I. Introduction 介紹

 

自1978以來，中華人民共和國已開始了一個巨大的經濟和社會轉型，向市場經濟轉型。社會科學家一直在研究其動因這一經濟變革對中國社會的不同方面，不平等，工資的確定、2等行為的遷移和人口的3。在這篇文章中我們將討論研究如何通過某些方面的經濟改革和政府的政策在上世紀80年代末在受影響的人口分布。特別是，我們專注于經濟條件如何，外國資本投資，以及農村企業存在影響省際人口遷移在改革時期。Since 1978, the People's Republic of China has embarked on a great economic and social transformation-the transition to a market- oriented economy. Social scientists have been studying the conse- quences this economic change is having on different aspects of Chinese society, including inequality,' wage determination,2 and demographic behaviors such as migration.3 In this article we extend this discussion by examining how certain aspects of economic reform and government policies affected population distribution in the late 1980s. In particular, we focus on how economic conditions, foreign capital investment, and the presence of rural enterprises influence interprovincial migration in the reform era. 

 

II. Migration Policy and Economic Reforms 移民政策和經濟改革

 

移民和人口分布一直是主要的控制問題，為政策制定者和社會科學家在中國。多年來，中國是其嚴格控制農村人口向城市遷移的眾所周知的，票面的遷移到大城市。4的官方政策，宣布在1980有三個要素：（1）嚴格限制大城市規模，合理發展中等城市（2），和（3）鼓勵小城市和城鎮的發展。然而5情況發生了變化，1978，隨著中國農村經濟改革開始，提高農業效率和產生的剩余勞動力。6還多，在上世紀80年代末城市改革削弱了戶籍登記時系統明顯，遷移到城市地區（特別是盟友臨時遷移）比以前容易多了。這些變化？1997由芝加哥大學。保留所有權利。0013-0079 / 97 / 4502-0004 o01.00美元發生之時，中國政府官方政策方面以人口遷移和城市化基本保持不變。為解決由此產生的農村剩余勞動力問題，中國政府資助的項目，促進了副活動。其中最系統的這一政策措施是激勵農村企業時代建立和具體（鄉鎮企業）由村、鎮、或個人。政府提供的技術前對這些新建立的鄉鎮企業專業技術和稅收優惠，7和作為一個結果，他們已經大幅增加。Migration and population redistribution have always been major con- cerns for policy makers and social scientists in China. For years, China was well known for its tight control of rural-to-urban migration, par- ticularly migration to big cities.4 The official policy that was declared in 1980 had three elements: (1) strictly limiting the size of big cities, (2) properly developing medium-sized cities, and (3) encouraging the growth of small cities and towns.5 The situation changed, however, with the beginning of economic reforms in rural China in 1978, which improved agricultural efficiency and generated surplus labor.6 Further- more, urban reforms in the late 1980s weakened the household regis- tration system significantly and made migration to urban areas (especi- ally temporary migration) much easier than before. These changes ? 1997 by The University of Chicago. All rights reserved. 0013-0079/97/4502-0004$O01.00 occurred at a time when the official Chinese government policy regard- ing migration and urbanization remained basically unchanged. To address the resultant rural surplus labor problem, the Chinese government sponsored programs that promoted sideline activities. One of the most systematic and concrete steps of this policy was to encour- age establishment of rural enterprises (xiang zhen qi ye) run by village, town, or private individuals. The government provides technical ex- pertise and tax privileges for these newly established rural enterprises,7 and as a result, they have increased dramatically. In 1979, there were 1.48 million rural enterprises that employed 29 million rural peasants. By 1993, however, there were 23.2 million rural enterprises that em- ployed about 112.3 million workers.8 Indirect evidence suggests that rural enterprises might indeed reduce migration from rural to urban areas.9 Nevertheless, there is little direct empirical evidence on the impact that rural enterprises have on the propensity to migrate. In other settings, rural development schemes have generally met with mixed results.' The post-1978 economic reforms have exacerbated regional in- equalities in population and income distribution. Regions with poorly developed markets exhibit slower growth rates relative to marketized areas." In 1987, Guangdong Province received $1.12 billion of foreign investment and loans, about 13.8% of the national total.'2 In contrast, one of the poorest provinces-Guizhou Province-received no foreign investment in 1987. Since foreign investments generate employment and salaries well above the average, those provinces that have a high level of foreign investment stand to gain migrants from other prov- inces.'3 Urban economic reforms not only have brought about changes in the production process, but these reforms have also enhanced a work unit's (dan wei) autonomy in hiring personnel. All individuals in China are recorded in a household registration system. To move from one province to another and still obtain entitlement and services, one needs permission from both province of origin and province of destination. The economic reforms have eroded the household registration system. A work unit now can hire anyone who has the know-how, regardless of registration status. Furthermore, the system of rationing food has been replaced by the cash market. As a result, the household registra- tion system is no longer effective as a way to control migration. Much contemporary research on interprovincial migration in China has described aggregate patterns of the flow of persons, but it suffers from underspecification of the mechanisms of interprovincial redistribution.A 4 The study of interprovincial migration has a particular advantage in that it is not subject to the boundary reclassification that compromises the study of rural-urban migration.'" Using the China 1990 Population Census information on reasons for interprovincial mi-  #p#分頁標題#e#

Zai Liang and Michael J. White 323 gration, Ellen Efron Pimentel and Liu Jinyun show that a very signifi- cant proportion of interprovincial migrants can be accounted for by those who reported the reason for migration as doing contract labor or business (such persons are also heavily present in the "floating population").16 The 10.8 million interprovincial migrants (1985-90) represent about 1% of the Chinese population, compared with the cor- responding U.S. 5-year interstate migration of about 10% during 1975-80, with about half of the U.S. movers changing regions.17 An- nual interregional migration in the United States was less than 2% in the mid-1980s.'8 In this article, we use unique micro-level data with linked prov- ince-level characteristics to examine two major questions: (1) To what extent does the market transition affect migration patterns in China? and (2) To what extent has the government policy of establishing rural enterprises reduced migration from rural areas? 

 

III. Data and Methods 數據分析

 

Our data are taken from a 10% random sample of the China 1988 2/1,000 Fertility and Birth Control Survey, conducted by China's State Family Planning Commission in July-August 1988.19 This is a nation- ally representative sample that contains' information on interprovincial migration for all members in the household surveyed. The response rate was 99.95%, with 29 refusals and 159 individuals not reached.20 One unique feature of the data set is its coverage of temporary migrants.21 Persons were counted in the province of survey, as long as they were in that province the night before the survey even though they were not officially registered in the province at the time of survey. We define an interprovincial temporary migrant necessarily broadly as someone who does not have a registration card at the province of destination. The survey asked, "When was the last time you made an interpro- vincial move?" In addition, respondents were asked reasons for last interprovincial migration, province of origin, and the province of household registration. Moreover, further information on rural or ur- ban nature of origin and destination provinces is also provided in the survey, for example, whether individuals are residing or registered in the city, town, or rural part of a province. To evaluate the impact of provincial-level characteristics on the patterns of migration, we augmented these individual data with provin- cial-level information from different sources.22 The rural enterprise variable can tap the extent to which well-developed rural enterprises stem migration out of the province. It is measured by the proportion of persons employed by either township or village enterprises divided by the total rural population in each province. Foreign investment in a province measures the degree of market transition that is taking place. We measure it as the sum of foreign loans, foreign direct invest- ment, and other types of foreign investment. Per capita industrial out- put (value in yuan of industrial output in 1982 divided by the 1982 population) is intended to measure the economic base of a province, which affects the rate of migration.23 The idea of using population size is derived from the classical gravity model of migration, which argues that the larger the population size in a region, the more likely people are to migrate.24 We also include province literacy rate (for all persons in the province 12 years old and older) as an indicator of social well- being. Our population universe includes persons who are between 15 and 64 years old at the time of survey. Our dependent variable indicates whether or not individuals migrated in the 5-year interval before the survey, 1983-88. This interval captures the later years of the 1980s during which urban reform accelerated. This is also in line with most census migration studies, which use a 5-year interval.25 We recognize that the province is a rather large unit to choose for the migration-defining boundary. What is important and consistent with previous thinking on the subject, however, is that there be appre- ciable heterogeneity between provinces, heterogeneity to which poten- tial migrants respond. Therefore, our statistical analysis can pick up the determinants of long-distance migration, even though it cannot capture more local movement. Overall, we have 25 provinces plus three municipalities (Beijing, Tianjin, and Shanghai), which are treated as provinces in our data. We exclude Tibet and combine the newly formed Hainan Province with Guangdong Province. We divide our analysis into two parts. First we predict out- migration from the province of origin, using the universe of labor- market-age residents, as defined above. Second, we predict destination choice for the pool of interprovincial migrants, using characteristics of the potential destinations. Out-Migration Methods Since our model of interprovincial out-migration uses both individual and provincial-level characteristics, we attach characteristics of each province of origin for migrants and characteristics of current province for nonmigrants. A logistic regression specification is used: ln[pil(1 - pi)] = 'Xi. (1) Here p, represents the probability of making an interprovincial move for person i between 1983 and 1988, and Xi represents individual- and province-level characteristics. For persons who make multiple inter- provincial moves in the 1983-88 period, our data allow us to measure only the most recent move.  #p#分頁標題#e#

Zai Liang and Michael J. White 325 We include certain individual-level variables that previous re- search has shown to be important: age, education, and marital status. However, we are especially interested in how migration probabilities are affected by the presence of rural enterprises and foreign invest- ment. Since the existence of rural enterprises is supposed to affect only the migration of people in the rural areas, we run the model for the rural origin sample and total sample separately. Following Griffith Feeney and Wang Feng, we also adopted a strict definition of rural and urban population.26 Urban population refers to people who are engaged in nonagricultural employment, reside in an urban district, and are registered in urban households. Individuals who fail to meet any one of the three criteria will be defined as rural. Previous research suggests that there is an interaction effect be- tween contextual-level variables (if we treat province as a social con- text) and individual-level variables.27 Following a similar logic, we test for interaction between the existence of rural enterprises and the edu- cational level of individuals. In-Migration Analysis The aims of the in-migration (destination selection) analysis are two- fold. First, we try to identify what kinds of characteristics of destina- tions (provinces) attract migrants. Second, we examine what types of migrants are attracted by provinces with different characteristics. To accomplish these goals, we selected only interprovincial mi- grants between the ages of 15 and 64 and estimated several conditional logit models. Technical discussions of the conditional logit model can be found in works by G. S. Maddala and by Saul D. Hoffman and Greg J. Duncan.28 Compared to more familiar models such as the multinomial logit model, which considers the impact of individual char- acteristics on alternatives, the conditional logit model focuses on the impact of characteristics of alternatives themselves. In our case, the alternatives are potential destination provinces. Formally, the condi- tional logit model may be written as PY = exp(P'Xij)/k exp(P'Xik), (2) where k indexes the possible choices of provinces and here runs from 1 to 28, and P refers to the parameter vector to be estimated. Here the probability of individual i choosing destination province j (P1j) depends on the characteristics of the province. We estimated sev- eral mixed models where Pij also depends on individual character- istics.29 To estimate the conditional logit model, we need to prepare the data in an appropriate format of person-province records. We replicate 28 records for each observed migrant, attaching province characteris- tics (potential destinations) to each person's record. A dummy variable (coded I if the province is selected and 0 otherwise) forms the depen- dent outcome in the in-migration analysis. 

 

IV. Results 結果

 

Description of Interprovincial Migration, 1983-88 Table 1 presents estimates of different measures of migration, calcu- lated from the survey. We examine first province destination propen- sity (PDP). This measure indicates how important each region is in terms of attracting migrants. Table 1 also shows that several provinces, including Beijing, Liaoning, Shanghai, Zhejiang, Hubei, Guangdong, and Shannxi, received more than 4% of China's interprovincial mi- grants and also exhibited positive net migration rates. These rates in- clude two of the three municipalities and some eastern provinces that also experienced rapid development. Out-migration rates over 0.03% are found in Heilongjiang, Jiangsu, Hebei, Henan, and Sichuan. Note that high out-migration rates are not limited to the poorest and most remote provinces. Furthermore, some provinces, such as Hebei (adja- cent to Beijing), experience relatively high in- and out-migration rates, an occurrence that has been found in other cases of internal mi- gration.30 Beijing not only has the highest value (15.0) for province destina- tion propensity but also has the highest in-migration rate (0.123) and net migration rate (0.097). The 5-year in-migration rate calculated using our data is 0.123, which is higher than the 0.071 figure from the 1990 China census.31 This is because the 2/1,000 survey captures a large number of temporary migrants. Table 2 compares the distribution of basic demographic and social characteristics for migrants and nonmigrants. Consistent with previous research findings, migrants are younger, with more than half of all migrants in the 20-29 age group. By contrast, only 6% of migrants are 50-64 years old, whereas for nonmigrants the figure is about 17%. There are more male than female migrants, and migrants are more likely than nonmigrants to be single, again reflecting a commonly found differential. In addition, there is a clear contrast in educational distri- bution between migrants and nonmigrants: a higher proportion of mi- grants have a college, senior high, and junior high school education. Only 7.6% of the interprovincial migrants are illiterate, whereas 20% of nonmigrants are illiterate. The occupations of interprovincial migrants are very diverse, ranging from agricultural workers to cadres and intellectuals. Nonmi- grants have a much larger proportion (59%) who farm or do other extractive work, whereas the corresponding proportion for migrants is only 24.4%. Given the small proportion of urban-to-rural migration, the migrants who do agriculture-related work probably represent rural  #p#分頁標題#e#

Zai Liang and Michael J. White 327 TABLE 1 MEASURES OF INTERPROVINCIAL MIGRATION, 1983-88 Region IMR OMR NMR PDP North: Beijing .1229 .0258 .0971 15.0 Tianjin .0294 .0126 .0167 3.1 Hebei .0231 .0567 - .0336 4.2 Shanxi .0134 .0155 - .0021 2.0 Inner Mongolia .0207 .0153 .0054 2.7 Northeast: Liaoning .0267 .0242 .0025 4.6 Jilin .0123 .0206 - .0083 1.8 Heilongjiang .0220 .0317 - .0096 3.2 East: Shanghai .0609 .0218 .0391 7.4 Jiangsu .0157 .0306 - .0148 3.2 Zhejiang .0270 .0254 .0016 4.6 Anhui .0142 .0219 - .0078 2.4 Fujian .0170 .0152 .0019 2.4 Jiangxi .0067 .0186 - .0119 .9 Shandong .0192 .0267 - .0075 3.9 Central and south: Henan .0162 .0325 - .0163 3.1 Hubei .0278 .0212 .0066 4.9 Hunan .0169 .0082 .0087 2.9 Guangdong .0238 .0215 .0023 4.1 Guangxi .0098 .0186 - .0088 1.5 Southwest: Sichuan .0198 .0346 - .0147 4.2 Guizhou .0131 .0131 0 1.9 Yunnan .0139 .0130 .0008 2.0 Northwest: Shannxi .0280 .0202 .0078 4.2 Gansu .0224 .0204 .0021 3.3 Qinghai .0234 .0048 .0186 1.8 Ningxia .0296 .0086 .0211 2.3 Xinjiang .0240 .0280 - .0040 2.5 SOURCE.-Ten percent sample of 1988 China 2/1,000 Fertility and Birth Control Survey. NoTE.-Migration information for Tibet is not available. IMR: in-migration rate; OMR: out-migration rate; NMR: net migration rate; PDP: province destination propensity = (in- terprovincial migrants choosing province)/(all interprovincial migrants). peasants migrating to a different province to engage in agriculture, forestry, and fishery activities. There is indeed some anecdotal evi- dence that in certain rural areas, where nonfarm employment is ample, peasants take nonfarm employment and hire peasants from other parts of the country to take care of their farms. Another large proportion of migrants work in manufacturing or construction, which have become booming industries. Cadres, intellectuals, and students are heavily rep-  

TABLE 2 BASIC CHARACTERISTICS FOR MIGRANTS AND NONMIGRANTS Migrants (%) Nonmigrants (%) Age: 15-19 14.1 17.9 20-29 54.7 27.8 30-39 17.9 23.3 40-49 7.3 14.4 50-64 6.1 16.6 Sex: Male 66.2 51.6 Marital status: Married 61.4 68.4 Divorced .3 0.5 Widowed .6 2.6 Not married 37.8 28.4 Education: College 13.1 2.0 Senior high school 22.2 12.6 Junior high school 35.5 29.8 Elementary school 18.4 29.0 No schooling, literate 3.2 6.6 No schooling, illiterate 7.6 20.0 Employment: Farmer in agriculture, for- estry, fishery 24.4 59.2 Farmer in industry, con- struction 15.0 5.5 Farmer in commerce, ser- vice industry 7.3 2.5 Mining workers in cities/ towns 11.2 8.5 Farming, forestry, animal husbandry (nonfarm) 1.0 1.0 Business, commerce, and service 5.0 4.0 Cadres, intellectuals 14.0 5.8 Student 9.4 5.9 Waiting for work assignment 1.9 1.0 Other 10.9 6.6 Reason for migration: Contract labor or circular migrants 24.9 Marriage 10.4 Worker transfer 7.3 Study 8.2 Newly assigned 4.2 Demobilization 18.0 Business 2.6 Other 24.3 N 3,271 137,761 SOURCE.-Ten percent sample of 1988 China 2/1,000 Fertility and Birth Control Survey. 328  

Zai Liang and Michael J. White 329 resented in the interprovincial migration streams. It is surprising that the unemployment rates ("waiting for work assignment") are not very different between migrants and nonmigrants. Of course, it is possible that migrants who did not get jobs simply returned home. Among reasons stated for migration, contract labor and circular migration account for nearly one-quarter of all interprovincial moves.32 Other reasons for migration cited with relative frequency are marriage (10%) and demobilization (18.0%). Further breakdown by sex (not shown in table 2) indicates that 28.7% of the female migrants gave marriage as the reason for migration, in contrast to only 0.7% of male migrants. The high proportion of the residual category, "other," sug- gests that the survey instrument was perhaps not well designed to capture some aspects of interprovincial migration.33 Results from the Interprovincial Out-Migration Analysis We provide some basic descriptive statistics in table 3. Table 4 shows the results from logistic regressions predicting provincial out-migration for persons from rural origins, sequentially building the model from individual-level to contextual characteristics. The dependent variable TABLE 3 MEAN AND STANDARD DEVIATION OF INDEPENDENT VARIABLES Mean SD Province-level characteristics: Population (in millions) 35.82 23.48 Per capita industrial output (yuan in hundreds) 7.88 10.85 Total literacy rate (%) 68.41 10.36 Foreign capital investment (US$100 million) 1.95 4.98 Rural enterprise (%) 6.73 7.26 N 28 Individual-level characteristics: Age (reference = 50-64): 15-19 .17 .38 20-29 .28 .45 30-39 .23 .42 40-49 .14 .35 Male .52 .50 Married .68 .47 Education (reference = no formal education): College .02 .15 Senior high school .13 .33 Junior high school .30 .46 Elementary school .29 .45 N 141,032 SOURCE.-Ten percent sample of 1988 China 2/1,000 Fertility and Birth Control Survey.  #p#分頁標題#e#

TABLE 4 LOGISTIC REGRESSION RESULTS PREDICTING PROVINCE OUT-MIGRATION: CHINA, 1983-88 (Rural Origin Sample) Model I Model II Model III Individual characteristics: Age (reference = 50-64): 15-19 1.213* 1.388* 1.413* (.153) (.155) (.155) 20-29 1.804* 1.941* 1.962* (.133) (.134) (.134) 30-39 .755* .834* .845* (.141) (.141) (.141) 40-49 .074 .143 .157 (.171) (.171) (.171) Education (reference = no formal education): College 2.339* 2.225* 2.451* (.172) (.175) (.186) Senior high school .314* .140 .355* (.107) (.110) (.126) Junior high school .369* .186* .400* (.082) (.085) (.105) Elementary school .0346 - .137 - .148 (.085) (.087) (.087) Male - .008 .040 .040 (.057) (.057) (.057) Married .077 .108 .108 (.073) (.074) (.074) Province-level characteristics: Population ... .013* .013* (.001) (.001) Per capita industrial output . . -.036* -.033* (.010) (.010) Total literacy rate . . . .028* .027* (.004) (.004) Foreign investment ... - .006 -.006 (.007) (.007) Rural enterprises ? ? ? .004* .005* (.001) (.001) Interaction term: (Rural enterprise) x (high education) ... . - .004* (.001) Intercept - 5.484* - 7.944* - 8.036* (.142) (.300) (.302) N 86,456 86,456 86,456 SOURCE.-Ten percent sample of 1988 China 2/1,000 Fertility and Birth Control Survey. NOTE.-Standard errors are in parentheses. * P < .05. 330  

Zai Liang and Michael J. White 331 equals 1 for an interprovincial move between 1983 and 1988 and 0 otherwise. Model I includes only individual-level sociodemographic charac- teristics. Migration propensity peaks among persons ages 20-29 and declines appreciably with age. This is consistent with what we show in table 2-that migrants tend to be concentrated in the 20-29 age group; it is noteworthy that a strong differential persists in the face of controls. Education has a significant positive effect on interprovincial migration, with college graduates having the strongest propensity to make an interprovincial move. It has been argued that people with high educational attainment tend to have the kinds of skills needed for jobs at the destination, that they participate in the national labor mar- ket, and have more information about opportunities in different prov- inces. Moreover, returns for more educated migrants also tend to be substantially larger than for less educated people. When other individ- ual characteristics are controlled, there is not so much difference be- tween married and nonmarried individuals, or between males and fe- males. In Model II, we add characteristics of province of origin. The effects of individual-level characteristics remain unchanged. However, provincial population size (of sending province) has a significant posi- tive impact on migration propensity. This is also consistent with the finding of Judith Banister.34 A high level of economic development (as measured by per capita industrial output) reduces the rate of migration out of the province. Of major interest is the impact of rural enterprises. It is surprising that people from provinces with a high proportion of peasants em- ployed in rural enterprises are more likely to make interprovincial moves. A 1% increase in the proportion employed in rural enterprises leads to the increase of log odds of migration by 4/1,000. This gives some initial indication that the government policy of preventing peas- ants from migrating by encouraging the establishment of rural enter- prises does not seem to work the way it was intended. On the contrary, migration is more likely to happen in places with a large proportion of peasants employed in rural enterprises, holding provincial social and economic development levels constant.35 Foreign capital investment in the province does not have a significant impact on interprovincial migration for the rural origin sample, probably because most of the foreign investment is located in urban China. In Model III of table 4, we added one interaction term between an individual's postelementary education and the presence of rural enterprises to see whether highly educated peasants are more likely to move or not. We include in this category those with college, senior high, and junior high school education. It is interesting that the results in Model III show that there is a significant negative interaction term- people with more education are less likely to make an interprovincial move if the origin province has well-developed rural enterprises. This is consistent with anecdotal evidence that rural enterprises are particu- larly attractive to local people with education. We also estimated almost the same model for the total (rural plus urban) sample (we omitted only the rural enterprise measure). All the individual-level characteristics tend to have effects similar to those found for the rural sample, whereas the effects of provincial-level char- acteristics change somewhat. Individuals are significantly less likely to move from provinces with more foreign capital investment. For the total sample, once we control for population and foreign capital investment, the measures of economic and social development are no longer important predictors. (These results are available from us on request.) Results from the In-Migration Analysis Table 5 displays the results from conditional logit models predicting province in-migration using both individual- and province-level charac- teristics. This approach is a "mixed" model.36 The sample consists of persons ages 15-64 who are interprovincial migrants. The major pur- pose in this part of the analysis is to examine what provincial-level characteristics are most attractive to migrants and how they operate for different types of migrants, such as temporary versus permanent and well-educated versus less educated. Again, we focus our discus- sion on the rural origin sample, but we did carry out companion estima- tions for the total sample. Model I, which includes only province-level measures, indicates that rural migrants are more likely to move to provinces with high literacy rates, high levels of foreign investment, and well-developed rural enterprises. Net of these factors, rural mi- grants are less likely to choose industrial and populous provinces. Turning to the interaction terms, we began to see some intriguing and complex patterns of migration. In Model II, we retain only interac- tion terms that are statistically significant. Provincial-level characteris- tics work differently for temporary versus permanent migrants, and for migrants with different levels of education. We first look at the kinds of places to which temporary migrants are likely to move. Com- pared to permanent migrants, temporary migrants are more likely to move to provinces with high levels of foreign investment and well- developed rural enterprises. Perhaps temporary migrants also perceive that rural enterprises or joint venture companies are less concerned about registration status. The high level of activity in these regions creates demands for construction and ancillary services, jobs that tem- porary migrants may seek. Once these characteristics are controlled, temporary migrants are less likely to move to populous provinces and provinces with highly developed economies (measured by per capita  #p#分頁標題#e#

Zai Liang and Michael J. White 333 TABLE 5 RESULTS FROM CONDITIONAL LOGIT MODEL PREDICTING PROVINCE IN-MIGRATION: CHINA, 1983-88 (Rural Origin Sample) Model I Model II Province-level characteristics: Population - .013* - .003 (.002) (.002) Per capita industrial output -.015* -.027* (.004) (.009) Total literacy rate .015* .016* (.004) (.004) Foreign investment .039* -.006 (.005) (.009) Rural enterprise .006* .003* (.001) (.001) Interaction terms: (Rural enterprise) x (temporary migrant) . . .006* (.001) (Industrial output) x (temporary migrant) - .020* (.008) (Foreign investment) x (temporary migrant) . . . .096* (.011) (Population) x (temporary migrant) . . -.030* (.002) (Rural enterprise) x (high education) . . . -.002* (.001) (Industrial output) x (high education) . .032* (.008) Intercept - 4.338* - 4.307* (.253) (.255) x2 625.07 949.94 N 39,368 39,368 SOURCE.-Ten percent sample of 1988 China 2/1,000 Fertility and Birth Control Survey. NoTE.-Standard errors are in parentheses. * P < .05. industrial productivity). Well-educated migrants are more likely to move to provinces with highly developed industries but less likely to move to provinces with well-developed rural enterprises.37 Taken together, we find that the impacts of rural enterprises are particularly interesting. Provinces with well-developed rural enter- prises become attractive places for individuals with less education and with temporary migrant status. Recall our results from the province out-migration model (table 4), where we find that people from prov- inces with well-developed rural enterprises are more likely to depart compared with people from provinces with less developed rural enter- prises. Moreover, better-educated individuals are more likely to stay in provinces with well-developed rural enterprises. Such differential effects across characteristics (themselves correlated) suggest a pattern of sorting of migration streams across locations. The possibility of underemployment in these provinces might provide some impetus for migration out of the provinces. Similar patterns and mechanisms have been observed in Mexico.38 

 

V. Conclusion and Discussion 總結和討論

 

Using nationally representative data froNotesm the 1988 China 2/1,000 Fer- tility and Birth Control Survey, this article examines the impact of economic reform and rural enterprises on the different forms of mobil- ity in China during 1983-88. Central to our endeavor is the measure- ment of the impact of economic development, foreign capital invest- ment, and rural enterprises on interprovincial migration. In general, our findings concerning individual-level characteristics are consistent with previous research in China and in other developing countries. Unique in our findings for China is the effect of province- level characteristics. Individuals are more likely to move out of prov- inces with a large population and a lower level of economic develop- ment. This is in line with classic arguments about migration and economic development. China's transition to a market economy adds a new dimension to the issue of migration: with so much foreign capital investment pouring into China, what role does foreign investment play in the process of internal migration? Research in other countries has stressed the impor- tance of foreign capital investment for international migration, but less attention has been paid to its impact on internal migration and urban- ization in developing countries. Our results indicate that foreign invest- ment, in fact, slightly reduces migration out of provinces receiving investment. In the meantime, migrants (temporary migrants especially) are more likely to choose provinces with high levels of foreign capital investment as destinations. Foreign investment, measured here, ac- counts not only for direct job opportunities but also for a host of secondary opportunities created by economic growth. Our findings provide some new insight on the role of China's rural enterprises in the process of Chinese urbanization and a mixed mes- sage about the role of rural enterprises (as well as small towns) in reducing migration to other provinces. We find that individuals from provinces with well-developed rural enterprises have a higher propen- sity to depart than others. However, more educated peasants are more likely to stay if provinces have well-developed rural enterprises. This implies that without rural enterprises, there could have been more of a "brain drain" from the province. These somewhat surprising results are in line with recent advances in the literature on migration and development. Douglas S. Massey argues that development initiatives in a rural area tend to increase migration rather than decrease it be- cause development by nature is disruptive. Development is also likely  #p#分頁標題#e#

Zai Liang and Michael J. White 335 to increase the efficiency of agricultural production, which releases even more workers from farm labor.39 Also, results from Zhejiang Province confirm our conclusion.40 It is possible that the effect of rural enterprises on migration differs in different stages. In the initial stage of rural enterprise development, rural enterprises are able to recruit local peasants and therefore reduce the chance of out-migration. As rural enterprises mature and promote an increase in the level of mecha- nization in agriculture, more peasants are released from farm work and out-migration increases. However, a test of such a complex hypothesis needs a longitudinal design that records the detailed history of rural enterprise development. Looking at the results from in-migration models, temporary and less educated migrants are more likely to move into provinces with well-developed rural enterprises. We argue that because of the success of the development of rural enterprises, these provinces are becoming more attractive for temporary interprovincial migrants and more mi- grants are moving in. At the same time, researchers began to notice the difficulties facing China's rural enterprises in the late 1980s and a decline in the rate of employment generation.41 This could create a pool of underemployed and a potential pool for out-migration, parallel to what happened in the Mexican Border Industrialization Program in the 1960s.42 This counterintuitive result for the effect of rural enterprises is worth some further thought. From our data we cannot discern the employment history of the migrants and comparison populations. The migrants may have worked in these rural enterprises. The prevalence of rural enterprises may proxy more access to information about poten- tial migration destinations because places with rural enterprises are likely to be in constant contact with the outside world. We may also be witnessing a multiphasic response, in which some persons are re- cruited to these rural enterprises, and the recruitment process itself frees other members to move to more distant locations. We suggest that further research be undertaken to identify the exact mechanism through which migration took place. From a policy perspective, the issue of rural enterprises is closely related to the Chinese government's overall policies concerning urban- ization and development. As we noted earlier, the Chinese government followed a policy of development of rural enterprises in the small towns, a policy aimed at stemming the perceived shift toward over- urbanization. This policy was supported by many well-known scholars in China,43 but others have begun to point out the limitations of treating rural enterprises as the only alternative." In this article we point out another unintended consequence and limitation of developing rural enterprises in small towns and rural areas-their limited ability to re- duce migration. We want to caution the reader that our results are based only on interprovincial migration. Future research could exam- ine the impact of rural enterprises on the pattern of intraprovincial migration and thus help complete the picture. As far as we know, such research has not been done. Perhaps the Chinese government has already begun to realize the limitations of depending on a small-town and rural enterprise strategy. In May 1993, there was a joint decision made by China's Department of Labor, Department of Rural Enterprise Management, and the State Council to undertake an experiment in eight provinces on an alterna- tive rural surplus labor absorption strategy. The program started in July 1993. The major elements of the program included developing labor-intensive secondary and tertiary industries in rural areas, speed- ing up the process of small-town development, and providing training for rural surplus laborers. The most significant aspect of the proposed plan is to open up the labor markets in the designated towns and cities.45 This is the most comprehensive and ambitious plan regarding rural surplus labor absorption and migration that the Chinese govern- ment has ever initiated. Its success or failure will probably determine the future of rural surplus labor absorption in other provinces of China. We are, like other students of China's urbanization and development, eager to see how this new experiment on such a large scale proceeds in the age of market transition. #p#分頁標題#e#

 

Notes

 

 * Earlier versions of this article were presented at the 1993 annual meet- ings of the American Sociological Association, the 1994 annual meetings of the Population Association of America, and the Demography Workshop at Brown University. We are indebted to participants at these meetings for their comments and suggestions. We thank the China State Family Planning Com- mission for permission to use the data and the East-West Center, especially Wang Feng, for kindly providing us with the data for this research. We are particularly grateful to Sidney Goldstein and Alice Goldstein for very helpful discussions and comments on previous versions of this article. Comments and suggestions from several anonymous reviewers and the editor helped improve the article. Financial support from the Rockefeller Foundation to Zai Liang is gratefully acknowledged. 1. Victor Nee, "Social Inequalities in Reforming State Socialism: Be- tween Redistribution and Markets in China," American Sociological Review 56 (1991): 267-82; and Andrew G. Walder, "Economic Reform and Income Inequality in Tianjin, 1976-1986," in Chinese Society on the Eve of Tianan- men, ed. Deborah Davis and Ezra Vogel, Harvard Contemporary China Series (Cambridge, Mass.: Council on East Asian Studies, Harvard University, 1990), pp. 135-56. 2. Yusheng Peng, "Wage Determinants in Rural and Urban Industrial China," American Sociological Review 57 (1992): 198-213. 3. Alice Goldstein, Sidney Goldstein, and Shenyang Guo, "Temporary Migrants in Shanghai Households, 1984," Demography 28 (1991): 275-91. 4. John D. Kasarda and Edward M. Crenshaw, "Third World Urbaniza- Zai Liang and Michael J. White 337 tion: Dimensions, Theories, and Determinants," Annual Review of Sociology 17 (1991): 467-501. 5. Tian Fang and Zhang Dongliang, Zhongguo renko qianyi xintan (An inquiry to migration in China) (Beijing: China Knowledge, 1989). 6. D. Gale Johnson, "Economic Reforms in the People's Republic of China," and Justin Yifu Lin, "The Household Responsibility System in China's Agricultural Reform: A Theoretical and Empirical Study," Economic Development and Cultural Change 36, suppl. (1988): S225-S245, S199-S224. 7. Depending on the types of rural enterprises, the tax privileges include tax exemption and tax reduction. The government also provides loans to rural enterprises; see Zhongguo xiangzhen qiye nianjian bianji weiyuanhui, Zhong- guo xiangzhen qiye nianjian (China rural enterprise yearbook) (Beijing: China Agricultural Publishing House, 1990). In addition, both provincial and central governments set up departments or bureaus to foster rural enterprises. 8. Zhonguo guojia tongjiju (China State Statistical Bureau), Zhongguo tongji nianjian (Statistical yearbook of China) (Beijing: China Statistics Press, 1993). 9. See Grant Blank and William L. Parish, "Rural Industry and Non-farm Employment: Comparative Perspectives," in Chinese Urban Reform: What Model Now? ed. R. Yin-Wang Kwok et al. (New York: Sharpe, 1990); and Sidney Goldstein, Urbanization in China: New Insights from the 1982 Census, Papers of the East-West Population Institute, no. 93 (Honolulu: East-West Center, 1985). 10. Nazli Baydar, Michael J. White, Charles Simkins, and Ozer Babakol, "Effects of Agricultural Development Policies on Migration in Peninsular Ma- laysia," Demography 27 (1990): 97-109; Francisco L. Rivera-Batiz, "Can Bor- der Industrialization Be a Substitute for Immigration?" American Economic Review 76 (1986): 263-73; and Sally E. Findley, "Rural Development Pro- grams: Planned versus Actual Migration Outcomes," Brown University Re- print Ser. 82-09 (1981). 11. Nee (n. 1 above). 12. Zhonguo guojia tongjiju (China State Statistical Bureau), Zhongguo tongji nianjian (Statistical yearbook of China) (Beijing: China Statistics Press, 1987). 13. There is an alternative argument that suggests that, for the case of international migration, foreign investment actually increases emigration. Saskia Sassen draws on examples of emigration from South Korea, Hong Kong, and the Philippines to the United States to argue this point. See Saskia Sassen, The Mobility of Labor and Capital: A Study in International Invest- ment and Labor Flow (New York: Cambridge University Press, 1988). 14. Li Menghua, "Zhongguo renkou qianyi de liuxiang" (The direction of population migration in China), Renkou yanjiu 18 (1994): 48-51. 15. Sidney Goldstein, "Urbanization in China: Effects of Migration and Reclassification," Population and Development Review 16 (1990): 673-701. 16. Ellen Efron Pimentel and Liu Jinyun, "Causes of Interprovincial Mi- gration in China" (paper presented at the International Seminar on the 1990 Chinese Census, Beijing, October 19-23, 1992). 17. Michael J. White and Peter R. Mueser, "Implications of Boundary Choice for the Measurement of Residential Mobility," Demography 25 (1988): 443-59. 18. Daniel T. Lichter and Gordon F. De Jong, "The United States," in International Handbook on Internal Migration, ed. Charles B. Nam, William J. Serow, and David F. Sly (New York: Greenwood, 1990). 19. See Guojiajihua shengyu weiyuanhui (State Family Planning Commis- sion), Zhongguo shengyujieyu choyang diaocha (China fertility and birth con- trol survey) (Beijing: State Family Planning Commission, 1988); and William Lavely, "China Unveils Its Monumental Two per Thousand Fertility Survey," Asian and Pacific Population Forum 5 (1991): 89-92, 116. 20. Li Honggui, "Overview of China Fertility and Birth Control Survey" (paper presented at the International Seminar on China Fertility and Birth Control Survey, Beijing, 1991). 21. Both the China 1987 1/100 Population Survey and the China 1990 Population Census miss many temporary migrants. The former uses 6 months as the minimum duration for nonregistered individuals to be counted as resi- dent at destination; the latter uses 1 year as the standard. Our data, however, are able to capture these missed temporary migrants. 22. The following variables are taken from Dudley Poston and Baochang Gu: population size, per capita industrial productivity, and total literacy rate- all measured in 1982. See Dudley Poston and Baochang Gu, "Socioeconomic Development, Family Planning, and Fertility in China," Demography 24 (1987): 531-51. Foreign capital investment by province in 1985 was taken from the China Statistical Yearbook (1985) and the first year complete information on foreign capital investment by province is 1985. Information regarding rural enterprises in 1985 is also obtained from China Statistical Yearbook (1985). See also Zhongguo guojia tongjiju (China State Statistical Bureau), Zhongguo tongli nianjian (Statistical yearbook of China), (Beijing: China Statistics Press, 1985, 1987, 1990); Zhongguo geshengshi zizhiqu zhixiashi lishi ziliao (Histori- cal statistics of province, autonomous regions, and municipalities, 1949-1989) (Beijing: China Statistics Press, 1990). 23. Michael J. White, "Three Models of Net Metropolitan Migration," Review of Regional Studies 7 (1977): 20-44. 24. William A. V. Clark, Human Migration (Beverly Hills, Calif.: Sage, 1986). 25. Lichter and De Jong. 26. Griffith Feeney and Wang Feng, "Parity Progression and Birth Inter- vals in China: The Influence of Policy in Hastening Fertility Decline," Popula- tion and Development Review 19 (1993): 61-101. 27. See Sally E. Findley, "An Interactive Contextual Model of Migration in Ilocos Norte, the Philippines," Demography 24 (1987): 163-90; and Douglas S. Massey, Rafael Alarcon, Jorge Durand, and Humberto Gonzalez, Return to Aztlan: The Social Process ofInternational Migration from Western Mexico (Berkeley and Los Angeles: University of California Press, 1987). 28. G. S. Maddala, Limited Dependent and Qualitative Variables in Econometrics (New York: Cambridge University Press, 1983); Saul D. Hoff- man and Greg J. Duncan, "Multinomial and Conditional Logit Discrete-Choice Models in Demography," Demography 25 (1988): 415-27. 29. Hoffman and Duncan. 30. Peter R. Mueser and Michael J. White, "Explaining the Association between Rates of In-Migration and Out-Migration," Papers of the Regional Science Association 67 (1989): 121-34. 31. Population Census Office of Beijing Municipality, Tabulation on the 1990 Population Census ofBeijing Municipality (computer tabulation) (Beijing: China Statistics Press, 1992). 32. "Circular migrant" (commonly known as the "floating population") refers to someone who lives in a locale less than 6 months and has not changed permanent residence. 33. One reviewer raised the issue of whether migrants who reported "other" for reason of migration differ significantly from people who gave  #p#分頁標題#e#

Zai Liang and Michael J. White 339 concrete reasons for migration. A one-way analysis of variance showed that migrants in this residual category are older than the remaining migrants, but their educational level and year of migration are about the same. To further ensure that inclusion of these "other" migrants does not bias our results, we estimated all statistical models (both out-migration and in-migration models) without this group of migrants. The results are almost identical. 34. Judith Banister, "Provincial Population Size and Interprovincial Mi- gration," in The Population of Modern China, ed. Dudley L. Poston, Jr., and David Yaukey (New York: Plenum, 1992). 35. Consistent with previous researchers (see White [n. 23 above]), we argue that per capita industrial output is a measure of the economic base of a province, which determines the employment opportunities. Following the suggestion of one reviewer, we reestimated all statistical models using per capita income instead of per capita industrial output in a province. We used three measures of income: per capita rural income, per capita urban income, and per capita total income (a weighted average of per capita rural and urban income). The results from out-migration models are almost identical to those in this article. The results from in-migration models changed only slightly. Overall, we are convinced that using per capita income in the model does not change our substantive conclusions. The details of all these alternative models are available from us on request. 36. Hoffman and Duncan. 37. We also tried another measure of rural enterprise by using the total number of rural enterprises in each province and adding an interaction term between the number of rural enterprises in the province and the education- level dummy variable; the results are very similar. 38. Rivera-Batiz (n. 10 above). 39. Douglas S. Massey, "Economic Development and International Mi- gration in Comparative Perspective," Population and Development Review 14 (1988): 383-414. 40. Xiushi Yang, "The New Economic Policy and Spatial Mobility in China" (paper presented at the annual meetings of the Population Association of America, Miami, May 5-7, 1994). 41. See John Bauer, "Demographic Change and Asian Labor Markets," Population and Development Review 16 (1990): 615-45; and Michael Douglas, "Surplus Labor, Rural-Urban Relations and Urban Policy in China" (paper presented at the International Conference on China's Urbanization, East-West Population Institute, East-West Center, Honolulu, January 15-27, 1989). 42. Rivera-Batiz. 43. Fei Hsiao Tung, Small Towns in China: Functions, Problems, and Prospects (Beijing: New World, 1986). 44. Goldstein, Urbanization in China (n. 9 above). 45. Zhang Jianjun, "Basheng liewei nongcun laodongli jiuye shidian" (Eight provinces became designated sites for experiment on employment for rural surplus labor), Renmin rebao haiwaiban (People's daily), overseas ed. (May 6, 1993). #p#分頁標題#e#