university ofgroningen

groningen growth anddevelopment centre

GGDC RESEARCH MEMORANDUM 147

Living standards in China between 1840 and 1912: A new estimate of Gross Domestic Product per

capita

Ye Ma, Herman de Jong and Tianshu Chu

July 2014

1

Living standards in China between 1840 and 1912: a new

estimate of Gross Domestic Product per capita1

Ye Ma, Herman de Jong & Tianshu Chu2

Groningen Growth and Development Centre /

Research Institute of Economics and Management

Abstract

This paper investigates China’s economic development between 1840 and 1912. We look at

living standards and general economic trends in the late Qing dynasty and discuss the

reliability of existing estimations of per capita GDP. Secondly, we introduce a new estimate of

long-term growth. Our estimation provides for the first time a continuous time series of both

nominal and real GDP per capita for the period 1840-1912. We see this as a starting point for

further research on 19th

century developments in levels of economic welfare and the

performance of the Chinese economy.

JEL classification codes: E23 E30 N15 N9

1 We are grateful to Harry Wu, Jan Luiten van Zanden and Bas van Leeuwen for giving us advice and suggestions about data

sources. In preparing the first draft, we have received valuable comments from many colleagues, including Pinghan Liang,

Dayong Zhang, and participants of the seminars at RIEM, Southwestern University of Finance and Economics, Chengdu, in

May 2012; participants of the Asian Historical Economics Society Conference at the Hitotsubashi University, Tokyo, 13-15

September 2012; and participants of the European Historical Economics Society Conference at the London School of

Economics, 6-7 September 2013. All errors are the responsibility of the authors.

2 Ye Ma, Groningen Growth and Development Centre, Faculty of Economics and Business, University of Groningen,

y.ma@rug.nl; Herman de Jong, Groningen Growth and Development Centre, Faculty of Economics and Business, University

of Groningen, h.j.de.jong@rug.nl; Tianshu Chu, Research Institute of Economics and Management, Southwestern University

of Finance and Economics, chutianshu@ymail.com.

2

1. Introduction

In this paper we present a new estimate of historical GDP and GDP per capita for the Chinese

economy in the 19th

and early 20th

century, i.e. in the late Qing Empire (1840-1912). During

this period, also known as the “years of troubles”, the Chinese population faced a combination

of wars, natural disasters, and economic reforms. Many scholars see this era as a break point in

China’s long-run economic development. On the one hand, it can be seen as a period of

economic distress, which stands in contrast with the relatively prosperous 18th

century. On the

other hand, it has been viewed as a period in which new social and economic conditions began

to replace old ones (Brandt, Ma, and Rawski 2012, p.27). China’s economy lost its position as

the world’s largest economy, but became involved in international trade, started to open up its

domestic market, and imported new technology from the rest of the world (Maddison, 2007).

Recent studies of China’s development in the late 19th

century support the long-lasting

influence of this era on China’s modern economy. Harry Wu concludes that China’s post-1949

state-led industrialization can be traced back by a development path that began in the late 19th

century (H. Wu, 2011). Keller et al. also attribute China’s present trade performance to its

experience in the 19th

century, rather than exclusively to the 1978 reforms (Keller, Li, and

Shiue, 2012). Brandt et al. relate the resilience of the Chinese modern economy to its

performance in the 19th

century, and argue that even in the mid-19th

century, the economy was

so resilient that “it had the inbuilt capacity not only to withstand certain shocks but also to

restore stability in the wake of potentially destabilizing disasters” (Brandt et al., 2012, p.49).

Although recent studies have found some implicit linkages between the economy of the 19th

century and the present one, more evidence is still needed to firmly substantiate these claims.

Understanding China’s economic history has broader implications. As the world’s largest

economy in the past and the second largest today, China always had the power to affect the rest

of the world, but it also has been influenced by global developments. To obtain a complete

understanding of world economic development since industrialization, China’s economic

situation in the late 19th

century is important. However, our knowledge about 19th

century

China is still rather limited. We need e.g. more solid information on cyclical and long-term

movements in national income. Moreover, previous GDP estimations show contradictory

outcomes, which prevent us from getting a good understanding of the performance of the total

economy.

This paper intends to solve the main puzzles around the historical GDP estimates for the late

19th

century. For this purpose, we reconstruct total production of the Chinese economy and

provide a new estimation of GDP, resulting in the first consistent time series for the period

1840-1912. The reconstruction necessarily will involve simplification and approximation, but

we see it as an important step in the study of the 19th

century Chinese economy. The structure of

the paper is as follows: in section 2 we analyze several studies on historical GDP estimations

for the 19th

century; in section 3 we elaborate on our own GDP estimation; section 4

summarizes the main findings and confronts them with earlier studies. The appendixes provide

a detailed explanation of our estimation procedures.

3

2. An overview of existing estimates of Chinese GDP per capita

This section summarizes the results of previous studies on Chinese historical GDP in the late

19th

and early 20th

century.3 The studies have been classified by us into two groups according to

the estimation methods used and we refer to them as macro and sector studies. We can also sort

these studies according to the estimation results: some studies provide an estimation for a

benchmark year (cross-section), others provide estimations for various years. This is shown in

Table 1 (a and b, respectively). In paragraph 2.2 we introduce two issues that need to be solved

before any in-depth investigation into the Chinese economy can take place. These questions are

related to the level and the growth rate of average national income respectively. Although it is

not easy to completely answer the two questions in one paper, we will try to solve some issues

and give suggestions for further research.

2.1 Macro and sector studies

Previous studies on China’s historical GDP provide us with benchmarks of income levels for

specific years and also time-series estimations of per capita GDP. Table 1 covers all the recent

historical GDP estimates for the period 1000-1933.

[Table 1]

We confine ourselves here to what we believe are the three most important studies. Firstly,

Chang (1962), cited in many studies, is the first GDP estimation for late 19th

century China.

This study concludes that per capita GDP in the 1880s was about 7.4 silver tael in current

prices, i.e. prices of the 1880s. The estimation applies a sectoral value-added approach, which

roughly follows the familiar three-sector classification in national accounting: agriculture,

industry, and services. The GDP estimates for the total economy are obtained by adding up the

output value of all three sectors. Sector information is retrieved from historical documents and

previous estimations. For instance, Chang uses the amount of cultivated land recorded in

official documents to reconstruct agricultural output.4 We refer to this type of estimations as

sector studies.

The second study is by Angus Maddison, generating a historical GDP dataset and providing a

long-term trend of Chinese economic development. For the period 1840-1912 five level

estimates were produced for the years 1850, 1870, 1890, 1900, and 1913. The average level of

3 Names of authors working in North America and Europe and publishing in English are presented in Western style (e.g. Debin

Ma). Names of Chinese authors based in Asia and writing primarily in Asian languages appear in the East Asian fashion, with

the surname first and capitalized for clarity (e.g. WU Chengming). The purpose of the classification is to distinguish between

English and Chinese texts, so that the reader knows where we rely on translations. 4 The definition of food crops is related to the principal food components in the Chinese diet (see Appendix B).

4

the GDP per capita estimates is 554 international Geary-Khamis dollars in 1990 prices (GK

dollars of 1990). Maddison calculated real growth rates for the period 1820-1995 by linking

previous GDP estimations, such as the GDP level for 1890 from Chang (1962), and a 1933

estimate from Ou (1947) and Liu and Yeh (1965). From the Chinese GDP level of 1990 (in

international GK dollars), historical GDP per capita levels for the 19th

century were generated

by a backward projection of the estimated rates of real growth. Maddison’s procedure thus

relies on existing GDP estimations for the total economy. We refer to this type of estimations as

macro studies. The macro and the sector approaches are intertwined in practice. For instance,

the macro approach can be based on GDP estimates from sector studies. Table 1a includes the

benchmark studies that Maddison employed in his estimation.

The third study that we mention here is a recent reconstruction by Broadberry, Guan and Li

(2013). Based on very detailed archival work, it is the first to provide a time series of per capita

GDP estimates covering three dynasties and more than 600 years of Chinese history. The

estimated average level of per capita GDP in the period 1600-1850, the early Qing Empire, was

715 international dollars in 1990 prices. In domestic currency the per capita average was 14.1

silver tael in 1840 prices (see Table 1). In their study, nominal GDP and real GDP in 1840

prices are estimated by using the sector approach. With an additional level estimate for the year

1840 in international dollars and a reconstruction of rates of real GDP growth, the authors are

also able to derive a time series of GDP estimates in international dollars for the early Qing

Empire. This outcomes of this sector study are more or less in line with Maddison’s estimation

for the per capita GDP level of 1850, which is about 600 GK dollars.

There are several GDP estimates for the period before 1850, 1850-1912, and the 1880s, but

expressed in different monetary units. Generally, the monetary unit “international dollars” is

comparable with the “GK dollars” used by Maddison. Both reflect a purchasing-power-parity

(PPP) adjusted value of official currencies. These internationally comparable monetary units

intend to measure and compare purchasing power of currencies across countries, by

considering not only market exchange rates but also domestic prices in different countries. We

will use the term “GK dollars” to emphasize the use of the Geary-Khamis PPP convertor

between the Chinese yuan and the dollar. The assumption is that there is no difference between

GK dollars and the international dollars as applied by Broadberry et al.

To compare the different estimates for the nineteenth century, the GDP values in terms of silver

tael need to be converted into GK dollars. Based on the official exchange rates between Qing

China and the U.S. and on the movement of the annual consumer price index for the U.S. from

1885 to 1990, the 1880s per capita GDP of China can be obtained, being 125.2 US dollars in

1990 prices (1990USD). By using the term “1990USD”, we emphasize that this value is

obtained through current market (silver) exchange rates. Chang also concludes that the level of

per capita GDP during the 1880s is about 32 Chinese yuan in 1930s prices. Following the

method in Fukao, Ma, and Yuan (2007), Chang’s 1880 estimate can be converted resulting in a

level of 337 international dollars. This is based on a PPP convertor. We find that the

exchange-rate-adjusted approach leads to serious underestimation; the resulted estimate here is

lower than one half of the bare-bones level of income (300 dollars). Still, the PPP-adjusted

5

estimate for the 1880s is over 35 percent lower than Maddison’s 1870 and 1890 estimate, 530

and 540 GK dollars respectively. It is hard to believe that the Qing economy decreased by over

35 percent and rebounded back to its initial level in the same decade.

Our conclusion is that a simple connection between the three estimations does not result in a

plausible understanding of the income level for late Qing China. Value conversion from silver

tael to international dollars clearly does not solve all the methodological differences between

the various estimations. The next section will discuss more extensively the problem of

inconsistency between previous estimations.

2.2 Why do we need a new estimate?

Table 2 compares per capita GDP estimates for the late 19th

and early 20th

century. It reveals the

discrepancies between the macro and the sector approach and the differences among sector

studies. These discrepancies make it difficult to get a good idea of the general living standards

in the late 19th

and early 20th

century. It is also difficult to generate plausible time-series

estimations that can be cross checked for certain years. We summarize these discrepancies into

two major points, leading to two questions that this paper intends to solve.

[Table 2]

The first point is related to the different outcomes of the level of per capita GDP before the

1890s. As mentioned above, the average per capita GDP estimate is around 550 GK dollars in

Maddison’s estimation, whereas the average is one third lower in the sector studies except

Broadberry et al. (2013). We need to check whether the approach of value conversion that we

apply to the 1880 estimates is a cause for the discrepancy in levels of GDP per capita between

the various macro and sector studies (See Table 1a). As said, this approach has been employed

by Fukao et al. (2007) to convert a GDP estimate in 1930 Chinese yuan to international dollars.

They derive a level estimate of 619 international dollars for 1934-6, which is 9 percent higher

than Maddison’s estimate for the 1930s of 570 GK dollars. If we apply a similar procedure to

translate Debin Ma’s level estimate for the 1910s (52.4 yuan), we arrive at 549.5 international

dollars. This is only 0.5 percent lower than Maddison’s estimate of 552 GK dollars (See Table

2). The approach of value conversion may explain some of the discrepancies, but it is not the

main reason behind the problem of inconsistency in the estimates for the period before the

1890s.

Differences can be found not only between macro and sector studies but also between some of

the sector studies. Aside from Broadberry et al. (2013), LIU T. (2009) also reports estimates in

international dollars and gives another benchmark estimate for 1840 with a value of 318

international dollars. Although following the same procedures of value conversion, the two

estimations differ by a margin of 50 percent (See Table 1a and 2). We find that the margin is

6

caused by variations in nominal GDP/cap levels as well as by the different PPP-adjusted

exchange rates between silver taels and international dollars: 10.8 tael and 29.4 international

dollars/tael in LIU T. (2009); 12.95 tael and 46 international dollars/tael in Broadberry et al.

(2013). Therefore, in our data reconstruction two steps need to be included: a reconstruction of

nominal GDP (per capita) and an estimation of the PPP convertor.

The second question concerns the long-term development of per capita GDP between 1840 and

1912. Maddison’s estimation shows hardly any significant changes in GDP per capita in this

period. The economy seems to be stable at a level of per capita GDP of 500-600 GK dollars.

However, we get a different view of the economy when the estimates of the various sector

studies are connected (See Table 2, column 2). Per capita GDP started from a lower level of 318

dollars during the 1840s, but increased by 20-50 percent in the following decades, and finally

reached a level of 500-600 international dollars in the 1930s. However, on this basis alone we

cannot get a more detailed picture of the economy. For instance, the data in Table 2 cannot give

us information about the general economic trend during the first Opium War and the Taiping

Rebellion (in the 1840s-1850s). This is due to the problem of inconsistency between macro and

sector studies. A re-estimation of nominal and real GDP would be the first step to reveal more

detailed fluctuations in the economy and to arrive at a consistent understanding of the growth

pattern of the Qing economy in the period 1840-1912.

2.3 The Great Divergence and the economy of the late Qing Empire

Solving these discrepancies in China’s historical GDP will also help to understand the long-run

comparative development of the Western and Asian economies. In this section, we will touch

upon the discussion of the Great Divergence between Europe and China. The important issues

around the widening income gap between both regions are clearly related to the questions

mentioned above.

Maddison’s long-run estimations reveal that China’s per capita GDP was higher than that of

Europe before the early 15th

century. After that the Chinese economy went into a long period of

stagnation, while some European economies began to take off, setting the stage for the

economic divergence between Europe and Asia. In Maddison’s research, the timing of the

divergence dates back to roughly the 16th

century (Maddison, 2007). Pomeranz (2000)

estimates the timing of the great divergence much later, starting in the 19th

century.

Recent studies show a potentially different picture of the long-run economic development in

pre-modern China. Broadberry et al. (2013) conclude in their estimation that around the 14th

century the Chinese economy had already fallen behind some European countries, such as

Italy, and the gap had become significantly large in the 18th

century. So it largely pushes back

the starting period of the economic divergence. More importantly, it interprets the great

divergence as a process lasting for several centuries. But we have to remember that there are

shifts in the relative economic position of countries (Broadberry et al., 2013, p. 36, Table 6).

Although it was lower than the levels in Italy and the Netherlands, Chinese GDP per capita in

7

the 16th

century is still comparable to that of England. After that, the Chinese economy

experienced a relative decline lasting for at least three centuries until the early 19th

century,

being the end point of the estimation by Broadberry et al. Note, however, that in their

estimation the most dramatic relative economic decline actually happened in the 18th

century,

which is generally seen as a rather favourable period for China, with a prosperous economy,

stable society, and an increasing population.

In the 19th

century general circumstances were much worse in China. The era has been labelled

as the “Turbulent Century” in China’s historiography. This refers especially to the period after

1840. Table 3 compares the incidence of wars and natural disasters between the early and late

Qing Empire (1640-1840 and 1840-1912, respectively). The general societal circumstances

probably worsened significantly after 1840.

[Table 3]

During the second half of the Qing Empire, there were many wars and natural disasters.

Around 100 million people lost their lives during this period because of domestic turmoil,

floods, as well as famines. The most serious population loss probably occurred in the 1850s,

largely caused by the Taiping Rebellion from 1851 to 1864. Large parts of the wealthy southern

area of Qing China had been devastated and approximately 20-30 million people died of the

war and the famines and plagues that went with it. At the same time, the Nien Rebellion

(1851-1868) burst out in northern China coinciding with a massive flood of the Yellow River in

1851. The large decrease in silver reserves of the central government indicates that the Qing

state had lost a large part of its control power (see Table 3, line C.). Its capacity to support extra

expenditures, such as excess military spending and relief funds, had been weakened

considerably.

But there were also political and economic reforms, which may have improved the standard of

living of the survivors. From the 1860s, forced by a deterioration of tax income, the Qing state

officially permitted the settlement of Han Chinese in Manchuria (Maddison, 2007, p.36).5 In

this period, the area under cultivation in China actually increased. After the First and the

Second Opium war (1839-1842 and 1856-1860, respectively), the country was forced to open

several new ports to western trade and residence. One outcome of the forced exposure to

international trade was specialization in agricultural production, especially for some cash

crops. After the period of wars Qing China tried to incorporate technologies from the more

advanced European economies and an early industrialization may have started at that time.

Probably the Chinese economy recovered gradually from the wars under the so-called Tongzhi

restoration during 1861-1875. The Qing economy started to export cotton goods, soy beans,

ore, and timber only from the 1890s.

5 This part of northeastern China was always seen as the original place for the royal blood of the Qing Empire and was a

restricted place for common Han Chinese.

8

However, in the historical records we find no indication that points to a significant economic

improvement in the late 19th

century. Brandt et al. (2012) speak of a paradox. They observe

economic reforms in this period, which should have benefited the economy. Any significant

and sustained economic growth, however, only took place a century later after the 1978

reforms. They characterize this with some understatement as the “delay of growth”.

Was there a tendency of decline and divergence of the Qing economy persisting in the 19th

century? Or can we trace signs of catch-up during this period? Was the level of output

sustainable, i.e. sufficient to be able to support the population with basic energy needs? These

questions justify a reassessment of the Chinese economy for this period.

3. A new estimation of historical GDP

In this section a reconstruction of Chinese GDP per capita is presented based on new

estimations of value added in the three major sectors of the economy. Details are provided in

Appendix B. Secondly, cross checks per sector will be offered to see whether these new

estimations are plausible. Finally, we discuss economic developments in late Qing China on the

basis of the new sector estimates.

3.1 The agricultural sector

Our reconstruction tries to cover China’s output of the major agricultural products during

1840-1912. As a typical agrarian economy, the country concentrated on crop production, such

as grains and textile fibers, rather than livestock products (Maddison, 2007, p.32). In order to

facilitate taxation, there are lots of official documents and records on crops. Two categories and

two types of crops are distinguished here. The two categories concern so-called basic-food and

non-food crops.6 The most important basic-food crop is rice, while cotton is an example of

non-food agricultural production. The type of crop discerns between the degrees of intensity of

land-use. Basic-food crops are mainly land-intensive, but fruits and vegetables are not.

Because we use different methods to calculate the value of crop production in different

categories and types, a summary of these crops and their categories is given in Appendix B,

Table B.1. In total, we have four groups of agricultural products and activities including cattle

farming.7 Adding up the four items results in an annual figure of total value added in the

agricultural sector.

The focus here is mainly on basic-food crop production because it is the most relevant to

6 Non-food crops are oil seeds, vegetables, and flowers. Here, we realize that soy beans and of course vegetables can also be

consumed as food, but we classify it as a non-food crop. First, in general Chinese do not consume soy beans as their principal

food, compared with rice and wheat; second, we follow the previous studies, where it has been calculated as a non-food crop,

for instance, XU and WU (2005, p.1098). 7 There are other ways of classification. The group of food crops in our classification is very similar to the category “cereals” in

Maddison’s research (Maddison, 2007, p.118). Perkins’ estimation uses “grain output” which includes the cereal group,

potatoes, and other tubers (Perkins, 1969, p.16-17). In Table B.1 we follow the same procedure and summarize the food crops

also as “grains”.

9

people’s living standards in China (Perkins, 1969, p.396). The estimation of crop production is

usually constrained by the unreliability of the information on cultivated area, yields, and

imports of new food crops (maize, Irish and sweet potatoes), especially for the 19th

century

(Perkins, 1969). In our calculation we will deal with these issues accordingly.

3.1.1 The amount of land under cultivation

How much of the cultivated land area was used for crop production in the late 19th

century? The

range of existing estimations of the total land area that was available for cultivation varies

between 0.8 billion and 2 billion mou, as is shown in Table 4 (1 mou = 666.5 square metres).

We believe that the most plausible level is ca. 1.2 billion mou. From the first land survey

organized by the People’s Republic of China (PRC) we know that the total land area available

for cultivation was around 1.6 billion mou in the 1950s. Another reliable survey which was

organized by the Republic of China shows that the level was around 1.4 billion mou in the

1930s. During the 1910s the level was around 1.2 billion mou (Perkins, 1969). See Appendix

B, Figure B.1.

[Table 4]

Based on the general level, a time series of cultivated land has been constructed. Figure B.1

shows our new estimation for cultivated land during 1840-1915. In this calculation, we rely on

three data sources: two benchmark estimates from recent historical studies and one time series

of cultivated land data from historical records. The two estimates are 1.15 billion mou in 1840

and 1.26 billion mou in 1914 (SHI, 1989; ZHANG Y., 1991). Both authors made amendments

on the official records from Qing China and the Republic of China respectively. To build a time

series, we also need reliable trends to link the two data points.

As is shown in the datasets compiled by YAN Zhongping, there was no increase in the

cultivated area in the main territory of the Qing Empire, except Manchuria, Xinjiang, and Tibet

(YAN Zhongping or Yen Chung-ping, 1955). Taking into account the immigration into

Manchuria from the middle part of China after the 1860s, we expect a mild increase in the total

area under cultivation. WU Hui provides a time series of the cultivated area from 1840-1915 in

the form of five-year averages, with special attention to Manchuria (see WU H., 1985, p.198,

although no detailed information on the data source are mentioned in this book). The general

level of this time series is higher than the two level estimates mentioned before, which we think

are more reliable. Therefore we use this third data source to provide estimations for the trend

only, and link it to the two data points for 1840 and 1912. The method and the calculation

procedure are explained in Appendix B.

Another issue here is the intensity of land-use during one year, e.g. in case of multiple

cropping, being the practice of growing two (or more) crops during a single growing season. To

10

estimate the actual land input involved in food crop production, it will be necessary to find out

how many times seeds were sown in one year, to calculate the multiple cropping ratio.8 The

ratio is neglected in many studies, but we add it to our estimation. The best solution is to find

the ratio for each food crop for each region. Here we rely on two levels mentioned in the

literature, 1.24 and 1.4 respectively (WU H., 1985, p.180; Maddison, 2007, p.36). We use the

average of 1.32 as an average for all food crops and for the country as a whole.

3.1.2 Average levels of yields

Next we turn to the estimation of yields per unit area of food crops. Official records about

yields per unit area began only in the 1930s. For the earlier period there is information in

regional documents, but they are too scattered to facilitate a good estimation for the whole

country. Two major data sources have been used for this: the first one is a study that covers a

period from the 1700s to the 1930s by LIU R. (1987); the second one is the official agricultural

survey for the 1930s, which is extensively discussed by YAN (1955). Appendix B gives the

ranges of the yields for different crops during the 1700s through the 1930s (See Table B.4).

In general, most studies assume that for the different crops the average yield in the 1930s is a

plausible estimate for the yield in the late 19th

century. Before directly applying the estimates to

our calculation, we need to justify this by comparing the underlying circumstances between the

two periods, i.e. the late half of the 19th

century and the 1930s. Both periods were characterized

by poor harvests because of the detrimental effects of wars and natural disasters. Setting the

result of a big harvest year at 100 percent and a normal harvest at 75, the level of output per unit

area was 50-60 during the 1840-1894 period, around 50 during 1894-1920, and 64 during

1927-1937 (LIU Y., 2001, p.25; LIU K., 2001, p.103). Although a complete comparison should

include details such as rainfall and temperature, we think that the impact of societal disorder

was much more significant. Looking at these estimates, is can be concluded that the yields

between the two periods were roughly similar; therefore we apply the yields of the 1930s to

estimate the yields for the late 19th

century.

We assume that the yields were constant over time, except for rice, wheat, and corn. The

underlying assumption is that agricultural technology in the late 19th

century was fixed. There

probably were some technological advances in the agricultural sector. However, considering

the negative societal and natural conditions displayed in Table 3, it is still safe to assume that

there was no broad technical progress. Thus, the only potential improvement on this estimation

would be an inclusion of weather and climate change. The unit yields of rice, wheat, and corn

are backwardly projected from their unit yields in the 1930s, using the annual harvest ratios

(CHEN, 1996, p. 419). Estimations of historical weather conditions were applied for the years

for which we found no harvest ratios in the official reports. After taking into account the

weather conditions, Figure B.2 shows relatively stable unit yields in the period concerned.

Probably the single most important contribution to agricultural production in the Qing Empire

is the introduction of new basic-food crops. To fight the problem of famine, the Qing governors

8 Multiple cropping ratio = Sown area in one year / Total cultivated area, from Maddison (2007), p.114, Table A.10.

11

encouraged the cultivation of sweet potatoes and maize. Specific adjustments to allow for these

two new crops were made (see details in Appendix B).

3.1.3 Cross checks of agricultural output

Based on these assumptions we calculated the annual output of food crops and levels per capita

during 1840-1912. On average, the per capita food output was 345 kilograms per year, with a

range between 305 and 366 kilograms. Figure 1a displays the annual trends of per capita food

output in the period concerned. The upper and lower limits were estimated by using the

maximum and minimum values of cultivated area and yields (see Appendix B, Figure B.1 and

Table B.4). Our results show that on average, the upper limit was 443 kilograms, while the

lower was 204 kilograms. Figure 1a shows an upward trend until 1870, after which a decline

set in. Presumably the economy gradually recovered from the First and Second Opium wars

(1851-1864) and then began to decline even before the Sino-Japanese war (1894-1895).

[Figure 1]

How can we verify whether this new estimation is plausible? Three cross checks of our

estimation of basic-food crop production will be given. First, our estimation of yields can be

compared with yields in other studies. In the present estimation, the annual output per unit area

was calculated at 114 kilograms in the late Qing Empire (annual output divided by the

cultivated area in mou). In previous studies, we find 128 kilograms in the Ming Empire and

183.5 kilograms in the Qing Empire (GUAN and Li, 2010, p.6; LIU R., 1987, p.109). In related

studies, the per unit area output in the second half of the Qing Empire was around 120-150

kilograms (Perkins, 1969; WU H., 1985; GUO, 1994, 1995).9 Because our upper limit is a level

of 120 kilograms, the present results might underestimate the total level of output.

Second, the estimated output level can be confronted with the results of previous studies. As

mentioned above, the average per capita output of food crops in the present estimation was 345

kilograms during 1840-1912. Deducting 40 percent of the total output for seeds, feed grains,

reserves, and alcohol production, the monthly per capita production was ca. 17.25 kilograms.

In Perkins (1969, p.398) the general annual level in the Qing Empire was around 250-375

kilograms. In WU H. (1985, p.192), the monthly per capita output in 1833 was 16.2 kilograms.

The estimated output per capita per year during the Qing Empire decreased from 576 kilograms

in 1753 to 314 kilograms in 1812 (WU H., 1985). In the middle of the Qing Empire, the level

was even lower than that in the 3rd

century (WU H., 1985, p.191). Compared with previous

studies the present estimation seems plausible. It confirms the statement that Qing China’s

agricultural production per capita was much lower than in previous empires.

9 In general it consisted of rice, wheat, and beans in southern China, and of wheat and beans in the northern area. Historians

estimate the national level of composite yields using the ratios of cultivated land in the two parts of China as weights.

12

The final question is whether the estimated levels of food production in the late 19th

century

could cover the minimum daily energy intake that is necessary to survive for a human being.

Grains, sweet potatoes, and maize were the main energy sources of food supply in Qing China.

Here we assume that the per capita output equals the per capita consumption. LIU R. (1987,

p.107) calculated the minimum food consumption per capita in the 18th

century at 290

kilograms. On average, our estimation is 19 percent higher than this subsistence level.

The available daily energy consumption per capita is estimated by using the annual food

production in the period 1840-1912.10

Taking account of an energy loss of 20 percent of the

total daily intake, the average daily energy consumption was 2600 kcal, with a range between

2290 and 2760 kcal for the whole period. In fact, the average is rather close to the situation in

modern China, with 2970 kcal in the period 2005-2007.11

Even if we assume that the energy

loss was 40 percent, the average daily energy consumption was 1950 kcal, which is still higher

than the minimum requirement of 1800 kcal, used by the FAO. There are two other studies that

give energy estimations. From Buck’s survey data (Buck, 1930, 1937), the general range of the

daily energy consumption in the 1930s was 1823-4434 kcal, which supports our estimation. In

Chang (1962), the estimated daily energy consumption was about 1800 kcal in the 1880s,

which is much lower than our estimation but still supports our finding that the general

biological living standard that we measure for the 1880s was sustainable.

To sum up, based on the cross check of production capacity, our estimation is lower than the

level provided by previous studies. Based on the cross check of average energy consumption

levels, our calculations might possibly overestimate the available amount of food output. In

general, our estimation passes two cross checks. Our estimation of food crop production in the

late 19th

century is plausible and supports sustainability.

For the remaining three categories of agricultural activities in Table B.1, value added in current

prices was estimated by a fixed ratio or proportion to the output value of basic-food crops (see

details in Appendix B, section 1.2-1.4). Figure B.5 presents the results. Average value added in

the agricultural sector during the period 1840-1912 is estimated at 11.25 tael per head of

population (in current prices), with a variation between 5.75 and 27.80 tael for the total period;

the average proportion of agriculture in total GDP was 75 percent. There was an annual

decrease of about 0.9 percent before the 1880s. This was followed by an increase of 4.2 percent

per year, mainly because of rising price levels. Low levels of value added were found in the

1850s and in the 1880s.

10 Energy data for the different food crops is from the Nutrient Data Laboratory, http://www.nal.usda.gov/ 11 http://www.fao.org/fileadmin/templates/ess/documents/food_security_statistics/FoodConsumption

Nutrients_en.xls

13

3.2 The industrial sector

There are relatively few data sources for the collection of quantitative information for every

industry. The existing literature, however, allows us to obtain a coherent picture of this sector.

In our estimation, we direct our attention in particular to the early development of modern

industrial production. One of the major economic characteristics in the late Qing Empire was

the start of industrialization and the so-called self-reinforcing movement between 1860 and

1895. This movement was characterized by the spread of foreign technology and the start of

state-owned factories in the iron, coal, textile, and transportation equipment industries. Later

on industrial development broadened through the spread of private-owned factories. Some of

the former state-owned factories were taken over by private owners. After 1894, the

coexistence of modern factories and traditional handcraft workshops was common in many

industries. Although traditional production was still dominant, we believe that the new

situation of foreign technology transfer is an important aspect of China’s early

industrialization. Therefore, our estimation not only focuses on the development of the proto

industry, but also on the results of early industrialization.

In the 1920s, the modernized sectors can be estimated at 20 percent of total industrial

production (XU and WU, 2005, p.1051). In our definition, a modern factory in China in the late

19th

century was characterized by new production technologies supported by European

machinery. In contrast, handcraft workshops kept the traditional way of production (see for the

same classification in Liu and Yeh, 1965). In practice, it can be difficult to make a distinction

between the two. Large and advanced workshops may have been just one step away from

modern factory production; on the other hand in factories without a well-trained labor force,

“new” production with novel technology may have existed only in name.

Our procedure goes in two steps. First, levels of value added in the traditional workshops and in

the factories using imported new technology are calculated for the years 1840, 1885, and 1920.

Next, we estimated a time series of output for both sectors. It shows that the proportion of

factory production in total industrial production increased from 5.4 percent in 1885 to 21.5

percent in 1920 (see Appendix B, Table B.7, column 1 for details). This indicates a rapid

increase of factory production in the early phase of industrialization.

3.2.1 Factory production

The reconstruction of value added produced in the modern factory system starts by making two

level estimates, for 1885 and 1920 respectively. These benchmarks are derived by adding up

the output of all major industries, as shown in Table B.7, column 1. We follow a procedure that

was used by Debin Ma (2008, p.367, Table 3) who estimated value added for the period

1914-1918 in the secondary and tertiary sector. Debin Ma projected the 1931-1936 industrial

sectors’ GDP levels backward to the period 1914-1918, using a growth rate estimated by

Rawski (1989). Here, we treat the year 1920 as the benchmark and back-project levels of value

added to 1885, using an average annual growth rate.

14

To estimate the annual growth rates during the period 1885-1920 we introduce a method based

on growth accounting (see details in Appendix B). In our Cobb-Douglas production function

we assume that the growth of the technology residual was close to zero at that time, because

technology was directly bought from the western world and took the form of capital

accumulation. Measuring growth means in this case that time series of capital and labor input

have to be estimated.

Two data sources have been combined to construct time series of capital input for factory

production. Table B.8, line 1, lists four estimates of capital input for 1885, 1894, 1913, and

1920, from XU and WU (2005, p.379 and p.1046). Their capital estimation includes

manufacturing industries, mining and production of basic metals using machinery,

transportation, and communication. They calculated the value of fixed capital on the basis of

quantity indicators, such as the number of spindles in textile mills, and tonnage of ships, etc.

For missing capital data they applied initial capital or net assets as proxies; for some industries

such as transport manufacturing, indicators like railway mileage or installed power-generating

capacity are applied as an approximation. They also distinguish factory production from

traditional workshops. To fill in all the missing data points between the benchmark years, we

added annual increments to initial capital during the period 1872-1912 (YAN, 1955, p.93-95).

Figure 2.4 shows that the rise in capital outlays accelerated after 1900.

With the estimated capital and labor inputs and the general price level for industrial products,

growth rates of factory production were calculated for the late 19th

century (see details in

Appendix B). Per capita output in factory production increased from 0.07 tael in 1885 to 1.84

tael in 1920. Our own time series generates a growth rate of 10.9 percent, which is reassuringly

close to the average growth rate of 9.5 percent that we get if we only use the two benchmarks.

The average growth rate of factory production, however, did not affect the total growth rate of

GDP significantly, because the initial proportion in GDP was less than 3 percent. As mentioned

earlier, agricultural production took up a major part of the whole economy but had an annual

growth rate lower than 1 percent. The momentum of early industrialization in late Qing China

was overshadowed by the sheer size of the agricultural sector.

We can do some cross checks. First, the consistency of our estimation can be evaluated. As an

upper limit, we quote Liu and Yeh’s estimation for 1933, which is another study that makes a

distinction between value added produced in the modern and in the traditional sector. Since we

have not included their data in our estimations for the benchmarks, it can safely be used as a

cross check. Liu and Yeh conclude that the proportion of the value added created by new

factory production was 5.4 percent of GDP in 1933. In our estimation for the period 1884-1912,

the proportion was less than 2 percent. The net value added of modern factory production per

capita is estimated by Liu and Yeh at 2.16 tael in the 1930s, whereas our average is only 0.66

tael for the whole period and 1.43 for the year 1912. Therefore, the present reconstruction of

new factory production in the late 19th

century may be an underestimation of the actual levels.

However, the increase of value added per capita may as well indicate a rapid process of

industrialization during the early 20th

century.

15

[Figure 2]

We can also check the reliability of the growth rates of industrial production that we have

found. Three figures (Figure 2.1-2.3) show that after the 1860s the economy experienced a

rapid increase in the imports of essential inputs for industrialization, such as machinery, coal,

and iron. The average growth rates were 10.2 percent for machinery imports between 1887 and

1916, 6.2 percent for coal imports in the period 1885-1920, and 4.7 percent for the iron, steel

and tin imports in the same period. These rates support the increase in output that we found. We

find that imports of these inputs accelerated after 1900, which looks quite similar to the

movement in capital inputs.

3.2.2 Cross checks for the total proto industrial sector

For the traditional manufacturing production during 1840-1920, use was made of

approximations that are shown in Table B.9. The question is whether we can rely on the general

trend given by the three existing benchmark estimates, which together implicitly indicate that

handcraft production decreased first, but then increased to an even higher level. This may be a

plausible development. But it is still difficult to support or reject the estimated trend because of

lack of data; more data research will be necessary. On average the net value in the proto

industrial sector during the period 1840-1912 was 1.61 tael per capita; the range was 0.52-3.41

tael; its average proportion in GDP was 10 percent; the average growth rate was 0.4 percent.

We have used an estimation for 1914-1918 from Debin Ma (2008, p.367, Table 3) as a cross

check. In his calculation, the per capita net domestic production of the industrial sector for

1914-1918 was 2.79 tael on average and its proportion in GDP was around 8 percent. The

present extrapolation based on the method explained before gives a level of 3.08 tael for the

period 1914-1918. Although still higher than Debin Ma’s estimate, our 1914-1918 estimate,

which is derived from a different method, is rather close. It supports our estimation for the

proto-industrial sector and the method used. However, we might overestimate net value added

for the entire sector. The estimate needs further improvements (see Appendix B, 2.3 for

details).

3.3 The service sector

Four sub-sectors have been included in our estimation for the service sector: public

administration, finance and housing, commercial activities, and the self-employed professions.

Here we concentrate on public administration, employing a new data source that distinguishes

16

our estimation for services from previous studies. For the three other sub sectors, we refer to

Appendix B for details (section 3.2-3.3).

3.3.1 Public administration

For a reconstruction of public administration we focus on government expenditure and

estimate its value added combining central government expenditure and the proportional share

of it in total governmental expenditure. The new dataset that we use in this section is compiled

from the historical document Hubu yinku huangce. As part of the governmental administrative

routine, the Qing state recorded in this document the annual central government income and

expenditure, and also its treasury reserves. The expenditure records mainly include military

spending, officials salaries, relief funds, administration costs, and royal funds. According to

SHI (2009), central government expenditure during the late Qing Empire was lower than 12

million tael, which had been the regular level in the early Qing Empire. But government

expenditure increased significantly after the Taiping Rebellion and after the Sino-Japanese

War.

We apply figures for the proportion of central government expenditure in the total government

expenditure of 13 percent before 1894 and 15 percent after 1894. These are much lower than

the level of 30 percent in the early Qing Empire (all figures based on SHI, 2009, p.102). This

may indicate the declining control power of the central government. To put this into a

long-term perspective: According to the official data from the China Statistical Yearbook

(2008), the proportion of central government expenditure in total government expenditure was

around 50 percent between 1978 and 1980 and about 25 percent in the recent decade.

In our estimation total government expenditure was also low; the per capita level was 0.27 tael

on average, and its proportion to GDP was 1.29 percent on average. According to the census in

1933, these levels were 1.1 tael per capita and 2.7 percent, respectively. In two previous studies

total government expenditure was estimated at 95 million tael in 1840 and 164 million tael in

the 1880s respectively (Chang, 1962; LIU T., 2009). However, our estimations are lower with

the 1840 level at 79.3 million tael, and the average level for the 1880s at 107.2 million tael

respectively. There can be two reasons for the difference: first, the original records that we used

may have covered only a part of central expenditure; second, we might have underestimated

the proportion of the central government expenditure in total government expenditure. Also

here further improvements will be necessary.

The documents show that in the late 19th

century the Qing state was in great trouble to maintain

its normal operations. As displayed in Figure 3, the proportion of public income and

expenditure to GDP decreased significantly in the second half of the Qing Dynasty. For most

years in this period expenditure was higher than income. In others years, the Qing state barely

sustained its balance. It means that public savings decreased continuously and possibly failed

to recover until the end of the empire.

17

One could easily ascribe these difficulties to the wars after 1840. However, we believe that the

crisis actually started in the early Qing Empire before the international wars broke out. The

decrease of the public income/GDP ratio and the public expenditure/GDP ratio can be traced

back to the late 18th

and early 19th

century (see Figure 3). Moreover, the tension between the

central and local controllers was likely to intensify. In most years of the Qing Empire, the

central government collected 25 to 33 percent of the total government revenue to support its

regular expenditure and it held high reserves before 1800 (SHI, 2009). The annual reserves in

the state treasury reveal that the state lost its control. E.g., to put down the White Lotus

Rebellion (1796-1805), the Qing state exhausted nearly 70 percent of the treasury reserves,

which decreased from 70 million tael in 1795 to 20 million tael in 1798. During the entire late

Qing Empire, the reserves remained around 10-20 million tael and never got back to a level of

35 million tael which was the regular level of reserves in the early Qing Empire. The

proportion of central revenue in the total declined from 22 percent before 1850 to 14.7 percent

in 1903 (SHI, 2009, p.43, 63).

[Figure 3]

3.3.2 Cross checks for the total service sector

By adding up the four sub sectors, we obtain total value added in the service sector for the

period 1840-1912: on average it was 2.71 tael per head, with a variation between 1.91 and 4.88

tael in the total period. The proportion of services in total GDP was around 18 percent; the

average growth rate was about 1 percent.

Also here previous studies for specific years can be used to compare our time-series

estimations with. The per capita value in this sector was 2.9 tael in 1840 according to LIU T.

(2009) and 2.3 tael in the 1880s according to Chang (1962). In the present reconstruction, the

1840 estimate is 2.35 tael, and the average level for the 1880s is 2.42 tael, respectively 20

percent lower and 5 percent higher than the existing figures. At the end of the Qing Empire, the

value added from the service sector reached 4.76 tael per capita in 1912, while previous

estimates arrive at 7.3 tael for the period 1914-18 (Debin Ma, 2008) and 6.81 tael for 1933 (Liu

and Yeh, 1965). We may have underestimated the growth rates of the service sector after the

1880s since these are based on movements in the population (see Appendix 3.2).

3.4 Nominal and real GDP between 1840 and 1912

The level of total nominal GDP according to our reconstruction is 6.00 billion tael in 1840 and

14.59 billion tael in 1912. The per capita nominal GDP estimates for both years are 14.57 tael

and 33.75 tael respectively (see Figure 4a). Taking a simple average for the whole period, we

18

arrive at a per capita level of 15.42 taels of silver, with a variation between 9.46 to 35.39 tael.

The growth rate of nominal per capita GDP between 1840 and 1912 was 1.2 percent per year on

average.

The Chinese economy in the late 19th

century consisted mainly of agricultural and service

activities, together between 83 and 95 percent of the total economy. The size of the proto

industrial sector was about 10 percent of GDP, but in the 1880s it was only 6 percent. This low

level is mainly caused by our use of the benchmark estimate for the proto industrial sector from

Chang (1962). In Maddison’s estimation for 1890, the percentage shares for the three sectors,

agriculture, industry, and services, were 68.5, 15.3, and 16.2, respectively (Maddison, 2007,

p.60). These figures were calculated by a backward extrapolation from the year 1933.

assuming a constant growth rate of sector GDP between 1890 and 1933 (p. 156), which is

highly unlikely. The percentage shares in our 1890s estimation are: 72.4, 8.1, and 19.5. These

differences show that there is room for more refinements in the calculation.

Figure 4a shows strong fluctuations in nominal GDP per capita during the 1850s-1860s and the

1900s-1910s. The main driving force is the price of rice. To give a better idea of the movement

of real GDP per capita, i.e. in volume terms, we need to adjust for price changes. To construct a

GDP deflator for the period 1840-1912, we compose a simple basket consisting of gold and rice

only, with weights of 28 percent and 72 percent respectively.12

The price of rice is

representative for basic-food crops. The rice price in late 19th

century China was volatile, while

the domestic price of gold in terms of silver was relatively stable (PENG, 1957; Wang, 1992).

The gold price is used as a proxy for the prices of non-agricultural products and services in the

economy. The weights of both items were established accordingly. The weight of the price of

rice is the average ratio of the value added of the agricultural sector in total GDP. We implicitly

assume that the gold price represents both industrial products and services. Adding other

commodities and services into the deflator, such as clothing and housing, will of course

produce a more precise indicator, reflecting the price behavior of more products. At this

moment, however, it is not possible to produce a better indicator because of lack of price series

for other products.13

Nevertheless, we believe that a simple deflator like the present one serves

our purpose to identify the broad trends in the movement of real GDP per capita.

In the period 1840-1912, real per capita GDP in 1840 prices is 14.53 tael on average, with a

variation between 12.53 and 16.38 tael for whole period. Figure 4b shows some major

movements in the trend. Real GDP decreased from an average level of 15 tael before the 1870s

to an average of 13 tael after the 1870s (in prices of 1840). Based on the beginning and end

points only we find a long-term annual decrease in GDP per capita of 0.08 percent. However,

an important turning point can be discerned around 1870. Before that year, GDP per capita

increased at an average annual rate of 0.35 percent. After 1870, there was a steady decline of

0.38 percent per year.

12 In a similar fashion, LIU T. used weights of 25 percent and 75 percent (LIU T. 2009, p.151). 13 There are estimations of general price levels for the late 19th century, such as WANG Y. (2005), but they only cover the

period after the 1860s. To avoid the problem of inconsistency, we define our own rice-gold price index for the total period.

19

[Figure 4]

How do these total economy estimates compare with other available estimates? In the

following we will confront our results with existing ones. LIU T. calculated a per capita level of

GDP in 1840 of 10.8 tael (LIU T., 2009, p.153). Our level estimate of 14.57 tael, is about 50

percent higher than his, which may indicate an overestimation in our results. Our estimation for

the 1880s is 11.78 tael, whereas Chang (1962) arrived at a level of 7.4 tael. The low figure of

Chang results from his low estimate of the area under cultivation. The minimum requirement

for a plausible GDP estimate is that people at least can survive on the estimated levels of per

capita GDP. It has been shown earlier that in the late Qing Empire food production could

support the daily energy requirement of the Chinese population. Here we infer the average

household budget in the Qing Dynasty. FANG X. (1996) estimates the average consumption

basket in the Qing Dynasty for a farmer family with five members living in the Jiang-Zhe

province. It covers food, cloths, rent, and fuel. He concludes that in the early Qing Empire the

level of annual household consumption was 32.6 tael and in the late Qing Empire 58.31 tael.

Following ZHANG Y. (2005), we add per capita education expenses to this estimation at an

amount of 1 tael or 5 tael per family. Thus, the estimated per capita consumption in the late

Qing Empire is 12.7 tael (63.31 divided by 5). This number probably overestimates the national

level because it relates to the rich Yangzi delta region. Furthermore price fluctuations in the

nineteenth century make it difficult to value this average consumption basket.

With a minimum level of 12.53 tael and an average above 14 tael in our real series, our

estimation of real per capita GDP from the production side is generally higher than the

estimated average consumption level. It follows that the aggregate output level was able to

sustain an above-minimum living standard for the population in the late 19th

century.

The new reconstruction provides a time series of annual GDP estimates. The second check of

our new estimation is whether it is consistent with the general circumstances of the late Qing

Empire. As mentioned before, the 19th

century is labeled in Chinese historiography as the

“years of troubles” or the “turbulent century”. Although real per capita GDP was high in the

1860s, the common theme in the historiography of the second half of the Qing Empire is one of

ongoing economic decline until the early 20th

century. Our estimation of real GDP supports this

general opinion. The study of Baten, Ma, Morgan, and Wang (2010) using real wages and

historical demographic data is in line with our results. They conclude that from the mid-19th

century there was a decline of living standards in South China followed by a recovery in the

early 20th

century.

The third and most important cross check for our estimation can be found in Broadberry et al.

(2013). The authors provide a detailed GDP reconstruction covering the period 980-1840. They

treat 1840 as their reference year for the price level (see their Table 3, p. 33). Their GDP per

20

capita estimate for 1840 is 12.95 tael, which is 11-12 percent lower than our estimate of 14.57

tael for the same year. The difference stems mainly from the applied assumptions of the

input/output ratio for grain crops. Broadberry et al. assumed a ratio of 18 percent, which in our

view is abnormally high (Broadberry et al., 2013, p. 38). We based our calculation on a ratio of

about 10 percent, which is more in line with existing studies (see Appendix B. 1.1.3).

Unfortunately, it is not possible to reconstruct their estimation by using our assumption of the

input/output ratio. Therefore, we applied their assumption to our study for a robustness check.

If we use the input/output ratio of 18 percent instead if 10 percent, our new 1840 estimate

would decrease to 13.61 tael. This would already explain already half of the difference between

our estimate and that of Broadberry et al. Because most studies apply a level of 10 percent for

the late Qing Dynasty we feel confident that this is a proper input/output ratio to base our

calculations on. It would also imply that there is only a small difference between the two

independent level estimates for Chinese per capita GDP in 1840.

4. Conclusion: main findings and remaining issues

Let us summarize the results of our estimate of Chinese GDP. We have employed new methods

and data sources, but derived our new estimation also from earlier studies on historical GDP.

One of the aims of our estimation was to construct a consistent time series of GDP by

improving and linking existing studies. For the agricultural sector a new time series of the land

area under cultivation was estimated and we added a multiple cropping ratio; for the industrial

sector, we focused on value added of factory production and applied a new method to estimate

its growth rates with a Cobb-Douglas production function; for the service sector, a new data

source was used to estimate government expenditure. On the basis of this a time series of

historical GDP was constructed for the late 19th

and early 20th

century in nominal and real

terms.

A major problem in the reconstruction of historical GDP is that the degree of autarky in

pre-modern China was probably high. The issue is in fact similar to the neglect of non-market

work in calculating GDP nowadays. For a low-commercialized economy, a

market-value-based GDP calculation may therefore generate imprecise results, especially if

households use their backyard for home production. Later improvements and more cross

checks for upper limits will therefore be necessary.

Now we will come back to the two questions that we raised in the second part of the paper. To

solve the inconsistencies between the various level estimates, it was necessary to get plausible

nominal GDP estimates and PPP convertors. This paper has focused mainly on the estimation

of nominal GDP and less on the international comparison of levels of welfare. Therefore, we

cannot completely solve the first question. Concerning the reconstruction of nominal GDP/cap

levels, we believe that both Chang (1962) and LIU T. (2009) underestimated levels of Chinese

GDP per capita. The nominal per capita GDP estimate for 1840 in LIU T. (2009) is 10.8 tael,

while the estimate for the 1880s in Chang (1962) is 7.4 tael. We arrived at higher levels for

corresponding years: 14.57 tael in 1840 and 11.78 tael for the 1880s. Our new estimation will

21

probably also solve a large part of the differences between the macro and sector studies in Table

2, once we have reliable international currency converters. This is something for future

research.

Our paper mainly intends to solve the inconsistency problem of growth rate estimates by

providing a consistent time series of GDP in both nominal and real terms. We believe that the

present reconstruction improves our understanding of the economy in late 19th

century China.

Figure 4b displays the graph of an economy that showed some per capita growth before the

1870s, but declined steadily afterwards until 1900. The economy started to recover in the

1910s, the end point of our reconstruction. The economic trends in our new estimation are

different from the stagnation that we see in Maddison’s estimation as well as from the pattern

of mild increase in the various sector studies. Maddison’s estimation indicates a turning point

in 1870. Per capita GDP declined from 600 GK dollars in 1850 to 530 GK dollars in 1870.

After that, the economy stagnated at a level around 550 GK dollars. Similarly, our estimation

indicates a turning point close to 1870, but instead our data show an economy that was in

ongoing decline after the 1870s. For the first half of the 19th

century, we find relative stable real

GDP/cap levels. This is quite in line with the corresponding estimates by Broadberry et al.

(2013). On the other hand, our data present a different picture than the real wage data

calculated by Allen et al. (2009), indicating a slow rise of real earnings of workers in the late

19th

and early 20th

century. Our real per capita GDP calculations suggest declining living

standards for the Chinese population after 1870.

It is also useful to make a comparison between two periods of economic shifts: the war period,

1840s-1850s, and the recovery period, 1860s-1870s (see the two shaded areas in Figure 4a and

3b). For the war period, the present estimation reveals an economy with slightly increasing

levels of real GDP per capita but with considerable nominal fluctuations. For the recovery

period, our estimation shows mainly decreasing trends in both nominal and real GDP per

capita. The comparison may indicate that the negative effect of wars and rebellions on the

Chinese economy was only regional and therefore possibly overstated. One way to check how

much the whole country was affected by the wars in the 1840s and 1850s is by studying the

records of the granary storage. Except for the regions that were directly affected by the Taiping

Rebellion (1851-64), the granary storage in other regions in the period 1840-1855 was stable

and similar to the levels in previous years. Even for the regions close to the rebellion area, such

as the Jiangxi province, the decline of granary storage only started after 1855, when the

rebellion was already four years underway (Will and Wong, 1991). The agricultural production

and the function and responsibilities of the political system were not affected in the early years

of wars and rebellions. Even with a high level of societal instability, the economy was able to

avoid an immediate decline. The real decline was delayed until after the 1870s.

Our new estimation suggests that the ongoing economic decline in the final decades of the 19th

century intensified the process of economic divergence between China and the modernized

Western economies. Figure 4b indicates that average levels of real output increased and

remained at a higher level in the 1850s-1860s. These were the decades when the economy was

increasingly involved in international trade, when new economic policies were introduced to

22

encourage knowledge transfer from western countries and technology absorption. At the same

time government investments were made in defense-related industries and in industries such as

textiles, mining, iron, and shipbuilding. However, the economy shifted quite suddenly into a

decline during the so-called self-strengthening movement (1860-1894), with social, economic,

and political reforms that were intended to catch up to the richer nations in the world. A deeper

investigation into the Chinese economy in the 1860s and 1870s will help us to explain the lost

chance of catching up in the 1860s, the “economic difficulty” in the 19th

century, and the

widening economic the gap vis-à-vis the West. Apparently opening up the economy came at a

cost.

Finally, our estimation provides a time series of GDP in current prices but also fixed-price GDP

estimates in silver taels of 1840. We believe that the real time-series may provide new avenues

for future research. First, we have not completely solved the first problem of inconsistency

between previous estimations. To compare our GDP estimates with estimates in international

dollars, a PPP convertor between silver taels and international dollars for late 19th

century

China will be necessary. We can apply the PPP convertor for the 1930s estimated by Fukao et al.

(2007), but a new PPP convertor for the 1910s, that is before the monetary disruptions caused

by WWI, would be more optimal. Second, it is possible to improve on the estimation for the

traditional handcraft sector by including detailed sub-sectors such as food processing, mining,

and construction.

23

Table 1. Historical GDP per capita estimates in previous studies, 1000-1933

a. Benchmark estimations

b. Time-series estimations

Sector Studies Macro studies

Period 1800-1900 1910s-1920s 1930s

1800 1840 1880s 1914-18 1920s 1933 1931-36 1934-36 1933

LIU

R.,

1987

LIU

T.,

2009

Broadberry

et al., 2013

Chang,

1962

Feuerwerker,

1969

Perkins,

1969

D. Ma,

2008a

XU and

WU, 2007

Ou,1947 Liu and

Yeh, 1965

Perkins

and Zhao,

1975

D. Ma,

2008e

Fukao et

al., 2007

Maddison,

2010

Benchmarks 6.7 10.8 12.95 7.4 39.3 112.6 52.4 35 46 59.8 123.4 57.4 60.5

Currency 1700

tael

1840

tael

1840 tael tael (current)

1933 yuan 1957 yuan 1930s

yuan

yuan

(current)

yuan

(current)

yuan

(current)

1957 yuan 1930s

yuan

1930s

yuan

Benchmarks 318 594 337.1b 411.6

b 221.4

c 549.5

b 129.2

c 482.4

b 626.6

b 242.6

c 601.9

b 619 579

d

Currency 1990

int.

dollars

1990 int.

dollars

1990

int.

dollars

1990 int.

dollars

1990USD 1990 int.

dollars

1990USD 1990 int.

dollars

1990 int.

dollars

1990USD 1990 int.

dollars

1990 int.

dollars

1990 GK

dollars

Macro studies Sector Studies

Period 1000-1933 1850-1913 1600-1850 1750-1850 1640-1840 1750-1840

Maddison, 2010 Maddison, 2010 Broadberry et al., 2013 Broadberry et al., 2013 LIU T., 2009 LIU T., 2009

Average 566 554 715e 618

e 352

f 317

f

Currency 1990 GK dollars 1990 int. dollars

Average 14.1 4

Currency 1840 tael 1600 tael

24

Source: See references, collected by the authors.

Most of the sector studies in Table 1a give results in terms of silver taels or Chinese yuan. We adjusted these estimates and presented them in terms of 1990 int. dollars

or 1990 USD. See the numbers with superscripts in Table 1a. a Here, the value represents net domestic product (NDP).

b The calculation is based on Fukao et al. (2007) , p. 17, Table 8. For these PPP adjusted estimates, we use the term “1990 int. dollars”.

c The calculation is based on the official exchange rates between China and the U. S. and the historical U.S. price levels. For these, we use the term “1990USD”.

The 1930s exchange rate: 1 U.S. dollar= 3.01Chinese yuan; 1 tael = 1.5 Chinese yuan .

The 1920s exchange rate: 1 U.S. dollar= 2.02 Chinese yuan. d The figure is not a direct benchmark estimate, but projected from Maddison’s 1990 estimation.

e The figure is calculated from Broadberry et al. (2013), p. 34, Table 4.

f The figure is calculated from LIU T. (2009), p. 153, Table 1.

Table 2. A comparison of average per capita GDP estimates in macro and sector studies, 1840s-1930s

Macro Studies

(1990 GK dollars)

Sector Studies

(1990 int. dollars)

Broadberry et al., 2013

1840s 318 594

(10.8 tael) (12.95 tael)

1850s 600 594

1860s

1870s 530

1880s 337.1-411.6

(7.4 tael)

1890s 540

1900s 545

1910s 552 549.5

1920s 562

1930s 570 482.4-626.6

1933 579 482.4-626.6

Source: Collected and calculated by the authors. The range of estimates from the sector studies is derived from Table 1a.

25

Table 3. Conflicts, disasters and the state treasury in Qing China, 1840-1912

1640-1840 1840-1912

(Early Qing Empire) (Late Qing Empire)

A. Number of conflicts

Total 60 54

Decadal average 6 8

B. Number of natural disasters

Total 2646 2698

Decadal average 15 30

C. Reserves in the state treasury

Average level (in million tael) 30 10

Source: Collected by the authors.

Table 4. Estimations of cultivated land area in various studies, Qing China, 1840-1912, in billion mou

LIANG F.,

1980

(official data)

Perkins,

1969

WU H.,

1985

SHI,

1989

ZHANG Y.,

1991

ZHANG Y.,

1991

LIU F. and

WANG Y.,

1996

ZHOU R.,

2001 GE et. al., 2008

Period 1851-1887 1873-1913 1840-1915 1840 1914 1851-1914 1840-1912 1840-1912 1873-1913

Average level 0.81 1.27 1.40 1.11 1.20 2.10 1.2

For a specific year 1.15 1.26

Source: Collected by the authors. 1 mou = 666.5 square metres

26

Figure 1. An estimation of food supply in Qing China, 1840-1915

a. Food crop production per capita in kilogrammes b. Daily energy intake per capita in Kcal

Sources: See the text, estimated by the authors.

27

Figure 2. Imports of industrial inputs and an estimate of capital input in Qing China, 1860-1920 2.1. Machinery imports, 1887-1916, in million yuan

2.2. Coal imports, 1867-1920, in million tael

2.3. Iron, steel, and tin imports, 1867-1920, in 1,000 tael

2.4. Capital input in factory production, 1885-1920, in 10,000 yuan

0

4

8

12

16

18

87

18

90

18

93

18

96

18

99

19

02

19

05

19

08

19

11

19

14

0

4

8

12

16

18

67

18

73

18

79

18

85

18

91

18

97

19

03

19

09

19

15

04812

16

186

7

187

2

187

7

188

2

188

7

189

2

189

7

190

2

190

7

191

2

191

7

051015202530

1885 1888 1891 1894 1897 1900 1903 1906 1909 1912 1915 1918

28

Source Figure 2: Collected and constructed by the authors.

Figure 2.1 is from CHEN Z. et al., 1957, p. 818. Figure 2.2 and 2.3 are from Hsiao, 1974. Figure 2.4 is constructed by the authors.

Figure 3. Public finance in Qing China, 1700-1912 (as a percentage of total GDP)

Source: Collected and calculated by the authors. The nominal GDP data before 1840 is derived from Broadberry et al., 2012, Figure 3. The data on public income and

expenditure is from SHI, 2009.

29

Figure 4. A new estimation of per capita GDP in Qing China, 1840-1912

a. Per capita GDP in current prices (in tael) b. Per capita GDP in constant prices of 1840 (in tael)

Sources: See the text, estimated by the authors. The two shaded areas represent the war period, 1840s-50s, and the period of recovery, 1860s-70s.

30

Appendix A . Background facts and figures

Table A.1. Empires in Chinese history

Empires Periods

Han Empire 206BC-220AD

Tang Empire 618- 907

Song Empire 960- 1279

Yuan Empire (Mongol) 1271- 1368

Ming Empire 1368- 1644

Qing Empire (Manchu) 1644– 1912

The early Qing Empire 1640- 1840

The late Qing Empire 1840- 1912 Republic China 1912- 1949

Source: Cambridge history of China

Table A.2. Major historical events in Qing China, 1796-1906

Wars White Lotus Rebellion 1796-1805

First Opium War 1839-1842

Second Opium War 1856-1860

Taiping Rebellion 1851-1864

Nien Rebellion 1852-1868

First Sino-Japanese War 1894-1895

Political or economic reforms Tongzhi restoration 1861-1875

Self-strengthening movement 1860-1894

The reform of 1901 1901-1906

Source: Cambridge history of China

Table A.3. Units of measurement used in the paper

Area 1 square kilometer= 1500 mou

Weight 1 jin = 500 grams

Exchange rates 1 Qing tael = 37.3 grams of silver.

Source: Perkins, 1969, p.1-2

31

Appendix B. A new estimation of historical GDP, 1840-1912

This appendix introduces and explains our GDP reconstruction. We present how we

estimated value added in the three sectors: the agricultural sector, the proto industrial

sector, and the service sector.

1. The agricultural sector

We have listed the major agricultural products and activities in Table B.1. In this section,

we will explain the calculation of the four categories of agricultural produce step by step

and concentrate mainly on the items in bold letters.

Table B.1. Four groups of agricultural products and activities

Groups Main products and activities

1 Food crops that were

land-intensive Rice

Wheat

(Grains)14

Millet

Kaoliang

Barley

Maize

Sweet potatoes

Other food crops

2 Non-food

but land-intensive crops Soy beans

Cotton lint

Mulberry and silkworm (raw silk)

Oil seeds

Peanuts

Astragalus root

Opium poppy

3 Non-food crops that were

not land-intensive Tea

Fruit and vegetable

Flower

4 Others Cattle farming, fishery, and forestry

Source: Chang, 1962; LIU R., 1987; XU and WU, 2005.

Items in bold type are treated in detail in the text.

14 See also footnotes 5 and 6.

32

1.1. Food crops

Food crops were calculated from the production side, summarized by equation (1.1):

�� = ∑ ������� (1.1)

, where Y is the gross value of food crop production measured in silver taels; A is the yield

per unit of the ith food crop in silver taels; K is the input of cultivated area for the ith food

crop (the crop area in a period); t is the year index. In total, we have eight food crops.

With the input/output ratio, we can then derive value added in food crop production. We

will first introduce how we reconstructed K, then A, and finally the input/output ratio.

1.1.1. Calculating land input K

We apply the following equation to estimate and construct a time series of K.

��� = �� ����� ����� ∗ ������� �� �� ���� ���� ∗ ��� ��� ���� �� ��� �

(1.2).

The three parts in the above equation were reconstructed as follows.

1.1.1.1 The cultivated land area

We applied a method where we combined two level estimates and an interpolation

procedure where we fitted a trend line in between both level estimates. The procedure of

fitting the trend can be explained with a simple example. It contains only two unknown

data points, not the 71 data points which are actually calculated in the paper.

Suppose there is a four-year time series of the cultivated area (Y), referred to as the old

time series, and for which we know the relative level for each year, that is, the annual

change in area (see Table B.2). Suppose also, that we cannot use this series directly, since

its average level is not a robust estimator for the total area. Next, we calculate two new

levels for the total area available for cultivation (X). Now we have only level estimates

for year 1 and 4, denoted as E1 and E4 in Table B.2. To fill in the two missing data points,

X2 and X3, we “fit” the trend of the old time series into the new one by assuming the

same rates of annual changes between the two, denoted as W3 and W4 in Table B.2. Here,

we actually assume that the second-order differences across years are the same for the

two time series. We give the details of this method of fitting in Table B.2.

Taking the rates of annual changes from the old time series (W3 and W4), we reinterpret

and present the relationship between the different time points for the new time series. To

calculate X1 and X2, the only unknown is v2, which can be derived from the relationship

between E1 and E4 (see Table B.2, last column). Following the same procedure for more

years, we obtain the estimated time series of the cultivated area in Figure B.1. Here, the

33

old time series is from WU H. (1985) which provides annual changes for our estimation,

and the new time series we construct connects the 1840 and the 1914 estimate given by

SHI (1989) and ZHANG Y. (1991), respectively (See Table.4.). We also provide the

maximum and minimum levels from the existing literature.

Table B.2. An interpolation procedure for the reconstruction of the area under cultivation

The old time series The new time series

Year Levels Annual

changes

Rates of

annual

changes

Annual changes Levels

1 Y1 E1

2 Y2 V2

=Y2/Y1

v2 X2 = v2* E1

3 Y3 V3

=Y3/Y2 W3

=V3/V2

v3 = v2* W3 X3 = v3* X2

= (v2* W3)*(v2* E1)

= (v2)^2* W3* E1

4 Y4 V4

=Y4/Y3 W4

=V4/V3

v4 = v3* W4

= (v2* W3)* W4

E4 = v4* X3

= ((v2* W3)* W4)*( (v2* W3)*(v2* E1))

= (v2)^3*(W3)^2* W4*E1

Source: Constructed by the authors.

Figure B.1. The estimation of the cultivated area (after interpolation), 1840-1914, in billion

mou

Source: Calculated by the authors.

In our estimation, the starting point of 1.15 billion mou in 1840, is from SHI (1989). The end point

at 1.26 billion mou in 1914, is from ZHANG Y. (1991). See Table 4. The annual growth rates are

from WU H. (1985).

1.1.1.2 The proportion of basic-food crop cultivation

Not all the cultivated area was used for food production; and some non-food crops were

also intensive in land use, such as soy beans. In our calculation, we believe that the proper

proportion of the basic-food crop cultivation in the late Qing Empire was around 85%. By

0

0,4

0,8

1,2

1,6

18

40

18

44

18

48

18

52

18

56

18

60

18

64

18

68

18

72

18

76

18

80

18

84

18

88

18

92

18

96

19

00

19

04

19

08

19

12

The lower limits: LIANG (1980)

Our estimation

The upper limits: WU Hui(1985)

34

surveying the existing literature, we find that the figure decreased in the course of time.

GUAN and D. Li use 92.35% as the plausible level for the Ming Empire (GUAN and Li,

2010, p.7). WU H. suggests that the proportion for the late Qing Empire should be below

85% (WU H., 1985, p.199). However, most studies apply a level of 90% for the early

Qing Empire (GUO, 1994; XU and WU, 2005; LIU T., 2009). We suppose that the eight

food crops in Table B.1 took up around 90% of the cultivated area in total. Based on this

assumption, we adjusted the proportions of the eight different food crops.

1.1.1.3 Relative proportions of different food crops

The two steps of our adjustments are documented in Table B.3. First, we take averages of

the three data sources from the 18th

century to the 1930s, and apply these averages to the

period 1840-1912 (see Table B.3, column 1-3). Liu and Yeh also use the similar averages

in their estimation for 1933 (Liu and Yeh, 1965). For instance, the proportion of rice

cultivation was 29.7% in the 18th

century, 29.3% in the 1910s, and 28.3% in the 1930s.

We apply their average of 29.1% in our calculation. Here, we assume that the proportions

did not change over time. In other words, we implicitly assume a fixed technology in the

use of the cultivated land during the period 1840-1912.

In a second step, we made adjustments for maize. We use the data collected by YAN

(1955), in which maize cultivation was 6% in the 1850s and 11% in the 1900s (YAN,

1955, p.359). We applied an average of 8.5% in our calculation.

Table B.3. The relative proportions of food crops used in previous studies and our

estimation, in %

LIU R., 1987 Perkins, 1969 Buck, 1930, 1937 Our estimation

1700-1800 1914-1918 the 1920s-1930s 1840-1912

Rice 29.7 29.3 28.3 29.1

Wheat 18.9 24.5 16.3 19.9

Millet 11.34 4.2 9.4 8.3

Kaoliang 11.7 9.9 4.7 8.8

Barley 9 8.5 4.4 7.3

Maize 3.6 5.58 3.7 8.5a

Sweet potatoes 1.8 1.8 2.4 2

Other food crops 3.96 11.5 3.7 6.4

In total 90 95 72.9 90

Source: Collected by the authors. a For maize, we have followed the records in YAN (1955).

1.1.1.4 The multiple cropping ratio

As mentioned in section 3.1.1, we apply a ratio of 1.32 in our estimation. Now we have

the amount of land input. Accordingly, we need to know the value of output per unit in

order to compute the gross value.

35

1.1.2. A calculation of the yield per unit (A)

A is the yield per unit measured in silver. For the ith food crop, we estimate A by equation

(1.3).

��� = � ����� × �� ��� (1.3)

We will treat the two parts of the equation separately.

1.1.2.1. Yields per unit area

For rice, wheat, and corn, we backwardly project their unit yields in the 1930s to the late

Qing period, using the yearly harvest ratios (CHEN, 1996, p. 419). The harvest ratios

used in our estimation are mainly based on historical records of local reports about food

crop harvests. For instance, local magistrates reported to the emperor that the output of

wheat in a certain year was equivalent to 90 percent of a full harvest. Estimations of

historical weather conditions were applied in case of missing reports. We used official

statistical records for the unit yields in the 1930s (see Table B.4, column 3).

Next, the following two steps were taken in our calculation. First, for each food crop we

took the average between the maximum and the minimum among the previous

estimations (Table B.4, column 1-2). We use these averages as the yield estimates for the

period 1840-1912. In the second step we include sweet potatoes and other food crops. For

these two, we directly apply the estimations from WU H. (1985) and Liu R. (1987) (See

Table B.4, column 4).

Figure B.2. Unit yields of rice, wheat, and maize, in Jin per mou

Source: Calculated by the authors.

0

100

200

300

18

40

18

44

18

48

18

52

18

56

18

60

18

64

18

68

18

72

18

76

18

80

18

84

18

88

18

92

18

96

19

00

19

04

19

08

19

12

Rice

Maize

Wheat

36

Table B.4. The yields per unit for different food crops and our estimation, in Jin per mou

The yield estimates

in the existing literature

Official records

(Republic China) Our estimation

The 1700s-1930s The 1930s 1840-1912

Minimum Maximum Averages

Rice 162 261 208

Wheat 114 160 139

Millet 159 180 168 170

Kaoliang 160 190 182 175

Barley 133 191 167 162

Maize 152 193 178

Sweet potatoes 185 263 242 250a

Other food crops 63 180 160a

Sources:

For rice, the minimum is from the estimation for the 1910s (Perkins, 1969); the maximum is from

the estimation for the 1880s (Buck, 1930 1937).

For wheat, the minimum is from the estimation for the 1920s-1930s (LIU K., 2001); the

maximum is from the estimation for the Qing China before 1840 (LIU T., 2009).

For millet, the minimum is from the estimation for the 1930s (Perkins, 1969); the maximum is

from the estimation for the 1880s (Chang, 1962).

For kaoliang, the minimum is from the estimation for the 1930s (Buck, 1930 1937); the maximum

is from the estimation for the 18th

century (LIU R., 198715

).

For barley, the minimum is from the estimation for the 1930s (Buck, 1930 1937); the maximum is

from the estimation for the 18th

century (LIU R., 1987).

The official data in column 3 is from YAN (1955), Liu and Yeh (1965), Perkins (1969), and

ZHANG Y. (1990). a For these two items, we use the estimation in WU H. (1985) and Liu R. (1987).

1.1.2.2. Prices

There are also two steps in our reconstruction of price information. First, we collected

time series of rice prices. We use the price in Yangzi delta to represent the country level,

since it takes the middle position in the prices of rice in six different regions and it is

comparable to the country-level price given by PENG, 1957 (see Figure B.3.). Second,

using the price ratios between rice and other food crops (PENG, 1957; LIU T., 2009), we

obtain the price series for other food crops. For example, we have three ratios from

previous studies: for rice, 1 shi = 150 jin; for wheat, 1 shi = 142 jin; wheat price per shi =

0.8* rice price per shi.16

We calculated the price ratio between wheat and rice at 0.84517

.

15 The yield estimates in LIU R., 1987, are based on Liu and Yeh, 1965. 16 Jin is a Chinese weight unit, 1 jin = 500 grams.

Shi refers to an ancient Chinese volume unit for measuring grains. For different crops, the weights for one shi are

different. 17 We calculate the price ratio between rice and wheat as follows. ����� !"#$���%�&

�����'()"���%�&= ����� !"#$���*+�/-��.+� !"#$���*+�

�����'()"���*+�/-��.+�'()"���*+�= ����� !"#$���*+�

�����'()"���*+� × -��.+�'()"���*+�

-��.+� !"#$���*+�= 0.8 × 234

256= 0.845

37

Here, we assume that the price ratio between rice and wheat is constant over time. Figure

B.4 approximately justifies our assumption. The price ratio between wheat and rice in the

province of Shandong remained approximately constant in the late 19th century.

Figure B.3 Regional rice prices, 1840-1915, in taels per shi

Sources: The price data for Yangzi Delta is from Wang (1992), Table 1.1. The price data for other

provinces is from “Grain Price Table Between 1821 and 1912”18 and the Qing-era Grain Price

Database19. Ten-year averages are from PENG (1957), p. 459.

Figure B.4 Rice and wheat prices, Shandong Province, 1840-1911, monthly, in taels per shi

Sources: The price data is from “Grain Price Table Between 1821 and 1912”.

1.1.3. Calculating the input/output ratio

To derive value added in food production, we need an input/output ratio. Table B.5

contains several estimates used in previous studies. As explained in section 3.4, we apply

18 We are grateful to Bas van Leeuwen (Utrecht University) for kindly providing the price data. 19 http://140.109.152.38/

38

the ratio used by LIU T. (2009), 0.096.

Table B.5. The input/output ratios used in previous studies

Ou, 1947 Buck, 1930, 1937 LIU R., 1987 LIU T., 2009 Broadberry et al.,

2013

The 1930s The 1920s-1930s 1700-1800 1600-1840 1640-1840

0.14 0.1 0.1 0.096 0.18

Source: Collected by the authors.

1.2 Non-food but land-intensive crops

The rest of the cultivated area was left for non-food crop production. As classified in

Table B.1, they are also land-intensive. In general, our method is as follows: first, we take

the most important crops and estimate their value; second, we estimate the relative

proportion of those selected crops in the entire group. We calculated these food items

from the consumption side, so we had to add exports and subtract imports.

We pick two items to simplify our calculation: cotton and raw silk. These two products

were respectively the main import and export goods for Qing China in the late 19th

century. The import of cotton and cotton goods took over 30 percent of total imports value

in the 1870s and 1880s. In the same period, the export of silk product amounted to around

25 percent of total exports (YAN, 1955, p. 76; XU and WU, 2005, p. 84). In 1913, the

import ratio of cotton and cotton goods was 32 percent, and the export ratio of silk

product was 25.3 percent (XU and WU, 2005, p. 722).

By studying the two products, we can cover a major part of the crop production in this

group. The price, domestic sale, and foreign trade data for the two products can be found

in YAN (1955), YAO (1962), Hsiao (1974), and XU and WU (2005). We interpolate

missing-data points.

Previous studies give us clues about the value proportion of the two products in this

category. The number was 26.8% in the 1910s (XU and WU, 2005, p.1098) and 21.4% in

the 1930s (Perkins and Zhao, 1975, p.385). The proportion of cotton and raw silk was

over 30% in the early Qing Empire (LIU T., 2009, p.147). We apply the average of the

1910s and the1930s, 24.1%, in our calculations. Using equation (1.4), we estimate the

value added for this category of agricultural product.

Value of Non-food but land-intensive crops

= (�����:��:& + �����<-*�=>)/24.1% (1.4)

1.3 Remaining non-food crops

The rest of the non-food crop production was less intensive in land use. Applying the

same method as in the previous section, we take tea as the most important product. The

proportion of tea export in total exports was around 50% in the 1880s (YAN, 1955, p.76).

39

Even though its export proportion had decreased since the 1870s, tea was still the most

important export good for Qing China (XU and WU, 2005). The price and output (exports

and domestic sales) data can be found in YAN (1955), YAO (1962), Hsiao (1974), and

XU and WU (2005).

In previous sections, we have obtained the value of food crops, non-food but

land-intensive crops, and tea. By adding up the three categories, we have the main

agricultural products for per-modern China, defined by Ou (1947) and XU and WU

(2005). The ratio of the remaining agricultural non-food products to the main agricultural

products was 10.09% in the 1930s, estimated by XU and WU (2005, p.1099). Based on

this ratio and equation (1.5), we estimate the value added for this category of agricultural

product.

Value of the remaining non-food products

= C����D::E��:�* + ����&:&FD::EGH�=<&EF�&��&*�I���:�* + ������<J × 10.09%

+������< (1.5)

1.4 Other agricultural production

Finally, we add other agricultural activities into the previous estimation, such as cattle

farming, fishery, and forestry. Based on Table B.6, we believe that they accounted for

about 12% of the value added in food crop production in the late Qing Empire as shown in

equation (1.6).

Value of other agricultural products= ����D::E��:�* × 12% (1.6)

By adding up the four categories, value added in the agricultural sector was 4.13 billion

taels of silver in 1840; 11.34 billion taels of silver in 1912. Figure B.5 shows per capita

estimates in current prices.

Table B.6. The value of other agricultural production

LIU R., 1987 LIU T., 2009 Perkins, 1969 Ou, 1947 Broadberry et al.,

2013

1700-1800 1640-1840 1910s 1930s Early Qing

12%a 10%

a 11%

b 18.13%

a 12%

b

Source: See the references, collected by the authors. a These ratios are relative to the main agricultural products which include food crops, non-food but

land-intensive crops, and tea. b These ratios are relative to food crops.

40

Figure B.5 Value added in the agricultural sector, per capita, 1840-1912 (in tael)

Source: Estimated by the authors.

2. The proto-industrial sector

The first step in our estimation is to separate factory production from handcraft

workshops. We do this for three benchmark years, 1840, 1885, and 1920 (see Tables B.7

and B.9). The total output values are respectively: 2254 million tael in 1840 (LIU T.,

2009); 485 million tael in 1885 (Chang, 1962); and 4029 million tael in 1920 (XU and

WU, 2005).

In fact, we only need to split the estimate for 1885 (Chang, 1962). XU and WU (2005)

already provide the value for the two types of production in 1920. There is no need to split

the estimate for 1840 as we assume that all industrial production was traditional by

nature; foreign technology entered China only after the 1860s. Chang (1962) does not

distinguish between the two types of production in his estimation for the 1880s.

Our separation for the 1885 estimate is based on the descriptive documents in XU and

WU (2005). After these adjustments, we obtain a new benchmark estimate for factory

production: 26.1 million tael in 1885.

41

Table B.7. Factory and handcraft workshop production in 1885 and 1920, in million tael

Total value of the industrial production in 1885

Factory Handcraft workshop Total

Manufacturing 22 302 323.86

Mining and manufacture of

basic metals 1.55 126.40 127.94

Transportation 2.00 30.00 32.00

Communication 0.77 0.00 0.77

Total 26.10 458.48 484.57

Proportion 5.39% 94.61% 100.00% Total value of the industrial production in 1920

Factory Handcraft workshop Total

Manufacturing 588.58 2840.39 3428.97

Mining and manufacture of

basic metals 70.44 123.23 193.67

Transportation 189.18 199.51 388.69

Communication 15.86 1.70 17.56

Total 864.06 3164.82 4028.88

Proportion 21.45% 78.55% 100.00%

Source: Part A is mainly from Chang, 1962. Since it does not directly provide the value of the

two types of production, we make the separation here and list the value of the new production

according to XU and WU, 2005. Part B is from XU and WU, 2005, p.1051.

1 tael of silver=1.5 Chinese yuan in the 1930s.

2.1 The total value of factory production during the period 1885-1920

Our estimation of the growth rates applies growth accounting techniques, but is limited to

one sector. We make the following assumptions. Equation (2.1) is the assumed production

function with the property of constant returns to scale,

(2.1)

, where Value is the total value of the factory output; P is the general industrial price level;

Y is the factory output level; K is fixed capital input; L is labor input in factory production.

The parameters, and A, are assumed to be constant for the period concerned.

We describe the data used in this section as follows. The dataset of the annual industrial

price levels during 1885-1920, i.e. , is from WANG Y., 2005,

p.471, Table.6. We have four data points of capital input and three data points of labor

input as listed in Table B.8. To deal with missing data, we use the initial capital inputs of

the new factories established annually as the increments (YAN, 1955, p.94). By

1

t t t t t tValue P Y P AK Lα α−

= × = ×

α

{ } , 1885, ......, 1920tP t∀ =

42

connecting the four points in Table.B.8, line 1, we get a rather smooth line of capital

inputs during 1885-1920 (see Figure 2.4). To obtain annual labor input, we assume that

the proportion of factory labor to the total population moved steadily over time. Then, we

construct the time series of capital and labor inputs during 1885-1920, i.e.

.

Table B.8. Capital and labor input in factory production, 1885-1920

1885 1894 1913 1920

Capital input (in million yuan) (K) 29.64 121.55 1786.73 2579.29

Labor input (L) 90,140 693,890 413,040

Proportion of the population 0.02% 0.16% 0.09%

Source: Capital data are from XU and WU, 2005, p.379 and p.1046. Labor data for 1885 are

from XU and WU, 2005, p.379; for the period 1912-1920, data are from Lieu, 1936, p. 91.

Population data are from Maddison, 2010.

Next, we turn to the unknown parameters in equation (2.1), and A. Because A is a

constant in our assumptions, it disappears from the growth rates according to equation

(2.1). We try to find the proper for the period concerned, given the information from the

two point estimates of factory production in Table B.7. Rearranging equation (2.1) with

the year 1920 as the base year, we have equation (2.2).

(2.2)

Using the data from Table B.7, column 1 and Table B.8, column 1 and 4, we derive the

parameter by equation (2.2), which equals 0.48.

We obtain equation (2.3) by putting the parameter into equation (2.2). Then, we

calculate the annual growth rates during 1885-1920, denoted by g.

(2.3)

By equation (2.3), we construct the time series of the total value of the factory production

during the period 1885-1920, i.e. . Figure B.6 shows our

results in current prices.

{ }, , 1885, ......, 1920t t

K L t∀ =

α

α

1885 1885 1885 1885

1920 1920 1920 1920

ln ln ln (1 ) lnValue P K L

Value P K Lα α= + + −

α

α

{ }

1920 1920 1920 1920

ln ln 0.48*ln (1 0.48)*ln

, where , , , 1885, ......, 1920

t t t t

t

t t t

Value P K Lg

Value P K L

P K L t

= = + + −

∀ =

{ } , 1885, ......, 1920t

Value t∀ =

43

Figure B.6. Total value of factory production, 1885-1920, in million tael

Source: Constructed by the authors.

2.2 Total value of handcraft production during the period 1840-1920

Starting from the three benchmark estimates in Table B.10, line 2, we assume that the per

capita value in traditional workshops moved steadily during 1840-1920. With the

estimated traditional output per capita and population data, we can derive a time series of

the total value of traditional production during the period concerned.

2.3 The proportion of net value

Ou (1947) shows that in 1933 the net value proportion for mining and metal manufacture

was around 50%; for textile manufacturing it was around 29%-30%; for ceramics

manufacturing it was 60%. In general, we take 40% for the industry to calculate value

added. Later improvements are necessary because the choice of the ratio here is a rough

average and takes no account of shifts in the structure of output. Finally, we have a time

series of value added in the proto industrial sector, from 0.9 billion taels of silver in 1840

to 1.2 billion tael of silver in 1912. Figure B.7 shows per capita estimates in current

prices.

Table B.9. Total value of traditional production, 1840-1920

LIU T., 2009 Chang, 1962 XU and WU, 2005

1840 1885 1920

Total value

(in million tael) 2250

20 458.48 3164.82

Per capita levels

(in tael) 5.46 1.23 6.71

Source: Collected by the authors. Population data is from Maddison, 2010.

20 LIU T. (2009) gives the net value directly, i.e. 900 million tael. We use the net value ratio, 40%, to obtain the total

value.

02004006008001000

1885 1890 1895 1900 1905 1910 1915 1920

44

Figure B.7. Value added in the proto industrial sector, per capita, 1840-1912 (in tael)

Source: Estimated by the authors. We keep the same scale as used in the figure of the agricultural

sector, Figure B.5.

3. The service sector

In this sector, we include four economic activities: public administration, finance and

housing, professional work, and commercial activities.

3.1 Public administration

Here, we mainly have to deal with the missing data problem in the archival data from SHI

(2009, p.222-253). In general, we put the average for an emperor period into the missing

data points in between the specific period ruled by the emperor. For the missing data

during 1821-1850, we impute the average expenditure under the Daoguang Emperor

(1821-1850), which was 11 million tael. Figure B.8 shows the archival records and the

estimates after an interpolation for central government expenditure. Using the proportion

of central government expenditure in the total government expenditure, we obtain the

value for this subsector (See section 3.3.1).

3.2 Finance, housing, and other professional work

We calculate the value of the early financial sector based on the three benchmark

estimates in Table B.10, including real estate and renting. For simplicity, we assume

constant annual growth rates for per capita values in this subsector during 1840-1933.

With the population data, we derive a time series of the net value. The same assumption

and method is applied to estimate the value created from professional work based on

Table B.11.

3.3 Commercial activities

For commercial activities, we refer to wholesale and retail trade. We use a ratio to derive

0

5

10

15

20

25

301

84

0

18

45

18

50

18

55

18

60

18

65

18

70

18

75

18

80

18

85

18

90

18

95

19

00

19

05

19

10

45

the value added in this subsector. In Chang’s estimation, the commercial/non-commercial

ratio was 0.086 in the 1880s (Chang, 1962); in Maddison’s estimation, it was 9.7% in

1890 (Maddison, 2007, p.254); in Ou’s estimation, it was 10.7% in 1933 (Ou, 1947,

p.12); for Liu and Yeh, it was 10.4% in 1933 (Liu and Yeh, 1965, p.66). Based on their

estimates, we suppose that the ratio was around 9% during 1840-1912. For

non-commercial activities, we refer to the agricultural and industrial production.

Equation (3.1) is used to calculate the value of this subsector.

Value of commercial activities = (����<.���H=�H�� + �����&EH*��L) × 9% (3.1)

Adding up the four subsectors, we get the value added levels in the service sector: 0.97

billion tael of silver in 1840 and 2.06 billion taels of silver in 1912. Figure B.9 shows per

capita estimates in current prices.

Figure B.8. Central government expenditure, 1840-1912, in million tael

Source: Constructed by the authors.

Table B.10. The net value in the financial and housing sector, 1840-1933

LIU T., 2009 Chang, 1962 Liu and Yeh, 1965

1840 1880s 1933

Net value (in million Tael) 320 238.6 826.7

Per capita levels (in Tael) 0.78 0.64 1.65

Source: Collected by the authors. Population data are from Maddison, 2010.

Table B.11. The value added from professional work, 1840-1933

LIU T., 2009 Chang, 1962 Liu and Yeh, 1965

1840 1880s 1933

Net value (in million Tael) 115 241.3 226.7

Per capita levels (in Tael) 0.279 0.645 0.453

Source: Collected by the authors. Population data are from Maddison, 2010. LIU T.’s estimation

for professional work follows Chang’s estimation. Liu and Yeh’s estimation for the value from

personal services is cited in the above table.

46

Figure B.9. Value added in the service sector, per capita, 1840-1912 (in tael)

Source: Estimated by the authors. We keep the same scale as used in the figure of the agricultural

sector, Figure B.5.

4. Nominal and real per capita GDP in the Chinese economy 1840-1912

By adding up the above estimates for the three sectors, we obtain a new estimation of

nominal GDP. With the population data from Maddison’s research and the estimated GDP

deflator from section 3.4 in our paper, we calculate real per capita GDP. We present our

estimation of nominal and real GDP in Table B. 12.

Table B.12. Nominal and real GDP per capita for Qing China, 1840-1912, in tael. Real GDP/cap

in prices of 1840

Per capita

GDP Value added per capita

GDP

deflator

Per capita

GDP

Agricultural

sector

Proto industrial

sector

Service

sector

current prices current prices current prices current prices 1840 prices

1840 14.57 10.03 2.18 2.35 1.00 14.57

1841 15.51 10.84 2.15 2.52 1.02 15.24

1842 15.43 10.84 2.11 2.48 1.01 15.26

1843 13.68 9.31 2.07 2.30 0.93 14.76

1844 13.84 9.52 2.03 2.28 0.96 14.38

1845 13.33 9.09 2.00 2.24 0.92 14.47

1846 11.88 7.75 1.96 2.17 0.82 14.47

1847 12.14 8.07 1.92 2.15 0.83 14.56

1848 12.38 8.28 1.88 2.22 0.84 14.69

1849 13.32 9.20 1.85 2.27 0.91 14.65

1850 14.04 9.89 1.81 2.34 0.97 14.54

0

5

10

15

20

25

301

84

0

18

45

18

50

18

55

18

60

18

65

18

70

18

75

18

80

18

85

18

90

18

95

19

00

19

05

19

10

47

1851 12.99 9.03 1.77 2.19 0.90 14.39

1852 9.46 5.75 1.73 1.98 0.63 15.01

1853 9.56 5.90 1.70 1.97 0.64 14.92

1854 9.51 5.95 1.66 1.91 0.65 14.62

1855 9.82 6.30 1.62 1.91 0.68 14.51

1856 10.21 6.69 1.58 1.93 0.68 14.93

1857 15.36 11.40 1.54 2.42 1.04 14.73

1858 16.38 12.41 1.51 2.46 1.08 15.17

1859 14.24 10.47 1.47 2.30 0.96 14.84

1860 11.29 7.73 1.43 2.13 0.75 15.07

1861 17.36 13.34 1.39 2.62 1.13 15.39

1862 23.09 18.59 1.36 3.15 1.46 15.82

1863 21.87 17.53 1.32 3.02 1.36 16.13

1864 22.49 18.10 1.28 3.11 1.39 16.17

1865 17.20 13.29 1.24 2.67 1.10 15.70

1866 18.29 14.32 1.21 2.76 1.14 16.00

1867 14.93 11.24 1.17 2.53 0.93 16.00

1868 13.07 9.58 1.13 2.35 0.80 16.38

1869 13.71 10.16 1.09 2.46 0.85 16.20

1870 14.33 10.80 1.06 2.47 0.89 16.17

1871 13.27 9.87 1.02 2.38 0.83 15.95

1872 12.82 9.49 0.98 2.35 0.80 16.04

1873 12.10 8.86 0.94 2.30 0.78 15.54

1874 11.48 8.33 0.90 2.25 0.78 14.71

1875 10.86 7.70 0.87 2.30 0.72 15.09

1876 11.29 8.12 0.83 2.34 0.72 15.60

1877 12.16 8.96 0.79 2.41 0.83 14.72

1878 14.44 11.08 0.75 2.61 0.97 14.91

1879 12.54 9.43 0.72 2.39 0.83 15.15

1880 10.86 7.86 0.68 2.32 0.73 14.92

1881 9.99 7.10 0.64 2.25 0.67 14.85

1882 11.48 8.50 0.60 2.38 0.78 14.73

1883 12.34 9.32 0.57 2.45 0.84 14.76

1884 11.57 8.65 0.53 2.39 0.81 14.35

1885 11.09 8.22 0.52 2.36 0.78 14.30

1886 12.90 9.79 0.59 2.52 0.92 14.08

1887 12.78 9.59 0.66 2.53 0.93 13.80

1888 11.93 8.72 0.73 2.47 0.86 13.87

1889 12.90 9.53 0.81 2.57 0.92 14.02

1890 13.80 10.26 0.89 2.65 1.02 13.47

1891 12.96 9.34 0.96 2.66 0.95 13.68

1892 13.49 9.82 1.02 2.66 1.00 13.44

1893 13.86 10.05 1.10 2.70 1.03 13.51

1894 13.77 9.80 1.17 2.80 1.03 13.36

48

1895 14.79 10.58 1.24 2.97 1.07 13.78

1896 15.31 10.98 1.31 3.02 1.17 13.12

1897 15.23 10.81 1.41 3.01 1.15 13.25

1898 18.77 13.88 1.50 3.38 1.46 12.83

1899 17.50 12.67 1.61 3.22 1.35 12.92

1900 17.60 12.64 1.69 3.27 1.34 13.09

1901 17.00 12.01 1.76 3.23 1.30 13.08

1902 21.66 16.14 1.89 3.63 1.60 13.51

1903 21.54 15.91 1.99 3.63 1.60 13.42

1904 21.85 16.10 2.08 3.67 1.60 13.70

1905 17.56 12.06 2.17 3.33 1.40 12.53

1906 19.56 13.78 2.27 3.51 1.49 13.14

1907 23.80 17.54 2.39 3.87 1.77 13.42

1908 24.92 18.47 2.47 3.98 1.80 13.83

1909 21.98 15.69 2.53 3.75 1.57 13.99

1910 25.49 18.83 2.61 4.05 1.76 14.49

1911 35.39 27.80 2.71 4.88 2.46 14.41

1912 33.75 26.24 2.75 4.76 2.45 13.78

average 15.42 11.26 1.45 2.71

14.53

max 35.39 27.80 2.75 4.88

16.38

min 9.46 5.75 0.52 1.91

12.53

Source: Constructed by the authors.

49

References

Allen, R.C., Bassino, J., Ma, D., Moll-Murata, C. & Zanden, J.L.v. 2011, "Wages, prices,

and living standards in China, 1738-1925: in comparison with Europe, Japan and

India", The Economic History Review, vol. 64, no. s1, pp. 8-38.

Baten, J., Ma, D., Morgan, S. & Wang, Q. 2010, "Evolution of living standards and

human capital in China in the 18–20th centuries: Evidences from real wages,

age-heaping, and anthropometrics", Explorations in Economic History, vol. 47, no.

3, pp. 347-359.

Brandt, L., Ma, D. & Rawski, T.G. 2012, From divergence to convergence: re-evaluating

the history behind China’s economic boom, Economic History Working Papers

41660, London School of Economics and Political Science, Department of

Economic History.

Broadberry, S., Guan, H. & Li, D.D. 2013, China, Europe and the great divergence: a

study in historical national accounting, Paper presented at the Swedish Economic

History Meesting, 4-5 October, Lund.

Buck, J.L. 1930, Chinese farm economy: a study of 2866 farms in seventeen localities and

seven provinces in China, University of Chicago Press, Chicago.

Buck, J.L. 1937, Land Utilization in China: a study of 16,786 farms in 168 localities, and

38,256 farm families in twenty-two provinces in China, 1929-1933, Oxford

University Press [etc.], Oxford [etc.].

Chang, C. 1962, The income of the Chinese gentry: A sequel to The Chinese gentry,

studies on their role in nineteenth-century Chinese society, University of

Washington Press, Seattle. (Chinese edition published in 1991)

CHEN, Z. & YAO, L. 1957, Zhongguo jindaishi ziliao [Statistics on the early

industrialization in pre-modern China], Sanlian shudian Press, Beking, Shanghai.

FANG, X. 1996, "Qingda Jiangnan Nongmingde Xiaofei [Study on the Farmers’

consumption in the Qing Dynasty in southern China]", Zhongguo jingjishi yanjiu, ,

no. 3, pp. 91-98.

Feuerwerker, A. 1969, The Chinese economy, ca. 1870-1911, Center for Chinese Studies,

University of Michigan, Ann Arbor.

Fukao, K., Ma, D. & Yuan, T. 2007, "Real GDP in Pre-War East Asia: A 1934-36

Benchmark Purchasing Power Parity Comparison with the US", Review of Income

and Wealth, vol. 53, no. 3, pp. 503-537.

50

GE, Q., DAI, J. & HE, F. 2008, Land Use Changes and Terrestrial Carbon Budgets in

China during the Last 300 Years, Science Press, Beijing.

GUAN, H. & Li, D.D. 2010, "China’s GDP in the Ming Dynasty", China Economic

Journal, , no. 2, pp. 787-828.

GUO, S. 1995, "Qingdai beifang hanzuoqu de liangshi shengchan", Zhongguo jingjishi

yanjiu, vol. 1, pp. 22-44.

GUO, S. 1994, "Qingqianqi nanfang daozuoqu de liangshi shengchan", Zhongguo

jingjishi yanjiu, , no. 1, pp. 1-30.

Hsiao, L. 1974, China’s Foreign Trade Statistics, 1864–1949, East Asian Research

Center, Harvard University : distributed by Harvard University Press, Cambridge,

Mass.

Institute of Economics, Chinese Academy of Social Science 2009, Grain Price Table

Between 1821 and 1912, Guangxi Normal University Press, Guangxi.

Institute of Modern History, Academia Sinica, The Qing-era Grain Price Database.

Available: http://140.109.152.38/.

Keller, W., Li, B. & Shiue, C.H. 2012, Shanghai's Trade, China's Growth: Continuity,

Recovery, and Change since the Opium War, NBER Working Paper No. 17754.

LIANG, F. 1980, Zhongguo Lidai Huko Tiandi Tianfu Tongji (Dynastic Data of China's

Households, Cultivated Land and Land Taxation), Shanghai People's Press,

Shanghai.

Lieu, D.K. 1927, China's industries and finance, Chinese Government Bureau of

Economic Information, Beking, Shanghai.

LIU, F. & WANG, Y. 1997, Jindai zhongguo de jingji fazhan [Economic development in

China between 1840 and 1949], Shandong renmin Press, Shandong.

LIU, K. 2001, "1927-1937 nian nongye shengchan yu shoucheng, chanliang yanjiu [The

agricultural production between 1927 and 1937]", Jindaishi yanjiu, , no. 5, pp.

59-112.

LIU, R. 1987, "Shiba shiji zhongguo renjun guomin shouru guji jiqi yu yingguo de bijiao

[Study on national income in 18th century China]", Zhongguo jingjishi yanjiu, , no.

3, pp. 105-120.

Liu, T. & Yeh, K.C. 1965, The Economy of the Chinese Mainland: National Income and

Economic Development, 1933-1959, Princeton University Press, Princeton, N.J.

51

LIU, T. 2009, "An Estimation of China's GDP from 1600 to 1840", Economic Research

Journal, , no. 10, pp. 144-155.

LIU, Y. 2001, "Zhongguo jindaide zhanzheng yu ziran zaihai duinongye shengchande

pohuai [Studies on the effect of wars and natural disasters on the agricultural

produciton in pre-modern China]", Gujin nongye, vol. 1, pp. 17-29.

Ma, D. 2008, "Economic Growth in the Lower Yangzi Region of China in 1911-1937: a

quantitative and historical analysis", The Journal of Economic History, vol. 68, no.

2, pp. 355-392.

Maddison, A. 2007, Chinese economic performance in the long run, OECD, Paris.

Maddison, A. 2010, Groningen Growth and Development Centre. Maddison Historical

Statistics. Available: http://www.ggdc.net/maddison/oriindex.htm.

Ou, P. 1947, China's National Income, 1933, Institute of Social Sciences Academia

Sinica Monograph Serie, Shanghai.

PENG, X. 1958, Zhongguo huobishi [History of Money in China], Shanghai People's

Press, Shanghai.

Perkins, D.H. 1969, Agricultural development in China, 1368-1968. University Press,

Edinburgh.

Perkins, D. H., & Zhao, G. (Eds.) 1975, China's modern economy in historical

perspective, Stanford University Press, Stanford, CA.

Pomeranz, K. 2000, The great divergence: China, Europe, and the making of the modern

world economy, 1st edn, Princeton University Press, Princeton, NJ.

Rawski, T.G. 1989, Economic growth in prewar China, University of California Press,

Berkeley.

SHI, Z. 2009, Qingda Hubu Yinku Shouzhi he Kucun Tongji [Statistical materials on the

government income and expenditure, and reserves in the state treasury in the Qing

Empire], 1st edn, People's Press, Fujian.

SHI, Z. 1989, "Qingdai qianqide gengdi mianji jiliangshi chanliang guji [studies on the

cultivated land in 1840]", Zhongguo jingjishi yanjiu, vol. 2, pp. 47-62.

Wang, Y. 1992, "Secular trends of rice prices in the Yangzi Delta, 1638–1935"

in Chinese history in economic perspective, eds. T.G. Rawski & L.M. Li, Berkeley, ,

pp. 35-68.

52

WANG, Y. 2005, "Zhonguo jindai de jingji zengzhang he zhongchangqi bodong

[Econimic development in China between 1840 and 1949]", Jingjixue, vol. 4, no. 2,

pp. 461-490.

Will, P.E. & Wong, R.B. 1991, Nourish the people: the state civilian granary system in

China, 1650-1850, Center for Chinese Studies, University of Michigan, Ann Arbor.

Wu, H.X. 2011, Rethinking China's path of industrialization, Working paper 2011-76

edn, World Institute for Development Economics Research.

WU, H. 1985, Zhongguo lidai liangshi muchanliang yanjiu [Study on agricultural

production in Chinese history], Beijing nongye Press, Beijing.

XU, D. & WU, C. 2005, Chinese Capitalism, 1522-1840. English translation by LI

Zhengde, LIANG Miaoru and LI Siping, St. Martin's Press, New York.

YAN, Z. et al. 1955, Zhongguo jindai jingjishi tongji ziliao xuanji [Statistical materials

on the Economic History of Modern China], Kexue Press, Beijing.

YAO, X. 1962, Zhongguo jindai duiwai maoyi shi ziliao (1840-1895) [A history of

Chinese international trade], Zhonghua Press, Beijing.

ZHANG, P. 1996, Zhongguo lishi qihou bianhua [A history of Chinese climate

change], Shandong Science and Technology Press, Shandong.

ZHANG, Y. 2005, "Study on the Farmers’ Consumption of Qing Dynasty in 1700s",

Gujin Nongye, vol. 4, pp. 80-90.

ZHANG, Y. 1991, "Jindai zhongguo renkou he gengdi de zaiguji [Studies on the

cultivated land in the period 1850-1915]", Zhongguo jingjishi yanjiu, no. 1, pp.

20-23.

ZHOU, R. 2001, "Qingdai qianqi gengdi mianji de zonghe kaocha he chongxin gusuan

[Study on the estimation of cultivated land in the Qing Dynasty]", Jianghan luntan, ,

no. 9, pp. 57-61.