In China, there are at least 50 million long-term migrants and a much larger volume of temporary counterparts who originated from the countryside, among which more than 20 million are inter-provincial movers.[1] The latest information indicates that in 1993 there were 100 million or more short-term migrants in China.[2] For some time now, peasant-related migration, or the 'peasant flood' as it is termed in China, has given rise to a number of social and economic issues new to the country. Leaving economic implications aside (see Sahota for a summary),[3] the huge volumes of temporary as well as long-term migrants have led to infrastructure, moral, security, family planning and other social and political problems throughout the nation. These problems attract a lot of attention from the government, the mass media and society at large.[4] Given the severity of the problems and the fact that this unstoppable migration process will inevitably accelerate as modernisation and industrialisation proceed, studies on internal migration in China, which are currently lacking, will be of the utmost urgency and importance, particularly for socioeconomic policy making and implementation.

Unlike many other less developed countries (LDCs), where rural-urban migration dominates the scene (see Nam, Serow and Sly and case studies therein[5]), significant inter-sectoral migration within small localities, or intra-township migration, has been a unique feature of the Chinese modernisation process. This inter-sectoral migration, which overwhelmingly depends on the growth of local industrial enterprises in various forms, has important implications for the urbanisation path in China and for the restructuring of the national economy as well as for the flow of resources. Any studies on Chinese population movement that do not consider the intra-township migration are incomplete, to say the least. On the other hand, inter- and intra-urban migration is relatively small because of the regulatory nature of the urban labour market coupled with the rigid wage, education, housing, medicare and other welfare systems. Thus, we leave inter-and intra-urban migration aside in this paper. This omission is also attributable to data unavailability.

Our focus on rural-rural, rural-urban and urban-rural migration can be further justified on the grounds that some 84% of the 1172 million Chinese population are rural residents and about 150 to 200 million rural labourers are in surplus in China.[6] Conversely, there are at least 3 million urban youth workers in the rural areas and the urban-rural migration trend is increasing.[7] Therefore, it is far more important to study rural-related migration. As economic reform deepens, the rural surplus human resource not only moves out of agriculture and enters non-traditional activities within localities (so-called litu bu lixiang---leaving land but not townships), it also penetrates tightly controlled urban labour markets and flows to distant rural and non-rural areas (so called litu you lixiang--leaving land as well as townships). Although the former has long been encouraged by the central government as a strategy of controlling the urban population explosion, the latter was rather unexpected with respect to its pace, scale and consequences.[8]

Briefly, the cause of the peasant-flood is recent economic reform and the associated social changes which have initiated and stimulated large-scale internal migration. Among other things, the abolition of the grain procurement system and gradual removal of the state monopoly over the grain trade has provided sufficient conditions for internal migration. Conversely, beyond the usual benefits of labour mobility as predicted by economic theory, migration does promote economic reform in the sense that the presence of migrants may provide the prerequisite for deregulating the urban labour market and instituting needed reform in industrial relations in the state enterprises. In addition, the emerging urban-rural (vs rural-urban) migration exerts pressure on the government and appropriate management to reform the state-owned segment of the economy if they wish to have some control over the outflows of the highly educated and skilled workers from the formal sector. As often reported in the media,[9] migrants also help transmit useful market information, disseminate technologies and bring both physical and human capital into the countryside, all of which are rather lacking but vital to the development of the rural economy.

Based on a set of micro-level survey data, this paper, one of the first of its kind, aims at revealing and discussing the facts and patterns of internal migration in China. Cross-country comparisons will be made when appropriate. It will become clear that many features of the Chinese case are unique, and some may be surprising. In particular, rural-rural migration is found to be positively related to distance while out- and net migration is negatively related to education. Female participation in migration is found to be extremely low and the tails of the age distribution of migrants are remarkably thin. Considerable effort will be devoted to the explanation of the major findings, though our ability to do so is limited in some instances because of the absence of relevant literature and data. Nevertheless, the information provided and the characteristics discovered regarding internal migration in China are useful in their own right, especially for comparative and further studies in this area.

The survey, undertaken by the Chinese Academy of Social Sciences in conjunction with relevant government authorities, was intended to discover the composition and patterns of rural population movement. To obtain meaningful information on intra-township migration, it was necessary to conduct the survey at a village level. The village (formerly called a production brigade) is an institutional unit right below the township (formerly called a commune) but above the production team, which is now virtually non-existent after the implementation of the family-based production responsibility system. Because of the enormous number of villages in China and the limited research resources available to the Academy, random sampling, which requires a substantial number of villages to be drawn in order to maintain its usefulness. was not employed. Instead a stratified survey procedure was adopted, which involved (i) choosing the representative provinces, autonomous regions or metropolitan cities (for convenience, they will all be referred to as regions hereafter); (ii) choosing representative counties within each representative region, and (iii) choosing representative villages within each representative county. Here, 'representativeness' is measured in terms of both development status and geographic location of the sampling population. A total of 230 villages was selected from 84 townships in 59 counties, scattered across the following 11 regions: Shanghai, Jiangsu, Zhejiang, Fujian, Helongjiang, Guangxi, Hebei, Shanxi, Le menggu, Ningxia and Qinghai. Information from eight villages was found to be incomplete by the surveyors, thus what will be used in this paper are data for 222 villages.

The survey documented 1986 population movement statistics through extensive interviews and record checking. Population flows were classified into three categories: (1) intra-migration or intraflow, which involves those who migrated within the boundary of their own townships; (2) out-migration, emigration or outflow, which involves those who left their townships; and (3) in-migration, immigration or inflow which involves those who moved into the 222 villages from outside the respective townships. To be counted as a migrant, an individual must have stayed in the new position or location for 30 days or more. Those who migrated for less than a year were classified as temporary, short-term or seasonal migrants; otherwise, they were referred to as long-term or permanent migrants. Age, sex, education, postmigration occupation and duration of stay (30-60 days, 61-180 days, 181-365 days, a year or more) details were obtained for every migrant. Origins for in-migrants and destinations for out-migrants were also recorded. Because relocation usually does not occur for within-township movers, no destination-origin information was gathered for the intra-migrants.

To show how representative the survey data are, Yu computed correlation coefficients between national totals and the aggregates from the surveyed villages for these variables (correlation coefficient in parentheses): agricultural labour force (0.98), rural GDP (0.98), per capita income (0.91) and number of labourers transferred out of agriculture (0.95). The high correlation coefficiencies indicate that drawing inferences from the survey data can be justified.[10]

Extent of migration

Based on the survey data, some 43 223 persons were involved in intraflows and 26 993 emigrated from the surveyed townships in 1986. The sum of 70 216 represents 37.15% of the corresponding labour force. Those who entered large cities accounted for 3.8% of the emigrants or 1.46% of the total migrants from the 222 villages (= intra-movers + emigrants). Meanwhile, 7793 persons immigrated into the surveyed villages, some 10% of whom were non-rural residents. This pattern is clearly in contrast with those experienced by most other developing countries, where the majority of rural migrants rush to big cities. The self-initiated emigration signifies the end of the long-standing tradition of living on the land. Although small in percentage terms, the impact of such a change in peasant ideology and behaviour on the social and economic system is far-reaching. For example, the population is no longer made of two distinct and geographically separated classes: the peasant class and the worker class (the leading class). A person can no longer be classified as a rural peasant (second grade citizen) or a city resident by his/her place of birth. In other words, the politically-enforced boundaries between urban and rural economies are disappearing. A national integrated labour market is being formed.[11]

Distance and migration

Contrary to earlier findings,[12] the volume of internal migration in China seems positively related to the distance between the origins and destinations of migrants, at least in the case of rural-rural and urban-rural migration. This phenomenon is believed to be attributable to the structural unemployment problem and unbalanced resource bases across rural China. For a detailed analysis, see the section on destinations of migrants in this paper.

Gender composition of migrants

Among the 70 216 migrants, about 74% were males and only 26% were females. This composition shows a remarkably low female participation in internal migration in China. Elsewhere, about half of the migrants are usually females. For example, in India females accounted for 69% of internal igration in the 1960s and 1970s. This percentage went up to 70.5% in 1981.[13] Even in Israel,[14] Botswana,[15] the Netherlands[16] and Egypt,[17] female migrants more or less matched their male counterparts. Reasons for the low female-male migrant ratio in China will be discussed later.

Occupations of migrants

The survey revealed that over 57.38% of the migrants from the 222 villages shifted into industrial (34.31%) and construction (23.07%) activities. The service sector (food providers, stall holders, door-to-door retailers, barbers, small shops and the like) attracted 9.73%. There were only 565 who engaged in cropping and 1614 in forestry, animal husbandry, sideline production and fishery. Adding them up, this group accounted for a negligible 3.11% of the total migrants from the surveyed villages. It is clear that internal migration in China exclusively involves people moving out of traditional activities. For comparison, in Egypt in 1976, 12.8% of active migrants were participating in primary activities, 8.1% in construction, 21.4% in industrial activities and 12.5% in service industries.[18] In Botswana at least 15.8% of total migrants were working as primary producers.[19]

Age structure of migrants

Breaking down the data into age groups, 2.4% of migrants (1688 heads) were aged 17 or under, nearly 88% (61 530) were aged between 18 and 45 and the remaining 9.9% (6999) were 46 and over (see Table I). This pattern is broadly consistent with that of Kenya,[20] Egypt[21] and Japan.[22] However, the tails of the age distribution are quite thin compared with those of the above-mentioned less developed countries and undoubtedly much thinner than those of developed countries. This reflects the nature of the Chinese case, where almost all migrants are essentially labourers. Family migration is rather uncommon in China. Hence, the terms of labour movement/mobility and internal migration can be used interchangeably in the case of China.

Short-term vs permanent migration

In 1986, seasonal migrants accounted for some 80% of the total emigrants from the surveyed villages. The corresponding figure for immigrants was 47%, significantly lower. Taking migrants and immigrants as a whole, the majority were temporary or seasonal movers. This pattern has mainly resulted from the underdevelopment or absence of an urban labour market, from a shortage of social services for migrants and the lack of an appropriate migration policy.

Education and migration

A surprising discovery is that education did not play a significant role in promoting migration in the mid-1980s in China. Table II indicates that the percentage in total migration for each education class more or less equalled its share in the total labour force. Rounding to two decimal points, 39% of the labour force received primary education and the same percentage of the out-mi-grants were primary school leavers. This result contradicts most previous theoretical and empirical findings, irrespective of the development status of the nations under study.[23] A thorough exposition of the relationship between education and internal migration in China is provided below.

Education and internal migration: are they related?

Herrick,[24] Levy and Wadycki,[25] Schwartz[26] and King,[27] among many others, concluded that education is positively related to migration. The argument behind such a conclusion is that the more educated are better informed about employment opportunities. Essentially, this is also what human capital migration models would predict.

What then do the Chinese data tell us? Using the 1986 year-beginning labour force as the base, rates of internal migration by education levels are computed and presented in Table III. Among those who had received tertiary education, 11.07% migrated from the 222 townships. The percentage was slightly higher for middle (11.88%) and high (11.94%) school leavers. It increased to 13.19% for primary school leavers and rose to 16.52% for the semi-illiterate and illiterate members of the labour force. Without attaching any statistical significance, the Chinese evidence is exactly contrary to previous findings. That is, education is inversely related to migration.

On the other hand, there was a positive connection between education and intra-migration if the first group (tertiary and above) is excluded from consideration (column 3 of Table 3). Also, unlike the case of out-migration, the intra-migration rates for middle and high school leavers are clearly different. This pattern, however, does not necessarily lend support to human capital migration models, as information on employment opportunities within a township are largely transmitted through person-to-person communication (which can be quite effective), in which case the educated are not really in a more advantageous position than others.

The most influential causes underlying the seemingly contradictory results, in the author's view, are the profound effects of reverse selectivity bias and the structural unemployment problem in China. In other words, it is not who chooses to migrate but who is in demand as migrant labour that matters most. It will become clear later that over 50% of the out-migrants have entered non-rural or urban areas (which include rural towns, county-level towns and cities of various sizes) where low skill labourers are in large demand. On the other hand, rural development now requires a better educated and skilled labour force than is available. Sweeping economic reforms have introduced market forces into the production, consumption, and marketing processes of individuals, households and various institutions in the rural areas. Thus, economic life there is no longer so routine. In particular, decision makers in non-urban areas not only face all kinds of risks but also have to bear the consequences of their decisions. All these need intellectual inputs into the multi-dimensional management system and basic operation of rural firms. It is in this sense that we can conclude that the unemployment problem in rural China is essentially a structural one. This explains the relatively high out-migration but low intra-migration rates for the less educated and the low out-migration but high intra-migration rates for the better educated. Since opportunities for the most qualified, who typically only accept secured and prestigious jobs, are rather limited in both urban and rural areas, their mobility is somewhat restricted. In fact, the better educated in the rural areas often hold technical, managerial and administrative positions, from which they enjoy an immense amount of respect and privilege. This naturally leads to the negative (positive) education-mobility relationship for emigrants (in-migrants). The reason underlying the clear differentiation between high and middle school leavers for intraflows is also attributable to education being a determinant in rural labour markets, while for outflow, education or skills do not count as much. In short, migration is largely demand-driven in China.

The above observations and arguments imply that the usual urban-pull effect is rather weak, if not absent, in China. Otherwise, urban areas would attract rural labourers from across the board if not favourably biased towards the better educated, regardless of whether or not a structural unemployment problem exists. Clearly, those who have obtained a better education do not appear to be disadvantaged compared to the less educated in the urban labour market. The fact that relatively more of the better educated stayed in the surveyed townships is obviously indicative of weak pull of the urban areas. It seems that the rural sector pushes the uneducated out on the one hand and pulls some of the best educated in on the other hand. The strong pushing effects, together with the structural unemployment problem can also be used to explain the surprising but positive distance-migration relationship, which will be addressed later.

The conclusions regarding reverse selectivity bias, structural unemployment and urban-pull and rural-push effects are supported by three other findings: (i) While the net outflow of migrants from the 222 villages was 7171 in 1986, tertiary educated in-migrants actually increased by 11 or 3.67%. The net migration rates increase as education levels decreases (Table IV), which may imply that the less qualified are facing stronger unemployment pressures in rural China; (ii) The inflows were better educated than the outflows. See the compositions of migrants by education levels in Figure 1. For example, nearly 72% of out-migrants only possessed primary education or below--the proportion is about 60% for intra-migrants and only 56% for in-migrants; (iii) Relatively more in-migrants were long-term stayers (see below), reinforcing the conclusion that the better educated are indeed demanded by the rural sector, despite the existence of a great amount of unemployment there.

Recognising the interdependence between growth and development and human capital, the future of the rural economy appears reasonably optimistic. The crucial question is: to what extent can the reverse pull and the lack of urban pull be attributed to the state-ownership nature of and absence of incentives in the urban formal sector? Foreseeing fundamental economic reforms in the state industries, how to sustain human capital inflow into or simply to keep the well educated in the rural areas is a question which must be seriously taken into consideration by policy makers at all levels, particularly by reform architects in China.

Why are females less mobile?

Calculation of sex ratios from the survey data shows that females were relatively less mobile irrespective of directions of flow (see Table v). In particular, the male-female ratio of 100:28 for out-migrants is extremely low by international standards. The ratio is slightly higher for intra-migrants (100:36) and even higher for in-migrants (100:59). There are several factors underlying the low mobility of females in China. These factors also contribute to the variations in the ratios among in-, intra-and out-migrants.

First of all, culture and customs are against female migration: virtually all decisions concerning a household or its individual members are made by the elderly and male members of the family. Females are often not allowed to travel far away from home without relatives or family companions, let alone to migrate by themselves. For thousands of years, females have traditionally been associated with housework, including washing, cooking, looking after the children and particularly the elderly. On the other hand, male members of the family (husband, sons, brothers) have been conventionally assigned to work outside. Whenever there is a need or opportunity to migrate, males are usually positioned ahead of females. This explains the across-the-board low female mobility. While these customs and traditions are undergoing changes, the traditional tie between females and housework, and the general acceptance by society and women themselves that females should stand behind the family, are still quite prominent, particularly in the vast poor areas. In the case of intra-migration, where it may not be necessary for females to break ties with their families, the female-male ratio improves. To elaborate, if a woman has children, the elderly or a husband to look after, a job change within her township might be acceptable, but out-of-township positions may not even be considered by such a family. The family tie issue is also reflected by the fact that sex ratios were considerably higher for the long-term migrants for all three categories (see Table V). The provision of child care, nursing homes and boarding schools, which may help break the female-family tie, should be looked at in promoting female mobility.

The second factor, which is not unrelated to the first, lies in the land plot assigned to each household mostly on a per capita basis, under the production responsibility system. Farmers consider land to be as important as their lives. They will not give it up even if it is more profitable to do so in the short run. Unless non-farming income is sustainable and reaches a high level which relieves the household from worrying about food shortages and price risks in bad harvest years, households prefer to be self-sufficient in their basic food supply, at least to a certain extent. Under this circumstance, customs and tradition come into effect to tie female labourers up with the plot, even if the elderly, children or housework problems are resolved. In many places, the small plot can be looked after on a part-time basis, but the labourer must stay within the township. This further explains why the intra-migrants had a higher female component. Stability or security of grain supply in the rural areas, and policies promoting corporative farming which allow individual families to invest in land only in the co-ops, would help increase female mobility. In addition, the incomplete nature of property rights over land in rural China must be rectified as it prevents peasants from selling their plots in order to become permanent migrants.[28]

The absence or shortage of social facilities available to migrants represents the third factor which is detrimental to family migration and thus female migration. This problem is particularly severe in the urban areas. As reported in Zheng et al.,[29] most construction workers resided in worksheds or big halls, peasant traders mostly sleep in the free market or bus/train stations. In addition, public utilities such as hospitals and schools are largely closed to migrants. This is why even some long-term migrants have to leave their families behind. The link of sex ratio to family migration can best be seen by comparisons between long-term and temporary migrants, as it is unusual for temporary migrants to take families with them. On the other hand, there is no reason, other than family migration, for long-term migration to have a larger female component than its temporary counterparts. Statistics presented in Table V confirm the conclusion here. For example, the male-female ratios were 100:88 and 100:35, respectively for long-term and temporary in-migrants. The issue of social facilities/services can also be used to explain the thin tails of age distribution of migrants, as discussed in the section on age distribution in this paper.

Finally, there may exist the sex-specific job factor. Bardhan suggests that sex-specific labour inputs are not perfectly substitutable in production.[30] Most emigrants engage in heavy, hard and even dangerous work and meanwhile live under appalling conditions. These are just not appropriate for women. Women are considered more suited to service industries, eg restaurants, hotels, shops. As is shown later, the absorption capacity of the service industry is rather small. The decreases in female mobility from inflow to intraflow to outflow correspond to the increases in the percentage of construction, building and transportation migrants. This may well confirm the sex-specific and non-substitutability argument put forward by Bardhan. As the service industry grows and as the urban industries other than construction and transportation open their doors to rural migrants, female mobility is expected to improve.

The above factors all contribute to the higher female-male ratio of incoming migrants as these consisted of some 10% urban residents, who are better educated. As is known, urban residents and the better educated are more liberalised with regard to customs and traditions. The urban-rural migrants are not in any way associated with land plots. Moreover, since some of the incoming migrants are sought after by the rural sector, and in fact normally have social facilities pre-arranged for them, family migration is definitely more prevalent for in-migrants. These mean a more balanced sex ratio and a more balanced age distribution. In a similar way, the intermediate sex ratio for intra-migrants can be explained.

Obviously these factors do not act independently and their relative impacts may change as the economy develops and as further reforms are introduced. Assessing the relative impacts of these factors awaits more data and further research, possibly in the form of econometric estimation of qualitative-response equations such as multivariate probit models.

Destinations/origins: a positive relationship between distance and migration

Looking at the destinations of the migrants from the surveyed villages, over 61% of them moved within their townships, 39% left the townships. Among the out-of-town migrants, 48.8% migrated to non-urban areas. The remaining 51.2% who entered urban areas were composed of 17.3% going to rural and county-level towns, 29.5% to small and medium cities, 3.8% to large cities and 0.6% to other countries. As far as origins of in-migrants are concerned, nearly 91% of them were from villages, with over 9% plus originating from urban areas, that is, towns and cities.

Human capital migration models propose that distance is a factor limiting internal migration because it is negatively related to information flow and positively related to travelling cost. Another reason might be the uncertainty and risks associated with social and cultural difference/conflicts in distant areas.[31] Surprisingly, the Chinese data do not support this proposition. Among the 26 993 out-of-town migrants, 1418 went to other rural towns in other townships, 3255 went to county-level towns, 7941 went to small and medium cities, 1024 went to large cities and 174 went overseas. The remaining 13 181 went to villages in other townships. Ignoring the big cities and overseas for the sake of argument, the figures trace out a positive distance-migration relationship, since distances to rural towns, county-level towns and small and medium cities can reasonably be assumed to be progressively increasing.

Readers may well question the validity of the positive distance-migration relationship revealed by comparing flows into rural towns, county-level towns and cities. And there is a good reason for them to do so, in that cities, towns and villages are heterogeneous in many aspects, among which distance is only one, and perhaps not the most significant. Of particular importance is the existence of many institutional constraints on internal migration in China; these constraints are different for migrating to villages, towns and cities. All else being equal, entering cities is most difficult and entering villages is relatively easy as far as government regulations are concerned. Therefore, the destination factor must be somehow controlled in order properly to establish the distance-migration relationship. This can be achieved if migrants to cities (or towns or villages) can be further disaggregated according to distances (for example, remote cities and nearby cities). Fortunately, the village-destined emigrants from the 222 villages can be further disaggregated into within-county and out-of-county movers. Also, the village-originated immigrants can be broken down into three groups: within-county, inter-county but within-region, and inter-region. Note that the village-originated or village-destined migrants as a whole can be viewed as being homogeneous, not only in terms of the group's origin and destination, but also in other aspects such as education, family background and so on. As a consequence, any conclusions extracted from such disaggregations are believed to be fairly convincing.

Among the 13 181 out-migrants who all emigrated from the surveyed villages and all went to villages or non-urban areas, 5606 were intra-county emigrants, while 7575 migrated outside the county boundaries. Meanwhile, among the 7060 in-migrants all of whom came from non-urban areas and immigrated to the surveyed villages, fewer than half (3169) came from within the county. Moreover, village-originated in-migrants from other regions totalled 2111, outnumbering the inter-county but within-region counterpart, which was 1780. In passing, it is noted that among the urban-rural migrants (sum of 733), over 68% came from cities (501) which are usually further away from villages than towns, and yet only 232 were from closer towns. The Chinese evidence undoubtedly shows a positive relationship between migration and distance. This relationship remained valid in 1993.[32]

The finding is certainly surprising but is firmly conclusive as well. The root cause of the positive relationship identified lies in the structural unemployment problem coupled with different resource bases not only between urban and rural economies but also across geographic locations within rural areas. The different skills and qualifies of labourers which were inherited or developed locally during the period of isolation and separation are often complementary. It is the author's first hand experience that Jiangsu school leavers are often trained as bricklayers, carpenters or dress-makers, while Zhejiang residents are reasonably known for their merchandising skills. The majority of hairdresser migrants are from Guangdong and the majority of urban baby sitters are from rural Anhui and Sichuan. Since there is a general unemployment problem in China, the further away the destinations are, the better the match of skills with the relevant demand, thus the more migration. The existence of reverse selectivity bias in China is reflected here again.

It is noted that the small and medium cities played a significant role in promoting internal migration in China. They absorbed nearly one-third of out-migrants. As is known to many China observers, sociologists and economists in China have long advocated the importance of small and medium cities in absorbing surplus labour. Since 1982 China has fostered the growth of rural marketing centres and hoped that they will form small cities, providing alternative urban settings for the absorption of surplus rural population. This has been called a rural urbanisation strategy adopted by the central and local governments. The results presented above indicate a degree of success of the government strategy and should help reinforce the views of sociologists and economists, though the success is very much an unconsolidated one in the sense that most of the migrants to small and medium cities were temporary stayers (see later discussion).

As announced by the Deputy Minister for Construction in late 1994, a more extensive experiment is underway focusing on the design and implementation of rural urbanisation policies. At the national level, the central government is in the process of selecting 500 key rural centres for trial. Regional and county governments are required to select their own key centres.[33] The analysis in this paper points to the need to provide public services to migrants in a flexible way when designing such experiments.

Occupational distribution of migrants

As mentioned earlier, there is little within-agriculture migration in China. The vast majority of migrants (97%) shifted from agricultural to non-farming activities. One may ask whether labour-exchange and within-village migration across farms are common in China. The answer is no. This conclusion is reached by analysing a set of survey data on household time allocation. The survey shows that only 3.1% of total labour utilisation was used for exchange or hire.[34] Further, this tiny 3.1% was not used in agricultural production because its distribution over time reversely matched the agricultural cycle. In China, the busiest farming periods are the third then the second followed by the low first and then fourth quarters, but the exchange and hiring of labour mostly occurred in the fourth (1.16%) and first (0.84%) quarters. The busier second quarter received 0.75% and the smallest 0.35% went to the busiest quarter. This pattern of exchange and hire labour usage closely matches the private house-building activities in the rural areas, which is consistent with the author's first hand observation. Thus, there is indeed little, if any, agriculture-agriculture migration in China. On the contrary, cross-farm migration is quite substantial in many other developing countries.[35]

Table VI presents the composition of migrants by occupation. At the aggregate level (see the last column of the table), the employment pattern of migrants seems broadly consistent with international evidence. That is, most migrants are involved in building and construction and industrial activities. An important difference, however, lies in the high percentage of intra- and in-migrants engaged in rural-based industrial activities (45.1% and 52.3%, respectively). By inference, the urban formal sector absorbed little rural surplus labour as only 17% of out-migrants participated in industrial activities and most of them were likely to be employed in rural enterprises. The role of township enterprises in absorbing surplus labour can thus never be overlooked, as strongly argued by Zhou, Dillon and Wan.[36] On the other hand, the absorbing capacities of agriculture and the service sectors appear quite low by international standards.

Given the importance of rural industry in China, it is useful to scrutinise its capacity of absorbing surplus labourers. This is achieved by further breaking down the industry-employed 19 499 intra-migrants. The largest share went to village-run enterprises which absorbed 33.42% (6517 heads). Conversely, the role of township-run enterprises is diminishing since they only absorbed 13.5% (2632). The recently established and flourishing family/private and cooperative enterprises in various forms accounted for 50.97% (9940). Of particular interest is the large in-taking capacity of family/private enterprises, which absorbed 25.52% (4977) of the intra-migrants. From an overall perspective it is likely that government policies in general, and the credit or lending policy in particular, which are currently biased towards collective (township-run and village-run) enterprises, have an effect on migration and should be reviewed.

Accompanying the rapid development of rural enterprises and further division of labour in rural areas, the service industry will germinate and prosper. The fact that a moderate proportion of intra- and in-migrants were active in the service sector signals the beginning of the formation of and the potential for growth of the rural service industry. The disappointing share of the service sector in out-migration has a lot to do with the nature of the urban informal sector and government policy. The urban service sector has been regarded as the major destination for the urban unemployed. In particular, the pressure on the city administration to solve urban unemployment problems had often produced discrimination against service outlets proposed or operated by rural migrants. The general absence of social facilities for rural migrants also put them at a disadvantage as the service sector, dictated by the nature of the industry, demands long-term residence. In fact, among the occupations for emigrants, the service sector possessed the highest rate of long-term stayers (see Table VII). Unless reforms in the urban housing, education, medical service and administration are complete and successful, the percentage of rural migrants participating in urban service activities is unlikely to reach the level witnessed in other LDCS.

The patterns of internal migration, in terms of occupational distribution, were quite consistent among out- in- and intra-migrants. Leaving the 'other' activities aside, industrial and construction and building sectors ranked first and second in attracting migrants, followed by the service sector. One noticeable difference between outflow and inflow or intraflow is the proportions of migrants under the row heading 'other'. The 10 percentage point difference is partially attributable to the fact that students progressing into high schools and above fall into this category and most of these schools are located in cities or large towns. On the other hand, it is rare for students at any level to migrate into village-level (almost exclusively primary and low-quality) schools.

Two interesting observations are worth mentioning. First, rural construction activity is found to be quite substantial and absorbed some 17.6% of intra-migrants and 13.9% of in-migrants. To the extent that building and construction mirror industrialisation, rural modernisation is well and truly underway. Second, the proportion of immigrants getting into enterprises has been very high, exceeding 52%. The underlying causes are: (i) immigrants include non-rural residents, who in general would not come to the countryside unless sacrificing their urban lifestyle was compensated through monetary gains;[37] (ii) immigrants are better-educated than the existing rural labour force or emigrants. As long as returns to human capital are positive, which is believed to be the case in rural China, it is reasonable for them to find employment positions with better financial rewards. Needless to say, rural enterprises provide the only opportunity for job security and high income in rural areas.

Age composition of migrants: the younger the more mobile

Thanks to the absence of age composition data in terms of either population or total labour force of the surveyed villages, percentages who have migrated within each age group cannot be calculated. The age compositions of migrants are, nevertheless, shown in Figure 2. If the two tails are ignored for the moment and each panel is examined independently, then the proportion of migrants declines as age goes up for intraflow, inflow and outflow. In all cases, the age group of 18 to 35 constitutes the largest component, exceeding 53%. This of course may be related to the whole population structure.

Comparing across panels in the figure, it seems that out-migrants possessed the youngest composition while immigrants had the oldest age structure. These differences remain true even if the group 18 and under is excluded from consideration. The relatively young age structure of the out-migrants signifies that the younger members of the population are more mobile. Many researchers have drawn the same conclusion. However, previous literature often attributes this finding to one or both of the following causes: (a) young people have a higher return to migration in the long run, (b) they seek education opportunities in urban areas. These are not really applicable in China as the door of the formal urban sector is not open to rural migrants and student migrants are generally under 18 (education opportunities can only be pursued through progression in China and cannot be sought at a later date). It can even be asserted that there are no long-run prospects for rural-to-city migrants and that these migrants therefore do not include any long-run ingredients into their decision-making processes. The fact that the emigrants possessed the lowest proportion of long-term stayers for every age group, except the youngest, speaks for itself (cf. Table X). From another perspective, the majority of rural-city migrants are construction and building workers (cf. Table VI) and yet this occupation held the least proportion of long-term migrants, second only to the typically seasonal agriculture (see Table VII). Consequently, the evidence from China points to the rejection of the earlier attribution regarding internal migration and age.

In the author's view, the young age structure of emigrants stems from both the supply and demand sides. On the supply side, young people are more active, more ambitious and more adventurous. They thus possess a higher propensity to move. On the demand side, relatively more of the emigrants are engaged in construction and transportation activities, which require strong and young labourers. Also, young people can easily fit into different and rather harsh sociocultural environments. The older age structure of in-migrants can be attributed to the fact that rural enterprises demand experienced and sometimes retired workers from the urban areas. These technical staff often bring with them product orders and new contracts. In fact, some retirees do go to country areas to start up enterprises of various forms. This explains why the in-migrants had the largest percentage of people over 55.

As discussed earlier, institutional arrangements and availability of social facilities for migrants, apart from job security which is a relevant factor in any case and in any country, determine the extent of family migration in China. This is why family migration is most extensive in the case of in-migration, and why out-of-town family migration is small. To a certain extent, these can be used to explain the relatively fat tails of the age distribution of in-migrants and the in-between age structure of the intra-migrants. One observation does need special attention--the age group 18 or under was smaller for intraflow than that for inflow. This is because family members who were dependants of the intra-migrants were not counted as migrants.

To reinforce the conclusion that young people are more mobile, the ratios between the intra- and out-migrants for all age groups are computed. The calculation shows that for the youngest age group (17 years or under), if 100 persons migrated within the township, more than 113 would migrate out of town. For the groups 18-35, 36-45 and 46-55, each 100 within-town migrants was accompanied by 70, 58 and 38 out-of-town migrants, respectively. In the eldest group, for every 100 in-town migrants, fewer than 17 left the town. Clearly, the younger the migrants, the more of them tended to move away from home. It is fairly reasonable to expect that younger people are more likely to be pulled by the urban environment, where a sense of adventure and the attraction of entertainment (not necessarily economic gains) play a role.

Temporary vs long-term migration

Given the huge surplus of rural population and the government intention to promote rural urbanisation through the development of towns and cities up to the medium size, one would logically expect the urban territories to accommodate a substantial proportion of long-term migrants with the exception of the large cities. Apart from the fact that migrant flows into large cities were indeed small, the survey data do not match government intentions or normal expectations. According to the survey data, nearly 80% of out-migrants were seasonal migrants. Conversely, over 50% of in-migrants and intra-migrants were long-term stayers. Worse still, Table VIII reveals that 65.72% of migrants into big cities were long-term migrants. In sharp contrast, long-term migrants into small and medium cities only accounted for 20.24%. It is also worth pointing out that urban-rural migrants tended to stay permanently (over 87%), while only 48.73% of rural in-migrants stayed permanently.

The across-the-board high long-term rates for in-migrants reflects easier access to accommodation, schooling and medical care, and the initiatives of authorities to assist migrants in rural areas. Provision of social facilities and reform of the welfare systems in towns, small and medium cities are definitely needed to facilitate rural urbanisation. It should also be mentioned that social and economic problems arising from internal migration, as reported in the media, are largely related to temporary migrants. Sensible solutions to these problems are difficult to find or implement, given the nature of the blindness and temporary nature of the flow (temporary migration is often referred to as blind flood or mangliu in Chinese). It is thus suggested that government policy should aim at increasing long-term migration rather than trying to manage the unmanageable, particularly in the medium and small cities.

Turning to distance and duration of stay, it is natural to propose a positive relationship between these factors for two reasons. First, seasonality of cropping dictates that those who are attached to farms must be temporary migrants; they would normally not migrate out of their counties of residence. Second, it is only economical to travel out of a region if a relatively long-term job is in sight. Once an individual or family has migrated to distant areas, job-seeking is the norm and can last for quite a while. There is, perhaps, a psychological factor here as well. Long-distance migrants, like gamblers, bet too much mentally and economically, thus they tend to insist on searching for opportunities of 'winning'. But not-so-distant migrants tend to return home quickly to await for future opportunities when a job is completed or a position is not clearly in sight. The Chinese data, however, do not provide a clear answer to our reasoning. When out- and in-migrants are added together, long-term migration rates are higher for intra-county (24.25%) than inter-county movers (22.39%), although inter-regional migrants did seem to possess a higher long-term rate (44.78%).

Table IX displays percentages of long-term migrants for each education level. It seems that terms of contract were positively related to education. This is definitely true for out-migrants. Also true is the fact that the best (worst) educated always had the highest (lowest) long-term rates across all three categories. Moreover, the data show relatively minor differences between high school and middle school leavers. This is understandable as they only differ by two years of schooling and current demand may fail to distinguish these two qualifications. Ignoring the two end-groups in Table IX, there seemingly exists a negative relationship between duration of stay and education for the inflows, for which we cannot offer a compelling explanation.

Long-term migrants as a proportion of total migrants in different age groups are presented in Table X. Except for intraflow, the two end-groups had the highest long-term rates. This is consistent with the earlier observation that family migration usually does not occur unless the migrants are long-term stayers. Naturally the children and elderly are most likely to be family migrants and thus are most likely to be long-term movers. Since intraflow does not involve relocation or family migration, the tails of the intraflow age distribution were understandably thin. In fact, the end-groups for intraflow had the lowest long-term rates. Leaving the eldest group aside, a positive relationship between age and period of stay does emerge, again with intra-migration as an exception. Clearly, the group aged 18-35 not only represented the largest group in all forms of migration, it also had higher long-term rates in the cases of out- and in-migration. In all cases, there is little difference between the age groups of 36-45 and 46-55.

By occupations, the underlying determinants of tenure of contracts for migrants must be the nature of the sectors. For example, manufacturing activities are usually continuous, with little seasonality; so are small shops and restaurants. Thus, migrants in the service and industrial sectors tend to be long-term stayers. Transport is in short supply in China, thus any seasonality may not affect the employment pattern in this sector. These arguments are all confirmed by the figures presented in Table VII. Construction and building workers possessed the smallest proportion of long-term stayers, whether or not they were involved in out- intra- or in-migration. This reflects the fact that few construction projects are large enough to last more than a year and sub-contracting is a common practice in China for achieving an earlier completion date, as required by the property developers.

Finally, female migrants are more likely to be long-term stayers. The proportion of long-term stayers among female migrants exceeds its male equivalent in all cases (see Figure 3). This exactly matches women's relative immobility and is a direct result of customs and culture. In addition, it is found that in-migrants not only had a higher rate of female participation in migration, they also had a more balanced gender ratio for long-term stayers. In fact, long-term gender ratios were always more balanced than their short-term counterparts. This is again attributable to females being more likely to be part of family migration.

In this paper, many aspects of internal migration in China have been analysed. This is attributable partly to the richness of information in the survey data and partly to the research objective, which was to provide an overall picture. The latter objective was set because of the general lack of such materials in the Western literature. As a consequence, the approach adopted had to be largely descriptive. Despite the absence of a vigorous quantitative model, the findings are interesting and the conclusions in many cases seem irrefutable.

It is undoubtedly true that many characteristics of migration in China are consistent with international experience. These include: (i) young people are more mobile than older people; (ii) females are less mobile than males; and (iii) a dominant proportion of migrants participates in non-farming activities, mainly industrial and construction work. But these consistencies may be superficial. For example, females are not so less mobile in the case of long-term migration. Also, their mobility increases in the case of rural-inward moves. As another example, although almost all migrants left agriculture, the majority remain in rural areas. In particular, family and cooperative enterprises in the countryside attract a substantial amount of migrants. On the other hand, rural-to-large-city migration was negligible in 1986. As for age structure, the in-migrants are much older than the out-migrants.

Surprising as it may seem, some of the important consensuses reached in the literature will have to be questioned in view of the findings in this paper. First, education does not promote migration, at least not significantly so. On the contrary, education acted as a deterrent to emigration and as stimulus to in-migration. Second, distance of travel displays a positive association with volume of migration in general. Everything else being equal, rural-rural migration increases as the origin and destination are further apart. Third, the informal sector absorbs only a small percentage of rural emigrants. Finally, rural-destined migrants are more likely to stay permanently and the opposite occurs for urban-destined migrants (except to large cities).

The above findings contradict what various migration models would predict. This implies that the human capital migration model and others must be modified substantially before they can be applied to China. Inappropriate application of Western frameworks to Chinese data are many, and misleading conclusions (some quite ridiculous) constantly appear in the literature. Policy recommendations drawn from such models and/or empirical applications will certainly be distorted. If implemented, the consequences could be disastrous.

To explain the seemingly counter-intuitive findings, the following hypotheses are advanced in this paper, which may be tested in future studies. First, urban reform lagged behind rural reform, which impedes the development of urban labour markets. This is the prime cause of negligible rural-city migration. It may also contribute to the negative education-migration relationship. Second, there is a significant structural unemployment problem in China, with the less educated being in huge surplus and the highly educated being in demand. This is the major factor determining the negative migration-education finding. When coupled with different resource bases (including human resources developed locally) across the vast geographic areas, this factor also helps explain the positive distance-migration relationship. Finally, there is a general shortage or absence of social services for migrants, especially in the non-rural areas. This is not only detrimental to migration in general, but more importantly adversely affects female and family migration, thus permanent moves.

Given that the Chinese government wishes to adopt a rural urbanisation strategy to solve the labour surplus problem through the development of rural enterprises and small and medium cities, the following policy implications from this study can be derived. First, in the design of and assistance to rural development, social facility provision must occupy some priority. This is necessary in order to sustain the rural-city (small and medium-sized) migration. At present, social facilities (nursing homes, boarding schools, etc) are almost absent in the countryside and unavailable to migrants in non-rural areas. This simply prevents family migration. Unless families (the elderly and the children) can be settled properly at their destinations, the majority of migrants will remain temporary residents. International as well as Chinese experience indicates that it is the short-term migrants who cause infrastructure, moral and other social problems. Furthermore, unless a large percentage of the migrants resides in non-urban areas on a permanent basis, the small and medium cities cannot prosper. In other words, the government strategy of rural urbanisation will not be realised. Second, in the design of urban reform, some mechanism should be introduced to facilitate two-way (urban-rural and rural-urban) flows. Apart from the city lifestyle, which cannot be changed, privileges attached to urban residents such as housing, social security payments, subsidised public transport, education and medical services must be removed. Migrants must be allowed to share and compete for any government-provided services or facilities. Meanwhile, such facilities or services must be gradually established in places other than big cities. This will help promote urban-rural migration and improve resource allocation (particularly human resource) between rural and urban sectors, and thus close up the urban-rural income gap. Third, property rights over land in rural areas must be clearly defined and implemented. To do so will increase female mobility and the volume of labour leaving farming. Also, it will lead to increases in agriculture productivity and farming scales. These will enable those remaining in the rural areas to exploit scale economies and raise their income levels. As is known, land fragmentation has been a serious obstacle to the growth of the foodgrain sector, which relates to the food security problem in China. Unless a substantial amount of the rural population emigrates from agriculture, the land fragmentation problem is expected to be difficult to resolve. Fourth, in order to promote rural urbanisation, locations of rural enterprises should be concentrated and planned in conjunction with the development of small and medium cities. This may take the form of providing rent assistance and publicly-funded infrastructure to those enterprises which relocate to these cities. In view of the rapid expansion of private/family (often small-scale) enterprises scattered all over the country, incentives must be initiated by the government to attract rural enterprises to the small and medium cities. A clear consequence is the promotion of rural urbanisation and family or permanent migration. What is more important is the reduction in costs of road construction, transportation and perhaps pollution control in the long run. Finally, a nationwide network must be set up by the government to provide information for migrants regarding their rights and obligations, job prospects and cultural differences in distant areas. The need for such a network is obvious given the number of migrants and their low education status.

Notes

During the revision of the paper, useful comments were received from the Editor of the journal, Professors Gordon MacAulay (University of Sydney) and Hong Li (Zhejiang Academy of Social Sciences), and Drs Zhangyue Zhou (University of Sydney) and Enjiang Cheng (University of Adelaide), for which I am very grateful.

1 Liu (Deputy Minister for Agriculture), 'Agricultural problems are of concern to the whole society' (Quanshehui duyao guanxing longyi wenti), Economic Daily, 28 June 1994, p 5; and Commentator of Economic Daily, 'Promoting organised peasant flow' (Qishi zhuahao mingong youxu liudong), Economic Daily, 21 January 1995, p 1.

2 G M Zhang, 'China facing three major population-related problems' (Zhongguo mianlin de shange zhuyao renkou wenti), China National Conditions and Power Monthly, December 1994, p. 14.

3 G Sahota, 'An economic analysis of internal migration in Brazil', Journal of Political Economy, 6 (2), 1968, pp 218-251.

4 p Cao, 'Peasant flood: how to alleviate problems arising from migration' (Minggongchao: haoniao xianxiang ruhe zhixi), Economic Daily, 19 March 1994, p 3; S K Zhao & P X Shun, 'Society at large must appropriately appreciate peasant migration' (Rang nongmin liudong hude guangfan shehui lijie), Economic Daily, 15 February 1994, p 7; and G M Zhang, 'China facing three major population-related problems'.

5 C B Nam, W J Serow & D F Sly, International Handbook on Internal Migration, New York, CT: Greenwood Press, 1990. This book contains many case studies.

6 J S Chen (State Councillor in charge of agriculture), 'Rural labour surplus and basic strategies' (Guangyu longchun laodongli shengyu he jiben duice wenti), Economic Daily, 28 January 1995, p 1; S L Chen & Z Q Yang, 'Township and village enterprises and rural industrialisation in China' (Wuguo xiangzheng qiyi yu longchun gongyihua), China National Conditions and Power Monthly, June 1994, pp 15-16; and Z Y Zhou, J L Dillon & G H Wan, 'Development of township enterprises and alleviation of the employment problem in rural China', Agricultural Economics, 6 (1), 1992, pp 201-215.

7 Y Xiao, 'New trend of employment: leaving cities for the country' (Jiuyi xingquxiang: qicheng xiaxiang), Economic Daily, 11 March 1994, p 8.

8 Zhao & Shun, 'Society at large must appropriately appreciate peasant migration'.

9 F C Li, 'Peasant migrants: vital force underlying the development of township and village enterprises' (Weichu longming: fazhan xiangzheng qiyi de shenglijun), Economic Daily, 8 February 1994, p 7.

10 D C Yu, An Analysis of the Survey Data on Labour Movement in China (Quanguo beichun laodongli qingkuang diaocha ziliao fengxi), Beijing: China Statistical Publishing House, 1990.

11 Y X Zhao, 'On mismatch between labour supply and demand in China' (Tan wuguo laodongli gongqiu shiheng), China National Conditions and Power Monthly, May 1994.

12 L A Sjaastad, 'The costs and returns of human migration', Journal of Political Economy, 70 (5), 1962, pp S80-S93; Sahota, 'An economic analysis of internal migration in Brazil'; A Schwartz, 'Interpreting the effects of distance on migration', Journal of Political Economy, 81 (5), 1973, pp 1153-1169; L Yap, 'The attraction of cities: a review of the migration literature', Journal of Development Economics, 4 (3), 1977, pp 105-127; and E M Falaris, 'The determinants of internal migration in Peru: an economic analysis, Economic Development and Cultural Change, 27 (2), 1979, pp 327-341.

13 M K Premi, 'India' in Nam, Serow & Sly, International Handbook on Internal Migration, 1990, pp 189-206.

14 D Friedlander & E Ben-Moshe, 'Israel' in Nam, Serow & Sly, International Handbook on Internal Migration, 1990, pp 225-238.

15 J Cobbe, 'Botswana' in Nam, Serow & Sly, International Handbook on Internal Migration, 1990, pp 17-30.

16 D Vergoossen, 'The Netherlands' in Nam, Serow & Sly, International Handbook on Internal Migration, 1990, pp 287-304.

17 M El-Attar, 'Egypt' in Nam, Serow & Sly, International Handbook on Internal Migration, 1990, pp 103-124.

18 Ibid.

19 Cobbe, 'Botswana'.

20 J O Oucho, 'Kenya' in Nam, Serow & Sly, International Handbook on Internal Migration, 1990, pp 275-286.

21 El-Attar, 'Egypt'.

22 A Otomo, 'Japan' in Nam, Serow & Sly, International Handbook on Internal Migration, 1990, pp 257-274.

23 R E B Lucas, Migration amongst the Botswana' Economic Journal, 95 (378), 1985, pp 358-382.

24 B H Herrick, Urban Migration and Economic Development in Chile, Cambridge, MA: MIT Press, 1965.

25 M Levy & W Wadycki, 'Education and the decision to migrate: an econometric analysis of migration in Venezuela', Econometrica, 42 (2), 1974, pp 377-388.

26 A Schwartz, 'Migration, age and education', Journal of Political Economy, 84 (4), 1976, pp 701-719.

27 J King, 'Interstate migration in Mexico', Economic Development and Cultural Change, 27 (1), 1978, pp 83-101.

28 D Perkins, 'Completing China's move to the market', Journal of Economic Perspectives, 8 (2), 1994, pp 23-46.

29 Q Z Zheng, S Y Guo, Y F Zhang & J F Wang, 'A preliminary inquiry into the problem of migration in Shanghai' (Shanghaishi liudong renkou ji youguan wenti de chubu diaocha), Population Research, March 1985, pp 2-7.

30 P K Bardhan, 'Work patterns and social differentiation: rural women of West Bengal', in H P Binswanger & M R Rosenzweig (eds), Contractual Arrangements, Employment and Wages in Rural South Asia, New Haven, CT: Yale University Press, 1984.

31 M R Rosenzweig, 'Risk, private information and the family', American Economic Review, 78 (2), 1988, pp 837-844.

32 Chen, 'Rural labour surplus and basic strategies'.

33 R B Mao (Deputy Minister for Construction), 'How to develop small and medium cities' (Xiaochengzheng jianshe ruhe zhou), Economic Daily, 21 June 1994, p 7.

34 Chinese Academy of Social Sciences, Households' Time Utilisation and Economic Behaviour (Nongfu jinji xinwei ji laodong shijian liyong diaocha zhiliaoji), Beijing: Statistical Publishing House, 1992.

35 M R Rosenzweig, 'Risk, private information and the family'.

36 Zhou, Dillon & Wan, 'Development of township enterprises'.

37 Xiao, 'New trend of employment: leaving cities for the country'.

                             TABLE I
  Age composition of intraflow and outflow migrants, 1986

Age group                  Migrants                 Percentage
BeLow 18                    1 688                      2.40
18-35                      39 322                     56.00
36-45                      22 207                     31.63
46-55                       5 860                      8.35
Over 55                     1 139                      1.62
Total                      70 216                    100.00

                         TABLE II
        Compositions of rural labour force and migrants
         (surveyed villages) by education levels, 1986

Education level              Labour force           Migrants
High school                      8.35                 8.85
Middle school                   25.16                26.37
Primary school                  38.62                39.40
Semi-illiterate
  and illiterate                27.71                25.25
Total                          100.00               100.00

                          TABLE III
   Rates of internal migration, by education levels, 1986

Education level                  Outflow              Intraflow
Tertiary and above                11.07                 19.72
High School                       11.94                 27.20
Middle school                     11.88                 26.40
Primary school                    13.19                 23.17
Semi-illiterate
  and illiterate                  16.52                 15.06
Overall                           15.16                 24.27

Labour force is used as the base in calculations because of the unavailability of population data. Inflow rates cannot be calculated because composition of in-migrants by education levels is unavailable.

                           TABLE IV
  Net migration from the surveyed villages, by education levels, 1986

                                                     Rate of net
Education level   Outflow   Inflow   Net migration    migration
                           (perons)                       (%)
Tertiary and above   32       43         -11            -3.67
High school        1 894     897         997             6.71
Middle school      5 730   2 479       3 251             7.23
Primary school    10 036   2 244       7 792            11.41
Semi-illiterate
and illiterate     9 301   2 130       7 171            14.59
Total             70 216   7 793      19 200            10.78

Labour force is used as the base in calculating the rates because of the unavailability of population data.

                           TABLE V
  Sex ratios (male:female) of long-term and short-term migrants, 1986

Term         Intraflow       Outflow       Inflow      Overall
Long-term      100:44         100:36       100:88      100:47
Short-term     100:26         100:26       100:35      100:27
Total          100:36         100:28       100:59      100:35

                          TABLE VI
      Composition of migrants by economic activities, 1986

Legend for Chart
A - Economic activities
B - Outflow--(persons)
C - Outflow--(%)
D - Intraflow--(persons)
E - Intraflow--(%)
F - Inflow--(persons)
G - Inflow--(%)
H - Overall--(persons)
I - Overall--(%)

A                   B            C           D               E
                    F            G           H               I

Agriculture       2 179         8.10           0           0.00
                    531         6.81       2 710           3.47

Industry          4 593        17.00      19 499          45.10
                  4 074        52.28      28 166          36.11

Construction      8 596        31.80       7 602          17.60
                  1 080        13.86      17 278          22.15

Transport         1 655         6.10       3 593           8.30
                    147         1.89       5 395           6.92

Service           2 249         8.40       4 583          10.60
                    557         7.15       7 389           9.47

Other             7 721        28.60       7 946          18.40
                  1 404        18.02      17 071          21.88

Total            26 993       100.00      43 223         100.00
                  7 793      100.00       78 009         100.00


                             TABLE VII
 Percentage of long-term migrants by economic activity, 1986 (%)

Economic activities    Outflow    Intraflow    Inflow    Overall
Agriculture             14.73        n/a        32.58     18.23
Industry                19.49       67.39       68.46     59.73
Construction            16.54       49.41       12.30     31.53
Transport               33.05       64.85       39.46     54.40
Service                 46.47       39.16       28.80     50.43
Other                   17.69       33.11       46.65     27.25

n/a = not available because of the absence of the relevant data.

                          TABLE VIII
Percentage of long-term migrants by origin/destination, 1986 (%)

Destination or origin       Outflow        Inflow        Overall
Villages                     14.92          48.73         26.71
Rural towns                  23.06           n/a          23.06
County towns                 26.24          89.22         30.43
Medium and small cities      20.24          87.83         23.62
Big cities                   65.72          87.80         67.36
Ove:eas                      97.13           0.00         97.13

n/a = not available because of the absence of relevant data.

                            TABLE IX
 Percentage of long-term migrants by education levels, 1986 (%)

Education levels     Outflow     Inflow     Intraflow    Overall
Tertiary and above    100.00      93.02       87.77       92.42
High school            49.74      51.39       59.94       56.14
Middle school          33.86      53.33       53.19       47.92
Primary school         18.43      59.85       60.88       46.56
Semi-illiterate
 and illiterate         8.95      43.29       51.73       30.83

                             TABLE X
   Percentage of long-term migrants by age group, 1986 (%)

Age group      Outflow      Inflow      Intraflow      Overall
Under 18        41.14        87.33        40.58         49.10
18-35           20.81        52.35        53.80         41.44
36-45           18.30        43.41        60.72         45.09
46-55           18.20        43.21        64.02         50.26
Over 55         45.30        82.66        50.77         58.09

GRAPH: FIGURE 1 Composition of migrants by education levels, 1986.

GRAPH: FIGURE 2 Age composition of migrants, 1986.

GRAPH: FIGURE 3 Percentage of long-term migrants by sex, 1986.

~~~~~~~~

By GUANG HUA WAN