Case Study Migrant Workers’ Residential Choices and China’s Urbanization Path: Evidence from Northeastern China Downloaded from ascelibrary.org by Chalmers University of Technology on 08/25/19. Copyright ASCE. For personal use only; all rights reserved. Jinqi Jiang, Ph.D. 1; Zhenhua Wang, Ph.D. 2; Wanzhen Huang 3; and Xiaohan Wei 4 Abstract: The controversy over whether the path of prioritizing large cities sprawling or the path of prioritizing small cities development is better for Chinese urbanization is unresolved. In northeastern China, extremely low fertility and increasing rural emigration are weakening the population foundation for industrialization and urbanization. Therefore, the approach to urbanization is key to a regional economic revival. Survey data from 2014 on 1,242 migrant workers in 6 cities in Liaoning Province were used to analyze residential choices and their determinants to respond to the controversy. The results found that midsized or large cities and cities near the hometown were preferred over small cities and cities far from the hometown; therefore, prioritizing large cities sprawling is more responsive to migrant workers’ choices in northeastern China. Gender, age, educational attainment, wages, employment quality, income satisfaction, length of employment, urban health insurance, history of family migration, and family land size predicted the choices. DOI: 10.1061/(ASCE)UP.19435444.0000523. © 2019 American Society of Civil Engineers. Author keywords: Residential choice; Migrant workers; Urbanization path; Northeastern China. Introduction Background and Research Goals Two distinct approaches to development are used for China’s urban expansion. Prioritizing large cities sprawling encourages expansion by forming megacity clusters, and prioritizing the development of small cities strictly controls large cities’ urban sprawl and encourages the development of small cities and, particularly, towns. In policy, the different urbanization patterns shape development strategies. China’s central government has chosen the latter pattern as its main approach to urbanization, but this policy has not garnered consistent support from research and application. Theoretical studies are mixed regarding which pattern China should adopt for urbanization. For example, economists who embrace the free market approach have argued that prioritizing large cities sprawling is necessary because China’s economy would likely benefit much more from the economic agglomeration and economies of scale of large cities and megacities (Lu 2016). On the other hand, sociologists and conservative economists have contended that prioritizing the development of small cities is better because so-called metropolitan diseases, such as air pollution, traffic 1 Associate Professor, Dept. of Agricultural and Resources Economics, College of Economics and Management, Shenyang Agricultural Univ., Shenyang, Liaoning 110866, PR China. Email: [email protected] 2 Assistant Professor, Dept. of Agricultural and Resources Economics, College of Economics and Management, Shenyang Agricultural Univ., Shenyang, Liaoning 110866, PR China (corresponding author). Email: [email protected] 3 Postgraduate Student, Dept. of Agricultural and Resources Economics, College of Economics and Management, Shenyang Agricultural Univ., Shenyang, Liaoning 110866, PR China. 4 Postgraduate Student, Dept. of Agricultural and Resources Economics, College of Economics and Management, Shenyang Agricultural Univ., Shenyang, Liaoning 110866, PR China. Note. This manuscript was submitted on September 13, 2018; approved on March 5, 2019; published online on August 19, 2019. Discussion period open until January 19, 2020; separate discussions must be submitted for individual papers. This paper is part of the Journal of Urban Planning and Development, © ASCE, ISSN 0733-9488. © ASCE congestion, and excessive population, are serious problems in many Chinese megacities, and rural and urban development is extremely unbalanced (Wen 2017). In the policy applications, huge regional socioeconomic differences suggest that China might need to adopt diverse modes of urbanization rather than a single approach. In the eastern coastal regions, such as the Beijing-Tianjin-Hebei region, the Yangtze River Delta, and the Pearl River Delta, populations are highly concentrated and the capacities of the central megacities in these regions have reached their limit. That situation has accelerated the urbanization of small satellite cities and towns to relieve the population concentration and strain placed on the services in the central cities. However, in the vast and sparsely populated midwestern and northeastern regions, it is more reasonable to promote population agglomeration in central cities to transform them into megacity clusters. Therefore, the extent to which China’s urbanization should focus on megacity clustering or small city/town development needs further study. In northeastern China (including Liaoning Province, Jilin Province, and Heilongjiang Province), the fertility rate is almost the lowest in the nation, and the populations, particularly working-age laborers, consistently emigrate from this region, which has weakened the population base for industrialization and urbanization. Therefore, the key issue for local governments in this region regarding economic revival is whether they should encourage people to live and work in the central cities, such as Shenyang, Changchun, and Herbing, to support the cities’ economic vitality and leading influences on the region’s economic growth, or encourage people to settle in the small cities to sustain the county-level economies. Although the theoretical debate on the best type of urbanization is ongoing, new urbanization is applying the small-cities approach. However, the low quality of urbanization in the midwest and northeast has not significantly changed, and the semiurbanization phenomenon has become very common among the small and midsized cities in these regions. Some scholars have proposed that the persistent problems might relate to the local governments’ lack of a clear understanding of the actual demands and residential choices of migrant workers (Sun 2015). Therefore, to effectively advance China’s urbanization, migrant workers’ residential choices 05019012-1 J. Urban Plann. Dev., 2019, 145(4): 05019012 J. Urban Plann. Dev. Downloaded from ascelibrary.org by Chalmers University of Technology on 08/25/19. Copyright ASCE. For personal use only; all rights reserved. and the factors that influence their choices should be investigated. Some of the specific questions to answer are 1. Do migrant workers want to permanently settle in cities? 2. Do migrant workers want to settle in large cities, midsized cities, or small cities? 3. Do migrant workers want to settle in cities far from their hometowns or near their hometowns? 4. What factors influence migrant workers’ residential choices? 5. How do these factors relate to migrant workers’ chosen types of residence? Literature Review Many previous studies have examined migrant workers’ residential choices and their determinants. The study results provide useful theoretical and policy insights into migrant workers’ willingness to settle permanently and the understanding of the progress of China’s urbanization. However, some aspects need further investigation. First, although migrant workers’ willingness to become urban citizens or to settle in cities has been deeply analyzed, the types of cities in which migrant workers prefer to reside has received less attention (Huang and Zhang 2013; Zhang 2011). Just three studies (Sun 2015; Xia 2010; Ye and Qian 2016) have considered it, which is insufficient for scientific policymaking. Second, a large regional socioeconomic development gap likely relates to migrant workers’ settlement behaviors across regions, implying that the urbanization approach should differ by region. However, all previous studies on migrant workers’ diverse residential activities used survey data from southeastern China. Northeastern China, a region important to China’s national economy, is different from southeastern China in that the population and the proportion of migrant workers are relatively less than in, for example, the Yangtze River Delta and Pearl River Delta regions. Migrant workers in the northeastern region mainly come from local rural areas and have relatively high agricultural production involvement (Yang et al. 2015). Thus, the northeastern region should be studied as a unique case. Because studies of the settlement choices of migrant workers in this region are rare, this study used 2014 survey data on migrant workers in six cities to investigate residential choices and their determinants. Using Liaoning Province as a typical region of northeastern China, this study responded to the question of which urbanization pattern would be best for northeastern China. The remainder of the paper is structured as follows. The theoretical framework that guided the analysis is presented next. The following section explains the sampling, data collection, variables, and regression model, and then the analytical results are reported. Conclusions and policy suggestions are provided in the last section. Theoretical Framework In contrast to some countries’ immigrants’ simultaneous transformation of occupation and identity from rural to urban, the process in China has two progressive phases: (1) the occupational change from farmer to migrant worker, and (2) the identity transition from migrant worker to citizen (Liu and Xu 2007). Migrant workers’ residential choices are mostly important to the second phase of this transition. In theory, the push-pull model (Bagne 1969; Lee 1966), the Harris–Todaro migration model (Harris and Todaro 1970), and the New Economics of Labor Migration theory (Stark and Bloom 1985) are important explanations of rural-to-urban migration. In previous studies, scholars usually integrated the findings of the © ASCE Harris–Todaro model with those of the New Economics of Labor Migration model into the push-pull model and used this modified push-pull model to analyze the residential choices of Chinese migrant workers (Ye and Qian 2016). This study followed that body of literature and adopted the modified push-pull model to frame its analysis. The pushing and pulling forces simultaneously might determine residential decisions; however, the potential power and influences of these two forces independently might vary depending on city size, location, and individual characteristics. In other words, migrant workers with similar residential choices and rural characteristics face similar pushing and pulling forces, whereas those with different residential choices and rural characteristics experience different forces. Moreover, the study considered whether the forces’ influence on migrant workers’ residential choices and their effects depended on aspects of the external environment, such as family characteristics, as well as the employment and individual characteristics of the migrant workers. According to the results of the previous studies, four types of factors might influence migrant workers’ residential choices as pushing or pulling forces. The first type was individual demographic characteristics and human capital, which indicate individual abilities and attributes. This type includes gender, age (Xia 2010; Wang et al. 2010), marital status (Bai and He 2002; Xia 2010), and educational attainment (Chen and Huang 2003; Xia 2010; Luo 2012; Sun 2015). Additionally, health status might influence migrant workers’ expectations regarding job tenure and employment status (Zhang 2006), which, in turn, might influence residential choices. The second type of factor relates to employment status and characteristics. Stable and high-quality employment is the premise for and foundation on which migrant workers settle in cities, and it is the most important type of push-pull factor. Previous studies have examined work environment (Luo 2012), job stability (Qi and Zhang 2012), employment quality (Wang 2013; Ye 2011), work experience in the city (Wang et al. 2010), and wages (Xia and Zhang 2011). Based on the studies’ findings, the present study introduced variables such as wages, job quality, employment stability, income satisfaction, work environment satisfaction, and length of employment to indicate employment status and characteristics. The third group of factors measured migrant workers’ social and labor security. The effects of social security on migrant workers’ settlement behaviors have been explored in many studies as institutional factors (Wang and Cai 2006; Ye 2011), and they even have been found to be the key factor in residential choices. However, as rural–urban integration advances, the influence of social security is weakening. However, it still deserves attention, and this study included the variables of urban medical insurances and pension insurances to examine the effects of social security. Furthermore, the extent of labor security in employment might influence migrant workers’ expectations about the stability of employment in the city, which might, in turn, influence their residential choices. Therefore, a measure of labor security was included in this study’s analysis. The fourth type of factor was family socioeconomic characteristics. A migrant worker’s migration and settlement decisions are usually based on the interests of the entire family; therefore, family factors might influence the intensity of a push-pull effect and, then, residential choices (Cai and Wang 2007; Liu and Su 2005; Wang 2006). Based on the results of previous studies, three aspects of the family were included in the analysis: family financial status, history of family migration, and family land size. 05019012-2 J. Urban Plann. Dev., 2019, 145(4): 05019012 J. Urban Plann. Dev. Data and Methods of Analysis Downloaded from ascelibrary.org by Chalmers University of Technology on 08/25/19. Copyright ASCE. For personal use only; all rights reserved. Data The data used to address the study’s objectives were derived from the migrant workers survey (MWS) in Liaoning Province conducted in January of 2014. The survey data were obtained through a two-stage stratified sampling process. In the first stage, six cities in Liaoning Province were selected as survey sites. Shenyang was representative of large cities; Fuxin and Jinzhou were typical midsized cities; and Kaiyuan, Tai’an, and Changtu were considered typical small cities based on geographical location, extent of economic development, and population size. Then, using the industrial distribution of the Monitoring Report of Migrant Workers (MRMW) (NBS 2013) and the regional distribution of migrant workers in Liaoning Province, the sample sizes by industry and city were determined. In the MRMW, the employment of migrant workers in 2013 was concentrated in the manufacturing, construction, and service industries, accounting for 31.4%, 22.2%, and 34.1%, respectively. Based on the industries’ distributions, we mainly selected a sample from the manufacturing, construction, and service industries with a ratio of 30∶25∶35. Based on the regional distribution of migrant workers in Liaoning Province, we obtained samples from each large, midsized, and small city using a 4∶2∶1 ratio. Altogether, 1,279 individuals were surveyed. From that, we excluded cases with less than three months’ employment in the city during the past year under the NBS definition of migrant workers. Then we dropped outliers and cases with missing values. The remaining 1,242 cases were analyzed. i i i i ð1Þ 24.88 72.12 Results Table 1. Sample characteristics (n ¼ 1,242) © ASCE Based on the theoretical framework, the following equation was developed to estimate the residential choices of migrant workers: X X X X PðY ¼ jÞ ¼ α þ β i Xi þ γ i Zi þ δ i Qi þ θi L i þ ε 8.29 20.69 28.66 24.96 17.39 Table 1 provides the descriptive statistics of the sample. The sample was 54.51% male and 45.49% female. Those born during or before the 1950s (age 55þ), 1960s (45–54 years), 1970s (35–44 years), 1980s (25–34 years), and 1990s (15–24 years) were 8.29%, 20.69%, 28.66%, 24.96%, and 17.39%, respectively. Categorized by generation using the year 1980 as the cutoff point, 57.65% comprised the older generation (born before 1980 and 35 years or older) and 42.35% comprised the new generation (born after 1980 and 15–34 years old). About three-quarters of the sample was married Gender Male Female Age group 55þ (born 1950s and before) 45–54 (born 1960s) 35–44 (born 1970s) 25–34 (born 1980s) 15–24 (born 1990s) Marital status Unmarried Ever married Immigration Intracity Intraprovince and intercity Interprovince City Small city (including county towns) Midsized city Large city Methods of Analysis where P = probability that residential choice Y has value j. First, the choice was indicated as the choice (desire or willingness) to settle in cities (urban areas) versus hometowns (rural areas). Then, based on the choice of city size and geographical proximity to the hometown, the choice indicator was categorized as: (1) hometown (rural areas), (2) small cities (and towns), or (3) midsized or large cities; and as (1) hometown (rural areas), (2) cities near the hometown, or (3) cities far from the hometown. To analyze the first variable, a binomial logistic regression was used for estimation, because the variable is a discrete dichotomous indicator. Because the latter two variables offered discrete choices with J (J > 2) alternatives, a multinomial logistic model, a conditional logistic model, and a mixed logistic model were available for estimation. When the characteristics of alternatives do not influence the choice or lack of data about the characteristics of the alternatives, a multinomial logistic model is often employed. However, in this study, although the characteristics of the alternatives influence the choice as well as the attributes of the individuals, a conditional logistic model or mixed logistic model should be used, because the data did not include characteristics of the alternatives of the residential choice. Eq. (1) only includes individual characteristics, and we used the multinomial logistic regression model to estimate. The symbols X, Z, Q, L in Eq. (1) were the explanatory variables representing the four types of factors (individual demographic characteristics, employment status and characteristics, social and labor security, and family socioeconomic characteristics) described previously. Table 2 lists the variables’ definitions and descriptions. Sample Characteristics Variable (75.12%). Most of the immigration was intraprovincial (83.90%): about 57.65% was intracity, and about 26.25% was intercity, intraprovincial immigration. Only about 16.1% had migrated across provincial borders. The distribution across city types was 28.82% in small cities (including county towns), 32.85% in midsized cities, and 38.33% in large cities. In terms of the representativeness of this sample, we found it to be consistent with the results of the 2013 MRMW by NBS and the sample of Liaoning Province in the 2013 China Migrants Dynamic Survey (CMDS) by the National Health Commission R.P. China. In the CMDS, 57.13% of the sample was male and 42.87% was female. In the MRMW, 53.4% comprised the older generation, and 46.6% comprised the younger generation, (these percentages were 49.3% and 50.7%, respectively, in the CMDS. Thus, the MWS sample is demographically representative of migrant workers in Liaoning Province. Proportion (%) 54.51 45.49 57.65 26.25 16.10 28.82 32.85 38.33 Migrant Workers’ Residential Choices Table 3 lists the residential choices of the respondents. Most of the respondents chose the city. Overall, 68.52% preferred the city, and only 31.48% preferred their hometowns. About 20.45% preferred small cities, and 48.07% preferred midsized or large cities. Almost 70% of those who preferred a city preferred midsized or large cities, 05019012-3 J. Urban Plann. Dev., 2019, 145(4): 05019012 J. Urban Plann. Dev. Table 2. Variable definitions and descriptions (n ¼ 1,242) Variable Definition Mean Downloaded from ascelibrary.org by Chalmers University of Technology on 08/25/19. Copyright ASCE. For personal use only; all rights reserved. Residential preference Residential Choice 1 Residential Choice 2 Residential Choice 3 Willing to permanently settle in the city (0 = hometown, 1 = city) City size (0 = hometown, 1 = small cities, 2 = large or midsized cities) Geographical proximity (0 = hometown, 1 = cities near hometown, 2 = cities far from hometown) Individual demographic characteristics Gender 1 = male, 2 = female Age Number of years Marital status 0 = unmarried, 1 = ever married Educational attainment Number of years of formal education Health status Self-assessed health (1 = low to 4 = high) Employment status and characteristics Wages Monthly wages (CNY) Value of Standard International Occupation Prestige Scale (SIOPS) Employment qualitya (1 = low to 9 = high) Income satisfaction Self-assessed satisfaction (1 = low to 5 = high) Work environment satisfaction Self-assessed satisfaction (1 = low to 5 = high) Employment stability Total number of job turnover after emigration Length of employment Number of years employed in current city Social and labor security Urban pension insurances (1 = yes, 2 = no) Urban pension insuranceb Urban medical insurances (1= yes, 2 = no) Urban health insurancec Labor contract Labor contract (1 = yes, 2 = no) Family socioeconomic characteristics Family land size The area of cultivated land (SI) Family migration Residing with family members in the city (0 = no, 1 = yes) Family financial status Logarithm of family net income in the past year Standard deviation 0.69 1.17 0.97 0.46 0.88 0.77 1.45 37.26 0.75 8.57 1.08 0.50 12.02 0.43 2.27 0.33 2,699.49 3.45 1,619.96 2.24 3.11 3.40 2.72 7.99 1.03 0.96 5.75 7.33 1.77 1.81 1.61 0.42 0.39 0.49 9,982,019.96 0.59 10.84 14,322,028.64 0.49 0.69 a Employment quality was measured with the SIOPS, constructed by Jiang et al. (2014) and Yang et al. (2015). Urban pension insurance includes urban social pension insurance and commercial pension insurance, but not the new rural social pension insurance. c Urban health insurance includes urban social health insurance and commercial health insurance, but not the new rural cooperative medical insurance. b Table 3. Residential choices (n ¼ 1,242) Size (percentage) Respondents Entire sample Within city choices Older generation (age 35 or older) (n ¼ 716) Within city choices Younger generation (age 34 or younger) (n ¼ 526) Within city choices Hometown Small city Midsized or large city Hometown City near hometown City far from hometown 31.48 — 39.66 — 20.34 — 20.45 29.85 19.13 31.71 22.24 27.92 48.07 70.15 41.20 68.29 57.41 72.08 31.48 — 39.66 — 20.34 — 40.5 59.11 38.83 64.35 42.78 53.70 28.02 40.89 21.51 35.65 36.88 46.30 suggesting that prioritizing megacities and large, sprawling cities is relatively more responsive to the choices of migrant workers in Liaoning Province. About 40.5% of the respondents preferred a city near the hometown, whereas only about 28.02% preferred cities far from their hometowns. Considering only those respondents who preferred to settle in cities, 59.11% preferred geographical closeness to and 40.89% preferred geographical distance from the hometown. The older generation (born before 1980 and 35 years or older) was compared with the younger generation (born in 1980 or later and 15–34 years old). The older generation was more likely than the younger generation to prefer and to want to return to the hometown (39.66% versus 20.34% in both cases). About 60.34% of the older respondents preferred the city, compared with 79.66% of the younger generation. The difference between the two groups was nearly 20% points. However, of the older people who preferred cities, about 68.29% preferred midsized or large cities, which, although less, was similar to the 72.08% of the younger people © ASCE Geographical proximity (percentage) who chose that option. This finding suggests that there were no large intergenerational differences regarding city size choices. As mentioned previously, geographical proximity to the hometown was preferred by a relatively large percentage of the older generation (39.66% versus 20.34%). Of the older respondents who preferred a city, 64.35% and 35.65% preferred to settle near and far from the hometown, respectively. These percentages were 53.70% and 46.30%, respectively, among the younger respondents, and the corresponding ratio was about 0.10. A comparison reveals that the differences by age group in geographical choices were larger than the differences by age group in choices for small versus midsized or large cities. Determinants of Residential Settlement Choices Cities versus Rural Areas Table 4 reports the logistic regression results of the analysis of residential choice as rural or urban. The analysis was a stepwise 05019012-4 J. Urban Plann. Dev., 2019, 145(4): 05019012 J. Urban Plann. Dev. Table 4. Hierarchical logistic regression results for rural versus urban residential choice (ref: rural residential choice) (n ¼ 1,242) Variable Model 4 Odds ratio Standard error 0.6417a −0.0349a 0.2671 0.0989a 0.2269 Individual demographic characteristics 0.3765b 0.4019a −0.0441a −0.0401a 0.26 0.3492c 0.0794b 0.0498 0.1323 0.1501 0.3390b −0.0385a −0.0901 0.044 0.174 1.404 0.962 0.914 1.045 1.190 0.1513 0.0080 0.2295 0.0321 0.2044 Wages Employment quality Income satisfaction Work environment satisfaction Employment stability Length of employment — — — — — — Employment status and characteristics −0.0003a −0.0003a b 0.0755b 0.0765 −0.0674 −0.0704 0.1639b 0.1285 0.0046 0.0053 0.0312a 0.0340a −0.0002a 0.0610a −0.085 0.1263 0.0074 0.0221b 1.000 1.063 0.919 1.135 1.007 1.022 0.0001 0.0326 0.0780 0.0834 0.0115 0.0105 Urban pension insurance (no) Urban health insurance (no) Labor contract (no) — — — Social and labor security — −0.6282c — −0.3525 — −0.0399 −0.5595c −0.3342 −0.0682 0.571 0.716 0.934 0.3326 0.3532 0.1638 — — — 0.5539 0.0587 90.88a Family socioeconomic characteristics — — — — — — 2.1665a 1.1106b 0.0993 0.1158 153.63a 179.16a −0.0089c 0.5044a 0.1045 1.1488 0.991 1.656 1.110 3.154 0.1244 192.42a 0.0051 0.1680 0.1094 1.2985 Downloaded from ascelibrary.org by Chalmers University of Technology on 08/25/19. Copyright ASCE. For personal use only; all rights reserved. Gender (female) Age Marital status (ever married) Educational attainment Health status Family land size Family migration (yes) Family financial status Constant Pseudo R2 Likelihood ratio χ2 Model 1 Model 2 Model 3 p < 0.01. p < 0.10. b p < 0.05. a c hierarchical approach using four models. The explanatory power of the models significantly increased across the models as the variables were progressively added, because the likelihood ratio chi-squared value increased with each model and was statistically significant (p < 0.01) in every model. This finding indicates that the four types of variables selected to explain variation in residential choices were appropriate. All the variables were tested together in Model 4. Female migrant workers were more likely than the males to prefer a city, and the younger respondents were more likely than the older respondents to prefer a city. Among the employment status and characteristics’ variables, wages negatively influenced (p < 0.01) residential choices, and employment quality (p < 0.10) and the number of years employed (p < 0.05) positively influenced residential choices. Based on the estimated coefficients, the respondents with relatively high wages preferred the hometown, but higher employment quality and longer employment tenure increased the likelihood of choosing to settle in the city. Results for the effects of social security and labor security found that respondents without urban pension insurance were relatively more likely to prefer settling in rural areas. Two of the family socioeconomic variables were significant. Respondents with relatively large family land size were more likely than their counterparts to prefer rural residence, and respondents residing with a family member were relatively more likely to prefer city residence. City Size Choices Table 5 presents the results of the multinomial logistic regression on city size choices. The overall model’s likelihood chi-squared test result was statistically significant (p < 0.01), indicating that the four categories of variables together significantly explained the difference in choices for city size in the sample. The significant © ASCE variables predicting a small city choice were gender, age, marital status, wages, income satisfaction, and length of employment. Urban pension insurance, urban health insurance, and history of family migration were also significant. The significant predictors of the midsized or large city choice were age, wages, employment quality, income satisfaction, length of employment, urban health insurance, family land size, and history of family migration. Comparing the estimated results for the two options (small cities versus midsized or large cities), age, wages, and history of family migration were similar in that they were significant predictors and the effects were in the same direction. Specifically, younger respondents, those with lower wages, and those residing with other family members in the city were relatively more likely to prefer cities. Second, income satisfaction, length of employment, and urban medical insurance were significant influences on both options, but the directions of the effects were opposite. These results mean that migrant workers with a high level of income satisfaction, a short term of employment, and without urban medical insurance preferred small cities, whereas those with a low level of income satisfaction, a long term of employment, and urban medical insurance preferred midsized or large cities. Third, gender, marital status, and urban pension insurance significantly influenced the choice of small cities, but they were not statistically important to the choice of midsized or large cities. On the other hand, employment quality and family land size were statistically significant predictors of choosing midsized or large cities, but not of choosing small cities. Specifically, females, respondents who were or had been married, and those with urban pension insurance were more likely than their counterparts to prefer small cities. Respondents with high employment quality and those with less family land were more likely than their counterparts to prefer midsized or large cities. 05019012-5 J. Urban Plann. Dev., 2019, 145(4): 05019012 J. Urban Plann. Dev. Table 5. Multinomial logistic regression results on city size choices (ref: hometown) (n ¼ 1,242) Small city Downloaded from ascelibrary.org by Chalmers University of Technology on 08/25/19. Copyright ASCE. For personal use only; all rights reserved. Variable Individual demographic characteristics Gender (female) Age Marital status (married) Educational attainment Health status Employment status and characteristics Wages Employment quality Income satisfaction Work environment satisfaction Employment stability Length of employment Social and labor security Urban pension insurance (no) Urban health insurance (no) Labor contract (no) Family socioeconomic characteristics Family land size Family migration (yes) Family financial status Constant Pseudo R2 Likelihood ratio χ2 Midsized or large city Coefficient Relative risk ratio Standard error Coefficient Relative risk ratio Standard error 0.5016a −0.0332b 0.6702c 0.0591 −0.1222 1.651 0.967 1.955 1.061 0.885 0.2010 0.0107 0.3288 0.0452 0.3060 0.2012 −0.0434b −0.3406 0.0367 0.2613 1.223 0.957 0.711 1.037 1.299 0.1615 0.0088 0.2425 0.0343 0.2122 −0.0003b 0.0452 0.2778c 0.1591 −0.0072 −0.0416c 1.000 1.046 1.320 1.172 0.993 0.959 0.0001 0.0444 0.1093 0.1151 0.0223 0.0163 −0.0002b 0.0705c −0.2335b 0.0960 0.0100 0.0445b 1.000 1.073 0.792 1.101 1.010 1.045 0.0001 0.0345 0.0839 0.0889 0.0121 0.0113 −1.1336b 0.9329c −0.2445 0.322 2.542 0.783 0.3868 0.4276 0.2131 −0.0664 −1.1028b 0.0442 0.936 0.332 1.045 0.3892 0.4016 0.1783 0.0004 0.8984b 0.1682 −2.8354 1.000 2.456 1.183 0.059 0.0060 0.2372 0.1497 1.7977 −0.0159b 0.3060a 0.0892 1.9155 0.1557 402.22b 0.984 1.358 1.093 6.790 0.0061 0.1809 0.1170 1.3861 p < 0.10. p < 0.01. c p < 0.05. a b Geographical Proximity Choices The results of the multinomial logistic regression on the residential choice of the hometown and near versus far from the hometown are listed in Table 6. The likelihood chi-squared test was statistically significant at p < 0.01, indicating that the model had strong explanatory power. Gender, age, wages, and length of employment were statistically significant, and the coefficients had the same sign for both options (near versus far from the hometown). Specifically, females, younger respondents, those with longer employment, and those with relatively low wages were more likely than their counterparts to prefer cities, regardless of the cities’ geographical proximity to the hometown. Second, educational attainment, employment quality, and history of family migration were only significant to the choice of cities near the hometown. Specifically, respondents with relatively high educational attainment, high employment quality, and those residing with family members were more likely to prefer cities near the hometown. Third, health status and satisfaction with income, the family’s land size, and family financial status had significant influences on the choice of cities far from the hometown. Specifically, those with relatively better health, dissatisfaction with income, less family land, and relatively better family finances were relatively more likely to choose cities far from the hometown. Conclusions and Policy Implications Although the Chinese government has clearly chosen to take the small city (particularly town) development approach and to limit large cities’ growth as its main focus for new urbanization, controversy continues over whether the large cities or small cities approach is better. Due to obvious regional socioeconomic differences in China, the best path for urbanization should be explored. © ASCE In northeastern China, low fertility rates and the net emigration of working-age populations have weakened the population foundations for urbanization at the regional level. In this context, this study was premised on the residential choices of migrant workers, because they are the major source of population increases in China’s urbanization process. Survey data collected in Liaoning Province in 2014 were used to analyze aspects of residential choice and their determinants to answer the question of which approach should be used to direct urbanization in northeastern China. The main findings were as follows: 1. The respondents significantly preferred permanent residence in cities as opposed to rural areas (68.52% versus 31.48%). 2. Midsized or large cities and cities near the hometown were preferred over small cities and cities far from the hometown. Among the respondents who preferred to settle in cities, 29.85% chose small cities and 70.15% chose midsized or large cities. About 59.11% of the respondents preferred to settle in cities near their hometowns, and 40.89% preferred to settle in the cities far from their hometowns. 3. Gender, age, wages, length of employment, and a history of family migration were similarly influential on residential choices. Their effects were almost invariant on size and proximity of cities. 4. Marital status, high income satisfaction, urban pension insurances, and lack of urban health insurances predicted the choice to settle in small cities, whereas high quality employment, low income satisfaction, urban health insurance, and less family land predicted the choice to settle in midsized or large cities. 5. Geographical proximity was important to residential choice in the sample. Educational attainment and quality of employment were positively related to the choice of cities near the hometown, and relatively good health, low income satisfaction, good 05019012-6 J. Urban Plann. Dev., 2019, 145(4): 05019012 J. Urban Plann. Dev. Table 6. Multinomial logistic regression results on geographical proximity choices (ref: hometown) (n ¼ 1,242) Cities near the hometown Downloaded from ascelibrary.org by Chalmers University of Technology on 08/25/19. Copyright ASCE. For personal use only; all rights reserved. Variable Individual demographic characteristics Gender (female) Age Marital status (married) Educational attainment Health status Employment status and characteristics Wages Employment quality Income satisfaction Work environment satisfaction Employment stability Length of employment Social and labor security Urban pension insurance (no) Urban health insurance (no) Labor contract (no) Family socioeconomic characteristics Family land size Family migration (yes) Family financial status Constant Pseudo R2 Likelihood ratio χ2 Cities far from the hometown Coefficient Relative risk ratio Standard error Coefficient Relative risk ratio Standard error 0.3060a −0.0419b 0.2035 0.0812c −0.0328 1.358 0.959 1.226 1.085 0.968 0.168 0.009 0.267 0.037 0.239 0.3136a −0.0376b −0.3659 −0.0103 0.3905a 1.368 0.963 0.694 0.990 1.478 0.178 0.010 0.260 0.038 0.228 −0.0004b 0.0648a 0.0417 0.102 −0.0027 0.0233c 1.000 1.067 1.043 1.107 0.997 1.024 0.000 0.036 0.088 0.094 0.015 0.012 −0.0001c 0.0571 −0.2372b 0.145 0.012 0.0224a 1.000 1.059 0.789 1.156 1.012 1.023 0.000 0.039 0.091 0.098 0.012 0.013 −0.5041 −0.1675 −0.1452 0.604 0.846 0.865 0.355 0.376 0.183 −0.5773 −0.601 −0.0021 0.561 0.548 0.998 0.398 0.411 0.196 −0.003 1.0043b −0.0036 1.0886 0.997 2.730 0.996 2.970 0.006 0.197 0.123 1.459 0.982 0.874 1.321 0.584 0.007 0.197 0.130 1.537 −0.0179b −0.1345 0.2783c −0.5379 0.1246 336.02b p < 0.10. p < 0.01. c p < 0.05. a b family financial status, and less family land significantly predicted the choice to settle in cities far from the hometown. Considered together, this study’s findings strongly suggest that giving priority to large, sprawling cities is the best urbanization approach in northeastern China, because it is relatively more responsive to the choices of the region’s migrant workers. Although the population of migrant workers in northeastern China is much smaller than in the southeastern regions, and intraprovincial migration, as the major migration pattern in northeastern China, is very different from that in southern regions, where interprovincial migration is the major pattern, the findings of this study lead to the same conclusions as the findings by Sun (2015) and Ye and Qian (2016), who identified that giving priority to large, sprawling cities was the best urbanization approach for southeastern China. Moreover, when the findings for city size choices are combined with the findings for proximity choices, it is clear that the municipal governments in northeastern China must not solely focus on the urbanization of their central megacities (e.g., Shenyang City, Dalian City, Changchun City, and Harbin City), but must also attend to their midsized cities. The authors recommend that provincial governments consider selecting two or three proximate midsized and large cities to develop into city clusters and then encourage migrant workers to permanently settle there as citizens. To improve policies that promote permanent settlement of migrant workers in midsized and large cities, we suggest that the municipal governments in northeastern China focus first on industrialization within urbanization and then drive urban development through the development of industries with inclusive employment creation (Lu et al. 2012). During the study, we deeply investigated the reasons that small cities and county towns did not attract migrant workers for permanent settlement. We concluded that the main reason was a scarcity of vigorous leading industries in the second- and third-tier sectors © ASCE and/or a lack of agglomeration of leading industries, which created insufficient employment opportunities for migrant workers. Thus, the first step should be to transform and upgrade the industrial structures of the midsized and large cities to fit the region’s actual resources, such as labor force and land. Assuming that effective industrial competitiveness could be ensured, excessive capitalization of the local secondary and tertiary industries should be prevented, and the industrial structure should be prevented from deviating from local resource endowment and comparative advantage. The second step should be that the abilities to absorb mediumskill and low-skill workers from the upstream and downstream industries should be considered when selecting leading industries for the region, because immigration from nearby agricultural sectors is expected to be the main fuel of urban population growth in northeastern China. Then, regarding employment, governments in northeastern China must also improve migrant workers’ urban employment skills and enhance their urban employment quality. The social security system should focus on integrating the urban and rural public healthcare systems to increase the attractiveness of settling in cities. In addition, because cultivated land resources are relatively abundant in northeastern China, decreasing the pull force of the land might promote migrant workers’ choices for settling in cities. Possible counterforces might be applied in the form of sizable rewards, subsidies, or urban housing compensations for migrant worker families that permanently leave their rural land, followed by programs to help them acclimate to urban lifestyles. Acknowledgments The authors are grateful for financial support from the National Natural Science Foundation of China (71303161, 71503173), the Social Science Foundation of Liaoning Province (L16BGL038), 05019012-7 J. Urban Plann. Dev., 2019, 145(4): 05019012 J. Urban Plann. Dev. the Program for Liaoning Excellent Talents in University (WJQ2015026), and the Youth Project of the Philosophy and Social Science Research, Ministry of Education of China (13YJC790057). Downloaded from ascelibrary.org by Chalmers University of Technology on 08/25/19. Copyright ASCE. For personal use only; all rights reserved. References Bagne, D. J. 1969. Principles of demography. New York: Wiley. Bai, N., and Y. He. 2002. “Return or emigration? A research on the return migration of migrants in Anhui province and Sichuan province.” [In Chinese.] Sociol. Res. 17 (3): 64–78. Cai, H., and J. Wang. 2007. “The research on migrant workers’ permanent migration intention.” [In Chinese.] Sociol. Res. 22 (6): 86–113. Chen, X., and Z. Huang. 2003. “Factor influencing labor migration to cities in economic developed areas; the case of Zhejiang Province.” [In Chinese.] Chin. Rural Econ. 18 (5): 33–39. Harris, J. R., and M. P. Todaro. 1970. “Migration, unemployment and development: A two-sector analysis.” Am. Econ. Rev. 60 (1): 126–142. Huang, Q., and G. Zhang. 2013. “An investigation of the willingness to settle in small and medium cities of new generation migrant workers.” [In Chinese.] World Surv. Res. 21 (7): 29–33. https://doi.org/10.13778/j .cnki.11-3705/c.2013.07.015. Jiang, J., D. Qi, and G. Zhang. 2014. “Prior job and present job quality and job mobility of migrant workers.” [In Chinese.] South China Rural Area 21 (7): 29–33. https://doi.org/10.15879/j.cnki.cn44-1099/f.2014 .0047. Lee, E. S. 1966. “A theory of migration.” Demography 3 (1): 47–57. https:// doi.org/10.2307/2060063. Liu, C., and J. Xu. 2007. “A study on the second generation peasant workers and their citizenization.” [In Chinese.] China Popul. Resour. Environ. 17 (1): 6–10. https://doi.org/10.1016/S1872-583X(07)60002-3. Liu, H., and Q. Su. 2005. “An empirical analysis of rural women labor’s intention of settling in cities; The case of Jiangsu Province.” [In Chinese.] Chin. Rural Econ. 21 (9): 42–47. Lu, M. 2016. Great country and big city. [In Chinese.] Shanghai: Shanghai People’s Publishing House. Lu, M., H. Gao, and S. Hiroshi. 2012. “On urban size and inclusive employment.” [In Chinese.] China Soc. Sci. 33 (10): 47–66. Luo, E. 2012. “The impact of employ ability on the willingness to live in urban areas of migrant workers: A case of Shanghai City.” [In Chinese.] Urban Prob. 31 (7): 96–102. https://doi.org/10.13239/j.bjsshkxy.cswt .2012.07.009. NBS (National Bureau of Statistics of China). 2013. “The monitoring report of Chinese migrant workers in 2013.” [In Chinese.] Accessed May 12, 2014. http://www.stats.gov.cn/tjsj/zxfb/201405/t20140512_551585 .html. © ASCE Qi, D., and G. Zhang. 2012. “Occupational mobility of migrant workers and the urban settlement willingness.” [In Chinese.] J. Agrotech. Econ. 31 (4): 44–51. https://doi.org/10.13246/j.cnki.jae.2012.04.006. Stark, O., and D. E. Bloom. 1985. “The new economics of labor migration.” Am. Econ. Rev. 75 (2): 173–178. Sun, Z. 2015. “Settlement intention of migrant workers in big cities and development path of China’s new-type urbanization.” [In Chinese.] Popul. Res. 39 (5): 72–86. Wang, D., and F. Cai. 2006. “Migration and poverty alleviation in China.” [In Chinese.] China Labor Econ. 3 (3): 46–70. Wang, G. 2006. “The macro model of rural-urban migration based on the maximization of farmers’ net revenue.” [In Chinese.] Chin. J. Popul. Sci. 20 (2): 48–57. Wang, G., G. Chen, and X. Wei. 2010. “Study on the influencing factors of rural-urban migrant workers’ willingness to be urban citizens in Shanghai.” [In Chinese.] Popul. Dev. 16 (2): 2–11. Wang, Y. 2013. “Settlement intention of rural migrants in Chinese cities: Findings from a twelve-city migrant survey.” [In Chinese.] Popul. Res. 37 (4): 19–32. Wen, T. 2017. “Rural construction is way to avoid economics crisis.” [In Chinese.] Dev. Small Cities Towns 45 (3): 6–10. Xia, X., and H. Zhang. 2011. “Study on citizenry willing of new generation migrant workers and its influencing factors.” [In Chinese.] Northwest Popul. J. 32 (2): 43–51. https://doi.org/10.15884/j.cnki.issn.1007 -0672.2011.02.003. Xia, Y. 2010. “Migrant workers’ willingness of residential place and its factors: A survey from Wenzhou City.” [In Chinese.] Chin. Rural Econ. 26 (3): 35–44. Yang, X., M. Li, and J. Jiang et al. 2015. Migrant workers’ employment and the urbanization in Liaoning Province [In Chinese.] Beijing: Economic Daily. Ye, J., and W. Qian. 2016. “Migrant workers’ urbanization willing in different scale cities and path-selected for new-type urbanization.” [In Chinese.] Zhejiang Soc. Sci. 32 (5): 64–74. https://doi.org/10 .14167/j.zjss.2016.05.008. Ye, P. 2011. “Residential preferences of migrant workers: An analysis of the empirical survey data from seven provinces/districts.” [In Chinese.] Society 31 (2): 153–169. https://doi.org//10.15992/j.cnki.31-1123/c .2011.02.001. Zhang, Y. 2011. “Migrant workers’ willing of hukou register and policy choice of China urbanization.” [In Chinese.] Chin. J. Popul. Sci. 35 (2): 14–26. Zhang, Z. 2006. “Settlement in city or return-migration: A life-cycle analysis of migrant workers’ decision.” [In Chinese.] Chin. Rural Econ. 22 (7): 21–29. 05019012-8 J. Urban Plann. Dev., 2019, 145(4): 05019012 J. Urban Plann. Dev.