Regional Differences in Chinese Agriculture: Results from the 1997 First National Agricultural Census

ROBERTO FANFANI and CRISTINA BRASILI

RECENT DEVELOPMENTS IN CHINESE AGRICULTURE The development of the Chinese agricultural sector and its capability to increase agricultural production is gaining a greater importance for food security and food trade on a worldwide scale. Considering the size of China, this eould influence international food priees., and thus determine major problems for food-importing developing eountries. The People's Republic of China (PRC) remains predominantly a rural country, but its economy is characterized by strong macro-regional differences owing to factor endowments {human capital, resources and infrastructure) and differences in the strueture of agriculture and other economic sectors. Economic development in China has registered uneven economic growth in favour of the coastal region {vs. inland and border ones), the urban areas (vs. rural areas) and the industrial seetor (vs. the agricultural sector). During its transition the Chinese economic system faced strong structural changes. One of the main changes is certainly the reduction of the share of the agricultural sector in relation to total GDP and labour force. Indeed, the rural enterprises (or township and village enterprises, TVEs) are the leading and most dynamic elements in China's economy. Therefore, the integration of agricultural activities with other economic activities in the rural areas is of great importance and interest [Biggeri. M., 1998; Giovannini. 2000].

Robedo Fanfani. Dept. of Statistics. University of Bologna, via Belle Ani. 41 - 40126 Bologna. Email: fanfani((J'stat.unibo.it. Tel. +39 051 2098212. Cristina Brasili. Dept. of Statistics. University of Bologna, via Belle Arti, 41 - 40126 Bologna. Email: brasili(tt;stai.unibo.it. Tel. +39 05! 2098260. Both authors share the responsibility for Ihis article. However. Roberto Fanfani wrote the sections 'General Characleristics and DitTerences among Provinces" and "The Complex Realiiy of Chinese Agriculture: .^ Firsl Analysis al County Level". Crislina Brasili wrote the section "The New Geography of Chinese Agricullure". The sections "Recenl Developments in Chinese Agriculture' and 'Concluding Remarks" were written jointly. The authors ihank also Mirko Bonctti for his valuable help in preparing ihe article.

REGIONAL DIFFERENCES IN CHINI^Sn AGRICULTURE

19

Despite its declining relative importanee, the agricultural sector has still a relevant and decisive role to play in the economic and social development of China, at least for the following three reasons. FirsL China is still a predominantly agrieultural country. Second, Chinese agriculture should try to meet the increasing domestic food demand caused by natural population growth and by the change in food consumption patterns. Finally, the agrieultural seetor is important in labour force and intermediate inputs supply to China's industrial and economic development. The Chinese government faces the problem of produeing enough food in order to avoid an increase in imports (mainly cereals) and to maintain food prices aecessible to urban consumers for reasons of political stability. There is a general consensus that China will import an increasing quantity of cereals in the near future. However there are two main different positions. Brown [1995] estimates a large net import of cereals (between 200 and 370 million tons) in 2030. Alexandratos [1996] estimates a net import of cereals of around 44 million tons. Brown's thesis assumes a rapid growth of food consumption, espeeially for livestoek products, and a substantial decline by 20 per eent of cereal produetion. On the other hand, Alexandratos claims that the reduction of cultivated land will be lower, owing to other crop production and aquaculture, not only to urban development. The most recent projections drastically revise down previous estimates. The USDA report [2000] estimates China's net import of wheat at only 2 million tons by 2005 and. surprisingly, a net export of corn of more than 3 million tons. The (UN) Food and Agriculture Organization [FAO., 2000] revised its long-term projections in view of the slowdown in population growth and the decrease of the erstwhile rapid growth in food consumption in the past decade. The development of the Chinese agricultural sector becomes relevant for its implications on the international food market. The 'definitive opening' of the PRC is something new for the international food market. In recent years, there has been a rather significant amount of research undertaken on China's food and agriculture prospects., and how these prospects relate to wider issues of world food trade and food security. It is no secret that such research has been hampered by uncertainties, providing opportunities for all sort of extreme positions concerning China's future role in world food trade, with some such positions verging on the outright alarmist for the food security of the world as a whole [FAO, 1998:3^]. The statisties are a crucial development in the empirical study of Chinese agriculture, and the results of the First National Agricultural Census, held in 1997, have played an important role recently in relation to

20

RURAL DEVELOPMENT IN TRANSITIONAL CHINA

the consideration of prospects for China's food security and for world food markets (espeeially for grain). In fact, the 1997 Census data suggest that China has significantly more cultivated land (30 per cent) than previously estimated by official data, and that the number of livestock is lower (-20 per cent) in comparison with previous estimates. Both these results indicate a greater potential for increases in yield and therefore in the domestic production of agriculture in China.' In this article we first analyse the main results of the 1997 Census to give evidence of the ditTerences between provinces (section 2). We then show the complex reality of agriculture with a brief analysis of some variables for the 2.300 Chinese counties (section 3). In the fourth section we will define a 'new geography' of Chinese agriculture defining first macro-agricultural areas at province level; then we analyse these areas more closely, down to county level. We mainly use cluster analysis methodology and data of the 1997 Census. Concluding remarks are reported in the last seetlon. GENERAL CHARACTERISTICS AND DIEFERENCES AMONG PROVINCES The 1997 Census is considered a milestone for gaining a better understanding of the Chinese rural system and for designing adequate policies targeted to the promotion of long-term soeio-economic development. From a statistical point of view, it can be considered a starting point for the revision of current and future statistical rural surveys of agricultural systems. It collected detailed data on the structure of the agricultural sector statistics as well as information on many other aspects of the rural economy. In this section we will consider some general characteristics of agriculture in China revealed by the 1997 Census, but in particular we will emphasize the differences between provinces in order to give evidence of the different geography of agriculture in China. The Abstract of the First National Agricultural Census in China [National Agricultural Census Office of China, 1999] gives data aggregated by province that could be further aggregated by economic-geographical regions. Census information shows important aspects of rural and agricuhural holdings (households and other holdings, purely agricultural household, etc.), the importance of the main activities of persons (agriculture and non-agriculture) and the amount of agricultural maehinery and equipment. The Census data, for instance, highlight that the number of rural households is about 214 million, with more than 193 million agricultural households. The number of persons involved is more than 873 million., with about 561 million engaged in production activities. The number of active

REGIONAL DIFFERENCES IN CHINESE AGRICULTURE

21

people is much higher than the previous official statistics on the rural labour force (453 million [SSB, 1997: 363]). However, the larger size of labour foree estimated by the Census eould be a result, at least in part, on the different definitions used. For instance, the Census definition considers those working less than two months a year as active persons. Aecording to the 1997 Census, the number of persons engaged in agriculture is very high. In rural households, over 75 per cent of the population are active in agriculture, while only 25 per cent are engaged in other productive activities. It is important to underline that the current statistics report a lower proportion (less than 70 per cent) for the agricultural sector. The difference between Census data and the previous statistics on the proportion of people engaged in agricultural activities varies considerably between provinces. This variable, as we will show later, is high correlated with per capita GDP Census results give important information on households classification according to their prevalent activity (agricultural and non-agricultural) and the typology of agricultural households (pure agricultural, mainly agricultural, mainly non-agricultural). The main findings show that in rural areas the degree of integration of agriculture with other economic activities is an important variable characterizing the differences between provinces. In fact, of the 214 million Chinese rural households, about 193 million (more than 90 per cent) are agrieultural households, aceounting for more than 799 million people. The agricultural households are differentiated as follows: Pure agricultural households, more than 126.7 million (65.6 per cent of households) accounting for more than 500 million people (61 percent). Mainly agricuhural households, more than 39 million (20.2 per cent of households) accounting for more than 179 million people. Mainly non-agricultural households, more than 27.3 million (19 per cent of households) accounting for 120 million people.

The total cultivated land is more than 130 million hectares, according to the Census, with more than 123.3 million hectares of sown land in household holdings. The average size of all households is only 0.6 heetares of cultivated land. The household holdings have 0.64 hectares of sown land. The agricultural households under 1 heetare are more than 90 per eent of the total agricultural households, but they have about 79 per eent of the sown land. The evaluation of total eultivated land is relevant because it is much higher than the previous estimate of 95 million hectares. This result could indicate that it will be possible to increase the productivity and intensity of agricuhural production, in terms of both the labour force and the amount of land worked.

22

RURAL DEVELOPMENT IN TRANSITIONAL CHINA

HGURI-: I CHINA AGRICULTURAL Cl'NSUS: PliRSONS ENGAGED IN AGRICULTURAL ACTIVITIES IN RURAL HOUSEHOLDS

person s engaged (%)

Source: Authors' elaboralion. SSB [1997]. FIGURE 2 CHINA: GROSS DOMESTIC PRODUCT PL-.R CAPITA 1996 {YUAN)

GDP per capita (yuan)

I 1

2093-5963

8000-23000

Source: Authors' elaboralion, SSB |I997].

REGIONAL DIFFERENCES IN CHINESE AGRICULTURE

23

The results and the analysis of data of the 1997 Census, as we have seen, are important and relevant for their economic and policy implications. This is particularly true if we consider the differences between provinces and counties. Since the future development pattern of China will result in a gradual reduction of the agricultural labour force, to a different extent amonj^ macro-regions, provinces and counties, it is worthwhile to make some first analysis on the changing importance of agricultural labour force. Considering the three macro-regions, the proportion of the labour foree engaged in agricultural activities is 66 per cent for the eastern area and about 80 per cent and 85 per cent for the provinces of the middle and western areas respectively (see Table 1 and Figure 1). There is also a large variation between provinees, from a minimum figure for agricultural employment in Shanghai (31 per cent) and Beijing (39 per cent), to a maximum in Tibet (95 per cent) (see Figure 1) We will see in the next section that such differences are more relevant at county level. !n order to obtain a better understanding of provincial development, we have analysed the correlation between the proportion of agricultural persons engaged in economic activities and the level of economic development for each province. We have measured such correlation using the GDP per capita (1996) as a synthetic variable for economic development for the Chinese province represented in Figure 3. The results show a very high inverse correlation between these two variables, with a high correlation coefficient (-0.81). However, the analysis demonstrates that the Chinese provinces can be divided into two main groups. One group includes the provinces characterized by a high percentage of people engaged in agricultural activities (more than 75 per cent) and by a low level of GDP per capita (less than 6,000 yuan). This group includes almost all the western and middle provinces, which are the main rural provinces of China (see Figure 3). The other group of provinces, with high levels of GDP per capita and lower levels of people engaged in non-agricultural activities, includes the municipalities of Shanghai and Beijing, the provinces in the eastern coast (Tianjin, Jiangsu, Zhejiang, Fujian and Guangdong). The only provinces with high levels of agrieuUural activities and a relatively high level of GDP per capita are Shandong and Liaoning, in the northern part of the eastern coast. The results of this analysis show that the provinces of China have very different levels of development (in terms of GDP and share of agricultural labour). The most developed ones are those loeated in the eastern area. However, the eastern part of the three economic-geographical regions of China have less developed provinces, both in the north (Liaoning and Hebei) and in the south (Guangxi). Our results are also different from the six main geographic regions of China. This result clearly indicates the

24

RURAL DEVELOPMENT IN TRANSITIONAL CHINA

TABLI-: 1

CHINA: NUMBER OV PtRSONS tNGAGKD IN AGRICULTURAL AND NONAGRICULTURAL PRODUCTION ACTIVITIES BY GEOGRAPHIC REGIONS Geographic regions Total persons Agricultural activities Persons % 424,995,487 75.7 Non-agricultural activities Persons % 136.484.003 24.3

China, Total North Region Beijing Tianjin Hcbei Shanxi Inner Mongolia Northeast Region Liaoning Jilin Heilongjiang East Region Shanghai Jiangsu Zhejiang Anhui Jiangxi Shandong Central South Region Henari Hiibci Hunan Guangdong Guangxi Hainan Southwest Region Chongqing Sichuan Gui7hou Yunnan Tibet Northwest Region Shaanxi Gansu Oinghai Ningxia Xinjiang

561.479,490

2.195.012 2.305.319 31.760.002 13.354.323 8.367.447

855.091 1.246.816 24.417,806 9.970.377 7,410,471

39.0 58.0 76.9 74.7 88.6

1,339,921 1,058,503 7,342,196 3,383,946 956,976

f.1.0 45.9 27.0 25.3 11.4

14,281.809 8,036,539 10,201.106

10.976,276 7,071,955 9,009,975

76.9 88.0 S8.3

3.305.533 964.584 1.191.131

27.0 12.0 11.7

2,861,042 36,417.453 22,867,225 14.602,387 19,160,861 45,384.910

892,583 20,868,549 10,232,612 9.021.487 14.112.097 35,263,460

31.2 57.3 44.7 61.8 73.7 77.7

1,968,459 15,548,904 12,634,613 5.580.900 5.048.764 10,121,450

68.8 42.7 55.3 38.2 26.3 23.0

47,774.836 24.016.439 31.932.654 27,556,073 22.328,674 2.390.188 16.863.591 46.793.756 18,501,594 21,191.199 1,217,889

38,553,743 19.170.020 24.918.200 15,777.116 18.073.078 2.026,855 13,619,697 37,341,931 16.619.576 18.920.874 1,162,910

S0.7 79.8 78.0 57.3 80.9 84.8 80.8 79.8 89.8 89.3 95.5

9,221,093 4,846,419 7,014,454 II.77X.957 4,255,596 363,333 3,243,894 9,451.825 1.882.018 2.270.325 54.979

19.3 20.2 22.0 42.7 19.1 15.2 19.2 20.2 10.2 10.7 4.5

16,504.193 12.060.126 2,067.124 2,219,375 4,911.188

13,696,170 10,521,324 1.871.359 1,887.506 4,569.593

83.0 87.2 90.5 85.0 93.0

2,808,023 1,538,802 195.765 331.869 341.595

17-0 12.8 9.5 15.0 7.0

Source: Authors' processing of .thxri-ncl of the Firsl National Agricultural Census of China [1998]

REGIONAL DIFFERHNCES IN CHINESE AGRICULTURE

25

necessity for a new geography of Chinese agriculture, eapable of taking into account the existence of differences within each macro-area. There are other significant differences between the main geographic areas, in particular between the eastern provinces and the rest of China. In fact, in the provinces of the eastern area, purely agrieultural households are only 57 per eent of the total, mainly agrieultural households are 22 per cent, and the mainly non-agricultural households are 21 per eent. If we consider that in the eastern area there are also 15 per eent of non-agrieultural households, these data emphasize the great importance that non-agrieultural activities have assumed in this area. The remaining Chinese provinees, both in the middle and in the western areas, have a mueh lower share of mainly non-agrieultural household (7.8 per eent in Southwest regions), and mainly agrieultural households (19 per eent Southwest regions); see Table 2. Another interesting aspeet is the size of Chinese rural households. The average number of aetive people (engaged in economic activities) per rural household is six (Figure 4), whieh is not a high figure(similarto that of Italy for example). It refleets the fact that, on average, the size of rural families is not particularly large, owing to birth control and family planning poliey. There are not big differenees between provinees for this variable, with a minimum of about two persons per household (Beijing), and a maximum of three people in Tibet and Qinghai. The ditYerenees between provinees in the household dimension are strictly related to the differenees of persons engaged in agriculture and in non-agricultural activities (see Figure 5). The National Agrieultural Census Offiee also reports data on numbers of selected types of machinery (traetors with less and more than 20 hp, harvester-threshers, thresher engines, transportation vehieles). The total
TABLE 2 CHINA: VALUES AND PHRCLNTAOKS OF THh DIFFERENT TYPES OF AGRICULTURAL HOUSLliOLDS. BY GEOGRAPHIC RhGION Geographic regions Agricultural households Number China, total North Northeast East Central South Southwest Northwest Agicultural households of which: Pure Mainly Mainly agricultural non-agricultural /o Number % Number Number % 65.6 66.0 77.3 56.8 64.6 72.9 75.4 39.011.614 3.578,774 1,632,511 12,604,790 U,623.203 7.152,725 2.419.611 20.2 17.1 13.3 22.3 21.9 19.2 I8J 27.357.707 3.535.041 1.149,878 11.796.912 7.124.540 2.910.304 841.032 14-2 16.9 9.4 20.9 13.4 7.8 6.3

193,0X8.158 126.718,837 20,947,554 13.833,739 I2,2H2.944 9,500.555 56.427.802 32.026,100 52,997.596 34.249.853 37.181.588 27.118,559 13.250,674 9.W0,O31

Source: Authors" processing of Abstract of the First Notional Agricultural Census of China [I998J.

26

RURAL DEVELOPMENT IN TRANSITIONAL CHINA

FIGURE 3 CHINESE PROVINCES AND MUN ICIPALITl KS; PERSONS ENGAGED IN AGRICULTURE (%( AND GDP PER CAPITA
TIBHI

Qua

1>
TianiL

Vim nan

Fiiiian Jiangsi]

X
ngdong

Hebe

Selling ^ S i

y= 102.986 "'' R' =0,7246

5000

10000

15000

30000

GDP per capita

Source: Authors' elaboration of Ahstraci of the First National Agricullurat Census of China [199S]. Including autonomous regions and municipalities.
FIGURE 4 PERSONS ENGAGED IN AGRICULTURAL ACTIVITIES (%) AND HOUSFHOLD DIMENSIONS (PERSONS PER HOUSEHOLD)
3,6'

TTie ongm is the Chinese average value

Qinghat

3.0

10,0

20.0

30.0
Shanglial

40,0

50.0

60.0

70.0

SOO

0.0

1000

110,0

Tin(in

Source: Authors" processing ai' Abstract of ihc First Nalioiial Agricultural Census of China [19981.

REGIONAL DIFFERHNCES IN CHINESE AGRICULTURE FlGtJRE 5

AVERAGE NUMBER OF PERSONS EMPLOYED PER HOUSEHOLD

27

0,5

1.0

1.5

2.0

2.5

3.0

3.5

^ 1 Pers. Aghe./Households

Pers. Non-Agric/Households

Source: Authors' processing of Abstract of the First National Agricultural Census of China

28

RURAL DEVELOPMENT IN TRANSITIONAL CHINA

number for China of these selected types of machinery is about 25 million. The most common types of machinery are small tractors with 20 or more hp (less than 12 million), followed by the operated threshing machines (7.5 million), and by the agricultural transport vehicles (4.6 million). In China, there are only 126 machines per 1,000 rural households, and only 58 machines per 1,000 persons engaged in agriculture. The differences are very large, with higher numbers for eastern provinces (172 per 1,000 households), which is more than three times the number in the western area (53). These differences become greater if we consider the numbers of machines per 1,000 people engaged in agriculture, with the number for the eastern area (85 per 1,000 people), about four times that for the westem area (22 per 1,000).

FlCiURE 6 UNITS OF SELHCTED AGRICULTURAL MACHINERY BY GEOGRAPHIC AREA

[Western area

Sotircf: Authors' processing of Ah.slracl of ihe First Nalional .i^riiiillural Census of China [1998].

REGIONAL DIFFERENCES IN CHINESE AGRICULTURE THE COMPLEX REALITY OF CHINESE AGRICULTURE: A EIRST ANALYSIS AT COUNTY LEVEL

29

For a better understanding of the complexity of Chinese agrieulture we have utilized data at county level. The Abstract of the First National Agricultural Census in China just shows seven maps representing data at county level. The seven variables represented in the maps are: per eapita possession of cultivated land; agricultural machinery power possessed by 100 persons engaged in agricultural activities; proportion of non-agricultural households to rural households; proportion of persons engaged in agricultural activities to rural persons engaged; proportion of effectively irrigated area to agricultural land; proportion of number of towns to total numher of towns and townships; educational level for rural labour foree in years. The maps presented at county level show that the reality of agriculture in China is mueh more diversified in comparison to that shown by data presented at province level. Thus, the analysis of data at county level could represent a great development to define a new geography for Chinese agriculture, not only at national level but also inside each province. However, although available variables are few, a first descriptive analysis at eounty level (2,392 counties) allows us to make some useful consideration of Chinese agriculiure' In order to analyse the relationships among these seven variables, we sorted all counties by the decreasing proportion of persons engaged in agricultural activities to rural persons engaged. We used this variable to classify the counties for its strict relation with eeonomic development, as we have shown in the previous section. The graphical analysis highlights, first of all, that the proportion of persons engaged in agricultural activities at county level shows a range of variability much larger than that among the provinces described in the previous section. By contrast, there are numerous counties with very high numbers of persons engaged in agriculture (in some eases more than 85 per eent); but it is worth emphasizing that, on the other hand, there arc also several counties with less than 40 per cent of persons engaged in agricultural activities. This result indicates that rural development in China is mueh less uniform than expected, and in particular there are many areas (counties) where agriculture is not the main economic sector. However, the most numerous group of counties in China have 60 85 per eent of persons engaged in agricultural activities; these counties represent the reality of the large part of rural China. The proportion of non-agrieultural households to rural households shows a elear inverse relationship with respect to the proportion of persons engaged in agricultural activities. Most counties have a proportion lower than 60 per cent, but, at the same time, there arc also counties with proportions between 20 and 40 per cent. This confirms the presence in

30

RURAL DEVELOPMENT IN TRANSITIONAL CHINA

China of a diversified situation of rural development at county level. In addition, as shou/n in Figure 7(b), the presence of non-rural households is characterized by a larger variability between counties with respect to the previous variable. The analysis of tbe educational level of rural labour force shows an interesting relationship with the other variables. The educational level is generally high: in most of the Chinese counties rural workers have from six to eight years of education. However, when the counties have a higher proportion of persons engaged in agriculture, the educational level becomes less homogenous, with an increasing number of counties with a very low edueational level (less than four years). The counties with low educational levels are not numerous. In fact the problem of ensuring an adequate level of education of the rural labour force interests several rural counties, which are mainly located in the provinces of Sichuan, Guangxi, Guizhou, Yunnan, Tibet and Gansy. The distribution of cultivated land (per capita) among the counties is much more uniform. In most of the counties the per capita possession of cultivated land is held at constant level (1, 2 or 3 mu), even if we can observe higher values in the counties with a higher concentration of agricultural labour. The distribution of cultivated land among counties seems to reflect, even now, more the criteria adopted in applying agrarian reform, than the degree of development of the counties. The fast transition and change of the Chinese eeonomy will probably lead to a substantial change in the distribution of land among households. Also, the distribution of agricultural machinery among counties does not change according to the proportion of persons engaged in agricultural activities. In almost all counties, the agricultural machinery power is less than 100 per 100 persons engaged in agrieultural activities. However we can note greater machinery power in counties with both low and high concentrations of agricultural labour. Hence, we can say that the distribution of machinery power too does not reflect completely the agricultural development of the counties. The proportion of effectively irrigated area to total agricultural land is the variable with the largest divergence between counties. The presence of irrigated areas is relevant {or otherwise) both in the less developed counties (with lower proportions of workers in agriculture) and in the most developed ones. The distribution of this variable is related mainly to the characteristic of natural resources, and up to now has only partially influenced the general degree of rural development. The analysis of the distribution of the variables considered at county level was carried out not only for China as a whole, but also for each Chinese province. The results of the analysis by provinee confirm the trends

REGIONAL DIFFERENCES IN CHINESE AGRICULTURE

FIGURE 7 MAIN VARIABLES FOR CHINESE COUNTIES THE COUNTIES ARE RANKED BY VARIABLE (a)
Prciportiiin of persons engageii in agricultural acliviiies to rural persons engaged (%) (a) Propcirtion ol non-aynculrural houseKolds lo njral households

31

fiOO

10CC

ifiOO

2000

2300

3OO0

9O0

1000

IMO

2CC0

2900

3000

dUL'alianal level tor rural labur lorce in years

Pel capita puiisession ol'cuhivaied land 10-2(1 mu|

(el

KM

10DO

1900

1000

2S0O

3000

SOO

1OO0

1900

2O0O

1900

3000

Agnculiumi machinery power possessed hy I Oil persons en^Li^ed m agncullural aelivities (counties from 0-500) (e)

I'mpDriion of eltcctivcly impaled area In agricullural land 1%) if)

600

1000

1500

iOOD

2500

3000

900

IQOO

2000

1300

3000

Source: Authors" processing of Abstract of the First National A^^ricultural Census of China

32

RURAL DEVELOPMENT IN TRANSITIONAL CHINA

described for the whole of China. Obviously, the shape of the distribution of each variable among the counties slightly differs across the provinces, and with respect to the national pattern. The results of this analysis arc available on request.
THE NEW GEOGRAPHY OF CHINHSH AGRICULTURE

The Macro-Agricultural Areas at Province Level The 1997 Census data give us a great opportunity to gain a better and deeper knowledge of the agricultural geography of China, The significant conclusion from the analysis in previous sections is that the economic geography of the China is very different from the agricultural geography, but these two important aspects are closely related. In faet, in section two we showed how GDP is inversely eorrelated to the ratio of persons engaged in agriculture. This means that we need to deepen the knowledge of the different agricultural systems in China, in order to identify those areas where the situation could be modified and those where the chances of modification are low, because the terrain can be extremely difficult. In order to obtain a new geography of agriculture we undertook some statistical muhivariate analysis. The analysis was carried out in three steps. In step one we chose 30 indicators and carried out the principal component analysis (PCA). In the second step we employed cluster analysis (CA) on data at province level, in order to identify the principal maero-agricuitural areas. Finally, in the third step, we carried out a cluster analysis inside each of the four macro-areas, identified in step two. We found 15 groups of counties very similar from a social and agricultural point of view. In the following paragraphs we will review the three steps in detail and the results obtained. In the first step we chose 30 indicators of the Chinese social and agricultural situation, as revealed by the First Agrieultural Census, for the 31 Chinese provinces. We employed the following indicators from the 30 (see Table 3): percentage of persons aged 7-25 years in rural households; percentage of persons aged 36-61 in rural households; rural households - percentage of persons aged seven and over engaged in crop planting; rural households - percentage of persons aged seven and over engaged in non-agricultural activities in rural areas; rural households - pereentage of persons aged seven and over engaged in economic activities in home townships; all holdings persons engaged in non-agricultural activities/persons engaged in agricultural activities; household holdings - pereentage of persons illiterate and semi-illiterate; household holdings - percentage of persons with superior educational level; household holdings - percentage of females with superior educational

REGIONAL DIFFERENCES IN CHINESE AGRrCULTURE

33

level; household holdings - tnales/femalcs with superior educational level; household holdings - percentage of persons engaged in agricultural activities six months and over; all holdings - percentage of agricultural land (area) for cultivated land; all holdings - percentage of agricultural land (area) for woodland; all holdings - percentage of agricultural land (area) for fisheries; all holdings percentage distribution of sown area for grain; all holdings - among cultivated land tractor ploughed area (per cent); all holdings - number of large animals for 100 persons engaged in agricultural activities; large and medium tractors/persons engaged in agricultural activities; small tractors/persons engaged in agricultural activities; all holdings - persons engaged in agricultural activities/persons engaged in economic activities. Then, to reduce the number of indicators, we implemented the principal component analysis (PCA). From the 30 indicators we extracted five principal components, which explain together 85 per cent of the total variances. The first component explains about 46.4 per cent, the second eomponent explains 15,6 per cent, the third eomponent explains 11.2 per cent, the fourth component explains 6.3 per eent and the fifth component explains 5.6 per eent. The first component, which explains a large proportion of variance, characterized the provinces in the sense of concentration of people engaged 'in agrieultural or in non-agricultural activities'. In eontrast, the second component is mainly linked to the level of agricultural mechanization. In the second step we earried out a cluster analysis on the five extracted components. We chose four clusters that identify the most important Chinese agricultural maero-areas.' The results are represented in the map at Figure 8. The four macro-areas have very different numbers of provinces: eluster I (mountain area) has two provinces; cluster 2 (peripheral area) has nine provinees; cluster 3 (urban area) has four provinees and cluster 4 (principally agricultural area) has 116 provinces. The means and the variances of the 30 original indicators for each cluster are reported in Table 3. The main characteristics of the four maero-agrieultural areas are as follows. Cluster 1. It constitutes only two provinces (Qinghai and Tibet) but they have a very large area, mainly mountainous. This cluster is largely characterized by traditional agriculture (67.8 per eent in erop planting), also by animal husbandry (25 per cent of the persons over age seven) with less use of machinery. There is a large number of young people in comparison to the other clusters and there is the highest number of semi-illiterate persons (61 per cent). The holdings are large in terms of surface area. In these provinees there are no fishery activities.

34

RURAL DEVELOPMENT IN TRANSITIONAL CHINA

Cluster 2. It includes nine provinces from the northern part of China and also from the south (Yunnan and Guizhou). It is the largest cluster in terms of surface area. It is also characterized by the largest holdings in terms of surface area (1.6 heetarcs). The proportion of persons engaged in agricultural activities is about 90 per cent of total persons engaged in eeonomie activities. The level of education is not high, with 21 per cent of persons being semi-ill iterate. Cluster 3. This eluster includes four provinces and municipalities (Beijing, Jangsu, Shanghai, Tiajin). It is the smallest eluster in terms of surface area. This cluster represents the biggest urban areas of China with remarkable food eonsumption. This cluster shows the smallest percentage of persons engaged in agrieultural aetivities (66.5 per eent), and the highest level of edueation (only about 13 per eent of persons are semi-illiterate and 7.6 per eent have superior education). The holdings have the smallest dimension. There is the highest level of mechanized agriculture: 82 per cent of traetor ploughed area, and 30 per eent of mechanieally harvested area. The agrieultural land for fisheries is the highest (6.7 per eent). Cluster 4. Tt includes the highest number of provinces (16), all located in the southeast, middle and south parts of China. This eluster includes also Siehuan province from the southwest part of China; this is probably due to the fact that Siehuan includes the Chongquing municipality, which has a large part of eultivated land (65 per eent of agrieultural area). This cluster is significant in terms of the number of persons engaged in agrieultural aetivities (80 per eent), which is greater than the previous eluster (pink). Generally, there is a good level of edueation and the lowest percentage of semi-illiterate persons (12 per cent). This large eluster represents the 'core' of Chinese agriculture. The agricultural land is characterized by a large presence of woodland (43 per cent). The four elustcrs emerging from our analysis are mainly ranked by the different levels of agrieultural development, as shown by the different levels of persons engaged in agricultural aetivities, associated with the different levels of education and mechanization. The four macro-areas derived from our analysis describe a geography of Chinese agriculture that is really different from the general eeonomie geography of China reported in the previous paragraph. The Internal Differentiation of Macro-Agricultural Areas (at County Level) The third step of our analysis has the objective of further differentiating the internal situation of each of the four maero-areas described before (except

REGIONAL DIFFERENCES IN CHINESE AGRICULTURE

FIGURE 8 MACRO-AGRICULTURAL AREAS IN CHINA

35

CluMer '^A I Mounlain areu S H 2.Peripheral area " Urban area *

l y i 4. PrJndpally agricultural area

Source: Authors' processing of AhMracr of ihc First Nulional Agrkuhural Census of China [1998].

for the urban area, which is small and unifbnn). We have used the same approach as in step two but in this step we utilize data at county level for eaeh macro-area. In this case there are available only seven indicators, as described and analysed in section 3. We standardized the indicators and then we carried out the cluster analysis with hierarchical methods on the eounties inside the macro-areas to choose the numbers of groups. Then we applied the non-hierarchical method and obtained the fmal results. The numbers for the clusters (representing groups of eounties inside eaeh macro-area) are five for cluster 1, five for cluster 2, whereas cluster 3 remains uniform. In fact, we found that each cluster has a very different numbers of eounties belonging to them (as reported in Table 4). The total number of Chinese eounties is 2,392 and the mean and variances for the seven indicators used are shown in Table 5 and Figure 9.'' This table immediately highlights the relevance of a differentiation inside each of the macro-areas previously defined. To underline the results for the most important reality of Chinese agriculture we have to note that inside the principally agricultural area there is a large homogeneous area that includes 1,050 counties. In Figure 9 these counties are represented in white. These

r-1 f^i

-Ji ra

^c o p T p

oc

o 3
p p d d

i ^ r, 00 00

r~
(N

rn O

r-1 o

r-i

00 ra 3 c .c ;

j=

lev vel

lev

V u

>
a
c
to TD

. , "ra u ra '' ra c ra

[BUOi

person witi or educ ati males /ith educat eriior edue ati female wit v'femal rior edu person :age. agricu!

-2

,0

ra

"2
S

y 2 -a T3 2 c c
LH

:= E
to

E T;
p
ra u

CL

LH

E S)
u U o C

2 '^
5

O.
r/l

ra (--

-C

O. 3

ars ii ears sons sons sons

5 e

-a - a -a 00 00 00

so
EJ
'f.

ra

ra

ID

u >.

Cl.

u u O, CL

T3 " a Z : oj ru 0(1 00 i
f.

y I Io
EH LH

in r/l

in

D H Cut; '>-'>-

r j -

r j -

r j -

o o
C C

o o o ra o ra o o c c 3
"J ^rfj 'J n1 J3 V d "i

i> r 0 ra

E
J^ j ^ 3

DO I .

5
p- '4J J V* L

c c c c c ^ f ^ c - ^ ^ - 0 0150 Ig T3 "O 5 E ^ 0 0
, JD -C -C -C -C
*^

^ ^ ^ ^2^ ;2 s M^MgJg' o 2 ;2
o o o o _ _ _ _ _ _ o u o o o c ' o ' o ' o o"5 3 3 3332-^-^-=-^-^ Q QOOO C ^ ~ ^ ^ ^
f^H ' 1 '

~ : : : _ : "e _ ra ra ra 3 3

; ^ __ := s

g 00

O^ O

r-l

"O

1"^

ler per rural rural ia d ( a rea) :ural la :ural la rea) ural la

c
T:

a. 3

a . E
-a

Olds cull i vat woodta grasslai fisherie

OJ

0 c

r/l

u (O 0
t4

i5 0 x:

2 0

LH

(B3J
c

ra ru LH ra ra ra
T:

a c

^ E .3 3 c
'^ Ob

. ^ . ^ ^. 3 3 3 u ^ u
^ ^

ra

ra ra ra ra i-H
0 0 0 0

oa

00

I r^ Q^ \C

in in ^ ''

3 S

0)

Mean

79.99 39.34 13.0! 12.35

00

114.7 830.8 213.7 143.5

o
m o I*"! so in oc in in

Van

ri u ^

III
III

fco050O0US

.2 c c c ^ .^
ngs
30 DO

ngs
a B

Iho

Iho ricu

38

RURAL DEVELOPMENT IN TRANSITIONAL CHINA

TA B L t,4 COUNTIES CLUSTERS IN THE MACRO-AREAS

Clusters - macro-areas Cluster 1 Cluster 2 Cluster 3 Cluster 4

Number of cluster

Number of counties inside the cluster 24, 16,8,21,46 427,206, 15, 14,75 100 42.333, 15, 1.050

Total counties 115 737 100 1.440

5 5

1 4

Source: Authors' processing of Ahstmcl of the First National Agricultural Census o/ Chino [1998].

counties are about 40 per cent of the tola! number of Chinese counties and they are eharacterized by a very low per capita possession of cultivated land (0.03). Further, this cluster has a high proportion of non-agrieultural households to rural households (16 per eent} eompared to the other maeroareas (except the urban area) but the lowest proportion in cluster 4. Also, these counties have the highest proportion of persons in agrieultural activities to rural persons engaged compared to the other three eluslers in the principally agricultural area (SO per eent). It has also a lower proportion of irrigated land (36 per cent of agricultural area). In these counties the number of towns is also lower. The counties included in this cluster (1,050) appear so similar that we could not produce a further differentiation, at least with the indicators available. Since this is the most important cluster of eounties inside the most significant agricultural macro-area (at province level, see step two), it will be really important to have new and more indicators at county level. Only then we could make more precise groups of homogeneous eounties for each maero-area or province. Inside the green macro-area there is another important cluster of 333 eounties. The agriculture of this eluster seems lo be less developed than that of the previous eluster. The proportion of non-agricultural households is much higher (38 per eent) and the proportion of persons engaged in agricultural activities is much lower (60 per cent). The proportion of irrigated land is high (more than 62 per eent), and the number of towns is greater

CONCLUDING REMARKS

The results of the First National Agrieultural Census of China provide a better knowledge of the structure of Chinese agriculture. The differenees between provinees and eounties are very large, but the potential for future growth of Chinese agriculture and other economie aetivities in rural areas is bigger than expected.

" ^ S c
O O ^ r-

o -o r^ ro i-i rj vi -^

g-E 2 2

Illfl
g.3|
00 ^
r-^ r--

rj 00

oo

<i
o u
u. O
(/I

a.
n1

C:.

icull hine

o
1-

"

-|

(--1

t
(N
'1

00 >

. ,

> <

z <

<

u a E

Cl.

o o cx
O

VI

on "B so CJ Qfi

r-- rn o vo 00
1'

o o

(N I ^ (N

o t-

d d d

CE:

Z o
d
11 'J-

mil
E 2 2 ^

.2 - "5 =

00 O O-

OJ

OJ

M y > < C CL CL

r--

S O 5
0 0 0 0 0 0 0 0

a. a. o J2

Z o

REGIONAL DIFFERENCES IN CHINFSE AGRICULTURE

FICIURt 9 COUNTIES CLUSTLRS INSIDE THE FOUR MACRO-AtiRICULTURAL ARF.AS

41

The Agricultural Census shows that there is a deereasing importance of persons active in agriculture, commensurate with the economic development of the different provinees. The proportion of persons engaged in agriculture in total in the main municipalities is less than 40 per cent (Shanghai and Beijing), and less than 60 per cent in most provinces of the eastern part. In these provinees, there are now many households whose revenue is mainly coming from non-agricultural activities, and only 55 per cent of total households are pure agriculture households. The inulti-sectorial activity (or pan-time) of the households is now a reality in many provinces and counties of China. The great availability of workers and mobility of people working in agricultural households (between villages and townships) could be the main impetus for future developments of non-agricultural activities in China. The larger availability of cultivated land, but also the lower numbers of livestock (cattle, pigs, poultry) relative to previous estimates indicate the possibility for increasing the productivity and intensity of agricultural production, in terms of both the labour force and the amount of land worked. The data available from the Agricultural Census at province level show that the differences between the provinces are relevant. The official classification of Chinese provinces into six geographical regions or three

42

RURAL DEVELOPMENT IN TRANSlTrONAL CHINA

main economic-geographical regions (west, middle and cast) will not be sufficient to describe a composite reality of Chinese agriculture. In most cases, the largest difTerenccs emerge from western and middle provinces, on the one hand, and the eastern provinces on the other hand. But we have found more profound differenees inside this large geographical area, between provinces and counties. The cluster analysis we have applied using variables of the 1997 Census at province and county levels confirms that the four maeroagrieultural areas of China are different from the previous elassification (eeonomie areas, eeonomie-geographieal areas). Generally, the cluster analysis revealed that there arc numerous provinces and counties belonging to the same group (cluster). The largest macro-agricultural area is characterized by significant agrieulture and a high proportion of persons engaged in agricultural economic activities. The level of education of people is generally high while mechanization is relatively important. The other two macro-areas are mainly mountainous regions or peripheral areas, characterized by poor endowment of natural and agricultural resources. The smaller macro-area represents the specific situation of most urbanized provinces and autonomous municipalities, where the persons engaged in agricultural activities are less relevant and agriculture is more developed. The cluster analysis carried out in respeet of the previous maeroagricultural areas, exploiting the few variables available at eounty level, show clearly that inside each maero-area there are different groups (elusters) of counties, with different degrees of agrieultural development. The results of our analysis show a totally 'new' geography of agrieulture in China, The reality of Chinese agrieulture is different at province and county level. The future utilization of Census data at province and especially at eounty level could contribute to a better definition of the reality of Chinese agrieulture. Speeiflcally, further analysis eould be done to specify the actual situation of agriculture in eaeh Chinese provinee, by utilizing more data at county level. Further analysis of Chinese agrieulture will be based not only on structural agrieultural characteristics, but also on natural resources, human resources and on the relationship between agriculture and the general economic development. The publication of the complete results of the First National Agrieultural Census of China will provide all the information needed to achieve a better understanding of Chinese agrieulture. The definition of a new geography for Chinese agriculture, with appropriate aggregation of provinces and counties, will be particularly important for the policy maker in order to defme appropriate strategies and poliey measures for future agricultural developments.

REGIONAL DIFFERENCES IN CHINESE AGRICULTURE REGIONS OF CHINA

43

1. Administrative Regions of China (3!): Municipalities (4): Beijing, Tianjin, Shanghai. Chongqing. Autonomous Regions (5): Inner Mongolia, Guangxi, Tibet, Ningxia, Xinjiang. Provinces (22): Hcbei, Shanxi, Liaoning, Jilin, Heilongjiang, Jiangsu, Zhejiang, Anhui, Fujian, Jiangxi. Shandong, Henan, Hubei, Hunan, Guangdong, Hainan, Sichuan, Guizhou. Yunnan, Shaanxi, Gansu, Qinghai. 2. Geographic Regions of China (6): North (5): Beijing, Tianjin, Hebei, Shanxi, Inner Mongolia. Northeast 0): Liaoning, Jilin, Heilongjiang. East (7): Shanghai, Jiangsu, Zhejiang, Anhui. Fujian, Jiangxi, Shandong. Central South (6): Henan, Hubei, Hunan, Guangdong, Guangxi, Hainan. Southwest (5): Chongqing, Sichuan, Guizhou, Yunnan, Tibet. Northwest (5): Shaanxi, Gansu, Qinghai, Ningxia, Xinjiang. 3. Eeonomic-Geographieal Regions of China (3): East (12): Beijing. Tianjin, Hebei, Liaoning, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Guangdong, Guangxi, Hainan. Middle (9): Shanxi, Inner Mongolia, Jilin. Heilongjiang, Anhui, Jiangxi, Henan. Hubei, Hunan. West (10): Chongqing, Sichuan, Guizhou, Yunnan, Tibet, Shaanxi, Gansu, Qinghai. Ningxia, Xinjiang. 4. Macro-Agrieutlural Regions (4): Mountain area (2): Quinghai. Tibet. Peripheral area (9): Gansu, Ningxia, Xinjiang, Inner Mongolia, Jilin, Heilongjiang, Henan, Yunnan, Guizhou. Urban area (4): Beijing, Tianjin, Shanghai, Jiangsu. Principally agricultural area (16): Hebei, Liaoning, Zhejiang, Fujian, Shandong, Guangdong, Guangxi, Hainan, Shanxi. Anhui, Jiangxi, Hubei, Hunan, Chongqing, Sichuan, Shaanxi.
NOTES 1. 2. For a description of the problem in design of the Agriuullural Census and in data collection see Biggcri. L, [2001 ], wherea:i for a first comparison of the previous official statistics and the census results sec Xiangdong [2001 ]. The data at county level represented by the maps in the Ahsiraci of the First Nationai Agricullural Census of China were made available courtesy of the State Statistical Bureau (SSB).

44
3.

RURAL DEVELOPMENT IN TRANSITIONAL CHINA

We have used for this elaboralioti the CLUSTER procedure of SAS system. We have used for lhi>i eliiboration the FASTCLUS procedure of SAS system. In particular, first we used a hierarchical approach, through the Ward Minimum Variance and Average Linkage Clustering methods, in order to choose fhe number of groups of provinces. The results of the analysis show four or five clusters. Then we applied the non-hierarchical method according to the statislics (pseudo F statistics and Cubic clustering criterion) and we chose four clusters. For more details see Fanfani. R. and C, Brasili, 2000. The complete representation of the 15 clusters at county level within the four main clusters at province level is shown in Figure 9, REFERENCES

4.

Alexandratos, Nikos, 199(i, 'China's Projected Cereals Deficits in a World Context', /igrkiillural FA-nnomics. Vol. 15, No.I, pp.l-i(i. Biggeri, L., 2001, China Agricullunil Cen.siis Design and Data CoUection: Issues and Lessons. Proceedings of the International Seminar on China Agricultural Census Results, Beijing: China Slatislics Press, Biggeri, M., 1998, 'I fatlori dello sviluppo deH'agricoltura cinese dopo le grandi riforme, analisi cmpiriche su un panel di dati a iivello provinciate', doctoral thesis. University of Siena. Brown, L., 1995, W'/PO mil Feed China'' Wakevp Call for a Small Planet, New York: W.W. Norton. Fanfani. R. and C. Brasili, 2000, 'The New Geography of Chinese Agriculture', paper presented at the International Seminar on China Agricultural Census Results, Beijing, 19-22 Sept, FAO, 1998, 'Technical Seminar on the First Chinese Agricultural Census Preliminary Results'. Rome, 26 Feb. FAO, 2000, 'Agriculture: Toward 2015/30, Technical Interim Report", FAO, Global Perspective Studies Unit, Rome, April. Giovannini, M.. 2000. The Chinese Rural Non-Agricultural Enterprises: The Evidence from the Agricultural Census", International Seminar on China Agricultural Census Results, Beijing, 19-22 Sept. National Agricultural Census Office of China. 1999, Ahslraci of the First National Agricultwiil Census in China. Beijing, China Statistics Press. SSB (State Statistical Bureau), 1997. China Slatisticcil Yearhoiik 1997, Beijing: Beijing Agricultural University Press. USDA, 2000, "China WTO Accession Would Boost U.S. Agricultural Export and Farm Income", Washington, USDA Agriculture Outlook series. Zhii, Xiangdong (ed.}, 1995, The Ci>ndse Aggregate Data of National Pilot Agricitllural Census. Beijing: Food and Agricultural Statistics Centre (FASC). Zhu, Xiangdong, 2001, "A Coneise Analysis of the Main Results of the First National Agricultural Census in China", in Proceedings of the International Seminar on China Agricultural Census Results, Beijing: China Statistics Press.