5)
The interaction term DC is our proposed instrument for Protestantism. Holding constant
the distance of retreat (D), we assume that the safety-concerned missionaries would
choose to flee to where C = 1 rather than C = 0, i.e., δ5 > 0.
25
IV.B.2 Validity of Instrument
Our hypothesized reasoning suggests that the safety-concerned missionaries would most
likely choose to flee to counties in the south because of the distinctly greater safety
it affords, which, if true, would duly affect the subsequent diffusion of Protestantism.
Equation (6) helps capture this intuition:
P = δ5 DC + δ1 D + δ2 C + δ3 B + δ4 CB + δ6 DB + δ7 DCB + W γ + v
(6)
We employ the interaction term DC, where D stands for the distance of retreat and C
the dummy variable of “Yangzi Compact” to instrument the diffusion of Protestantism.
Based on Equation (6) we regress the diffusion of Protestantism and report the pertinent results in Table 3. The most striking finding is that the effect of the distance of
retreat is significantly different between counties on opposite sides of the boundary. This
lends support to our hypothesis that the missionaries are more likely to flee to counties
lying to the south of the boundary, reflecting the condition of δ5 > 0.
Table 3 about here
A key assumption underlying our identification strategy is that, except for Protestantism, our interaction term DC is not correlated with other factors that might be
correlated with urbanization. To ensure that our instrument is valid, we regress Roman
Catholicism on the same set of independent variables, but assume δ5 to be insignificant.
Column (3) of Table 3 indeed shows that, in the instance of Catholicism the distance of
retreat (D ) is not significantly different between counties to the north and counties to
the south of the boundary. Likewise, to ensure that the counties on both sides of the
boundary of equal distance to the Boxer Uprising (D) are not significantly different due
to differences in Western economic influences and initial economic conditions, we repeat
the same exercise with Roman Catholicism, and find that the effect of D on the tested
variables is not significantly different between counties in the safe region and those in
the dangerous region (columns (4) and (5) of Table 3). This suggests that DC is not
26
correlated with variables that may be correlated with economic prosperity.
IV.B.3. Effect of Instrumented Protestantism on Economic Prosperity
Now we can apply our instrument to correct for the possible endogenous relationship
between Protestantism and economic prosperity. Based on Equation (7), we report our
instrumental-variable (two-stage least squares) estimates in Table 4:
y = ρP
P =
+ η1 D + η2 C + η3 B + η4 CB + η6 DB + η7 DCB + W β + ε
(7)
δ5 DC + δ1 D + δ2 C + δ3 B + δ4 CB + δ6 DB + δ7 DCB + W γ + v
The pertinent result shows that the instrumented Protestantism does have a significant and positive effect on urbanization. Unlike the baseline results in Table 2, in which
the magnitude of the effect of Protestantism decreases somewhat after controlling for
initial economic conditions and Western economic influences, the pertinent magnitude
increases from 0.797 in the estimation where the covariates of initial economic conditions, Western economic influences and geographic factors are all excluded (column (1.2),
Table 4), to 0.818 where they are fully accounted for (column (2.2), Table 4).
These results are very encouraging because they show that our instrument is correlated with only Protestantism but not with other possible co-determinants of economic
prosperity, suggesting that Protestantism is most certainly the only channel through
which our instrument DC affects economic prosperity. Moreover, in light of the much
larger marginal effect of instrumented Protestantism (0.818, column (2.2), Table 4) than
the baseline estimate of 0.159 (column (5), Table 2), the instrumental evidence also suggests that we have successfully corrected for the measurement error in the proxy for
Protestantism.
Table 4 about here
27
V. TESTING THE “PROTESTANT ETHIC” - DID THE CALVINIST COUNTIES GROW FASTER?
Given the positive effect of Protestantism on economic prosperity, we now turn to examine
the effect of individual denominations on economic prosperity. A distinct advantage of
exploiting the variation in the effect of individual denominations on economic prosperity is
that it allows us to identify the precise channel through which Protestantism had an effect
on economic outcome - specifically whether it was through Weber’s “Protestant ethic” or
some other channels such as human capital (e.g., Becker and Woessmann, 2008).44
V.A. Protestant Ethic as a Possible Channel
If the effect of Protestantism on economic prosperity comes from the “Protestant ethic”,
we should expect counties dominated by the Calvinists to achieve the fastest rate of urbanization, simply because of its distinctly strong emphasis on “double predestination”
or unconditional election to both salvation and damnation - a doctrine that bears the
closest resemblance to Weber’s “Protestant ethic”. Stated more simply, while the Calvinists believe they are God’s chosen ones, they consider it an imperative to attain “worldly
achievements” in order to “get rid of the fear of damnation” (Weber, 1930, p.115).45
Unlike the Calvinists, however, followers of the other four denominations do not view
worldly achievements such as business successes as carrying the same importance. In
the instance of Lutheranism (Engelder, 1934) and Anglicanism (Cross and Livingstone,
1997),46 for instance, who subscribe to the belief in unconditional election to salvation
only, they are not driven by the “fear of damnation”. The same applies to the Armini44
To our knowledge, few empirical studies have focused on the effect of Protestant denominations
on economic prosperity. An exception is Cantoni (2012), who finds no significant difference between
Lutheranism and Calvinism on economic outcome in Germany.
45
Thus, according to Weber (1930), “(F)or even the elect remain beings of the flesh, and everything
they do falls infinitely short of the divine standards, nevertheless, they are indispensable as a sign of
election. They are the technical means, not of purchasing salvation, but of getting rid of the fear of
damnation” (p. 115).
46
According to Article 17 of the Thirty-Nine Articles of Religion established in 1563, the historic
defining statements of Anglican doctrine, which refers to the predestination to life, the doctrine of
double predestination is rejected (Cross and Livingstone, 1997).
28
ans;47 comprised chiefly of the Wesleyans and the Baptists (the Methodists), they believe
that even election to salvation is not unconditional but has to be premised on foreseen
faith or unbelief (Lange, 2006). In sum, with the exception of Calvinism, believers of all
other Protestant denominations do not subscribe to the belief that the Protestant ethic
is a condition specifically required for ridding themselves of the fear of damnation.
Hence, if Weber was indeed correct with his interpretations of the various doctrines of
Predestination associated with the different Protestant denominations, then Calvinism
should have a more dominant effect than the rest of the denominations without double predestinations on the economic behavior of those who subscribe to this particular
Protestant doctrine. More specifically, if the Protestant ethic indeed worked in China,
then Calvinism - the reformed churches and Presbyterianism - should have a distinctly
significant effect on economic prosperity than either Lutheranism or the Arminian-based
denominations such as the Wesleyans and the Baptists.48
V.B. Empirical Results
As mentioned in the beginning we divide the missionaries into six denominations: the
China Inland Mission (CIM), the Lutherans, the Calvinists, the Baptists, the Wesleyans
and “Others” (mainly the Anglicans and other non-sectarian missionaries). To test the
effect of each denomination on economic prosperity we regress urbanization on the share
of a denomination and its interaction with the level of Protestantism, as specified in
Equation (8):
y = (α0 +
5
P
ρi di +
i=1
5
P
= α0 P +
3
P
j=1
3
P
ρi Di +
i=1
5
P
λj hj ) ∗ P + W β +
3
P
ρ̃i di +
i=1
λj Hj + W γ +
j=1
5
P
ρ̃i di +
i=1
47
λ̃j hj + ε
j=1
3
P
(8)
λ̃j hj + v
j=1
Arminianism is a school of soteriological thought within Protestant Christianity based on the theological ideas of the Dutch Reformed theologian Jacobus Arminius (Wikipedia).
48
While some Baptists - especially the Particular Baptists and Reformed Baptists - advocate Calvinism
(Dongell and Walls, 2004), those surveyed by Stauffer (1922) were all either the General Baptists and
Free Will Baptists, both of whom advocate Arminianism.
29
in which di is the share of the ith denomination in a county; specifically, d1 stands for
the CIM, d2 the Lutherans, d3 the Calvinists, d4 the Baptists, and d5 the Wesleyans. In
addition, we also control for the country from which the missionaries came (in terms of
share) (hj ). These countries include the United Kingdom, the United States of America,
Europe, with other international denominations serving as the reference group.
The coefficients that we are interested in are ρ1 , ρ2 , ρ3 , ρ4 and ρ5 , which represent the
additional effects of the five denominations in comparison with the reference group - the
“Others” category (mainly the Anglicans and other non-sectarian missionaries). According to the “Protestant ethic” hypothesis, ρ3 (namely Calvinism) should be significantly
positive relative to both the reference group and the other denominations. Specifically,
we expect ρ3 , ρ3 − ρ1 , ρ3 − ρ2 , ρ3 − ρ4 and ρ3 − ρ5 to be significant and positive.
We then estimate the difference between pairs of denominations with respect to their
effects on economic prosperity based on Equation (1), and report the results graphically in
Figure 6. While Calvinism does have a significantly larger effect on economic prosperity
than both Lutheranism (ρ3 − ρ2 ) (0.36 at the 1 percent level of significance) and the CIM
(ρ3 − ρ1 ) (0.48, also at the 1 percent level of significance),49 thereby lending support to
the “Protestant ethic” hypothesis, it enjoys no significant additional effect over those of
the Baptists, the Wesleyans and the “Others”. Hence Weber’s thesis is disconfirmed. In
short, the results are mixed.
Figure 6 about here
49
The Lutherans and the CIM were outperformed not only by the Calvinists but also by the rest of
the denominations. For instance, the effect of Lutheranism is smaller than the Baptists by 0.51 at the
1 percent level of significance; it is also smaller than the Wesleyans by 0.37 at the 5 percent level of
significance. Moreover, the respective effects of the Lutherans and the CIM on urbanization are even
smaller than that of the Anglicans and other non-sectarian missions. The fact that the Lutherans and
the CIM were the two denominations with the smallest effect on urbanization is well borne out in Figure
3, which shows that they had indeed invested the least in both middle schools (panel B) and hospitals
(panel C).
30
VI. KNOWLEDGE DIFFUSION - CAN DENOMINATIONAL DIFFERENCES
BE EXPLAINED BY THE DIFFUSION OF WESTERN KNOWLEDGE?
VI.A. Knowledge Diffusion
Although we’ve established that Protestantism has had a significant effect on economic
prosperity, we still have not ascertained whether economic growth may come from channel(s) other than the “Protestant ethic”. Jonathan Spence’s (1990) neatly summarizes a
whole range of possibilities beyond that of the “Protestant ethic”: “Through their texts,
their presses, their schools, and their hospitals, the efforts of missionaries affected Chinese
thought and practice. The strength of that influence is impossible to calculate, but the
missionaries did offer the Chinese a new range of options, a new way of looking at the
world” (p. 208). It is to these other possible channels that we now turn.
Indeed, the introduction of Western science and a variety of applied disciplines into
China took on disproportionate importance when placed in the context of China’s antiquated educational system. For many centuries - beginning from as early as the Han
Dynasty and through Ming times - the Chinese had placed huge emphasis on reciting and
memorizing the classic Chinese texts which had little, if any, relevance for both scientific inquiry and the practical administrations of government. Articulated by Alexander
Williamson in an 1857 publication (of what was a common theme among both missionaries and reformers), he noted that while the Chinese were very skilled, they wasted
time on civil examination essays (Elman, 2006, p. 117; see also Huff, 2003; Lin, 1995).50
The expansion of a missionary education system was undergirded by the introduction of
a new curriculum of which Mathematics, Astronomy, English, History, and Geography
(in addition to Chinese) were made crucial, all of which were intentionally tied to some
putative flaw in the Chinese education system. For example, the study of astronomy
could allegedly help dispel and correct many superstitious beliefs of the Chinese, whereas
50
Not surprisingly, Williamson was noted to have also “disdained Chinese literature and poetry” (Elman, 2006, p. 117). Under the classical Chinese education system, students were expected to memorize
the sounds and characters in the classical texts without knowing what they meant. As such, the system
did not reward independent thinking or creativity. Because such system provided the state with an ample pool of potential civil servants (tested by the Imperial Civil Service Examinations), the traditional
education in China was aimed at securing “the repose of the state”, according to Alice Gregg (1946, p.
6). Elman (2000) provides an erudite analysis of the classical Chinese examination system.
31
mathematics taught “the faculties of reasoning and analysis in which the Chinese are so
deficient” (Graham, 1995, p. 42).51 By demonstrating the superiority of Western science
and culture, the missionaries hoped that the Chinese would open up (or would be more
open) to Christian teachings. We, however, are more concerned with how this uncanny
diffusion of knowledge at the school level may very likely have produced an educational
outcome that was far superior in economic terms.
Although the Chinese led in earlier scientific discovery, they had fallen vastly behind
for many centuries (Needham, 1954). And although the Jesuits attempted to introduce
Western science to the Chinese in the seventeenth century, their compromising attempt at
“fusing Chinese science with universal world science” had limited effects (Elman, 2005).
By funding the translations and publications of Western science in Chinese and the
erection of schools that introduced and disseminated it, the Protestant missionaries played
a singularly important role in the “introduction of modern science” to China from around
the 1850s (Elman, 2006: 101), a time when China badly needed such knowledge for
advancing its economy.
In addition to the positive role of education, the practice of Western medicine and the
diffusion of such knowledge to a wider populace may also likely result in better health
conditions and accordingly greater economic prosperity. Mokyr (2002, p. 167), for example, shows how an increase in household knowledge of medicine and health had effectively
prevented infectious diseases from driving up mortality rates in the industrialized West
after 1870, long before effective cures to such diseases came about (only after 1945). Likewise, by analyzing a unique dataset on the human resources of a Chinese corporation in
the nineteenth century, Yuchtman (2009) links the superior labor market outcome of the
group of employees to the Western curriculum that they received in school.52
51
Similarly, history was taught to show the Chinese students “what has been and is now her (China’s)
real position amongst the nations of the earth”, whereas geography was to show “what a small space
China occupied on the map of the world and what other countries have achieved in comparison” (Graham,
1995, p. 42).
52
There is some support of this claim from the Chinese officials, who were trying to develop a new
“Examination Scheme” in Western subjects throughout the winter of 1899-1900, amidst the growing
persecution of Christians in the North, “so that successful candidates might possess certificates having
‘some recognized value’ ” (Gregg, 1946, p. 25). If Yuchtman’s (2009) finding is reliable, the value of a
Western curriculum would be recognized not only on paper, but also in reality.
32
The crux of all these is whether the great variety of new knowledge that was not only
spread through texts but also diffused in the everyday context of schools and hospitals
served to enhance economic efficiency or productivity in the Chinese context. Yuchtman’s
(2009) preliminary evidence tends toward the affirmative. Mokyr (2005), who stresses
the efficiency implications of knowledge, also tends to endorse the kind of linkages that
we are attempting to establish in this context.53 So do those who are able to substantiate
the view that (the inseparable) knowledge and technological progress matter for economic
growth (Aghion and Howitt, 1992; Arrow, 1962; Grossman and Helpman, 1991; Nelson
and Phelps, 1966; Romer, 1987, 1990; Solow, 1956; Uzawa, 1965). Thus, if knowledge
diffusion was the real, underlying channel through which the significant effect of Protestantism impacts economic outcome, we should expect denominations with significantly
greater involvement in schools and hospitals to have a larger effect on economic prosperity. Alternatively stated, by controlling for knowledge diffusion activities, the difference
in the effect previously found between pairs of denominations on economic prosperity
should disappear. This is precisely what we hypothesize and test with empirical data in
the sub-section that follows.
VI.B. Empirical Results
In order to test the above hypothesis it is necessary for us to control for knowledge
diffusion activities in education and health. For our purpose we focus on two kinds of
knowledge diffusion activities: the construction of schools and the building of hospitals.
In the case of education we generate a dummy variable each for five different kinds of
schools at the county level. They are the lower primary schools (K1 ), the higher primary
schools (K2 ), the middle schools (K3 ), the Bible schools (K4 ) and the teacher-training
schools (K5 ). In the case of health we similarly enumerate the number of hospitals for each
county (K6 ). Altogether we include these six variables and their respective interaction
53
“Technology is knowledge. Knowledge, as is well known, has always been a difficult concept for
standard economics to handle. It is at the core of modern economic growth” (Mokyr, 2005: 119).
33
terms with Protestantism in our estimation based on the following specification:
y = (α0 +
5
3
6
5
3
6
X
X
X
X
X
X
ρ i di +
λj hj +
κn Kn ) ∗ P + W β +
ρ̃i di +
λ̃j hj +
κ̃n Kn + ε (9)
i=1
j=1
n=1
i=1
j=1
n=1
The difference between Equations (8) and (9) is that the latter includes knowledge
diffusion and its interaction term with Protestantism. Similar to what we did in Figure
6, in Figure 7 we compare the denomination-specific effects of Protestantism on economic
prosperity. After controlling for knowledge diffusion we find that there is no significant
difference in the effects of Protestantism on economic prosperity between pairs of denominations. Recall earlier that, in the absence of such control six out of 15 pairs of
denominations are significantly different at the 1 percent level, and two are significantly
different at the 5 percent level. For instance, previously Calvinism had a significantly
larger effect than Lutheranism (ρ3 − ρ2 ) at the 1 percent level of significance, but after
controlling for knowledge diffusion the difference decreases precipitously from 0.36 to 0.04
and is no longer significant. These results squarely suggest that the difference in effect
among pairs of denominations on economic prosperity was basically caused by differences
in knowledge diffusion than in the “Protestant ethic”.
Figure 7 about here
Moreover, the argument that knowledge diffusion was the primary determinant of economic prosperity is further supported by the positive and significant relationship between
both the intensity and kind of knowledge diffusion and the overall effect of Protestantism.
Panel A of Figure 8, which summarizes the coefficients of the interaction terms between
Protestantism and the various components of knowledge diffusion, clearly shows that the
effect of Protestantism is significantly positive only in counties with a strong presence of
middle schools (κ3 ) but not of either higher primary schools (κ2 ) or lower primary schools
(κ1 ). These differences between the three levels of schools are indeed consistent with our
“knowledge diffusion” hypothesis. While the elementary schools, which consisted of both
lower and higher primary schools, now required a sound training in the fundamental tools
34
of education, including reading, writing, and the elementary operations in arithmetic,54
it was the middle schools, with their newly invigorated curriculum which incorporated a
wide array of Western subjects ranging from Algebra, Astronomy and Physics to English,
History, and Geography, that provided the source of what Simon Kuznets (1965) called
“useful knowledge” in the context of economic development.
Similarly there is evidence to support the claim that the kind of knowledge transmitted, in addition to the intensity of knowledge diffusion, matters. Indeed, what Panel A
of Figure 8 also shows is that, unlike the Bible schools (κ4 ), which significantly decrease
the effect of Protestantism on economic prosperity, the teacher-training schools (κ5 ) have
a positive and significant effect. This statistical difference can be accounted for by the
radically different goals between the two levels of schools in terms of human capital training. Whereas the specific goal of the Bible schools was to produce students who would
become devout Christians dedicated to undertaking the missionary duties, the aim of the
teacher-training schools was to train qualified teachers for both middle and elementary
schools. More specifically, whereas the Bible schools were erected to diffuse religious
knowledge, the teacher-training schools were established to diffuse Western knowledge especially knowledge in the spheres of science and technology - that is distinctly more
practical. Small surprise, therefore, that the effect of Protestantism on economic prosperity is significantly larger in counties with a stronger presence of teacher-training schools
than in counties predominated by the Bible schools. Last but not least, and consistent
with our thesis of knowledge diffusion, is the finding that the effect of Protestantism was
also significantly greater in counties where more hospitals had been built (Panel A, Figure
8).
Figure 8 about here
54
“By the end of the lower primary the child should be able to read with ease and pleasure Mandarin or
the local vernacular, to write ordinary letters and business forms, and to use his knowledge of arithmetic
in the ordinary processes of his daily life. On the completion of the higher primary he should have
acquired a habit of and love for reading” (Committee of Reference and Counsel of the Foreign Missions
Conference of North America, 1922, p. 75). In addition to reading, writing and arithmetic, whose
occupational values were now emphasized, the higher primary school also offered some occupational
training (such as wood-work, metal-work, leather-work, and designing).
35
VI.C. Estimating Knowledge Diffusion Activities in China
Although we have included in our estimations indices of knowledge diffusion and their
interactions with Protestantism, our estimations may still be biased if the level of knowledge diffusion activities was selected a priori by the missionaries based upon a county’s
socioeconomic characteristics. For instance, a missionary society would likely have built
more primary schools in a poor county than in one that was economically more advanced.
To address this concern we construct another dataset that allows us to employ the information on the kind and intensity of knowledge diffusion activities undertaken by the
same missionary societies in other parts of the world to estimate the magnitude of their
commitments in China, based on the assumption that their behavior and strategy are
likely highly correlated across countries.
In this context, it is fortunate that we are able to construct a dataset based on
Beach’s (1903) A Geography and Atlas of the Protestant Missions, a compendium that
contains detailed information on the size of each missionary society (people) and its
involvement (in terms of number) in three types of knowledge diffusion activities, viz.
the day (religious) school, the higher education institution and hospital in 19 countries
(or regions, see Appendix V) in the world.
The first step is to normalize a mission’s (n) involvement in each of these activities in
a country or region j by dividing the pertinent activities (number) with the size of the
n
(the ratio of day schools to the size of
missionary (people). We denote these ratios as k1j
n
missionary society in country or region j), k2j
(the ratio of higher education institutions
n
to the size of missionary society in country or region j), and k3j
(the ratio of hospitals to
the size of missionary society in country or region j), respectively, we then calculate the
average of a mission’s involvement in these three activities in all the countries or regions,
19
P
n
and denote these averages as k̄1n , k̄2n , and k̄3n , respectively. Specifically, k̄1n =
k1j
,
j=1
k̄2n =
19
P
n
k2j
, and k̄3n =
j=1
19
P
n
k3j
.
j=1
Given that our unit of analysis is the county, we sum up the averages of a given activity
(e.g. day school (K̃1 )) from all the missionary societies to estimate the overall supply of
36
the same activity by missionary bodies in a Chinese county.55 For instance, there are λ
λ
P
missionary societies in one county, then K̃1 =
k̄1n . We repeat the same procedure for
n=1
other knowledge diffusion activities - in this case the higher education institutions (K̃2 )
and hospitals (K̃3 ). Doing so allows us to obtain three indices of knowledge diffusion
activities (K˜m , m = 1,2,3) from which to estimate the same activities for each and every
Chinese county.
Similar to Equation (9), we control for the three separate indices of knowledge diffusion
and their interactions with Protestantism in the following specification:
y = (α0 +
5
3
3
5
3
3
X
X
X
X
X
X
ρi di + λj hj +
τm K̃m )∗P +W β + ρ̃i di + λ̃j hj +
τ̃m K̃m +ε (10)
i=1
j=1
m=1
i=1
j=1
m=1
The results thus obtained are consistent with our hypothesis regarding the effects of
knowledge diffusion on the Protestantism’s effect on economic prosperity as well as the
earlier findings based on Equation (9). For convenience we plot only the pertinent coefficients (τ1 , τ2 and τ3 ) in Panel B of Figure 8. Of foremost interest is the finding that the day
schools are negatively correlated with the effect of Protestantism on economic prosperity, whereas the higher education institutions have a significantly positive effect. These
findings are consistent with our expectations regarding the effect of educational content
on economic outcome. Although the day schools now emphasized reading and writing
and the “rudiments of Western learning”, the Bible remained the “staple of instruction”
(Beach, 1901, p. 279). By contrast, higher education institutions provided more “useful knowledge” (Kuznets, 1965) than did the day schools.56 As expected, hospitals also
significantly increase the effect of Protestantism on economic prosperity.
To see if the effects of different denominations on economic outcome may disappear
55
Based on Stauffer (1922), a list of missionary societies in China is provided in Appendix II, which
is then used to match with the list of missionary societies elsewhere in the world (provided by Beach,
1903).
56
Beach (1903) does not give a precise definition of higher education institutions. The term definitely
includes colleges and universities, as it was suggested that “a Higher Educational Council or Senate
(should be established) in each area in which there is a group of colleges (for coordinating) the work of
all the colleges and universities” (Committee of Reference and Counsel of the Foreign Missions Conference
of North America, 1922, p. 59). Given that higher education was often used to compare and contrast with
lower (elementary) education, we are inclined to think that the term may likely include, in addition to
colleges and universities also the middle schools, teacher-training schools, technical (vocational training)
institutes and the likes.
37
after the inclusion of knowledge diffusion activities we plot the effect of Protestantism on
economic prosperity between pairs of denominations in Figure 9, and find that Calvinism
no longer has any significant additional effect over those of the other denominations,
suggesting once again that the seeming difference in denominational effects on economic
prosperity is caused by knowledge diffusion and not by the “Protestant ethic”.
Figure 9 about here
VII. CONCLUSION
This paper sets out to test the “most famous link” between culture and economic development in a country ruled out by Weber himself for the possible development of a
Western religion and accordingly capitalism at a time most suited for testing this subtle
link. In sharp contrast to Weber’s prediction, we find that Protestantism was in fact able
to thrive on the soil that for thousands of years had been predominated by the ethics of
Confucianism, and, as confirmed by our instrumental evidence, with positive economic
consequences. As many have found, however, the superior performance of the economies
with cultural affinity to Protestantism is actually attributable to a variety of channels
other than the “Protestant ethic”. Our study adds to this list of alternative channels
the diffusion of a completely new paradigm of knowledge embodied in a qualitatively
new curriculum in education, in particular in the middle schools in China. Likewise, the
introduction of Western medicine, reflected in the erection of hospitals and healthcare
provisions, also had a distinctly positive effect on economic prosperity.
A distinct contribution of our study is that we are able to test the “Protestant ethic”
hypothesis by decomposing the overall effect of Protestantism into its denominationspecific constituents and show that while Calvinism dominated the counties where Lutheranism
and the China Inland Mission had a strong presence, it had no distinct significant effect
over counties predominated by the Wesleyans and the Baptists. Moreover, any observed
differences among the denominations on economic outcome can be fully accounted for by
the effect of knowledge diffusion activities in general, and by more advanced education
38
(middle versus elementary schools), specific educational content (religious versus human
capital training), and healthcare (hospital) provisions in particular. It is this latter exercise that allows us not only to confirm the significant and positive relationship between
Protestantism and economic prosperity but also, more importantly, to be more precise
about the channel through which Protestantism actually worked in the historical context
of China.
References
[1] Acemoglu, Daron, Simon Johnson, and James Robinson. 2002. “Reversal of Fortune:
Geography and Institutions in the Making of the Modern World Income Distribution,” Quarterly Journal of Economics, 117 (Nov.): 1231-1294.
[2] Acemoglu, Daron, Simon Johnson, and James Robinson. 2005. “The Rise of Europe:
Atlantic Trade, Institutional Change and Growth,” American Economic Review, 95
(Jun.): 546-579.
[3] Aghion, Philippe and Peter Howitt. 1997. Endogenous Growth Theory. Cambridge,
MA: MIT Press.
[4] Arruada, Benito. 2010. “Protestants and Catholics: Similar Work Ethic, Different
Social Ethic,” Economic Journal, 120 (547), 890-918.
[5] Arrow, Kenneth J. 1962. “The Economic Implications of Learning by Doing,” Review
of Economic Studies, 29 (3): 155-173.
[6] Austin, Alvyn. 2007. China’s Millions: the China Inland Mission and Late Qing
Society, 1832-1905. Grand Rapids, Mich.: William B. Eerdmans Publisher.
[7] Barnett, Suzanne Wilson and John King Fairbanks. 1985. Christianity in China:
Early Protestant Missionary Writings. Cambridge, MA: Committee on AmericanEast Asian Relations, Dept. of History, Harvard University.
[8] Barro, Robert J. 2004. “Spirit of Capitalism,” Harvard International Review, 25(4):
64-66.
[9] Barro, Robert J. and Rachel M. McCleary. 2003. “Religion and Economic Growth
across Countries,” American Sociological Review, 68: 760-781.
[10] Bays, Daniel H. 2012. A New History of Christianity in China. Wiley-Blackwell,
John Wiley & Sons, Inc.
[11] Beach, Harlan P. 1901. A Geography and Atlas of Protestant Missions, Vol I: Geography. New York: Student Volunteer Movement for Foreign Missions.
[12] Beach, Harlan P. 1903. A Geography and Atlas of Protestant Missions, Vol II: Statistics and Atlas. New York: Student Volunteer Movement for Foreign Missions.
39
[13] Becker, Sascha O. and Ludger Woessmann. 2009. “Was Weber Wrong? A Human
Capital Theory of Protestant Economic History,” Quarterly Journal of Economics,
124(2): 531-596.
[14] Becker, Sascha O. and Ludger Woessmann. 2010. “The effect of Protestantism on
education before the industrialization: Evidence from 1816 Prussia,” Economics
Letters, 107(2): 224-228.
[15] Bendix, Reinhard. 1977. Nation-Building and Citizenship. Berkeley: University of
California Press.
[16] Berger, P.L. 1967. The Sacred Canopy: Elements of a Sociological Theory of Religion.
Garden City: Doubleday.
[17] Broomhall, A. J. 1984. Hudson Taylor and China’s Open Century, Book Four: Survivors’ Pact. London: Hodder and Stoughton and Overseas Missionary Fellowship.
[18] Cantoni, Davide. 2010. “The economic effects of the Protestant Reformation: Testing
the Weber hypothesis in the German lands,” working paper, Universitat Pompeu
Fabra.
[19] Chaves, M. 1994. “Secularization as Declining Religious Authority,” Social Forces,
March: 749-774.
[20] CHGIS, Version 4. Cambridge: Harvard Yenching Institute, January 2007.
[21] Cohen, Paul A. 1978. “Christian Missions and their Impact to 1900,” The Cambridge
History of China, Vol. 10, Late Ch’ing, 1800-1911, Part I, ed. By John K. Fairbank.
Cambridge: Cambridge University Press.
[22] Committee of Reference and Counsel of the Foreign Missions Conference of North
America, 1922. Christian Education in China. New York City: Committee of Reference and Counsel of the Foreign Missions Conference of North America, Inc.
[23] Cross, F. L. and E. A. Livingstone. 1997. The Oxford Dictionary of the Christian
Church (3rd ed.). USA: Oxford University Press.
[24] Delacroix, Jacques, and Francois Nielsen. 2001. “The Beloved Myth: Protestantism
and the Rise of Industrial Capitalism in Nineteenth-Century Europe,” Social Forces,
80: 509553.
[25] Dongell, Joseph and Jerry Walls. 2004. Why I Am Not a Calvinist. Downer’s Grove:
Intervarsity Press.
[26] Eisenstein, Elizabeth L. 1979. The Printing Press as an Agent of Change. New York:
Cambridge University Press.
[27] Elman, Benjamin A. 2000. A Cultural History of Civil Examinations in Late Imperial
China. Berkeley,CA: University of California Press.
[28] Elman, Benjamin A., 2005. On their Own Terms: Science in China, 1550-1900. MA:
Harvard University Press.
40
[29] Elman, Benjamin A. 2006. A Cultural History of Modern Science in China. Cambridge, MA: Harvard University Press.
[30] Engelder, T.E.W. 1934. Popular Symbolics. St. Louis: Concordia Publishing House.
[31] Esherick, Joseph. 1987. The Origins of the Boxer Uprising. Berkeley: University of
California Press.
[32] Feuerwerker, Albert. 1958. China’s early industrialization: Sheng Hsuan-huai (18441916) and Mandarin Enterprise. Cambridge, Mass.: Harvard University Press.
[33] Gallego, Francesco and Robert D.Woodberry. 2010. “Christian Missionaries and Education in Former African Colonies: How Competition Mattered,” Journal of African
Economies, 19(3): 294-329.
[34] Glaeser, Edward L., and Spencer Glendon. 1998. “Incentives, Predestination and
Free Will,” Economic Inquiry, 36: 429443.
[35] Glaeser, Edward L., and Sacerdote, Bruce. 2008. “Education and religion,” Journal
of Human Capital, 2(2): 188-215.
[36] Glover, Archibald Edward. 1918. A Thousand Miles of Miracle in China: a Personal
Record of God’s Delivering Power from the Hands of the Imperial Boxers of Shan-si.
Glasgow: Pickering & Inglis.
[37] Graham, Gael. 1995. Gender, culture, and Christianity: American Protestant mission schools in China, 1880-1930. New York: P. Lang
[38] Gregg, Alice Henrietta. 1946. China and educational autonomy: the changing role of
the Protestant educational missionary in China, 1807-1937. NY: Syracuse University
Press.
[39] Grossman, G. and E. Helpman. 1991. Innovation and Growth in the Global Economy.
Cambridge, MA: MIT Press.
[40] Guiso, Luigi, Paola Sapienza, and Luigi Zingales. 2006. “Does Culture Affect Economic Outcomes?” Journal of Economic Perspectives, 20(2): 23-48.
[41] Habermas, Jurgen. 1989. The Structural Transformation of the Public Sphere. Translated by T. Burger and F. Lawrence. Cambridge, MA: MIT press.
[42] Hsu, Immanuel C. Y. 2000. The Rise of Modern China. New York: Oxford University
Press, sixth edition.
[43] Hoffman, Philip. 2006. “The Church in Economy and Society,” in Stewart J. Brown
and Timothy Tackett, eds., the Cambridge of History of Christianity, Vol II: Enlightenment, Reawakening and Revolution 1660-1815. Cambridge: Cambridge University
Press.
[44] Huff, Toby. 2003. The Rise of Early Modern Science: Islam, China, and the West.
Cambridge. UK: Cambridge University Press.
41
[45] Hyatt, Jr., Irwin T. 1966. “Protestant Missions in China, 1877-1890: The Institutionalization of Good Works,” American Missionaries in China: Papers from Harvard Seminars, ed. Kwang-Ching Liu. Cambridge, MA: East Asian Research Center,
Harvard University.
[46] Iannaccone, Laurence R. 1998. “Introduction to the Economics of Religion,” Journal
of Economic Literature, 36(3): 1465-1495.
[47] Iyigun, Murat. 2008. “Luther and Suleyman,” Quarterly Journal of Economics,
November, 123(4): 1465-1494.
[48] Kuznets, Simon. 1965. Economic Growth and Structure: Selected Essays. New York:
Norton.
[49] La Porta, R., F. Lopez-de-Silanes, A. Shleifer, and R.W. Vishny. 1998. “Law and
Finance,” Journal of Political Economy, 106(6).: 1113-1155.
[50] Lange, Lyle W. 2006. God So Loved the Word: A Study of Christian Doctrine.
Milwaukee: Northwestern Publishing House.
[51] Lin, Justin Yifu. 1995. “The Needham Puzzle: Why the Industrial Revolution Did
Not Originate in China,” Economic Development and Cultural Change, 43(2): 269292.
[52] Lipset, S.M. and G.S. Lenz. 2000. “Corruption, Culture, and Markets,” in L.E.
Harrison and S.P. Huntington, eds., Culture Matters: How Values Shape Human
Progress. New York: Basic Books.
[53] Luo, Zhongping. 1955. Selected Statistic Materials of Modern China Economy (Zhong
Guo Jin Dai Jing Ji Shi Tong Ji Zi Liao Xuan Ji ). Beijing: Ke Xue Presses.
[54] Lutz, Jessie G. 2001. “China and Protestantism: Historical Perspectives, 1807-1949,”
in Stephen Uhalley, Jr. and Xiaoxin Wu, eds., China and Christianity: Burdened
Past, Hopeful Future. Armonk, N.Y.: M.E. Sharpe.
[55] McCleary, Rachel M. and Robert J. Barro. 2006a. “Religion and Economy,” Journal
of Economic Perspective, 20(2): 49-72.
[56] McCleary, Rachel M. and Robert J. Barro. 2006b. “Religion and Political Economy
in an International Panel,” Journal for the Scientific Study of Religion, 45(2): 14975.
[57] Mokyr, Joel. 2002. The Gifts of Athena: Historical Origins of the Knowledge Economy. Princeton: Princeton University Press.
[58] Mokyr, Joel. 2005. “Long-term Economic Growth and the History of Technology,”
in Philippe Aghion and Steven N. Durlauf, eds., Handbook of Economic Growth,
vol.1B. Amsterdam: Elsevier.
[59] Nelson, Richard R. and Edmund S. Phelps. 1966. “Investment in Humans, Technological Diffusion, and Economic Growth,” American Economic Review, 56(1/2):
69-75.
42
[60] Needham, Joseph. 1954. Science and Civilization in China. Cambridge: Cambridge
University Press.
[61] Putnam, Robert. 2000. Bowling Alone: The Collapse and Revival of American Community. New York: Simon and Schuster.
[62] Romer, Paul M. 1987. “Increasing Returns and Long-Run Growth,” Journal of Political Economy, 94: 1002-1037.
[63] Romer, Paul M. 1990. “Capital, Labor, and Productivity,” Brooking Papers Economic Activity, 1990: 337-367.
[64] Rozman, Gilbert. 1973. Urban Networks in Ch’ing China and Tokugawa Japan.
Princeton, N.J.: Princeton University Press.
[65] Sachs, Jeffrey D. 2000. “Notes on a New Sociology of Economic Development,” in
Lawrence E. Harrison and Samuel P. Huntington, eds., Culture Matters: How Values
Shape Human Progress. New York: Basic Books.
[66] Samuelsson, Kurt. 1993. Religion and Economic Action: The protestant Ethic, the
Rise of Capitalism, and the Abuses of Scholarship. Toronto: U. Toronto Press.
[67] Saunders, Alexander R. 1900. In Weariness and Painfulness: Being the Account by
Rev. A.R. Saunders of the Flight from Ping-iao to Han-kow of Himself and Party.
London : China Inland Mission, Newington Green.
[68] Solow, Robert M. 1956. “A Contribution to the Theory of Economic Growth,” Quarterly Journal of Economics, 70: 65-94.
[69] Song, Jiaheng and Wei Li. 1995. Canadian Missionaries in China. Dongfang Publishing House.
[70] Spence, Jonathan D. 1990. The Search for modern China. New York: Norton.
[71] Stauffer, Milton T. 1922. The Christian occupation of China: a general survey of the
numerical strength and geographical distribution of the Christian forces in China.
Shanghai: China Continuation Committee.
[72] Stulz, R.M. and R. Williamson. 2001. “Culture, Openness and Finance,” NBER
working paper no. 8222, April.
[73] Uzawa, H. 1965. “Optimum technical change in an aggregative model of economic
growth,” International Economic Review, 61:1831.
[74] Weber, Max. 1930. The Protestant Ethic and the Spirit of Capitalism, translated by
Talcott Parsons. London, Allen & Unwin.
[75] Weber, Max. 1964. The Religion of China. Translated and edited by H. H. Gerth.
New York: Collier and Macmillan.
43
[76] Wiest, Jean-Paul. 2001. “From Past Contributions to Present Opportunities: The
Catholic Church and Education in Chinese Mainland during the Last 150 Years,” In
Stephen Uhalley, Jr. and Xiaoxin Wu, eds., China and Christianity: Burdened Past,
Hopeful Future. Armonk, N.Y.: M.E. Sharpe.
[77] Wilson, B.1966. Religion in Secular Society: A Sociological Comment. London:
Watts.
[78] Woodberry, Robert D. 2006. “Reclaiming the M-Word: The Consequences of Missions for Nonwestern Societies,” the Review of Faith and International Affairs, 4(1):
3-12.
[79] Woodberry, Robert D. 2007a. “Missions and Economics,” Encyclopedia of Missions
and Missionaries. Jonathan J. Bonk (ed.). New York: Routledge.
[80] Woodberry, Robert D. 2007b. “The Social Impact of Missionary Higher Education.”
pp. 99-120 in Christian Responses to Asian Challenges: A Globalization View on
Christian Higher Education in East Asia. Philip Yuen Sang Leung and Peter Tze
Ming Ng (eds.). Hong Kong: Centre for the Study of Religion and Chinese Society,
Chung Chi College, Chinese University of Hong Kong.
[81] Woodberry, Robert D. 2010a. “Religious Roots of Economic and Political Institutions,” Forthcoming in Oxford Handbook of the Economics of Religion. Rachel
McCleary (ed.). Oxford: Oxford University Press
[82] Woodberry, Robert D. 2010b. “Weber through the Back Door: Protestant Competition, Elite Dispersion and the Global Spread of Democracy,” Project on Religion
and Economic Change Working Paper #002.
[83] Yang, Senfu. 1968. China History of Christianity (Zhongguo Jidujiao Shi). Taibei:
Taiwan Shang Wu Press.
[84] Yuchtman, Noam. 2009. “Teaching to the Tests: An Economic Analysis of Educational Institutions in Late Imperial and Republican China,” Harvard University,
Working paper.
44
Figure 1: The Development of Protestantism in China
100.00
Percentage of counties with missionary work
90.00
80.00
70.00
60.00
50.00
40.00
30.00
20.00
10.00
0.00
1840 1850 1855 1860 1865 1868 1871 1874 1877 1880 1883 1886 1889 1892 1895 1898 1901 1904 1907 1910 1913 1916 1919 Year
Source: The Christian occupation of China: a general survey of the numerical strength and geographical distribution of the Christian forces in
China.
45
Figure 2: The Presence of Missionaries in China, by County
Panel A: The Distribution of the Number of Missionaries, by County
800
Number of counties
700
600
500
400
300
200
100
0
0
1
2
3
4
5
6
7
8
9
11
25
Number of missionaries
Panel B: Counties with Missionary Presence, by Denomination (number and percentage)
50
35.00
45
Number
35
25.00
30
20.00
25
20
15.00
15
10.00
10
Percentage
30.00
40
5.00
5
0
0.00
CIM
Lutheranism
Calvinism
Baptist
Wesleyanism
Others
Denomination
Number of missionaries
Percentage of counties with missionary presence
46
Figure 3: The Diffusion of Western Knowledge into China
Panel A: The Distribution of Various Kinds of Schools, by County
Number
800
700
600
500
400
300
200
100
0
Lower Primary Higher Primary
School
School
Middle School
Bible School
Teacher-training
School
Panel B: The Distribution of Middle Schools, by Denomination
140
120
Number
100
80
60
40
20
0
Erected by
CIM
Lutherans
Calvinists
Baptists
Wesleyans
Others
Wesleyans
Others
Panel C: The Distribution of Hospitals, by Denomination
70
60
Number
50
40
30
20
10
0
Erected by CIM
Lutherans
Calvinists
47
Baptists
Figure 4: The Sample Counties
48
Figure 5: Boxer Uprising - A Safe (South) China versus an Unsafe (North) China
49
Figure 6: The Effect of Protestantism on Economic Prosperity, by Denomination Pair (Baseline Results)
1.20
1.00
0.80
0.64***
0.60
0.51***
0.49***
0.48***
0.40
0.37**
0.36**
0.20
-0.60
95% Confidence Interval
Notes: (1) with 95% confidence interval. (2) Data source: Appendix VI
50
vs
.
Ba
pt
ist
ism
an
i sm
W
es
le
y
vs
.
Ca
l
vs
.
The effect of denominations on urbanization
W
es
le
ya
ni
sm
ist
Ba
pt
-0.80
-1.00
Ca
lv
in
vi
ni
sm
ism
ut
he
ra
n
vs
.L
ut
he
ra
n
ist
.L
vs
vs
.L
ut
he
Ba
pt
W
es
le
ya
ni
sm
-0.57***
ism
ra
ni
sm
vs
.C
IM
in
ism
Ca
lv
W
es
le
ya
ni
sm
vs
.
CI
M
IM
Ba
pt
ist
vs
.C
Ca
lv
in
ism
vs
er
an
i
Lu
th
le
ya
W
es
sm
vs
.
ni
sm
vs
.
ist
Ba
pt
-0.45***
.C
IM
s
O
th
er
O
th
er
s
rs
th
e
vs
Ca
lv
Lu
th
-0.40
in
ism
sm
er
an
i
M
CI
vs
vs
.O
-0.20
.O
th
e
.O
th
er
s
rs
0.00
Figure 7: The Effect of Protestantism on Economic Prosperity, Controlling for Knowledge Diffusion, by Denomination Pair
0.80
0.64***
0.60
0.51***
0.49***
0.48***
0.40
0.36**
0.37**
0.20
-0.60
Ba
pt
ist
v
Ca
l
vin
is
m
vs
.
m
vs
.
C
Lu
th
er
an
i
IM
s.
CI
M
W
es
le
ya
ni
s
Ba
pt
ist
v
vin
is
m
vs
.
C
IM
vs
.C
IM
Ca
l
m
O
th
er
s
Lu
th
er
an
is
W
es
le
ya
ni
s
m
vs
.
s.
O
th
er
s
O
th
er
s
Ba
pt
ist
v
m
vin
is
Ca
l
-0.45***
sm
s
.L
W
es
ut
he
le
ya
ra
ni
ni
sm
sm
vs
.L
ut
he
ra
ni
Ba
sm
pt
ist
vs
.C
W
es
al
vi
le
ni
ya
sm
ni
sm
vs
.C
W
al
vin
es
le
is
ya
m
ni
sm
vs
.B
ap
t is
t
-0.40
vs
.
vs
.O
th
er
s
m
-0.20
Lu
th
er
an
is
CI
M
vs
.O
th
er
s
0.00
-0.57***
-0.80
Without controlling for knowledge diffusion
Controlling for knowledge diffusion
Notes: (1) with 95% confidence interval. (2) Data source: Appendix VI
51
Figure 8: The Effects of Knowledge Diffusion on Protestantism’s Effect on Economic
Prosperity
Panel A: The Effects of Education and Health on Protestantism’s Effect on Economic
Prosperity
0.40
0.30
0.271*
0.236***
0.20
0.142***
0.10
0.012
0.00
-0.10
-0.048
Lower
Primary
School
Higher
Primary
School
Middle
School
Bible School
Teachertraining
School
Hospital
-0.20
-0.30
-0.338***
-0.40
Panel B: The Effects of Estimated Education and Health on Protestantism’s Effect on
Economic Prosperity
0.08
0.072***
0.07
0.06
0.047***
0.05
0.04
0.03
0.02
0.01
-0.003***
0.00
Day-school
Higher
Educational
Institution
-0.01
Data source: Appendix VI
52
Hospital
Figure 9: The Effect of Protestantism on Economic Prosperity, Controlling for Estimated Knowledge Diffusion, by Denomination Pair
0.80
0.64***
0.60
0.48***
0.51***
0.49***
0.40
0.36**
0.37**
0.28*
0.21**
0.20
-0.60
s.
W
O
es
th
le
er
ya
s
ni
sm
vs
.O
th
Lu
er
th
s
er
an
is
m
vs
.C
IM
C
al
vi
ni
sm
vs
.C
IM
Ba
pt
is
tv
s.
W
C
es
IM
le
ya
ni
sm
C
al
vs
vi
.C
ni
sm
IM
vs
.L
ut
he
Ba
ra
ni
pt
sm
is
tv
s
.L
W
es
ut
he
le
ya
ra
ni
ni
sm
sm
vs
.L
ut
he
ra
ni
Ba
sm
pt
is
tv
s.
W
C
es
al
vi
le
ni
ya
sm
ni
sm
vs
.C
al
W
vi
es
ni
le
sm
ya
ni
sm
vs
.B
ap
tis
t
th
er
s
is
m
vi
n
C
al
-0.40
Ba
pt
is
tv
vs
.
O
vs
.O
m
ra
ni
s
-0.20
Lu
th
e
C
IM
vs
.O
th
er
s
th
er
s
0.00
-0.45***
-0.57***
-0.80
Without controlling for knowledge diffusion
Controlling for knowledge diffusion
Data source: Appendix VI
53
Controlling for estimated knowledge diffusion
Table 1: Definition of Variables and Data Sources
S.D.
Data
Observation Mean
Sources
Share of urban population in
A
Urbanization
total population
1175
4.309 (12.666)
Protestantism Total Communicants/population
A
1175
8.394 (15.161)
Total Christian
A
constituencies/population
1175
13.790 (26.271)
Knowledge
Lower primary school
A
1175
0.612
(0.488)
diffusion
Higher primary school
A
1175
0.267
(0.443)
Middle school
A
1175
0.091
(0.288)
Bible school
A
1175
0.051
(0.220)
Teachers school
A
1175
0.018
(0.133)
Hospital
A
1175
0.187
(0.526)
Share of
China Inland Mission
A
1175
0.244
(0.401)
Lutheranism
A
1175
0.081
(0.250)
Calvinism
A
1175
0.230
(0.377)
Wesleyans
A
1175
0.102
(0.253)
Baptist
A
1175
0.106
(0.265)
Share of
from England
A
1175
0.183
(0.340)
missionary
from USA
A
1175
0.430
(0.433)
from European continent
A
1175
0.124
(0.308)
Roman
Number of Catholic churches per A
Catholicism
1,000,000 population
1175
0.630
(1.269)
Western
D
economic
Treaty ports
influence
1175
0.179
(0.383)
Railway
D
1175
0.345
(0.475)
Initial economic
C
1175
Small city (1840s)
conditions
0.031
(0.175)
Middle city (1840s)
C
1175
0.022
(0.147)
Large city (1840s)
C
1175
0.007
(0.082)
Geographic
B
Prefectural government location
factors
1175
0.150
(0.357)
Grand Canal
A
1175
0.037
(0.188)
Changjiang (Yangzi) River
A
1175
0.048
(0.213)
Coast
A
1175
0.108
(0.311)
Population density (1850s)
A, B
1175
189.399 (211.683)
Size
B
1175
1914.911 (1218.037)
Boxer Uprising Incidence of Boxer Uprising
B
1175
0.195
(0.396)
Distance to Boxer Uprising
B
1175
0.341
(0.474)
A: Stauffer, Milton T. 1922. The Christian Occupation of China: A General Survey of the
Numerical Strength and Geographical Distribution of the Christian Forces in China.
B: "CHGIS, Version 4" Cambridge: Harvard Yenching Institute, January 2007
C: Rozman, Gilbert. 1973. Urban Networks in Ch’ing China and Tokugawa Japan.
D: Luo, Zhongping. 1955. Selected Statistical Materials of Modern China Economy.
Variables
Variables Definition
54
Table 2: The Effects of Protestantism on Urbanization
(1)
Protestantism
0.215***
(0.055)
Catholicism
0.637**
(0.307)
Initial economic conditions:
Small city
(2)
0.195***
(0.052)
-0.237
(0.273)
(3)
0.162***
(0.050)
0.009
(0.250)
6.491**
(3.069)
18.466***
(4.222)
44.449***
(7.711)
(4)
0.187***
(0.052)
-0.181
(0.273)
(5)
0.159***
(0.050)
0.014
(0.250)
6.260*
(3.233)
Middle city
18.370***
(4.349)
Large city
43.670***
(7.801)
Treaty ports
2.791**
1.582
(1.099)
(0.993)
Railway
0.855
0.245
(0.827)
(0.762)
Prefectural government
10.348***
4.810***
10.521***
5.061***
(1.448)
(1.416)
(1.437)
(1.436)
location
Grand Canal
8.864***
5.796**
8.909***
5.895**
(3.169)
(2.388)
(3.120)
(2.377)
Changjiang (Yangzi) River
2.074
-0.529
0.917
-0.896
(1.822)
(1.917)
(1.812)
(1.841)
Coast
2.561*
2.261*
1.389
1.779
(1.469)
(1.302)
(1.346)
(1.184)
Population density (1850s)
4.493***
2.648***
1.329**
2.518***
1.320**
(0.725)
(0.594)
(0.524)
(0.590)
(0.523)
Size
2.347***
0.391
-0.291
0.448
-0.221
(0.730)
(0.636)
(0.581)
(0.643)
(0.589)
Province dummies
Yes
Yes
Yes
Yes
Yes
1175
1175
1175
1175
1175
Observations
R-squared
0.15
0.24
0.34
0.25
0.34
Note: Robust standard errors in parentheses. * significant at 10%; ** significant at 5%; *** significant at 1%. Constant terms are not reported.
55
Table 3: Validity of Instrument
Distance to Boxer Uprising
* safety region
Prefectural government location
Grand Canal
Changjiang (Yangzi) River
Coast
Population density (1850s)
Size
Control variables in Equation (5)
(1)
Communicants
/population
6.386***
(1.778)
3.916***
(1.073)
0.239
(1.698)
1.102
(1.681)
4.380*
(2.473)
-1.773
(1.144)
-2.404**
(1.002)
Yes
(2)
Constituency
/population
8.597***
(2.739)
(3)
Catholicism
-0.044
(0.167)
-0.029
(0.042)
0.039
(0.061)
(6)
Urbanization
(1840s)
-0.053
(0.087)
7.533***
(1.996)
-0.064
(3.510)
3.319
(3.856)
8.004*
(4.427)
-2.636
(1.978)
-4.699***
(1.799)
0.981***
(0.154)
0.197
(0.121)
-0.122
(0.174)
0.059
(0.088)
-0.103
(0.064)
-0.106
(0.065)
-0.050*
(0.028)
-0.056
(0.065)
0.395***
(0.072)
0.331***
(0.058)
0.037**
(0.016)
0.002
(0.019)
-0.052
(0.032)
0.113
(0.072)
0.059
(0.067)
0.094*
(0.050)
0.078***
(0.016)
-0.060***
(0.022)
0.568***
(0.057)
0.274***
(0.100)
0.229***
(0.084)
0.046
(0.048)
0.113***
(0.022)
0.068***
(0.016)
Yes
Yes
(4)
Treaty ports
Yes
(5)
Railway
Yes
Yes
Province dummies
Yes
Yes
Yes
Yes
Yes
Yes
Observations
1175
1175
1175
1175
1175
1175
R-squared
0.14
0.13
0.19
0.33
0.41
0.38
Note: Robust standard errors in parentheses. * significant at 10%; ** significant at 5%; *** significant at 1%. Constant terms are not reported.
Control variables in Equation (5) include distance to Boxer Uprising, distance to boundary, interaction between distance to boundary and safety
region, interaction between distance to Boxer Uprising and distance to boundary, and the interaction between distance to Boxers Uprising,
distance to boundary and safety region (for estimators, please refer to Appendix IV for details).The dummy of safety region is dropped due to
collinearity since we control for province dummies in the regressions.
56
Table 4: Instrumental Evidence
(1)
The 1 step
The 2nd step
0.797**
(0.367)
6.545***
(2.201)
st
Protestantism
Distance to Boxer Uprising
* safety region
Initial economic conditions
Small city
(2)
The 1 step The 2nd step
0.818**
(0.370)
6.700***
(2.180)
st
4.032
(2.684)
5.493*
(3.175)
18.254***
(5.305)
4.487***
(1.322)
Middle city
Large city
Treaty ports
1.548
3.964
(3.012)
13.909***
(4.671)
32.011***
(9.412)
-1.204
(1.940)
0.180
(1.071)
3.495*
(1.558)
-0.685
(2.418)
-1.492
(2.241)
2.745
(1.804)
0.433
(0.359)
-2.462***
(0.708)
-2.598***
(0.708)
(1.901)
6.946***
(2.650)
0.031
(1.981)
0.268
(2.317)
-0.211
(0.553)
2.576**
(1.252)
1.355
(1.328)
Railway
Prefectural government
location
Grand Canal
Changjiang (Yangzi) River
Coast
Catholicism
Population density (1850s)
Size
0.545
(0.348)
-0.866*
(0.658)
-1.432
(0.658)
0.270
(0.549)
4.673***
(0.976)
3.044***
(1.076)
Control Variables in
Equation (5)
Yes
Yes
Yes
Yes
Province dummies
Yes
Yes
Yes
Yes
Observations
1175
1175
1175
1175
R-squared
0.144
0.103
0.176
0.324
Note: Robust standard errors in parentheses. * significant at 10%; ** significant at 5%; ***
significant at 1%. Constant terms are not reported. Control variables in Equation (5) include
distance to Boxer Uprising, distance to boundary, interaction between distance to boundary
and safety region, interaction between distance to Boxer Uprising and distance to boundary,
and the interaction between distance to Boxers Uprising, distance to boundary and safety
region (for estimators, please refer to Appendix IV for details).The dummy of safety region is
dropped due to collinearity since we control for province dummies in the regressions.
57
APPENDIX I: DOCUMENTATION OF DATA SOURCES – Table A1
Variables
Data sources
Protestantism
The various measures of Protestant missionaries in China in the period 1918-1921 were jointly surveyed by the
Special Committee on Survey and Occupation and the China Continuation Committee (Source: “The Christian
Occupation of China: A General Survey of the Numerical Strength and Geographical Distribution of the
Christian Forces in China”, Stauffer, 1922).
Appendix A of Stauffer (1922) contains information on the following two variables at the county level: (the
number of) Christian communicants around 1920 and (the number of?) Christian communicants per 10,000
population/people. We match the county list in 1922 with that in 1840, and arrived at 1175 counties in 14
provinces.
Denominational information
Appendix A of Stauffer (1922) contains information of the missionary societies in each county (names). Stauffer
(1922, p. 26) lists the missionary societies with their denominations and host countries. Based on such
information, we first count the total number of missionary societies in each county and the number of
missionaries in each denomination and from each host country. Then we calculate the share of each
denomination: d1 the China Inland Mission (CIM), d 2 the Lutherans, d 3 the Calvinists, d 4 the Baptists, and
d 5 the Wesleyans, with the Anglicans and other non-sectarian missions serving as the reference group d 6 ) and
the share of missionary societies from each country ( h1 the United Kingdom, h2 the United States of America,
h3 the European Continent, and h4 the group of international denominations).
Knowledge diffusion
Appendix A of Stauffer (1922) contains information on the number of lower and higher primary student
enrollments in 1920 at schools erected by the Protestants at the county level. Based on such information, we
58
construct two dummies: the lower primary school ( K1 ) and the higher primary school ( K 2 ).
Appendix E of Stauffer (1922) enumerates the middle schools and hospitals by location. Based on this
information, we compute the dummy of middle schools of each county ( K 3 ).
Stauffer (1922, p. 1111) enumerates bible schools by location, which is used to construct the dummy of Bible
schools ( K 4 ).
Stauffer (1922, p. 1100) enumerates teacher-training schools by location, which is used to construct the dummy
of teacher-training schools ( K5 ).
Appendix E of Stauffer (1922) enumerates hospitals by location. Based on this information, we compute the
dummy of middle schools and hospitals for each prefecture.
Roman Catholicism
Appendix C of Stauffer (1922) contains maps of the provincial distribution of Roman Catholic Mission Stations.
Based on these maps, we count the number of Roman Catholic Mission Stations for each prefecture, and
normalize it by the total population of the prefecture.
Urbanization in 1920
Appendix G of Stauffer (1922) contains a list of cities with urban population larger than 25,000. Based on the
information on total prefectural population (China Population History, Ge, 2000), we obtain the urbanization
ratio for each prefecture by dividing the urban population by the total population.
Western influence
The duration of treaty ports and railway is calculated from the initial year of the establishment of a treaty port or
the railway, the information of which is provided by Luo (1995) in Selected Statistical Materials on the Modern
Chinese Economy (Zhongguo Jindai Jingji Ziliao Xuanbian).
Urbanization before 1840
In his book “Urban Networks in Ch’ing China and Tokugawa Japan”, Rozman (1973) classifies cities in China
into four size categories. Since the highest category contains only one city (Peking), we reclassify all cities into
59
only three size categories, namely Big, Mid-sized, and Small.
Geography
The China Historical Geography Information System (CHGIS) of Harvard Yenching Institute provides detailed
information on the coastline and rivers of China, from which we can derive two dummy variables to indicate
whether a prefecture is located on the coast and along the Changjiang River (Yangtze River).
Also, based on the data of CHGIS, including latitude and longitude of the prefecture capital, we can compute the
distance to the Boxer Uprising for each prefecture.
Population density (1850s)
China Population History (Ge, 2000) contains information on population density as well as the size of each
and regional size
prefecture in 1850.
Boxer Uprising
Annex 8 of the Boxer Protocol provides a list of prefectures with Boxers.
Distance to Boxer Uprising
Generated based on “CHGIS, Version 4”
(and to Boundary)
Data on World Missionaries
A Geography and Atlas Protestant Missions (Beach, 1903) lists the missions in 1900 in 19 countries (or
regions), with which we match the missionary societies in China and construct three indices of knowledge
diffusion (Please find the details in Appendix V).
60
APPENDIX II: MISSION LISTS – TABLE A2
Missionary societies in China 1
Abbreviation
AAM
American Advent Mission Society
(Advent Christian Mission)
ABCFM
American Board of Commissioners for
Foreign Missions
ABFMS, or
American Baptist Foreign Mission
ABF
Society
AFM
Apostolic Faith Missionaries
AFO
American Friends' Mission (Ohio
Yearly Meeting)
AG
General council of the Assemblies of
God
B
Evangelische Missionsgesellschaft zu
Basel
BMS
Baptist Missionary Society
BN, or Bn
Berliner Missionsgesellschaft
CCACZ
CCAu
CGM
Christian Catholic Church in Zion
Federal Foreign Mission committee of
Churches of Christ in Austrialia
Church of God Mission
CHM
ChMMS
Cannadian Holiness Movement
China Mennonite Missionary Society
CIM
CM
CMA
China Inland Mission
Christians' Mission
Christian and Missionary Alliance
Missionary societies over World 2
Abbreviation
AA
American Advent Mission Society
Anglicanism
U.S.A
ABCFM
American Board of Foreign Missions
Calvinism
U.S.A
ABMU
American Baptist Missionary Union
Baptist
U.S.A
HFM
AFFM
Hephzibah Faith Mission
American Friends' Board of Foreign
Missions
NA
NA
International
U.S.A
NA
U.S.A
Lutheranism
Switzerland
Baptist
Lutheranism
U.S.A
German
NA
NA
U.S.A
England
NA
U.S.A
B
BMS
Bn
ZCCC
FCA
CG
MenEB
CIM
CA
Evangelische Missions-Gesellschaft,
Basel
Baptist Missionary Society
Ges. Z. Beforderung d. ev. Missionen
unter d. Heiden
Zion Catholic Christian Chruch
F. M. Com., Fed. Ch's of Australia and
Tasmania
Board of Missions, Gen. Eldership,
Churches of God
Denomination 1 Host country 1
NA
Mennonite Evangelizing and Benevolent Baptist
Board
China Inland Mission
CIM
NA
Christian and Missionary Alliance
NA
61
Canada
England
International
England
U.S.A
CMML
Christian Missions in Many Lands
CMS
Church Missionary Society
CMS
CMS+CEZMS Church of England Zenana Missionary CEZMS
Society
CNTM
China New Testament Mission
CPW
Cumberland Presbyterian Mission
CP
CSFM
EA
EbM
ELAug
ELMo
EPM
EvM
FCMS
FFC(CIM)
FMA
FMS
GBB
Church of Scotland Foreign Mission
CSFM
Committee
Missionary Society of the Evangelical EA
Association of North America
Ebenezer Missions
Augustana Synod of the Evangelical
ELGC
Luteran Church of North America
Evangelical Lutheran Synod of Missouri ELGS
and other States
English Presbyterian Mission
PCE
Evangel Mission
Foreign christian Missionary Society
FCMS
Fria Missionen I Finland (Finnish Free
Church Mission)
American Free Methodist Mission
FMA
Finska Missions-Sallskapet
FMS
General Mission Board of the Church of GBB
the Brethren
GCAM or
German China Alliance Mission in
ND
GCAM(CIM) Barmen
Heb
Hebron Mission
HF(CIM)
Helgelse-Forbundet(Swedish Holinese HF
Church Missionary Society
Church of England Zenana Miss'y
Society
NA
Anglicanism
Anglicanism
Baptist
Board of Missions and Church Erection, Calvinism
Cumb. Prb.
Church of Scoltalnd Foreign Mission
Calvinism
Committee
Missionary Society, Evagelical
Wesleyans
Association
NA
Board of F. M., Gen. Counc., Ev. Luth. Lutheranism
Ch., N. A.
Board of For. Miss., Evangelical Church, Lutheranism
Gen. Syn.
Foreign Missions Com., Presb. Church of Calvinism
England
NA
Foreign Christian Missionary Society
NA
CIM
Gen. Miss. Board, Free Methodist
Church. N.A.
Finska Missions-Sallskapet
Gen. Miss. And Tract Com., German
Baptist Brethren
Gesellschaft fur innere und aussere
Mission
Helgelseforbundet
62
England
England
England
U.S.A
U.S.A
Scotland
U.S.A
U.S.A
U.S.A
U.S.A
England
U.S.A
U.S.A
International
Wesleyans
U.S.A
Lutheranism
Baptist
Finland
U.S.A
CIM
German
NA
CIM
U.S.A
Swedien
ILM
Ind
KCM
L(CIM)
LB
LBM
LMS
LUM
MBM
MEFB
MES
MGC
MP
MSCC
NCM
NFEM
NHM
NKM
NLF
NLK or
NLK(CIM)
Union)
Independent Lutheran Mission
Independent Missionaries
Kieler China Mission
Liebenzeller Mission
American Lutheran Brethren Mission
Lutheran board of Missions
London Missionary Society
Lutheran United Mission
Mennonite Brethren Mission
Board of Foreign Missions of the
Methodist Episcopal Church
Methodist Episcopal Mission, south
IU
Independent and Unconnected
LFC
LMS
UN
Lutheran Board of Missions
London Missionary Society
United Norwegian Lutheran Church in
America
MEN
Missionary Society of the Methodist
Episc. Church
Board of Missions of the M. E. Church,
South
General Conference of Free Baptists
MES
General Conference of the Mennonites FB
of North America
Methodist Protestant Mission
MP
Missionary Society of the Church of
CEC
England in Canada
Tsjilimissionen (North Chihli Mission)
Norges Frie Evangeliske
Missionsforbund
National Holiness Mission
IBM
Northwest Kiangsi Mission
Mission of the Evangelical Lutheran
Free church of Norway
Norwegian Lutheran Mission
SchA
Sch
Lutheranism
NA
Lutheranism
Lutheranism
Lutheranism
Lutheranism
Calvinism
Lutheranism
U.S.A
International
German
German
U.S.A
U.S.A
England
U.S.A
Baptist
Wesleyans
U.S.A
U.S.A
Wesleyans
U.S.A
NA
U.S.A
Board of Foreign Missions, Methodist
Calvinism
Prot. Church
Dom. and For. M. S., Church of England, Anglicanism
Canada
NA
NA
U.S.A
Independent Baptist Missionary
Movement
NA
U.S.A
NA
Lutheranism
England
Norway
Lutheranism
Norway
Synod of the Norwegian Evang. Luth.
Ch. In Amer.
Norske Kirkes Mission ved Schreuder
63
Canada
Norway
Norway
NMC(CIM) or
NMF(CIM)
NMS or
NMS(CIM)
PCC
Norske Mission I Kina (Norwegian
Missino in China)
Norwegian Missionary society
NLK
Norske lutherske Kinamissionsforbund
Lutheranism
Norway
NM
Norske Missionsselskab
Lutheranism
Norway
PCC
Calvinism
Canada
PBW
Foreign Missions Committee, Presb.
Church, Canada
Pentecost Bands of the World
NA
U.S.A
PCNZ
Presbyterian Church of New Zealand
Calvinism
New Zealand
PE
Domestic and Foreign M. S., Prot. Episc.
Church
Board of Foreign Missions, Presb.
Church, North
Exec. Com. Of For. Missions, Presb.
Church, South
Board of For. Miss., Reformed Church in
America
Board of For. Miss., Reformed Church in
the U.S.
Anglicanism
U.S.A
Calvinism
U.S.A
Calvinism
U.S.A
Calvinism
U.S.A
Calvinism
U.S.A
PE
Board of Foreign Missions of
Presbyterian church in Canada
General Missionary board of the
Pentecostal Church of the Nazarene
Foreign Mission Committee of
Presbyterian Church of New Zealand
Protestant Episcopal church, U.S.A.
PN
American Presbyterian Mission, North PN
PS
American Presbyterian Mission, South PS
RCA
Ricksha
RM
RPC
Board of Foreign Missions of the
Reformed Church in America
Board of Foreign Missions of the
Reformed Church in U.S.A
Shanghai Ricksha Mission
Rhenish Mission
Reformed Presbyterian Mission
RM
RP
SA
SAM(CIM)
Salvation Army
Scandinavian China Alliance Mission
SA
SAM
SBC
Foreign Mission board of the Southern SBC
Baptist convention
Swedish Baptist Mission
SBM
South China Boat Mission
MS
PCN
PCNZ
RCUS
SBM
SCBM
RCA
RCUS
NA
Rheinische Missionsgesellschaft
Lutheranism
Synod of Reformed Presb. Church in N. Calvinism
America
Salvation Army
NA
Scandinavian Alliance Mission of North CIM
America
Foreign Mission Board, Southern Baptist Baptist
Convention
Sallskapet Svenska Baptist Missionin
Baptist
Missions to Seamen
NA
64
International
German
U.S.A
International
Sweden
U.S.A
Sweden
U.S.A
SCHM
SCM
SDA
SDB
SEFC
TSM
UB
South China Holiness Mission
PMS
South Chihli Mission
Seventh-Day Adventist Mission Board SDA
Seventh-Day Baptist Missionary Society SDB
Swedish Evangelical Free Church
(Swedish American Mission)
Swedish Evangelical Mission
SEMC
Convenant of American
Svenska Missionen I Kina (Swedish
SKM
Mission in China)
Swedish Missionary Society
SMF
Society for the Propagation of the
SPG
Gospel
Tsehchowfu Mission
United Brethren in christ Mission
UB
UE
United Evangelical Church Mission
UE
UMC
United Methodist Church Mission
UMFC
SEMC
SMC or
SMC(CIM)
SMF
SPG
WMMS
WU
Wesleyan Methodist Missionary Society WMS
Woman's Union Missionary Society of WU
American
YM
Yale Foreign Missionary Society
YMCA
Young Men's Christian Association of YMCA
China
YWCA
Young Women's Christian Association YWCA
of china
Data sources: 1. Stauffer (1922); 2. Beach (1903)
Peniel Missionary Society
NA
NA
Mission Board of Seventh-day Adventists NA
Seventh-Day Baptist Missionary Society Baptist
NA
U.S.A
U.S.A
U.S.A
U.S.A
Norway
Swedish Evangelical Mission Covenant
of America
Svenska Kyrkans Mission
Lutheranism
Sweden
CIM
Sweden
Svenska Missionsforbundet
Lutheranism
Society for the Propagation of the Gospel Anglicanism
Home, Front. And For. M. S., Unit.
Breth. In Christ
Home and For. Miss'y Soc., United
Evang. Church
United Methodist Free Churches H. and
F. Missions
Wesleyan Methodist Missionary Society
Woman's Union Missionary Society
NA
Calvinism
England
U.S.A
Wesleyans
U.S.A
Wesleyans
England
Wesleyans
NA
England
German
NA
Foreign Department, Y. M. C. A., North NA
America
World's Young Women's Christian
NA
Association
65
Sweden
England
U.S.A
International
International
APPENDIX III: SAUNDERS’ RECOLLECTION OF HIS RETREAT DURING THE
BOXER UPRISING
Date
June 26
Location
Pingyao
Records
“June 26th, when our place was attacked and looted, and we with
our four children, had to escape…”(Saunders, 1900: 5)
June 28
Taiyuan
“June 28th, we got within seven miles of T’ai-Uen (Taiyuan),
where we met a convert who told us that we had better not go
there as the large compound of the Sheoyang Mission had been
burned the night before, and Miss Coombs burned to death.”
(Saunders, 1900: 7)
July 5
Luan
“So we turned southward again and set our faces towards LuCh’eng – a city in the Lu-an district … We arrived at the city on
Thursday, July 5th, and stayed until Saturday, the 7th, when the
station also was rioted.” (Saunders, 1900: 9)
July 12
Zezhou
“We pushed on and reached Tseh-chow (Zezhou) Fu, the border
city of Shanxi, about 11 pm (July 12th). We passed through many
towns and villages and it was in this district our suffering
reached its climax.” (Saunders, 1900: 20)
July 16
Huaiqing
“We reached Huai-ching Fu and were treated tolerably well by
and
the officials there and at our next stopping place, U-chi Hsien.”
Yuanqi
(Saunders, 1900: 24)
July 20
Chenqi
“Money was supplied us at U-chi Hsien freely and we began to
hope for a rapid journey to Han-kow. Disappointment met us,
for at Chen-Chi Hsien…” (Saunders, 1900: 25)
July 22
Chenzhou “We were sent on by cart, but as common criminals…”
(Saunders, 1900: 27)
July 28
Queshan
“This treatment continued for six days, till we reached Ch’iohshan Hsien, where we were taken to a temple, and the
Mandarin’s wife sent round sweetmeats for the children.”
(Saunders, 1900: 28)
July 30
Xinyang
“At Sin-yang Chow, the border city of Ho-nan, which we
Zhou
reached on Monday, July 30th … We stopped at Sin-yang Zhou
three days, because soldiers were passing on route for Pekin(g),
and it was feared that if we met them on the road trouble might
arise.” (Saunders, 1900: 28)
August 2
Yingshana “We found in passing through Ho-nan that it was our greatest
protection from the wrath of the people to let them know that we
were Protestants…” (Saunders, 1900: 29)
August 14 Hankou
“Now we had come to the Hu-pen Province, ruled over by
Chang-chih Tung, and we were treated well by all the officials,
and instead of travelling as before, in carts…and the rest of our
journey to Han-kow was accomplished in comparative comfort.
We arrived at our mission house on Tuesday morning, August
14th, in all 49 days since we left Ping-iao.” (Saunders, 1900: 29)
a
Yingshan is a county in Hubei Province but is contiguous to Xinyang Zhou in Henan
Province.
66
Figure A1: The Retreat Route of Saunders, a Missionary
Source: Constructed based on Saunders (1900)
67
APPENDIX IV: INSTRUMENTAL EVIDENCE (COEFFICIENTS NOT REPORTED IN THE MAIN TEXT)
Panel A: Supplementary results for Table 3 –validity of instrument
(1)
(2)
Communicants Constituency
/population
/population
Distance to Boxer Uprising
0.768
0.746
(0.974)
(1.701)
Distance to Boundary
-0.106*
-0.191*
(0.063)
(0.114)
Distance to Boundary
0.287***
0.382***
* safety region
(0.079)
(0.134)
Distance to Boxer Uprising
0.025
0.036
* Distance to Boundary
(0.016)
(0.030)
Distance to Boxer Uprising
-0.069***
-0.085**
* Distance to Boundary*safety region
(0.020)
(0.033)
Panel B: Supplementary results for Table 4 – instrumental evidence
(1)
st
The 1 step
Distance to Boxer Uprising
0.299
(1.374)
Distance to Boundary
-0.110
(0.069)
Distance to Boundary
0.286***
* safety region
(0.086)
Distance to Boxer Uprising
0.024
* Distance to Boundary
(0.019)
Distance to Boxer Uprising
-0.064***
* Distance to Boundary*safety region
(0.021)
Notes: * significant at 10%; ** significant at 5%; *** significant at 1%.
68
(3)
Catholicism
0.113
(0.134)
-0.008
(0.015)
0.009
(0.015)
0.001
(0.004)
-0.002
(0.004)
nd
The 2 step
-4.502**
(1.958)
0.133
(0.102)
-0.184
(0.155)
-0.036
(0.026)
0.052
(0.038)
(4)
Treaty ports
(5)
Railway
0.019
(0.020)
-0.000
(0.001)
0.000
(0.001)
-0.000
(0.000)
0.000
(0.000)
0.069*
(0.041)
-0.006***
(0.002)
0.003
(0.002)
0.001
(0.001)
-0.001
(0.001)
st
The 1 step
0.537
(1.380)
-0.092
(0.068)
0.271***
(0.086)
0.023
(0.018)
-0.068***
(0.021)
(2)
(6)
Urbanization
(1840s)
-0.026
(0.063)
-0.000
(0.003)
0.000
(0.003)
-0.000
(0.001)
0.000
(0.001)
The 2nd step
-3.158*
(1.822)
0.150*
(0.089)
-0.181
(0.141)
-0.036
(0.024)
0.048
(0.036)
APPENDIX V: WORLD MAP AND SAMPLED REGION WITH PROTESTANTISM DIFFUSION – Figure A2
69
APPENDIX VI: DIFFERENCES IN THE EFFECTS OF PROTESTANTISM
(1)
(2)
(3)
Without controlling for knowledge Controlling for knowledge diffusion Controlling for estimated knowledge
diffusion activities (Equation )
activities (Equation )
diffusion activities (Equation )
Coef.
S.E.
Coef.
S.E.
Coef.
S.E.
CIM vs. Others
-0.569***
0.129
-0.187
0.132
-0.114
0.140
Lutheranism vs. Others
-0.447***
0.148
-0.115
0.151
0.014
0.155
Wesleyanism vs. Baptist
-0.091
0.102
-0.077
0.105
-0.046
0.103
Calvinism vs. Others
0.068
0.119
0.075
0.122
0.019
0.119
Wesleyanism vs. Others
-0.075
0.111
0.045
0.111
0.167
0.117
Wesleyanism vs. Calvinism
0.122
0.100
0.072
0.097
0.128
0.104
Baptist vs. Others
0.478***
0.151
0.110
0.150
0.068
0.160
Lutheranism vs. CIM
0.636***
0.163
0.262
0.162
0.133
0.174
Baptist vs. Calvinism
0.494***
0.156
0.232
0.154
0.281*
0.158
Calvinism vs. Lutheranism
0.356**
0.155
0.038
0.155
-0.060
0.164
Wesleyanism vs. Lutheranism
0.514***
0.168
0.190
0.169
0.005
0.179
Calvinism vs. CIM
0.371**
0.160
0.160
0.161
0.153
0.164
Wesleyanism vs. CIM
0.159
0.141
0.152
0.134
0.065
0.138
Baptist vs. Lutheranism
0.016
0.106
0.122
0.103
0.213**
0.107
Baptist vs. CIM
-0.143
0.120
-0.030
0.119
0.148
0.123
Lower primary school
Higher primary school
Middle school
Bible school
Teachers school
Hospital
-0.048
0.012
0.236***
-0.338***
0.271*
0.142***
Day school
Higher Education Institutions
Hospital
Notes: * significant at 10%; ** significant at 5%; *** significant at 1%.
70
0.081
0.058
0.075
0.086
0.154
0.045
-0.003***
0.047***
0.072***
0.001
0.015
0.015