Urbanization with Chinese Characteristics DOMESTIC MIGRATION and URBAN GROWTH in CONTEMPORARY CHINA

Urbanization with Chinese Characteristics DOMESTIC MIGRATION AND URBAN GROWTH IN CONTEMPORARY CHINA

Nicholas Eberstadt and Alex Coblin, with Cecilia Joy-Pérez and Kangyu Mark Wang OCTOBER 2019

AMERICAN ENTERPRISE INSTITUTE Urbanization with Chinese Characteristics DOMESTIC MIGRATION AND URBAN GROWTH IN CONTEMPORARY CHINA

Nicholas Eberstadt and Alex Coblin, with Cecilia Joy-Pérez and Kangyu Mark Wang

OCTOBER 2019

AMERICAN ENTERPRISE INSTITUTE © 2019 by the American Enterprise Institute. All rights reserved. The American Enterprise Institute (AEI) is a nonpartisan, nonprofit, 501(c)(3) educational organization and does not take institutional positions on any issues. The views expressed here are those of the author(s). Contents

Executive Summary...... 1

I. Introduction...... 3

II. Background...... 6

III. Data on Population and Urbanization in Contemporary China: Availability and Reliability ...... 9

IV. The Geography and Demography of Migration in China 2010...... 27

V. Chinese Urbanization in International Perspective...... 40

VI. The Arithmetic (and Politics) of Migration and Urbanization with Chinese Characteristics...... 51

VII. Educational Attainment: The Urban-Rural Divide, Migration, and Development...... 66

VIII. Concluding Observations...... 79

Notes ...... 84

Appendix A. China’s Population Policy U-Turn: A Tougher New Battle Ahead...... 91

Appendix B. Changes in the Classification of Urban and Rural Population in China’s National Statistics...... 102

Appendix C. The Political Economy of China’s Evolving Hukou Regime...... 112

Appendix D. Explaining the Migration Patterns of Tertiary-Educated Migrants: Higher Education Overcapacity and the Economics of Labor...... 120

About the Authors...... 137

iii

Executive Summary

n this report, we use data from China’s 2010 census a national labor market and that a large portion of I to examine China’s urbanization and migration the working-age population in cities is made up of patterns from an international perspective. First, we migrant labor. present an overview of the historical use of the city Despite the rise of populous and densely popu- as a means for control and administration across lated cities in China, a look at China’s urbanization Chinese history. Second, we examine the compara- in an international perspective suggests that China bility and reliability of official demographic data and is under-urbanized. Subsequently, we examine what find that definitions of urbanization change dramat- China’s human resource profile in urban and rural ically across China’s official censuses. Delving into areas portends for its ongoing urbanization drive. the contours of urbanization and internal migration Finally, we offer some concluding observations on during the post-Mao era, we find that, even within how modern technological advances may affect the constraints of the hukou system, China truly has migration and urbanization in China.

I. Introduction

n the post–World War II era, the most basic demo- As a simple matter of arithmetic, this worldwide Igraphic rhythms of human existence have been fun- postwar rise of cities was fueled by migration—mainly damentally recast across the planet. For the global domestic migration. In fact, migration was the source population today, patterns of death, birth, and resi- of the overwhelming bulk of postwar urban popu- dence are radically different from those ever previ- lation growth if we include both migrants and their ously known. descendants. Migration and urbanization are insepa- With respect to death, worldwide mortality rates rable in the modern era: There would have been no have plunged and continue to fall to ever lower lev- great rise of cities without migration, and without els. At this writing, humanity’s overall life expectancy the rise of cities, postwar migration would have been at birth appears to be higher than the highest level vastly more limited in scope and scale. achieved in any country before World War II.1 With The conjoint trends of migration and urbanization, respect to birth, a worldwide revolution in childbear- furthermore, were integral to the economic developing patterns halved fertility levels over the half cen- ment processes that almost quadrupled planetary per tury 1960/65 and 2010/15, with no end to these global capita output between 1950 and 2010 alone.4 Migra- declines in sight.2 And with respect to residence, tion and urbanization have been part and parcel of human beings are now, for the first time in our history, the great structural transformation of economies more likely to live in urban than rural communities. globally. This complex dynamic has generated sub- The emergence of urban life as the modal form of stantial and sustained increases in productivity and human existence is an epochal development, but it took living standards as the overall structure of production place in just six decades. In 1950, by UN Population and employment shifted from agriculture to indus- Division (UNPD) estimates, the share of the world’s try and then to the services and knowledge-intensive population in urban settlements was under 30 percent; undertakings.5 by 2010, it was more than half (51 percent).3 Humans’ There is reason to believe that urbanization is not sudden and rapid urbanization was due to an explosive only a handmaiden of modern economic growth but worldwide growth in cities. Between 1950 and 2010, the an engine of it as well. It is not just that cities trans- globe’s urban population nearly quintupled, surging form daily life, occupations, expectations, and ways from about three-quarters of a billion to over 3.5 billion of thinking. The urban environment affords potential and to nearly four billion by 2015. improvements in productivity through economies of Even more extraordinarily rapid than the overall scale based on population density; better “inputs” of growth of urban areas was the pace of growth for big health, education, and other facets of human capi- cities (or, as the UNPD calls them, “urban agglomera- tal that are characteristic of the “urban edge” over tions”)—locales with populations of 300,000 or more. rural localities; a setting inherently more conducive And the fastest-growing urban agglomerations were to knowledge production and rapid dissemination of the largest in size: the settlements with populations innovation and ideas; and other economic advantages of a million or more. Today, even as global population intrinsic to the very nature of the modern city itself.6 growth continues to decelerate from its postwar apo- In the broadest of terms, the postwar story of gee in the 1960s, these megacities’ populations (espe- migration and urbanization in China reads much like cially in low-income areas) continue to burgeon. that of the rest of the world. For one thing, China

3 URBANIZATION WITH CHINESE CHARACTERISTICS

experienced truly explosive postwar urban growth. By over a million and none with five million residents. official Chinese census figures, the proportion of the Sixty years later, in 2010, China had 85 cities with over population living in urbanized settlements rose from one million people, 14 with over five million, and four just 13 percent in 1953 to nearly 50 percent in 2010.7 with over 10 million—far more huge cities than any (We discuss the provenance and reliability of Chinese other nation. By 2015, according to UNPD estimates, urbanization data later.) China had more than a hundred million-plus cities, According to its official population sample survey, with a total of nearly 380 million inhabitants—seven China crossed the urban-majority threshold in 2011, times as many as resided in its 22 million-plus cit- and by 2015 China was 56 percent urban.8 By year-end ies back in 1985, just three decades earlier, and over 2017, that ratio was reportedly approaching 59 per- 25 times as many as in 1950.11 cent.9 All around the world, urban growth has been But it is not just the quantitative—say, China’s especially explosive in the biggest cities—so, too, in sheer size or the spectacular speed of its structural China. transformation and economic growth over the past Just as elsewhere in the world, domestic migra- four decades—that distinguishes the processes of tion has played a major role in augmenting China’s urbanization and development in modern China from urban numbers. And just as elsewhere, urban growth those of other postwar societies. Scarcely less dra- in China not only accompanied overall national socio- matic are the qualitative differences—most notably, economic development but also played a major role the extraordinary reach of the Chinese state in shap- in catalyzing material advance. China’s government ing and managing domestic migration and urban regards urbanization as so central to the country’s growth. (No less should be understood from Beijing’s prospects for continued rapid economic growth that claim to have authority for administering an urban- Beijing is implementing an “urbanization drive,” as ization drive.) Vestiges of China’s totalitarian past, as it is informally known: a long-term policy for accel- well as aspects of its highly authoritarian present, are erating the rise of cities and the increase of urban deeply impressed into the basic features of migration population.10 and urbanization in China today—and with powerful, Yet for all these similarities, China’s patterns of pervasive consequences that virtually no would-be urbanization have also been distinctive—profoundly migrant can hope to avoid. different from those of almost all other contemporary Chinese leadership since Mao’s death has striven societies in important and far-reaching respects. We to evince sustained rapid economic growth—and with examine some of these differences and their implica- remarkable success. But it has been no less vigilant in tions in this report. demanding, and enforcing, social control to maintain Naturally, the scale of things Chinese makes Chi- political stability and unchallenged one-party rule. na’s urbanization process different from that of any Of its manifold instruments for social control, one of other country almost by definition. In absolute terms, the most distinctive is surely the hukou system, the to no surprise, contemporary China’s urban growth national registration system for all mainland Chinese has been unrivaled. Between 1985 and 2015, China’s nationals. urban population increased by over half a billion— We will have more to say about the hukou system. accounting for not only more than a fourth of all For now, three observations about it will suffice. worldwide urban growth over those decades but also First, in its current form, hukou identification well over twice as much urban population growth as paperwork determines whether any person in China the next largest contributor, India. is within or outside their officially designated place of Likewise, the proliferation of megacities in China residence. It provides municipal officials with grounds is a phenomenon unmatched in absolute terms by any to deny social and other services to men and women other country in the modern world. In 1950, by UNPD without locally approved hukou. From a juridical estimates, China had just eight cities with populations and police perspective, each and every out-of-hukou

4 NICHOLAS EBERSTADT AND ALEX COBLIN

migrant’s existence in an “unauthorized” county, implications for China’s economic, social, and politi- town, or city is wholly provisional and contingent. cal outlook. Second, this Chinese migration-control arrange- The analysis undertaken here is quantitative and ment is unlike anything in the modern non-Sinitic analytical, not prescriptive. That is, we focus on world. Strictly speaking, China’s current hukou sys- understanding trends and phenomena revealed by tem is not entirely unique—Vietnam and North Korea statistics and data and refrain from recommenda- maintain similar social control mechanisms—but the tions. Much has been written in more qualitative similarity is because Hanoi and Pyongyang copied terms about modern China’s urbanization and migra- their versions from the original Maoist prototype. tory processes. There is also a large literature assay- Third, Chinese authorities evidently have no ing the economics of migration and urbanization in intention of ending this peculiar institution any time modern-day China and still another corpus of stud- soon—quite the contrary. Talk of “reforming” (i.e., ies offering policy analysis and advice on migration, dismantling) the hukou system has been entertained urbanization, and development in China. in university, research institute, and even official pol- In this report, we instead attempt to extract mean- icy circles in China for the better part of two decades. ingful information about Chinese urbanization and But the urbanization drive Beijing is now administer- migration from demographic and socioeconomic sta- ing envisions no end to the hukou-ization of the Chi- tistics, including official Chinese population data. We nese population. If that policy works out as planned, pay special attention to China’s 2010 census, the most China will enter the coming decade with some scores recent detailed official compendium of data concern- of millions of migrants finally granted local residence ing urbanization and migration. (Although a more rights in their new abodes—but with hundreds of mil- recent 2015 mini-census was conducted, its results lions of migrants still designated as out-of-hukou res- were not readily accessible at this writing.)12 idents working and living in unauthorized localities, This report has seven sections. The first offers some effectively illegal aliens in their own country. historical background on migration and urbanization To be sure, not all the distinctive features of what in China. The second addresses the comparability and might be called “urbanization with Chinese charac- reliability of the official Chinese demographic data teristics” can be explained in terms of the Chinese that form the basis of our study—that is, the ques- state’s exceptional ambition to oversee (or super- tion of what we actually know and how we know it. vise) basic decisions in its subjects’ lives that would The third presents the basic contours of migration in be considered private affairs in almost any other the post-Mao era as reported in official Chinese data, country. On the other hand, without an apprecia- centering on the 2010 census. It also examines inter- tion of the extraordinary degree of politicization provincial migratory flows as reported by China’s and state involvement in residential movement, 2010 census. urban planning and development, and other matters The fourth section looks at the Chinese urbaniza- that pertain broadly to population control or man- tion process from an international perspective: how agement, some of the most important differences China differs from what might have been expected between patterns of urbanization in contemporary from a “typical” country and what this may mean for China and the rest of the world cannot be under- economic performance. The fifth discusses the arith- stood or explained. metic and politics of the great migration to Chinese This report examines the great urbanization in cities. The sixth examines the contrasting human China over the post-Mao era, a tremendous demo- resource profiles of urban and rural areas in China as graphic transformation that was largely abetted reported in China’s 2010 census and what this may by internal migration. In this report, we outline portend for the ongoing official urbanization drive the dimensions of this great change, detail some of and thereafter. The final section offers some conclud- the dynamics in play, and point to some of the key ing comments.

5 II. Background

ore or less by definition, urban settlements in Historians and economists have debated this great M China are as old as Chinese civilization itself, paradox,18 but it need not detain us here. and urban life runs back as far as Chinese recorded Two distinctive features of urbanization in pre- history. Given ancient China’s high levels of tech- modern China, though, merit mention and consider- nological attainment, economic sophistication, and ation for the perspective they cast on contemporary administrative organization, sizable cities were char- circumstances. First, there is the long history of rel- acteristic of the Chinese empire even in its earliest ative geographical mobility for ordinary people in years. The level of overall urbanization at that time China. China has witnessed not only great move- was possibly unsurpassed anywhere else in the world. ments of population across its territories over time Although such estimates are necessarily impre- but also—thanks to opportunities afforded by the cise, Kang Chao of RAND Corporation used surviv- long-standing commercialization of its economy and ing (non-demographic) records to estimate that the attendant division of labor—considerable volun- about 14 percent of China’s population lived in set- tary individual and family migration between coun- tlements of 2,000 persons or more in the third cen- tryside and city, both temporary and permanent. tury BC, on the eve of the Qin Dynasty.13 By 2 AD, the Second, and hardly less relevant, is the enduring date of Imperial China’s first census, Chao estimated shadow that an overarching Chinese state has cast China’s population to be about 17 percent urbanized. over city life and urbanization throughout almost all Around that time, according to studies by Princeton’s of Chinese history. In other places, cities may have Gilbert Rozman14 and German urban planner Alfred arisen more or less spontaneously. But in China since Schinz,15 the Han Dynasty capital of Chang An (now at least the beginnings of the Qin Dynasty, imperial Xi’an) might have had a population of a quarter of a rulers took it for granted that the city was first and million—and as Chao notes, “Chang An may not even foremost an instrument for control and administra- have been the largest [Chinese] city at that time.”16 tion. Urban China was deliberately built in accor- Over the next millennium, urbanization in China dance with that understanding. appears to have waxed and waned, reaching an apo- In China, as Kang Chao observed, “cities were gee under the Song Dynasty (on the eve of the Mongol built by governments.”19 This extended well beyond conquest). In the early 13th century, by Chao’s careful the general bureaucratic imperatives of establishing a reckoning, over 20 percent of the population of Song vast network of outposts to manage a geographically (southern) China was urbanized, and the Song capital far-flung empire, reaching down into cities of any size of Lin An (present-day Hangzhou) may have had as “so that a more or less standard layout obtained in many as 2.5 million people. To the north, in Jin China them,”20 thanks to ubiquitous official urban planning (then under rule by the Jurchen), Chao suggests the and detailed regulation of urban land use. overall level of urbanization was only slightly lower.17 Thus, in the Chinese tradition, even more than in Strikingly, however, urbanization ratios in China others, the city has been both a symbol of power and and even the size of China’s largest cities markedly control by central authorities and the nerve center for declined over subsequent centuries—until the begin- it. Conversely, Chinese history can also be read as a ning of the industrial era in China—even though Chi- story of what happens to rulers who do not maintain na’s total population surged over that same period. order and stability in their biggest cities. Time and

6 NICHOLAS EBERSTADT AND ALEX COBLIN

again over successive millennia, the end of a dynasty movement by tying rations (in earlier years of the is heralded when the erstwhile ruling directorate Communist regime) or social benefits (today) to loses control over the capital and other major urban one’s hukou locality and by exposing China’s out-of- centers. hukou populace to a panoply of penalties, both formal Since most of China’s population was rural, impe- and unofficial. At the discretion of authorities, Chi- rial rule necessitated administration and control of nese subjects may be detained, punished, and forci- the countryside, which in turn presupposed, among bly returned if they are found outside their authorized other things, an enormous bureaucracy for surveilling place of residence. and policing peasants in small, distant localities. This The disastrous Great Leap Forward campaign of imperial bureaucracy of course evolved and did not 1957–59, which brought catastrophe and famine to the function effectively (or at all) in large parts of China Chinese countryside, put the PRC hukou system to during times of upheaval, chaos, and war. the test. But for the regime’s mechanisms for “influx Certain aspects of this historical system are highly control,” China’s capital and other key cities would relevant to China today, insofar as the lineage of have been overwhelmed by desperate and destitute extant People’s Republic of China (PRC) institu- peasants in search of sustenance. The consequences tions for keeping track of people and managing pop- for political stability would have been unpredict- ulation movement can be traced directly back to able. Thanks to ruthless administration of hukou and them. Strictly speaking, after 1949 the PRC scrapped urban rationing, Chinese authorities prevented urban the old, imperial baojia arrangements for commu- growth during this crisis and actually sent city-based nity (multi-household) surveillance and monitoring, migrant workers back to their erstwhile rural hukou sometimes translated as the “mutual responsibility to reduce the financial strain of food costs for urban system.”21 But one key element of baojia was retained: centers on the planned economy. hukou, the compulsory, official registration and iden- In the era of strict hukou-enforced controls on tification system for individuals and their households. migration under Mao, urbanization was held to a vir- Under Maoist Communism, hukou became an tual standstill (at least according to official statistics), even more important tool for social control than despite the regime’s ambitious plans for rapid mod- under imperial rule. As Fei-Ling Wang put it, “The ernization and industrialization. (Migration controls PRC hukou system had its distinctive totality, central did break down temporarily in Maoist China for much planning, and Soviet-style ideological coating.”22 Per- of the country during the turmoil of the Cultural Rev- fected in the 1950s, the PRC hukou system was run olution in the late 1960s, but they were fully restored under the auspices of the Ministry of Public Secu- after the military suppressed the Red Guards.) There rity (by “hukou police”). It recorded every Chinese are serious issues surrounding urbanization data in citizen’s identity and personal details in official files modern China, as we will see shortly. Taken on their readily accessible by party and state hierarchy. Every face, however, official Chinese statistics report that person was given a class status and assigned a specific China’s urbanization ratio declined between 1960 and residence, and they needed explicit state permission 1976, the year of Mao’s death: from 19.8 percent to to travel from their approved locality. Migrating or 17.4 percent.23 traveling without express government authorization To go by these numbers, China’s urbanization became a crime and could incur severe penalties. ratio in 1980 (19.3 percent) would have been not only Hukou officially registers each person’s household slightly lower than two decades earlier but also lower and establishes his or her legal place of residence. By than Kang Chao’s corresponding calculations for so doing, it also establishes whether a Chinese man, Song Dynasty China more than seven centuries ear- woman, or child is venturing (or residing or working) lier.24 There are multiple reasons for this, not least illegally in any given location where authorities might of these totalitarian vigilance against urban disor- question or challenge them. It controls population der, but poor agricultural performance was surely a

7 URBANIZATION WITH CHINESE CHARACTERISTICS

primary factor here. Careful research suggests per provisionally superseded a worker’s original and capita grain output in China under Mao was no higher permanent hukou but would almost never actually than it had been centuries earlier.25 This means that replace it. the inefficiencies of agricultural collectivization and In practice, China’s hukou system works very dif- the dysfunction of the commune system placed posi- ferently today from Maoist times. In the 1950s, it was tive limits on the surplus rural areas could generate to used to prevent peasants from moving to cities or to support urban populations. deport them back home if they somehow managed China’s current and ongoing surge in urbanization to secure an uncountenanced urban foothold. Now- commenced only after the economic shift inaugu- adays local authorities typically turn a blind eye to rated three years after Mao’s death, at the landmark non-hukou residents, so long as those migrants are December 1978 3rd Plenum of the 11th Congress of earning money, paying taxes, and not causing trou- the Chinese Communist Party, presided over by Deng ble. Nevertheless, the hukou system endures—as does Xiaoping. The following years would see (among other the Chinese state’s continuing reliance on this mech- things) the demise of the commune system, increas- anism of social and political control, which helps ing tolerance of private enterprise, and an outward account for the peculiarities and contradictions of orientation of the economy, including export pro- urban development in modern China. motion. These policy changes simultaneously freed Thus arose a major new demographic feature up labor in the countryside and jump-started new of contemporary China: the floating population of demand for workers in cities and townships—a new out-of-hukou migrants that figures so centrally in our economic situation that necessitated mass migration. report.26 In 1982, according to China’s Third Popu- The huge new demand for migrant labor could lation Census conducted that same year, just under not be squared with rigid Maoist hukou rules. But seven million Chinese were living outside their offi- rather than scrap the hukou system, Chinese author- cial residence. By the Sixth Population Census in ities preserved the system in principle while dramat- 2010—the main reference point for this report—their ically adjusting its operations in practice. Starting in ranks had swollen into the hundreds of millions. And the early 1980s, would-be migrants were permitted the floating population has reportedly grown since to apply for “temporary hukou” residential authoriza- then—standing out as a significant feature of Chi- tion in a new locality if, for example, they had secured na’s population in arithmetic, but also economic and employment there. This temporary authorization political, terms.

8 III. Data on Population and Urbanization in Contemporary China: Availability and Reliability

efore we attempt any quantitative analysis on underreporting births, incentives that may also vary B population, urbanization, and migration in mod- over space and time.27 ern China, we must have some sense of the inherent China does not yet have complete and accurate uncertainties in—or limitations to—the numbers we vital registration for births and deaths, so its popula- wish to analyze. In other words, as in any other statis- tion data come primarily from national censuses and tical investigation, we need to begin by assessing what nationwide population surveys (or “mini-censuses”). we know and how we know it. The official time series numbers for China’s popula- This means looking at the reliability of Chinese tion are interpolations, reconstructions, or projec- population data overall, the changing definitions of tions based on these data.28 urban and rural areas over time (and thus the com- But Chinese census estimates for newborns, chil- parability of urban and rural data within Chinese time dren, and even youth from the era of the One-Child series), and the discrepancies between official Chi- Policy have proved to be highly error-prone. The com- nese estimates of rural and urban populations and pulsory anti-natal program commenced unannounced those reconstructed by international demographic in 1979 and was publicly proclaimed the following authorities for the most recent Chinese census (2010). year. It was formally recalibrated into a two-child Understanding the strengths, weakness, and peculiari- policy in 2015, and at this writing it is evolving once ties of the information one is working with is essential again—perhaps even toward pro-natalism. (For clues for any statistical investigation about any country. But about these most recent changes and what they may it is all the more important when one is attempting to portend for population policy in China, see Appen- place the patterns or circumstances for a country in an dix A.) The accuracy of Chinese demographic data in international perspective, as we wish to do. years to come may not be disfigured by the incentive Before delving into what we know—and do not for parents to conceal or misrepresent the true num- know—about China’s rural and urban population ber of children in their family. But in the 2010 cen- profiles, we need to understand the overall limits of sus, these pressures were clearly present—as was the China’s demographic data at the national level. So we case for the previous three censuses that traced Chi- should start with the question of the quality of nation- na’s demographic development during the nation’s wide or all-China population numbers. Although Chi- now-famous post-1978 economic transformation. na’s demographic capabilities now may be better Figures 1–4 underscore the problem. These figures than those of other low-income countries, the “data” contrast the counts and implied sex ratios for the in China today are by no means a given—especially Chinese population born since 1976 according to for those born after the advent of the One-Child successive head counts by birth year from China’s Policy, which created considerable incentives for post-Mao censuses.

9 URBANIZATION WITH CHINESE CHARACTERISTICS

Figure 1. Population Count by Birth Year from Successive PRC Censuses: Males and Females, 1976–2010

24 Census Year

22 1982 1990 20 2000 18 2010

Population (Millions) 16

76 78 80 82 84 86 88 90 92 94 96 98 00 02 04 06 08 10 19 19 19 19 19 19 19 19 19 19 19 19 20 20 20 20 20 20 Birth Year Source: State Statistical Bureau, Department of Population Statistics, “1982 Population Census of China,” 1985; National Bureau of Sta- tistics, Department of Population and Employment Statistics, “China Population Census: Tabulation of the 1990 Population Census of the People’s Republic of China,” 1992; National Bureau of Statistics, Department of Population and Employment Statistics, “China Pop- ulation Census: Tabulation of the 2000 Population Census of the People’s Republic of China,” 2002; and National Bureau of Statistics, Department of Population and Employment Statistics, “China Population Census: Tabulation of the 2010 Population Census of the Peo- ple’s Republic of China,” 2012.

Figure 2. Population Count by Birth Year from Successive PRC Censuses: Females, 1976–2010

12 Census Year 1982 11 1990 10 2000 9 2010

Population (Millions) 8

1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 Birth Year

Source: State Statistical Bureau, Department of Population Statistics, “1982 Population Census of China,” 1985; National Bureau of Sta- tistics, Department of Population and Employment Statistics, “China Population Census: Tabulation of the 1990 Population Census of the People’s Republic of China,” 1992; National Bureau of Statistics, Department of Population and Employment Statistics, “China Pop- ulation Census: Tabulation of the 2000 Population Census of the People’s Republic of China,” 2002; and National Bureau of Statistics, Department of Population and Employment Statistics, “China Population Census: Tabulation of the 2010 Population Census of the Peo- ple’s Republic of China,” 2012.

10 NICHOLAS EBERSTADT AND ALEX COBLIN

Figure 3. Population Count by Birth Year from Successive PRC Censuses: Males, 1976–2010

12 Census Year

11 1982 1990 10 2000 9 2010

Population (Millions) 8

1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 Birth Year

Figure 4. Sex Ratios by Birth Year from Successive PRC Censuses and Mini-Censuses, 1982–2010

130

125

120 Census Year

115 1982 1990 110 1995 Sex Ratio 105 2000 2005 100 2010 95

1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 Birth Year Source: State Statistical Bureau, Department of Population Statistics, “1982 Population Census of China,” 1985; National Bureau of Sta- tistics, Department of Population and Employment Statistics, “China Population Census: Tabulation of the 1990 Population Census of the People’s Republic of China,” 1992; Judith Banister, “Shortage of Girls in China Today: Causes, Consequences, International Com- parisons, and Solutions,” 2003, https://www.slideshare.net/anilkr123/shortage-of-girls-in-china; National Bureau of Statistics, Depart- ment of Population and Employment Statistics, “China Population Census: Tabulation of the 2000 Population Census of the People’s Republic of China,” 2002; National Bureau of Statistics, Department of Population and Employment Statistics, “China 2005 1% Popula- tion Survey Data Assembly,” 2007; and National Bureau of Statistics, Department of Population and Employment Statistics, “China Pop- ulation Census: Tabulation of the 2010 Population Census of the People’s Republic of China,” 2012.

11 URBANIZATION WITH CHINESE CHARACTERISTICS

Figure 5. 2010 Chinese Population Structure: US Census Bureau Estimates vs. PRC Census

84 79 74 69 64 59 54 49 44 39 34 29 24 19 14 9 4

15 10 5051015 Male Female Population (Millions) Census International Data Base PRC Census Males PRC Census Females

Source: US Census Bureau, International Data Base, “Population by Single Age Groups,” September 2018, https://www.census.gov/ data-tools/demo/idb/informationGateway.php; and National Bureau of Statistics, Department of Population and Employment Statis- tics, “China Population Census: Tabulation of the 2010 Population Census of the People’s Republic of China,” 2012.

As Figure 1 shows, there are huge discrepancies differ by nearly one million—that is, by over 10 per- in head counts for total population in a given birth cent. For 1999, the discrepancy is 1.2 million, which is year. To pick an especially egregious example, only nearly 20 percent. Since we do not yet have access to 23 million babies were counted in the 1990 census, data from the 2015 population survey, we cannot tell but 20 years later, 28 million 20-year-olds were tabu- what sorts of discrepancies we may be facing for the lated in the 2010 census. By the same token, 11.5 mil- past decade (i.e., since the 2005 mini-census). lion babies were counted in 1999 for the 2000 census, Because boys and girls are being undercounted— but there were 13.9 million children for that same or concealed—the confounding effects of tabulation birth year in the 2010 census—over 20 percent more, errors could be considerable. Indeed, given the con- despite intervening mortality. current discrepancies in head counts for boys and Not surprisingly, female births are drastically girls, the year-by-year estimates of sex ratios for the undercounted (Figure 2). The discrepancies for populations of any given birth year in One-Child Pol- females born in 1990 in the 1990 versus 2010 censuses icy China vary substantially (Figure 4). To be sure, amount to 26 percent—even before mortality adjust- there is plenty of evidence that sex ratios at birth have ments. For girls born in 1999, the gap in the 2000 cen- been going up since the One-Child Policy began. But sus versus 2010 census is 24 percent—again, before by exactly how much? According to the 2010 census, mortality adjustments. (International migration is a the sex ratio for those born in 1991 was 106 males per negligible factor here.) 100 females, while according to the 1995 mini-census, However, even boys seem to have been apprecia- it was 115 males per 100 females. Discrepancies of bly undercounted (Figure 3). For birth year 1994, for similar magnitudes can be identified for other birth example, the 2000 census and the 2010 census totals years as well. As some scholarship has suggested, the

12 NICHOLAS EBERSTADT AND ALEX COBLIN

Figure 6. 2010 Chinese Population Structure: UN Population Division World Population Prospects Estimates vs. PRC Census 100+ 96 92 88 84 80 76 72 68 64 60 56 52 48 44 40 36 32 28 24 20 16 12 8 4 0 15 10 5051015 Population (Millions) UNPD PRC Census Males PRC Census Females Source: UN Population Division, World Population Prospects: The 2017 Revision; and National Bureau of Statistics, Department of Population and Employment Statistics, “China Population Census: Tabulation of the 2010 Population Census of the People’s Republic of China,” 2012. concealment or misrepresentation of births in con- sex ratio for the young men and women born between temporary China may have played a considerable 1982 and 1989 ranged between 99 and 102—distinctly role in skewing reported sex ratios in China since the lower than would be expected in a “normal” human One-Child Policy began—a statistical artifact com- population with China’s mortality levels and little pounding the genuine distortions generated by mass international migration (relatively speaking). female feticide.29 Two authoritative research bodies on global demog- Especially noteworthy here are the sex ratios for raphy—the US Census Bureau and the UN Population babies born in the first decade of theOne-Child Pol- Division (UNPD)—attempt to reconstruct China’s icy—or, more specifically, 1982–89. These young men official population returns to adjust for inconsistencies and women would be 30–37 years old today. Accord- and errors in official reporting. These are ongoing ing to the 1982 PRC census, the country’s sex ratio efforts: The UNPD’s latest updates come from its World at birth (boys per 100 girls) was about 108. The 1990 Population Prospects (WPP) series, most recently census and 1995 mini-census reported similar sex revised in 2017. For its part, the Census Bureau’s Inter- ratios for that birth cohort. Yet the 2000 and 2010 national Data Base (IDB) was most recently updated censuses and the 2005 mini-census reported dramati- in September 2018. The contrast between official 2010 cally lower sex ratios for birth year 1982; according to PRC census figures and these two organizations’ recon- the 2010 census, the sex ratio for that group was just structions of 2010 PRC population profile can be seen 102. Indeed, according to the 2010 PRC census, the in Figures 5 and 6 and Tables 1 and 2.

13 URBANIZATION WITH CHINESE CHARACTERISTICS

Table 1. Comparing the PRC Census to the US Census Bureau and UN Population Division: Absolute Population Difference by Age, 2010

US Census Bureau UN Population Division Male Total Female Total Male Total Female Total Age Differences Differences Differences Differences 0–4 1,199,882 1,347,373 3,412,606 3,636,210 5–9 2,212,245 1,998,021 3,871,089 4,223,601 10–14 6,497,177 5,605,688 2,851,494 3,374,399 15–19 2,695,887 921,662 1,736,020 227,914 20–24 1,564,303 –623,850 3,444,340 –413,926 25–29 1,122,852 –996,448 1,970,211 –238,867 30–34 –1,363,692 –1,627,893 110,801 –541,244 35–39 –655,928 –573,072 878,065 685,692 40–44 –827,930 –1,252,846 209,740 662,657 45–49 –2,808,737 –2,618,839 –763,474 –1,062,876 50–54 333,699 38,940 1,522,668 1,748,721 55–59 –760,124 –1,013,668 1,189,150 683,327 60–64 –1,256,970 –1,361,949 –1,102,800 –666,997 65–69 –611,736 –710,730 –114,603 –576,754 70–74 –479,663 –325,008 –85,051 –386,802 75–79 –242,853 –224,515 –676,598 –652,405 80–84 –217,369 –196,700 –666,761 –749,908 85–89 –157,795 –188,274 –223,186 –444,080 90–94 –126,595 –209,455 25,306 –49,021 95–99 –74,684 –134,697 –33,921 –63,736 100+ –6,711 –18,928 –1,961 –4,802 Total 6,035,258 –2,165,188 17,553,135 9,391,103

Source: UN Population Division, World Population Prospects: The 2017 Revision; US Census Bureau, International Data Base, “Mid- Year Population by Single Year Age Groups,” https://www.census.gov/data-tools/demo/idb/informationGateway.php; and National Bureau of Statistics, Department of Population and Employment Statistics, “China Population Census: Tabulation of the 2010 Population Census of the People’s Republic of China,” 2012.

Note, however, that the UNPD and Census IDB persons—meaning the Chinese census total should estimate population at midyear (July 1), while have been higher than the UNPD and Census IDB 2010 PRC census was conducted “with zero hour estimates. However, both Census IDB’s and UNPD of November 1, 2010 as the reference time.”30 WPP’s China 2010 estimates were higher than the Since China’s population was growing by roughly PRC census head count—in aggregate, by about seven million persons per year in 2010, this tim- four million and 27 million, respectively. Thus, ing difference would account for about 2.3 million Census IDB implicitly took the 2010 PRC census

14 NICHOLAS EBERSTADT AND ALEX COBLIN

Table 2. Comparing the PRC Census to the US Census Bureau and UN Population Division: Percentage Population Difference by Age, 2010

US Census Bureau UN Population Division Male Female Male Female Percentage Percentage Percentage Percentage Age Differences Differences Differences Differences 0–4 2.9% 3.9% 3.9% 29.0% 5–9 5.8% 6.2% 6.2% 30.6% 10–14 16.1% 16.2% 16.2% 24.5% 15–19 5.2% 1.9% 1.9% 11.8% 20–24 2.4% –1.0% –1.0% 6.4% 25–29 2.2% –2.0% –2.0% 5.2% 30–34 –2.8% –3.4% –3.4% 4.2% 35–39 –1.1% –1.0% –1.0% 6.3% 40–44 –1.3% –2.0% –2.0% 4.4% 45–49 –5.2% –5.1% –5.1% 2.3% 50–54 0.8% 0.1% 0.1% 9.1% 55–59 –1.9% –2.5% –2.5% 5.1% 60–64 –4.2% –4.7% –4.7% –0.4% 65–69 –2.9% –3.5% –3.5% 1.3% 70–74 –2.9% –2.0% –2.0% –1.5% 75–79 –2.2% –1.8% –1.8% –15.7% 80–84 –3.7% –2.6% –2.6% –29.6% 85–89 –7.2% –5.5% –5.5% –42.4% 90–94 –23.8% –20.0% –20.0% –46.9% 95–99 –63.4% –53.4% –53.4% –66.8% 100+ –75.8% –69.9% –69.9% –74.6% Total 0.9% –0.3% –0.3% 7.6%

Source: UN Population Division, World Population Prospects: The 2015 Revision; US Census Bureau, International Data Base, “Mid-Year Population by Single Year Age Groups”; and National Bureau of Statistics, Department of Population and Employment Statistics, “China Population Census: Tabulation of the 2010 Population Census of the People’s Republic of China,” 2012. returns as an undercount of about six million, or But the discrepancies between Census IDB and about 0.4 percent, while the UNPD implicitly held the UNPD estimates and China’s own 2010 census fig- undercount to be nearly 30 million, or about 2.2 per- ures are much larger once we disaggregate by age cent. (By way of comparison, the US Census Bureau’s and sex. Leaving aside the timing difference, Census 1990 count of the American population was officially IDB’s estimate for China’s 2010 under-30 population deemed to have been a net under-enumeration of is more than 23 million higher than China’s official 1.6 percent.31) count. The UNPD’s is 28 million higher, suggesting an

15 URBANIZATION WITH CHINESE CHARACTERISTICS

undercount for the entire One-Child Policy genera- an overcount of almost four million for men and tion of 5 percent. women in their 60s. UNPD reconstructions imply For particular age groups in the One-Child Pol- smaller overcounts for these same groups; even so, icy generation, the discrepancies look all the more they suggest an overcount of nearly one million for extreme. The latest Census Bureau reconstructions 40-somethings and an overcount of about two and imply a 2010 undercount of 16 percent for boys and a half million for 60-somethings. In proportionate girls age 10–14. Even more strikingly, the latest UNPD terms, both Census IDB and the UNPD imply that the reconstructions imply that girls under age 15 were population overcounts were most dramatic for older undercounted by more than 25 percent—and girls Chinese, those 70 and above. 5–9 years old were undercounted by slightly more Real-world undercounts are in general much eas- than 30 percent! ier to explain than presumptive overcounts—though If the IDB or UNPD reconstructions are closer to some of these China 2010 overcounts could be rea- reality than the original 2010 PRC census, the implica- sonably explained. But notional overcounts can also tions would be of more than just demographic inter- be the consequence of demographic reconstructions est. All Chinese citizens, recall, are required to submit that attempt to reconcile the observations from two to hukou registration for their identity documenta- or more seemingly inconsistent sets of census returns. tion. On the strength of the 2010 census, Chinese So much for international assessments of the authorities concluded that 13 million persons— overall reliability and quality of the 2010 PRC cen- mostly children born above the population policy sus. Next we need to examine the official Chinese birth quota—were without hukou at that time.32 But data on urban and rural population as reported in the that estimate was based on the difference between 2010 census. We defer for the moment the far-from- the total hukou population and the total population trivial question of China’s official definitions of rural enumerated in the census count. and urban areas, focusing instead simply on the reli- If the 2010 PRC census missed as many children as ability of the 2010 census numbers for the places then the IDB and the UNPD imply, China’s non-hukou pop- designated as “rural” and “urban.” ulation that year would have been much larger than We can start by comparing 2010 PRC census esti- Chinese authorities believed—perhaps three times as mates for rural and urban populations to the cor- large. That would be a major oversight for a state that responding estimates from the UNPD’s World jealously commits to surveilling its entire popula- Urbanization Prospects (WUP), most recently tion. If China were losing track of its population, this revised in 2018.33 The UNPD’s WUP is a separate would have ramifications for migration and urbaniza- initiative from the UNPD’s WPP and is not directly tion, the topics of our report, but for a great deal else coordinated with it—WUP country-level population as well. (Lest it go unsaid: There are no available esti- estimates are not identical to those of WPP—but the mates as yet from international demographic organi- two projects come out of the same offices, from an zations on the size of China’s floating population or expert staff fully cognizant of one another’s research. out-of-hukou migrants.) So there is considerable consonance in the overall And what is striking in Figures 5 and 6 and Tables assessment of population profiles and outlooks and 1 and 2 is not only the presumptive undercounts. The the assessments underlying them.34 UNPD and Census IDB also seem to conclude that the At this writing, the UNPD presents data sets for 2010 PRC census has population overcounts for many two versions of WUP on its website, indicating that birth cohorts—including most birth cohorts before it judges both of these revisions to be reliable and the One-Child Policy went into effect. For example, current. The latest is their 2018 revision of WUP; the Census IDB implies China actually had seven million earlier version is the 2014 WUP revision. The latest fewer men and women in their 40s in 2010 than were version is the most recently updated assessment— counted in the 2010 PRC census. It likewise implies presumably reflecting insights that informed the 2017

16 NICHOLAS EBERSTADT AND ALEX COBLIN

Table 3. Population Estimates for China 2010 from Various Sources

Total Total Urban Total Rural Population Population Population Males Females 2010 PRC 1,332,810,869 670,005,546 662,805,323 682,329,104 650,481,765 Census US Census International 1,336,680,939 — — 688,364,362 648,316,577 Database UN Population Division World Population 1,359,755,102 — — 699,882,239 659,872,868 Prospects 2017 Revision UN Population Division World 1,359,755,000 669,354,000 690,402,000 — — Urbanization Prospects 2018 UN Population Division World 1,359,821,465 669,386,225 690,435,240 704,183,055 655,638,410 Urbanization Prospects 2014 China National Bureau of 1,340,910,000 669,780,000 671,130,000 687,480,000 653,430,000 Statistics Source: UN Population Division, “Urban and Rural Population by Age and Sex, 1980–2015,” http://www.un.org/en/development/ desa/population/publications/dataset/urban/urbanAndRuralPopulationByAgeAndSex.shtml; National Bureau of Statistics, Depart- ment of Population and Employment Statistics, “China Population Census: Tabulation of the 2010 Population Census of the People’s Republic of China,” 2012; National Bureau of Statistics of China, China Statistical Yearbook 2017 (Beijing: China Statistics Press, 2018), Table 2-1, http://www.stats.gov.cn/tjsj/ndsj/2017/indexeh.htm; US Census Bureau, International Data Base, “Population by Single Age Groups,” https://www.census.gov/data-tools/demo/idb/informationGateway.php; UN Population Division, World Population Prospects: The 2017 Revision, http://esa.un.org/unpd/wpp/index.htm; and UN Population Division, World Urbanization Prospects: The 2018 Revision, 2018. round of revisions for WPP. But for our purposes, China National Bureau of Statistics (NBS) revision of perhaps the most important difference between census 2010 numbers that has superseded the original these two evaluations is that the 2018 version offers returns in the China Statistical Yearbook and elsewhere. only aggregate estimates of urban and rural popula- We also include the UNPD WPP 2017 and Census tion, whereas the 2014 version provides breakdowns Bureau IDB numbers for comparative purposes; these of rural and urban population by age and sex as well. provide no insight into China’s internal population dis- We will use both the 2014 and 2018 WUP revisions to tribution, but their breakdowns of overall population place official Chinese census estimates of the coun- totals by sex for China 2010 may be usefully compared try’s urban and rural population in perspective. against the corresponding estimates from WUP 2014, Table 3 presents an overview of the various available PRC census 2010, and China NBS 2017. estimates of urban and rural population for 2010 China. As seen in Table 3, Census IDB, UNPD, and even In it we contrapose the official 2010 census results China NBS now all regard the 2010 PRC census as an for China, the WUP 2014 and 2018 estimates, and the undercount, although by different margins. Further,

17 URBANIZATION WITH CHINESE CHARACTERISTICS

Figure 7. China 2010 Urban Population Structure, PRC Census vs. UN Population Division

80+ 75–79 70–74 65–69 60–64 55–59 50–54 45–49 40–44 Age 35–39 30–34 25–29 20–24 15–19 10–14 5–9 0–4 60 40 20 02040 Millions UNPD PRC Census Males PRC Census Females

they all conclude that the 2010 census undercount countryside. China NBS implies an undercount of for China was greater for males than for females—a the rural population by some eight million—enough somewhat surprising conclusion, perhaps, in light to mean China would still have been slightly more of the evidence of widespread and disproportionate rural than urban in 2010, contrary to what the 2010 underreporting of female births. (The Census IDB’s census suggests. The UNPD indicates a rural popula- implicit judgment that the 2010 census slightly over- tion undercount of over 27 million, which would be counted China’s females may be noted in passing a net undercount of 4 percent. The UNPD seems to with curiosity.) be fairly confident about these results, since the dif- As for rural and urban populations, all sources ven- ferences among the three UNPD series for estimated turing such estimates concur that the 2010 PRC cen- all-China 2010 population totals, totals for male and sus is basically on the mark in estimating the country’s female populations, and totals for urban and rural urban population. The 2010 census placed that total populations are never more than some hundredths of at 670 million. Both UNPD estimates and the China 1 percent. If UNPD estimates are close to correct, Chi- NBS estimate of urban population for China that year na’s 2010 urbanization ratio would be a bit lower than differ from the original count by less thanone-tenth the China NBS revision suggests. of a percentage. The scale of possible underreporting and misre- Appreciable differences, on the other hand, are porting for urban and rural China in the 2010 census evident in the estimates for the rural population— may be greater than estimates of net urban or rural all of them pointing to an undercount in the Chinese under-enumeration by themselves would suggest. We

18 NICHOLAS EBERSTADT AND ALEX COBLIN

Figure 8. China 2010 Rural Population Structure: PRC Census vs. UN Population Division

80+ 75–79 70–74 65–69 60–64 55–59 50–54 45–49 40–44 Age 35–39 30–34 25–29 20–24 15–19 10–14 5– 9 0– 4 40 30 20 10 010203040 Millions UNPD PRC Census Males PRC Census Females

Source: UN Population Division, “Urban and Rural Population by Age and Sex, 1980–2015,” http://www.un.org/en/development/ desa/population/publications/dataset/urban/urbanAndRuralPopulationByAgeAndSex.shtml; and National Bureau of Statistics, Department of Population and Employment Statistics, “China Population Census: Tabulation of the 2010 Population Census of the Peo- ple’s Republic of China,” 2012. see this by comparing the age-sex composition esti- If accurate, that would be a noteworthy oversight mates for rural and urban areas in China for 2010 for a government that places a high premium on fol- from the 2010 PRC census with those from the UNPD lowing—and ultimately being able to police—its WUP 2014. (See Figures 7 and 8 and Tables 4 and 5.) urban population. It would also imply that the over- Discrepancies are largest for rural areas: WUP 2014 all proportion of “hidden” or “out of quota” children implies that the Chinese census missed something in urban China would be rather higher than most like one in eight rural Chinese children that year and observers in recent years have presumed.35 nearly every sixth rural girl under age 5. (If we had This overview of data on China’s population in age-sex breakdowns for WUP 2018, these discrepan- 2010 by sex, age, and residence is not intended to favor cies might loom even larger, considering the implied one particular data source over any other. Instead it nationwide undercount of children in WPP 2017.) illustrates the range of uncertainty surrounding key And even though the UNPD’s total population 2010 demographic figures for China in 2010, the reference estimate for urban China tracks closely with the 2010 year for this report. Generally, the range of uncer- PRC census, fairly large differences are apparent for tainty appears to be greater for China’s rural popu- some urban cohorts. WUP 2014, for example, implies lation than its urban population, and it also appears that the census missed about nine million urban Chi- to be disproportionately high for China’s children and nese children that year, which would mean that nearly youth population. every ninth urban child was missed.

19 URBANIZATION WITH CHINESE CHARACTERISTICS

Table 4. Contrasting Estimates of China 2010 Urban Population Structure

Absolute Comparison of 2010 PRC Relative Comparison of 2010 PRC Census Urban Estimates to UN Census Urban Estimates to UN Population Division Urban Estimates Population Division Urban Estimates

Age Male Female Male Female

0–4 1,733,566 1,691,190 10.3% 11.9% 5–9 1,216,369 1,289,573 7.3% 9.2% 10–14 1,400,239 1,380,619 7.9% 9.1% 15–19 2,306,159 1,087,583 8.4% 4.2% 20–24 2,479,252 –321,100 6.9% –0.9% 25–29 1,278,770 –101,405 4.4% –0.4% 30–34 –1,248,197 –1,551,872 –4.4% –5.6% 35–39 –610,544 –508,028 –1.8% –1.6% 40–44 –1,283,078 –1,829,727 –3.9% –5.9% 45–49 –1,184,737 –1,543,485 –4.3% –6.0% 50–54 348,698 –230,198 1.7% –1.2% 55–59 –74,173 –1,053,436 –0.4% –5.6% 60–64 –949,255 –989,491 –7.3% –7.6% 65–69 –502,418 –638,027 –5.7% –7.1% 70–74 –448,139 –476,362 –6.2% –6.3% 75–79 –329,807 –351,459 –6.5% –6.4% 80+ –296,014 –310,389 –7.5% –6.1% Total 3,836,692 –4,456,013 1.1% –1.4%

Thus far, we have focused on only the accuracy of Appendix B details the definitional and conceptual “point estimates” for China’s population for 2010—in changes in the treatment of “urban” and “rural” local- other words, on the reliability of the reported rural, ities between the PRC’s First Census (1953) and Sixth urban, and nationwide head counts based on Chi- Census (2010). Suffice it to say that standard and abid- na’s own administrative designation of “urban” and ing criteria have not been the hallmarks of “urban” “rural” locales for that particular year. But if we are and “rural” designations. Under these six censuses, going to place China’s patterns of migration and Beijing has maintained the same largely traditional urbanization in historical and international perspec- taxonomy of “cities” (shi), “towns” (zhen), and “vil- tive, we also need to examine definitions for rural and lages” (xiangcun)—the first two defined as urban— urban areas: their coherence, consistency over time, but the particulars of what is officially accepted as and comparability with those from other countries. “urban” has changed. And these definitions have also

20 NICHOLAS EBERSTADT AND ALEX COBLIN

Table 5. Contrasting Estimates of China 2010 Rural Population Structure

Absolute Comparison of 2010 Relative Comparison of 2010 PRC PRC Census Rural Estimates to UN Census Rural Estimates to UN Population Division Rural Estimates Population Division Rural Estimates

Age Male Female Male Female

0–4 3,427,712 3,193,908 14.1% 15.7% 5–9 2,435,785 2,411,810 11.1% 13.1% 10–14 2,665,739 2,538,014 11.8% 13.0% 15–19 3,161,424 1,926,179 13.0% 8.8% 20–24 3,299,763 1,003,254 11.8% 3.5% 25–29 2,032,588 929,807 9.3% 4.3% 30–34 26,453 –223,689 0.1% –1.1% 35–39 697,351 724,297 2.6% 2.8% 40–44 27,682 –629,093 0.1% –2.1% 45–49 –77,983 –563,669 –0.3% –2.2% 50–54 1,176,369 529,261 5.8% 2.7% 55–59 734,808 –452,410 3.3% –2.1% 60–64 –661,592 –674,534 –3.9% –4.3% 65–69 –279,605 –417,986 –2.4% –3.7% 70–74 –257,422 –258,886 –2.8% –2.9% 75–79 –194,986 –217,801 –3.1% –3.1% 80+ –196,826 –205,803 –4.1% –2.9% Total 18,017,259 9,612,658 5.3% 3.0%

21 URBANIZATION WITH CHINESE CHARACTERISTICS

connotation of “urban” population for China is dif- buildup during this period meant that large rural ferent from what is commonly understood elsewhere. areas were being incorporated into city administra- These differences can have paradoxical ramifi- tive divisions. Thus, the definition for “urban” in the cations. In its 2014 report on Chinese urbanization, 1990 census was radically tightened, employing far for example, the World Bank demonstrated that con- more rigorous criteria for which parts of city admin- version of rural land to “urban” land has led to fall- istrations counted as urban to limit the inclusion of ing population density among prefecture-level cities rural areas. from 2003 to 2011.37 The same report observes that if While the 2000 census used population density Guangzhou conformed to the same geographical gra- to determine which locales were urban,40 the most dients of population density as Seoul, a city of compa- recent 2010 census dropped this population density rable numbers, it would accommodate an additional criteria and allowed local-level administrative units to six million persons.38 define urbanized areas. It was argued that using lower Population density is considerably lower in Chi- administrative units, such as residential and village na’s cities than in other East Asian countries because committees, to report urbanization would improve Chinese cities include less built-up areas—largely urbanization estimates. because farming or agricultural areas are incorporated Beijing raised and lowered the bar for qualify- into the official boundaries of “cities.” But this is also ing as “urban” from one census to the next with because of the legacy of Communist urban planning, almost pendulum-like regularity between 1953 and which did not prioritize land use in major population 2000: tightened in the Second Census (1964), then centers in accordance with its scarcity value. Thus, loosened in the Third (1982), then tightened again large portions of prime real estate in China’s big cities in the Fourth (1990), and then loosened once more were left underutilized. in the Fifth (2000). Zhou Yixing and Laurence J. C. The second source of confusion stems from Ma estimated the impact of these continuous redef- the changing definition of “urban.” As the Chinese initions on the reported urbanization rate in China National Statistics Bureau indicates,39 three defini- from one census to the next.41 (See Table 6.) By their tions of urban and rural population have been used. reckoning, none of these revisions were minor—but Initial census data employed administrative regions, the most astonishing revision came with the Fourth such as “cities,” to calculate the urban population. Census (1990). Due to limited migration before Deng’s Reform and In effect, the new guidelines for 1990 cut Chi- Opening, the National Statistics Bureau (then called na’s reported urbanization rate in half. Zhou and Ma the State Statistical Bureau) used urban hukou res- point out that if the 1990 census had employed the idents to estimate the urban population at 14.1 per- 1982 definition of “urban,” then the urbanization cent in 1964. However, these estimates may be biased rate in 1990 would have been 53 percent, as opposed down since there was actually migration to cities to the 26 percent actually reported for 1990. With during this period. this breathtaking revision, as Reiitsu Kojima of the In the Third Census (1982), urban populations Institute for Developing Economies noted at the were deemed to be anyone, regardless of hukou sta- time, the China Statistical Yearbook cut its count for tus, living in administrative city divisions, as previ- China’s 1989 urban population by over a quarter bil- ously described. Due to the relatively low level of lion persons (280 million) between its 1990 and 1992 economic development and the strict criteria for editions!42 city administrations at this time, it is believed that And what of the latest (2010) official redefini- the use of city administrations was representative of tion of urban and rural territory in China? Nothing de facto urban areas. as spectacular as the 1990 revision took place, but However, over the 1980s, the criteria for city the tradition of continually redefining “urban” from locales was lowered to stimulate investment. Urban one census to the next was continued. Unlike the

22 NICHOLAS EBERSTADT AND ALEX COBLIN

Table 6. Incomparability of the Census Data on the Levels of Urbanization as Reflected by the Percentage of “Urban” Population

Census Date Percentage of Urban Population in National Population

June 30, 1953 13.26

June 30, 1964 (18.40) 14.10

July 1, 1982 (14.40) 20.55

July 1, 1990 (53.21) 26.23

November 1, 2000 (31.39) 36.09

Note: Figures without parentheses are official census data, while the figures in parentheses are percentages for the years listed using the criteria of the preceding census. Vertical arrows indicate the resulting levels of urbanization if the criteria of the preceding census were used. Figures to the left and right of the horizontal arrows show that they lack continuity and temporal compatibility because they are based on different sets of criteria. Source: Zhou Yixing and Laurence J. C. Ma, “China’s Urbanization Levels: Reconstructing a Baseline from the Fifth Population Census,” China Quarterly 173 (March 2003): 176–96, https://www.cambridge.org/core/journals/china-quarterly/article/chinas- urbanization-levels-reconstructing-a-baseline-from-the-fifth-population-census/6EE598E6BE4B8F64CA1F3D75510F4B7F.

previously extant cycle of tightening and loosening “urban population”—but it is a task far beyond the the criteria for qualifying as “urban,” however, the scope of this report.44 Sixth Census marked the first time that the definition We might take some comfort from the fact that, was loosened twice in a row. at least for big cities, China’s Sixth Census count After a thorough 2014 overview of the different and the UNPD’s WUP 2018 revision come up with definitions used and incorporated in the 2010 census, fairly similar numbers for 2010: the former count- Qin Bo of Renmin University and Zhang Yu of Ohio ing 404 million, the latter 425 million. Since the State conclude that no less than 24 percent of the UNPD’s count is for cities of 300,000 population or increase in urban population between the 2000 and more, we might infer that China’s 2010 definition of 2010 census “should be attributed to urban reclassi- “city” pertains to agglomerations slightly larger. Alas, fication and the readjustment of urban definitions.”43 the apparent congruence here appears to be noth- In other words, by Fifth Census standards, China’s ing more than a coincidence: China does not clas- reported urbanization rate in the Sixth Census (2010) sify its cities based on that population threshold, and might be lower than what was actually reported— its specific enumeration of cities varies considerably possibly markedly lower. from this UNPD list. Even from this summary overview, a glaring and The demographic techniques that the UNPD’s unavoidable conclusion must be recognized: Modern researchers and other trained demographers use China’s figures for “urbanization rates” and “urban are helpful for detecting inconsistencies in reported population” are not measuring the same thing from population data and reconstructing plausible demo- one period to the next. With enough time and care, graphic trends over time based on separate census time series data could be constructed for China under counts or sample surveys for a given country. But alternative and temporally consistent definitions for those techniques rely on a head count approach for

23 URBANIZATION WITH CHINESE CHARACTERISTICS

Figure 9. Percentage of Population Living in Urban Areas: PRC Census vs. UN Population Division, 1950–2016

30 Percentage

0 1950 1960 1970 1980 1990 2000 2010 2020

PRC Census Population Living in Urban Areas UNPD Population Living in Urban Areas

Source: UN Population Division, World Urbanization Prospects 2017; China 2010 Census, https://population.un.org/wpp/; and China National Bureau of Statistics, Tabulation of the 2010 Population Census of the People’s Republic of China, http://www.stats.gov. cn/tjsj/pcsj/rkpc/6rp/indexch.htm.

what is in effect a closed system: In principle, popu- country. The definitions are generally those used by lation totals today should precisely equal last year’s national statistical offices in carrying out the latest population, after factoring in births, deaths, and available census. When the definition used in the lat- migration and aging everybody by a calendar year. est census was not the same as in previous censuses, There is no comparable set of tools to estimate the data were adjusted whenever possible so as to urban and rural population trends with perfect pre- maintain consistency. . . . United Nations estimates cision, even in principle. This is because there is no and projections are based, to the extent possible, on biological regularity to urban versus rural population actual enumerations.45 composition in any given country. In every national population, the reported size of the urban population As we have seen, the UNPD does adjust official as a fraction of the total population depends on voli- reported Chinese urbanization data. But these are rel- tion of residents and definitions of settlement types, atively small adjustments; the UNPD basically works matters far beyond the powers of population mathe- with the latest time series on urbanization that Bei- matics to replicate or predict. jing reports (Figures 9 and 10). This is why the UNPD accepts, and works with, For better or worse, we follow the same approach every country’s own particular definition and enu- in our report—having offered what we hope to be meration of urban and rural population. As WUP 2018 due-diligence caveats regarding the quality and reli- explains: ability of official data on contemporary Chinese urbanization. Since most of our analysis focuses We do not use our own definition of “urban” popu- on Chinese population data for 2010, furthermore, lation but follow the definition that is used in each we will mostly not be plagued by the difficulty of

24 NICHOLAS EBERSTADT AND ALEX COBLIN

Figure 10. Urban Population: PRC Census vs. UN Population Division, 1950–2016

900

800

700

600

500

400 Millions 300

200

100

0 1950 1960 1970 1980 1990 2000 2010

PRC Census Urban Population Urban Population UNPD

attempting to enforce consistency on what appears “large” city center.47 Such standardization could make to be a manifestly inconstant time series for China’s for big revisions in some of the international data urbanization data. that demographers, economists, and geographers are China’s urbanization data are beset by more quirks accustomed to citing and using for urbanization. and peculiarities than those for many other—likely Initial work on such an index, for example, sug- most—contemporary societies. But by examining gests that such global standardization would signifi- these shortcomings in isolation, we are by no means cantly and systematically increase “agglomeration implying that there is a generally accepted standard ratios” in Sub-Saharan Africa in relation to currently in the rest of the world to which China is simply fail- reported urbanization ratios. On the other hand, it ing to adhere. The voluminous country notes in the would generally have the opposite effect on the Latin “Source and Documentation” section of WUP 2018 America and Caribbean region. attests strongly to the contrary.46 For better or worse, What about China? For what it is worth, the test- there is no single standard in practice for measuring drive for the agglomeration index produced a point urbanization across countries today—or for measur- estimate for China for 2000, before the most recent ing it within countries over long periods of time. adjustments in definitions for “urban areas.” Never New research today is contemplating a univer- theless, for that specific year, one variant of the notional sally relevant “agglomeration index”: an internation- agglomeration index—the one using a 50,000-person ally consistent measure of population concentration threshold for “large city”—nearly matched the urban- based on indicators for population density, the size ization ratio reported in China’s Fifth Census (both for a “large” city, and the time required to travel to a estimates coming in at approximately 36 percent).

25 URBANIZATION WITH CHINESE CHARACTERISTICS

Such experimental work may be expanded and per- of such a new index. And in any case, the constant haps further refined with the assistance of new data change in human lifestyles, preferences, and techno- sources, including satellite imagery and GIS tech- logical capabilities could mean that the criteria and niques for identifying changes in built-up areas. But thresholds for agglomeration indexes will themselves we are still far from the rollout and global embrace be constantly evolving, too.

26 IV. The Geography and Demography of Migration in China 2010

n this section, we offer a broad overview of the erstwhile most populous province. At the same time, Igeography and demography of migration (focusing the population of other parts of the country was on the floating population) in contemporary China as reportedly burgeoning. In just 10 years, Guangdong reflected in the 2010 census. Crucial here is the con- province’s population jumped by almost 20 million— vention of hukou: the aforementioned legal require- or 22 percent. Guangdong’s population increase is ment for a registered place of residence in China. dramatic in a country where total population rose The hukou system is not unique to Communist by just 7 percent between 2000 and 2010. Just four China, or Communist countries; it is found inter alia provinces—Guangdong, Zhejiang, Shanghai, and in democratic Taiwan as well—and as already noted, Beijing—accounted for nearly half of China’s over- long predates the People’s Republic of China in Chi- all population growth between 2000 and 2010, even nese history. In practical and administrative terms, though they composed less than a seventh of the though, there is a world of difference in the portent of total Chinese population in 2000. As shown in Fig- this registration system for ordinary people in China ure 12, interprovincial migration explains these big and Taiwan. In China today, a person living outside increases in regional population—as well as the his or her hukou is in violation of the law and poten- decreases registered. tially subject to both fines and penalties and to relo- Figures 13–15 illuminate the migration dynamic cation back to the registered place of residence. While more directly by depicting the net reported interpro- one’s hukou can legally be reassigned, this takes place vincial flows of migrants in 2010. Figures 14 and 15 for only a tiny proportion of contemporary migrants. show the tally of in-migrants minus out-migrants For 2010, Chinese census protocols counted as for each of China’s 31 provinces and province-level migrants those persons who had resided outside their municipalities. By 2010, more than 85 million men hukou for six or more months before the November 1 and women—about one out of every 15 Chinese— national population count. Since we have no readily were reportedly interprovincial migrants, living out- available data by which to cross-check the accuracy of side their home province. (The true number may these hukou migration data, we take them as given in have been even higher. Some migrants may have this analysis. (For additional information on develop- resided outside their assigned hukou for less than six ments in the hukou system, see Appendix C.) months at the time of the census, and others may not The dimensions of the ongoing movement of peo- have answered that question accurately.) The share ple in China are hinted at in Figure 11, which reports of interprovincial migrants was up sharply from just inter-censual provincial population change for 2000– 10 years earlier, when just one in 29 Chinese was 10. For the first time since the Maoist famine, major reportedly an interprovincial migrant. Net migration geographic areas in China experienced population flows by province speak to the geography of opportu- decline—perhaps most notably Sichuan, China’s nity—or the lack of it—in China.

27 URBANIZATION WITH CHINESE CHARACTERISTICS

Figure 11. Total Population Net Change by Province, 2000 vs. 2010

5 Population (Million s) 0

–5 i a g g ian be i j hui Jilin na n nan na n i ubei ans u angx i izhou ianji n zang) hanxi Fu njang H chuan He An G Hu T Jiangxi He haanxi Beijin S Ha Jiangsu heijian g Xi Yunnan Ningxi Si S Qingha Gu Gu hanghai ongoli a Liaonin Z angdon g longjiang handong S i hongqing M S Gu C He Tibet (Xi Inner Source: China 2000 Population Census, “Population by Sex, Age and by Region,” China Data Online, 2011; and China 2010 Population Census, “1.7 Population by Sex, Age and by Region,” 2010.

Figure 12. Absolute Net Migration Between Provinces, 2010

5 illions M 0

–5

–10

–15 i i i i i i i a u u u n n n n n g g g ia ai xi ji n an ng ng ng ng ng ili xi gx ha gx hu be be J jia ol gs jin ns ua na na an ho on in ia ia ia gh ni an nn an aa nx do ng an an zang) iz ng Fu An gq in ej an nj ch ng Hu He gj Ti Ga Ji Sh He Hu an Bei gd ao Ji Ha Yu an Ni Sh Xi Si Qi Li Gu Gu on Zh on an Sh Mo il Sh r Ch Gu He ne Tibet (Xi In

Source: National Bureau of Statistics, Department of Population and Employment Statistics, “China Population Census: Tabulation of the 2010 Population Census of Beijing Municipality,” 2012.

28 NICHOLAS EBERSTADT AND ALEX COBLIN

Figure 13. Net Migration Between Provinces, 2010

< –6 Million –6 to –3 Million –3 to 0 Million 0 to 6 Million 6 Million to 12 Million 18 Million+

Figure 14. Net Migration as a Percentage of Nonmigrant Provincial Population, 2010 40%

30%

20%

10%

–10%

–20% i i i i i i i a u u u n n n n g g g ia ai xi ji n li an ng ng ng ng ng xi ha gx gx hu be be Ji ji an ol gs jin ns ua na na an ho on in ia ia ia gh ni an nn an aa nx do ng an an zang) iz ng Fu An an nj ej gq in ch ng He Hu gj Ti Ga Ji Sh Hu He an Bei gd ao Ji Ha Yu an Ni Sh Xi Si Qi Li Gu Zh on Gu on an Sh Mo il Sh r Ch Gu He ne Tibet (Xi In Source: National Bureau of Statistics, Department of Population and Employment Statistics, “China Population Census: Tabulation of the 2010 Population Census of Beijing Municipality,” 2012.

29 URBANIZATION WITH CHINESE CHARACTERISTICS

Figure 15. Migrants as a Reported Percentage of Provincial Population, 2010 PRC census

60%

50%

2010 Nationwide Average: 19.6% 40% 2000 Nationwide Average: 11.6%

30% 1990 Nationwide Average: 3.0%

20%

10%

0% Jilin zang) Fujian Anhui Hubei Hebe i Tianjin Gansu Jiangxi Shanxi Hunan Henan Beijing Hainan Jiangsu Yunnan Shaanxi Ningxia Sichuan Xinjian g Qinghai Liaoning Guangxi Guizhou Zhejian g Shanghai Shandong Chongqing Guangdon g Heilongjiang Tibet (Xi Inner Mongolia Interprovincial Intraprovincial

As of 2010, the net inflow of migrant population from depressed Anhui amounted to over one-sixth of to booming Guangdong exceeded 20 million persons, the province’s population. Four additional provinces— the net inflow for coastal Zhejiang was reportedly Jiangxi, Hunan, Sichuan, and Guizhou—reported net 10 million, and Beijing, Shanghai, and Jiangsu (the other emigrant populations equivalent to 10 percent or more province bordering Shanghai besides Zhejiang) netted of their 2010 residential populations. roughly nine million, seven million, and four million, On the other end of the spectrum, net in-migration respectively. On the other hand, five Chinese provinces was equivalent to almost 40 percent of Shanghai’s res- reportedly experienced a net loss of over five million idential population, 35 percent of Beijing’s, and over through migration: Jiangxi (five million), Hunan (over 20 percent of Tianjin’s. Beyond the province-level six million), Sichuan (nearly eight million), Henan municipalities, net in-migration was equivalent (over eight million), and Anhui (nine million). The to almost 20 percent of residential population for geography of net migration, highlighted in Figure 13, Guangdong and Zhejiang.48 shows massive movements to the coastal areas plus Figure 15 casts further light on regional migration Beijing and more modest movement to the northern patterns in modern China. Whereas Figures 11–14 and western regions, with central and inland China’s concerned the “flow” of migrants, Figure 15 examines provinces furnishing most of the net out-migration. the “stock” of migrants. It shows the proportion of Another take on interprovincial migration comes the total population comprised of migrants in 2010 from casting these figures in proportional terms against for China’s 31 provinces and municipalities—further the reference province or province-level municipality. indicating whether these migrants originated within (See Figure 14.) The reported fraction of net out-migrants or outside the province.

30 NICHOLAS EBERSTADT AND ALEX COBLIN

Figure 16. Migrant Proportion of Total Population by Reported Residence Type, 2010

100%

90%

80%

70%

60%

50%

40%

30%

20%

10%

0% City Town Village Migrant Population Total Population

Nationwide 2010 Migrant Total: 261 Million (33% Interprovincial, 67% Intra-Provincial)

All in all, roughly two-thirds of all enumerated (city, town, or village) in China in 2010. According to migrants in 2010 reportedly were intra-provincial these data, nearly two-thirds of migrants were living migrants. But there were huge variations in patterns in big cities (“cities”), and nearly seven-eighths were nationwide. As may be seen, migrants accounted for in urban areas (“cities” and “towns”). In terms of the the majority of the total population in both Shanghai composition, migrants accounted for just 5 percent and Beijing and nearly 40 percent of the overall popula- of the rural population (villages), but over one-fifth tion in China’s third province-level municipality, Tian- (21 percent) of the total population of smaller cities jin. In thriving Guangdong and Zhejiang, migrants were and over two-fifths (42 percent) of the total popula- 35 percent of the total population. On the other hand, tion of China’s bigger cities. For urban areas overall, in remote and forbidding Tibet (Xizang), migrants migrants accounted for just over a third (34 percent) were less than a tenth of the total population. of the total population by 2010. For the country as a whole, over 19 percent of the Dependency ratios—conventionally defined as the population were migrants—nearly one person in five. proportion of older (65+) and younger (0–14) popula- This was twice the proportion just 10 years earlier and tions in relation to the working-age (15–64) population— seven times the proportion 20 years earlier. Migration are of course shaped by long-term fertility and mortality has been growing explosively in modern China, and trends in all societies. But in China today, they are also migrants are now a major component of China’s over- powerfully affected by migration patterns, as may be all population. appreciated in reviewing the differences in dependency Figures 16 and 17 offer an overview of the migrant ratios among village, town, and city populations across population’s distribution by residential location type the nation’s provinces. (See Figures 18–20.)

31 URBANIZATION WITH CHINESE CHARACTERISTICS

Figure 17. Migrants by Reported Current is well over 80 percent. In effect, big cities in China Residence Type, 2010 today are workshops, relying heavily on imported labor to produce their wares. China today has a truly national labor market— and a strikingly large number of people have relocated 13.4% in pursuit of economic opportunities. According to China’s 2010 census, 261 million people—including 223 million people of conventional working age (15–64)—were residing outside their assigned hukou (or official registered place of residence). Of these 21.3% 223 million, nearly three of every fiveworking-age 65.3% persons (57 percent) claimed they migrated to “work in a business” or “change jobs.” Those who did not cite job concerns may also have moved for economic considerations, but may not have said so.49 According to the 2010 Chinese census, the mainland’s 15–64 population totaled 993 million. Thus, the 223 million working-age migrants enumerated in the 2010 census would have accounted for slightly over City Town Village 22 percent of the country’s conventional working-age manpower that year. This means roughly every fifth Source: National Bureau of Statistics, Department of Population person of conventional working age was living out- and Employment Statistics, “China Population Census: Tabula- side his or her assigned place of residence. tion of the 2010 Population Census of the People’s Republic of Census data allow us to look at the relationship China,” 2012. between migration and education nationwide—that is, both interprovincial and intra-provincial migration according to educational status.50 (See Figures Not surprisingly, nationwide China 2010 depen- 22–24.) Dramatically different patterns of migration dency ratios were highest in the countryside and low- in China are evident for persons of differing educa- est in the big cities. But in rural parts of Chongqing, tional attainment. the total dependency ratio was twice as high as in For men and women with no more than primary rural parts of Shanghai, while the dependency ratio schooling, migration is disproportionately reflected was nearly three times as high in rural Guizhou as it in the western hinterlands (Guizhou, Qinghai, Tibet, was in rural parts of Beijing. The dependency ratio for and Yunnan). Beijing appears to be the least welcom- children, similarly, is almost always highest in rural ing for such unskilled migrants. For persons with sec- areas. (It is four times as high in rural Guizhou as in ondary education, by contrast, Guangdong appears rural parts of the Beijing municipality.) Correspond- to be the special magnet; middle and high school– ing, though less regular, trends may also be seen for educated migrants in Guangdong are far overrepre- old-age dependency in China in 2010. sented in relation to the national average. They are In 2010, 75 percent of the population of China’s most underrepresented in Tibet and Qinghai. smaller cities (“towns”) and 80 percent of the nation’s Finally, for migrants with at least some tertiary big cities (“cities”) were composed of working-age education, Beijing is disproportionately the venue for men and women (15–64). (See Figure 21.) In big-city migrants—no surprise here. Somewhat surprisingly, Guangdong, Zhejiang, and Fujian—vibrant coastal though, Shanghai does not seem to hold the same areas—the proportion of the working-age population attracting power. Yet other places, such as inland

32 NICHOLAS EBERSTADT AND ALEX COBLIN

Figure 18. Total Dependency Ratio by Province, 2010

Dependency Ratio 20

0 Jilin Total zang) Fujian Anhui Hubei Hebei Tianjin Gans u Jiangxi Shanxi Hunan Henan Beijing Hainan Jiangsu Yunnan Shaanxi Ningxia Sichuan Xinjian g Qinghai Liaoning Guangxi Guizhou Zhejian g Shanghai Shandong Chongqing Guangdon g Heilongjiang Tibet (Xi Inner Mongolia Village/Rural Town City

Figure 19. Youth (0–14) Dependency Ratio by Province, 2010

50 45

Dependency Ratio 15

0 i i i l i i i a u u u n n n n n g g g xi ai li lia ta ji n an an ng ng ng ng ng xi ha hu gx be be Ji ns ji ua na na gs jin anxi ho on in ia gh ni ia an ji an To nna aan angx ng an zang) iz do ng An Fu gqi an ch ej Hu He nj Ga Ti Sh Ji Hu He an ao Bei Ha Ji Yu ng Ni Sh Si an Xi Qi Li Gu on Gu Zh angd Sh Mongo ilo Sh r Ch Gu He Tibet (Xi Inne Village/Rural Town City

33 URBANIZATION WITH CHINESE CHARACTERISTICS

Figure 20. Old-Age (65+) Dependency Ratio by Province, 2010

10 Dependency Ratio

Figure 21. Working-Age (15–64) Population as a Percentage of Total Population by Province, China 2010 90

75 Percentage 70

60 i i l i i i i a u u u n n n n n g g xi ai lia ta ji n an an ng ng ng ng ng ng xi ha hu gx be be Jili ns ji na ua na gs jin anxi ho on in ia gh ni ia an jia To nna aan angx ng an zang) iz do ng An Fu gqi an ch ej Hu He nj g Ga Ti Sh Ji Hu He an ao Bei Ha Ji Yu Ni Sh Si an Xi Qi Li Gu on Gu Zh angd Sh Mongo il on Sh r Ch Gu He Tibet (Xi Inne Village/Rural Town City Source: National Bureau of Statistics, Department of Population and Employment Statistics, “China Population Census: Tabulation of the 2010 Population Census of the People’s Republic of China,” 2012.

34 NICHOLAS EBERSTADT AND ALEX COBLIN

Figure 22. Percentage of Migrants in and Figure 23. Percentage of Migrants in and Between Provinces, Primary Schooling (2010) Between Provinces, Secondary Schooling (2010)

0–0.13 0–0.45 0.13–0.14 0.45–0.5 0.14–0.16 0.5–0.55 0.16–0.18 0.55–0.6 0.18–0.20 0.6–0.65 0.20–0.25 0.65–0.7 0.25–1 0.7–1

Source: National Bureau of Statistics, “China Population Census: Source: National Bureau of Statistics, “China Population Census: Tabulation of the 2010 Population Census of the People’s Repub- Tabulation of the 2010 Population Census of the People’s Repub- lic of China,” 2012, Table 7-6, Table 7-7. lic of China,” 2012, Table 7-6, Table 7-7.

Figure 24. Percentage of Migrants in and Figure 25. Migration to Guangdong from Between Provinces, Tertiary Schooling (2010) Province of Registered Hukou per Million Residents, 2010

0–0.13 0.13–0.15 0–5,000 0.15–0.17 5,000–10,000 0.17–0.19 10,000–50,000 0.19–0.21 50,000–100,000 0.21–0.24 100,000–250,000 0.24–1 250,000–500,000 500,000+

Shaanxi and Henan, are highly overrepresented for today! This may suggest that the burgeoning demand highly educated interprovincial migrants. The rea- for labor inside and outside their export processing son is something of a puzzle—but there appears to zones is not tilted toward higher-skilled labor. be an answer to this puzzle, as argued in Appendix D. The 2010 PRC census does not provide data on (That answer in turn provides insight into the logic the migrant population’s exact place of origin, but it and dynamics of migration for skilled manpower in does report the province of origin. This allows us to China today.) trace patterns of interprovincial migration. Generally At the other end of the spectrum, Tibet looks to (though of course not always), these interprovincial be one of the least inviting venues for migrants with migrants have traveled farther than intra-provincial higher education—again, no surprise. But the same is migrants in search of work or a new home. In true of bustling coastal Guangdong and Zhejiang— 2010, one in three migrants lived outside his or her the two top provinces for net in-migration in China home province. We can learn about this aspect of

35 URBANIZATION WITH CHINESE CHARACTERISTICS

Figure 26. Migration to Zhejiang from Figure 27. Migration to Shanghai from Province of Registered Hukou per Million Province of Registered Hukou per Million Residents, 2010 Residents, 2010

0–5,000 0–5,000 5,000–10,000 5,000–10,000 10,000–50,000 10,000–50,000 50,000–100,000 50,000–100,000 100,000–250,000 100,000–250,000 250,000–500,000 250,000–500,000 500,000+ 500,000+

Figure 28. Migration to Beijing from Province of Figure 29. Migration to Jiangsu from Province of Registered Hukou per Million Residents, 2010 Registered Hukou per Million Residents, 2010

0–5,000 5,000–10,000 0–5,000 10,000–50,000 5,000–10,000 50,000–100,000 10,000–50,000 100,000–250,000 50,000–100,000 250,000–500,000 100,000–250,000 500,000+ 250,000–500,000 500,000+

Figure 30. Migration to Guangdong from Figure 31. Migration to Zhejiang from Province of Registered Hukou per Million Province of Registered Hukou per Million Residents, 2010 Residents, 2010

0–100 0–100 100–200 100–200 200–500 200–500 500–1,000 500–1,000 1,000–7,000 1,000–7,000 7,000–15,000 7,000–15,000 15,000+ 15,000+

36 NICHOLAS EBERSTADT AND ALEX COBLIN

Figure 32. Migration to Shanghai from Figure 33. Migration to Beijing from Province Province of Registered Hukou per Million of Registered Hukou per Million Residents, Residents, 2010 2010

0–100 0–100 100–200 100–200 200–500 200–500 500–1,000 500–1,000 1,000–7,000 1,000–7,000 7,000–15,000 7,000–15,000 15,000+ 15,000+

Figure 34. Migration to Jiangsu from Province Figure 35. Migration from Jiangxi in the Past of Registered Hukou per Million Residents, Five Years per Million Residents in Hukou 2010 Province, 2010

0–100 0–300 100–200 300–600 200–500 600–1,000 500–1,000 1,000–1,500 1,000–7,000 1,500–3,000 7,000–15,000 3,000–6,000 15,000+ 6,000+

contemporary migration in China by examining the of migrants, irrespective of its distance from the characteristics and directions of these interprovincial “importing” province. population flows. Figures 25–29 show absolute flows—but prov- In Figures 25–34, we look at aggregate interpro- inces can vary considerably in size in China today. vincial migrations flows, focusing on the top five What about relative or proportionate out-migration provincial migration destinations in 2010. These to the major migrant-attracting provinces? These figures depict a sort of “gravity map” for interpro- patterns are displayed in Figures 30–34. This does not vincial migration, with the pool of migrants gener- appreciably change the patterns revealed in the pre- ally located inland and flowing toward the nearest of vious set of figures—except for relatively populous the “magnet provinces” in question. In all five cases, Sichuan, whose import as a source of out-migrants however, Sichuan stands out as a major “exporter” is correspondingly diminished.

37 URBANIZATION WITH CHINESE CHARACTERISTICS

Figure 35. Migration from Jiangxi in the Past Figure 36. Migration from Hubei in the Past Five Years per Million Residents in Hukou Five Years per Million Residents in Hukou Province, 2010 Province, 2010

0–300 0–300 300–600 300–600 600–1,000 600–1,000 1,000–1,500 1,000–1,500 1,500–3,000 1,500–3,000 3,000–6,000 3,000–6,000 6,000+ 6,000+

Figure 37. Migration from Sichuan in the Past Figure 38. Migration from Henan in the Past Five Years per Million Residents in Hukou Five Years per Million Residents in Hukou Province, 2010 Province, 2010

0–300 0–300 300–600 300–600 600–1,000 600–1,000 1,000–1,500 1,000–1,500 1,500–3,000 1,500–3,000 3,000–6,000 3,000–6,000 6,000+ 6,000+

And where do migrants from the major “sending” The reason is not immediately clear to us and perhaps provinces end up? We can see this in Figures 35–39. merits further investigation. Generally, the answer tends to be the coastal areas— We have demonstrated that there are distinctive, Shanghai most certainly included—plus Beijing. But predictable, and (in greater or lesser measure) eco- differences and particularities also stand out. Sich- nomically rational patterns of regional migration uan, Henan, and Hubei, for example, also send signifi- in China today. But we should not wish to obscure cant flows of migrants into adjacent inland provinces, the important point with which we began: China’s even though those venues are themselves relatively labor market—and its attendant domestic migra- poor or underdeveloped. Sichuan and Henan, further- tion network—is truly national in nature. Migrants more, also account for a proportionately high flow of from every one of China’s provinces can be found people into the westernmost Xinjiang borderland. in virtually every other province—and large cities in

38 NICHOLAS EBERSTADT AND ALEX COBLIN

Figure 39. Migration from Anhui in the Past China include migrants with hukou from every part of Five Years per Million Residents in Hukou the nation. Province, 2010 This “churn” has economic implications, but it has political ramifications as well. We should never forget that, for policymakers in Beijing, acceptable internal migration is conditioned by not only its potential economic contributions but also its impact on social sta-

0–300 bility and regime security. 300–600 600–1,000 1,000–1,500 1,500–3,000 3,000–6,000 6,000+

Source: National Bureau of Statistics, Department of Population and Employment Statistics, “China Population Census: Tabula- tion of the 2010 Population Census of the People’s Republic of China,” 2012.

39 V. Chinese Urbanization in International Perspective

n this section, we examine the dynamics of Chinese extra urban residents. We further surmise this is Iurbanization from an international development one reason the Chinese government was prompted perspective—that is, how China’s transformation into to adopt a deliberate policy of positively stimulating a majority urban society compares with transforma- urban population growth—the so-called urbanization tions in countries with comparable income levels and drive unveiled in 2014. social development indicators. We show that China in In this section, we use international data sets to 2010 appeared to be markedly under-urbanized, given undertake our statistical analysis—not official Chi- its levels of per capita income, educational attain- nese statistics. These large international data sets per- ment, and public health. Furthermore, compared mit us to investigate international patterns of postwar with other contemporary societies, China’s appar- socioeconomic development in some detail and with ent under-urbanization may have been retarding the some confidence (though, of course, with all the cave- country’s economic performance. ats that must attend such exercises). This may sound surprising, considering the tre- The estimates for international urbanization come mendous surge of urbanization since the Deng-era from the UN Population Division’s (UNPD) World reforms were unleashed at the Third Plenum of the Urbanization Prospects.51 Estimates on educational 11th Party Congress in December 1978. Yet any coun- attainment are drawn from the massive International try that had experienced the same surge of incomes, Institute for Applied Systems Analysis and the Vienna education, and health that China recorded since those Institute of Demography data set on postwar educa- historic reforms would have been predicted to urban- tional patterns, an undertaking convened under the ize even more than China did. Wittgenstein Centre for Demography and Global We surmise that China’s lagging urbanization Human Capital in Vienna.52 We also use the Barro-Lee progress was a consequence of policy—more specif- Educational Attainment Dataset, a widely referenced ically, the hukou system. Although the hukou system source on the same topic assembled by Professors has been significantly relaxed since the Mao era, it Robert Barro of Harvard and Jong-Wha Lee of Korea still distorts labor market flows, restricting the rural University.53 population’s movement to urban locales. And we esti- For the estimates for annual levels of real GDP per mate that this distortion has cost China economic capita, we use both purchasing power parity (PPP)– potential. adjusted and non-PPP-adjusted series. The estimates By our preliminary reckoning, China’s urban pop- come from the World Bank’s World Development ulation in 2010 was roughly 70–100 million smaller Indicators (WDI)54 and from the Maddison Project than general international patterns of development Database’s historical estimates of PPP-adjusted per would have predicted. By those same international capita GDP.55 World Bank PPP-adjusted GDP esti- patterns, China would have been expected to gener- mates go back to only 1980. The Maddison Project ate a gross domestic product (GDP) 10–15 percent Database, begun by the late Angus Maddison and con- larger than it actually produced with 70–100 million tinued after his death by his students and colleagues,

40 NICHOLAS EBERSTADT AND ALEX COBLIN

Figure 40. China’s Urbanization Ratios, 1950–2016

50 49.68

36.22 30

26.44 Percentage 20 20.91 18.3 10 13.26

0 1951 1971 1981 1991 201 1 1950 1953 1955 1960 1964 1965 1970 1972 1973 1974 1975 1976 1977 1978 1979 1980 1982 1983 1984 1985 1986 1987 1988 1989 1990 1992 1993 1994 1995 1996 1997 1998 1999 2001 201 0 201 2 201 3 201 4 201 5 201 6 2000 2002 2003 2004 2005 2006 2007 2008 2009

China Statistical Yearbook Population Living in Urban Areas (%) UNPD Population Living in Urban Areas (%) PRC Census

Source: UN Population Division, “Urban and Rural Population by Age and Sex, 1980–2015,” http://www.un.org/en/development/ desa/population/publications/dataset/urban/urbanAndRuralPopulationByAgeAndSex.shtml; National Bureau of Statistics of China, China Statistical Yearbook 2017 (Beijing: China Statistics Press, 2018), Table 2-1, http://www.stats.gov.cn/tjsj/ndsj/2017/indexeh. htm; and National Bureau of Statistics, Department of Population and Employment Statistics, “Basic Statistics on National Population Census in 1953, 1964, 1982, 1990, 2000 and 2010,” 2012, http://www.stats.gov.cn/tjsj/Ndsj/2011/html/D0305E.HTM. begins in 1 AD and extends to 2010. We also use WDI Over those three decades, China’s urban popula- for annual estimates on life expectancy at birth for tion, by the UNPD’s reckoning, shot up by nearly overall population.56 30 percentage points: from 20 percent to slightly over We use two data sets that offer (inescapably sub- 49 percent. (See Figure 40.) jective) readings on the business climate (or, one The UNPD World Urbanization Prospects’ esti- might also say, on the quality of local institutions and mates for China’s urbanization ratios almost per- policies): the Fraser Institute’s Economic Freedom of fectly match the latest official time series from the World Index57 and the Heritage Foundation/Wall Beijing. But this is because the UNPD basically takes Street Journal’s Index of Economic Freedom.58 These official data by member states as given—and not data sets permit us to conduct panel data analysis of because it is vouching for their accuracy, much less international patterns of development from 197059 to standardizing China’s urbanization data against the the present, covering more than 100 countries and data for other countries. often including well over 1,000 observations. Section II went into considerable detail outlining As we know, China’s urbanization since the end the inherent problems and constantly changing defi- of the Maoist era has been monumental, unprece- nitions for “urban” in China’s population statistics. dented by almost any historical measure. According Consequently, we have rather less confidence in the to the UNPD, between 1980 and 2010 alone, China’s reliability of the long-term trends depicted in Fig- urban population soared from about 190 million to ure 40 than UNPD appears to have. Obviously there 670 million—a leap of nearly half a billion people. was a phenomenal movement of people into cities

41 URBANIZATION WITH CHINESE CHARACTERISTICS

Figure 41. Urbanization Ratios, 1950–2010

100

Percentage 40

0 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010

China China, Taiwan Province of China Japan Republic of Korea India Indonesia

Source: UN Population Division, World Urbanization Prospects: The 2018 Revision, 2018.

and urban areas in China between 1980 and 2010, Indonesia. Per capita output was substantially higher and this urban surge has continued since 2010. We, in China in 2010 than in Indonesia.60 All other things however, cannot be confident about estimating the being equal, this would suggest China should have precise dimensions and tempo of these changes. Con- been more urbanized than Indonesia, not less. sequently, we will focus most of our attention in this If we examine the income trajectories of postwar section on the reference year 2010 and how China’s urbanization for these same six countries, the degree urbanization looked in an international perspective at of apparent under-urbanization in China is much that time. clearer (Figure 42). China may have been slightly Paradoxically, despite the breathtaking structural more urban than India in 2010 when it reached India’s transformation of the Chinese economy between 1980 estimated 2010 level of PPP-adjusted GDP per capita. and 2010—and the scarcely less remarkable surge in But China was reportedly about 10 percentage points urban population over those same decades—a wealth less urbanized than Indonesia when it had achieved of development indicators suggests that urbanization Indonesia’s 2010 levels of GDP per capita, and in 2010 was actually lagging in contemporary China. Compar- China was roughly 20 percentage points less urban ing the Chinese experience with that of other selected than either Japan or South Korea when they reached Asian societies shows the degree to which China China’s 2010 level of GDP per capita.61 In this Asian today is an under-urbanized society. perspective, China certainly looked under-urbanized Unsurprisingly, China in 2010 was reportedly less in 2010. urban than Taiwan, Japan, or South Korea and more Furthermore, these figures mayunderstate the com- urbanized than India. The surprise in Figure 41 is that parative dimensions of under-urbanization in China. China was reportedly slightly less urbanized than As mentioned, an apples-to-apples agglomeration

42 NICHOLAS EBERSTADT AND ALEX COBLIN

Figure 42. Urbanization vs. Maddison 1 GDP per Capita, 1950–2010

100

Percentage 40

0 100 1,000 10,000 100,000

China India Indonesia Japan Republic of Korea Taiwan, China

Source: UN Population Division, World Urbanization Prospects: The 2018 Revision, 2018; Maddison Project Database, 2018; and Jutta Bolt et al., Rebasing “Maddison”: New Income Comparisons and the Shape of Long-Run Economic Development, University of Gron- ingen, January 2018, https://www.rug.nl/ggdc/html_publications/memorandum/gd174.pdf. index would help standardize measures of urban pop- perspective. If we so compare China to the rest of the ulation concentration and circumvent the defini- world, China still looks under-urbanized, given its tional difficulties and peculiarities. At the moment, level of development in 2010. we have only international agglomeration index esti- Figures 43–52 can give us a further sense of just mates for 2000. how under-urbanized China appeared to be in 2010. In 2000, India’s estimated agglomeration ratio These figures present simple international correla- was more than 24 percentage points higher (at the tions for 2010, “predicting” urbanization ratios for 50,000 big-city threshold) than its UNPD-estimated countries the world over based on income (per cap- urbanization ratio. For Indonesia, the correspond- ita GDP), educational attainment (estimated mean ing agglomeration index estimate was more than years of schooling), life expectancy at birth, or some 17 points higher than the estimated urbanization combination of these variables, in some cases adding ratio. Japan, South Korea, and Taiwan all likewise gar- our indicators for business climate. By any combi- nered higher agglomeration ratios than urbanization nation of indicators, China’s estimated urbanization ratios for 2000 by this same threshold.62 An objec- ratio always falls short of what would be predicted for tive international measure of urbanization, conse- a society with its other developmental characteris- quently, would almost certainly accentuate China’s tics—often markedly short. under-urbanization in relation to these Asian com- China is not the only country to appear under- parators; the same would probably hold true for 2010. urbanized, of course. Indeed, by definition, Until more work is done on agglomeration indexes, roughly half the countries in the world would look we must rely on the conventional urbanization data under-urbanized in the sorts of ordinary least squares to place China’s 2010 performance in international regressions we are using. India—the world’s other

43 URBANIZATION WITH CHINESE CHARACTERISTICS

Figure 43. Predicted Urbanization Rate (UN Figure 45. Predicted Urbanization Rate Population Division) by World Development (UN Population Division) by Mean Years of Indicators PPP GDP, 2010 Schooling 15+ (Wittgenstein), 2010

100 100

90 90

80 80

70 70

60 60

50 50

40 40 Actual Urbanization Rate Actual Urbanization Rate 30 30 20 20 10

10 0 0102030405060708090100 0 Predicted Urbanization Rate 020406080 100 Rest of World China Linear (Rest of World) Predicted Urbanization Rate Rest of World China Linear (Rest of World) Source: UN Population Division, World Urbanization Prospects: Source: World Bank, World Development Indicators, accessed The 2018 Revision, 2018; and Wittgenstein Centre for Demog- October 2017, https://datacatalog.worldbank.org/dataset/ raphy and Global Human Capital, Wittgenstein Centre Data world-development-indicators; and UN Population Division, Explorer, http://www.wittgensteincentre.org/dataexplorer. World Urbanization Prospects: The 2018 Revision, 2018.

Figure 44. Predicted Urbanization Rate Figure 46. Predicted Urbanization Rate (UN Population Division) by Life Expectancy (UN Population Division) by Mean Years of (World Development Indicators), 2010 Schooling 15+ (Barro-Lee), 2010

100 100

90 90

80 80

70 70

60 60

50 50

40 40 Actual Urbanization Rate Actual Urbanization Rate

30 30

20 20

10 10

0 0102030405060708090100 0 0102030405060708090100 Predicted Urbanization Rate Predicted Urbanization Rate Rest of World China Linear (Rest of World) Rest of World China Linear (Rest of World)

Source: World Bank, World Development Indicators, accessed Source: UN Population Division, World Urbanization Prospects: The October 2017, https://datacatalog.worldbank.org/dataset/ 2018 Revision, 2018; and Robert Barro and Jong-Wha Lee, “A New world-development-indicators; and UN Population Division, Data Set of Educational Attainment in the World, 1950–2010,” World Urbanization Prospects: The 2018 Revision, 2018. Journal of Development Economics 104 (2013): 184–98.

44 NICHOLAS EBERSTADT AND ALEX COBLIN

Figure 47. Predicted Urbanization Rate Figure 48. Predicted Urbanization Rate (UN Population Division) by Mean Years of (UN Population Division) by Mean Years Schooling 15+ (Wittgenstein) and World of Schooling 15+ (Barro-Lee) and World Development Indicators GDP PPP, 2010 Development Indicators GDP PPP, 2010

100 100

90 90

80 80

70 70

60 60 50

50 40 Actual Urbanization Rate 40 30 Actual Urbanization Rate 30 20

10 20

0 10 0102030405060708090 100 Predicted Urbanization Rate 0 Rest of World China Linear (Rest of World) 0102030405060708090100 Predicted Urbanization Rate Source: World Bank, World Development Indicators, accessed Rest of World China Linear (Rest of World) October 2017, https://datacatalog.worldbank.org/dataset/ world-development-indicators; UN Population Division, World Source: World Bank, World Development Indicators, accessed Urbanization Prospects: The 2018 Revision, 2018; and Wittgen- October 2017, https://datacatalog.worldbank.org/dataset/ stein Centre for Demography and Global Human Capital, Witt- world-development-indicators; UN Population Division, World genstein Centre Data Explorer, http://www.wittgensteincentre. Urbanization Prospects: The 2018 Revision, 2018; and Robert org/dataexplorer. Barro and Jong-Wha Lee, “A New Data Set of Educational Attain- ment in the World, 1950–2010,” Journal of Development Eco- nomics 104 (2013): 184–98. demographic giant—also looks to be under-urbanized We do not wish to assign too much import to this in relation to its level of development. But China’s finding—but it seems interesting enough at least to low level of urbanization cannot be explained by its merit mention. population size or population density. Just how under-urbanized was China in 2010? We For reasons already explained, we are reluctant can use simple cross-sectional data for that year to to analyze time series data on Chinese urbanization. offer a first approximation. If we “predict” urbaniza- However, if we were to include these in an internation ratios using education and per capita GDP with tional analysis of postwar urbanization ratios, China’s PPP adjustments, China’s reported level looks to be reported levels would be far lower than for countries 7.8 percent lower than expected. If we use the same with comparable levels of health and education, even approach but instead substitute exchange-rate-based after taking population size, population density, or per capita GDP figures (i.e., without PPP adjustments), arable land per capita into account. The “China effect” the estimated gap would be about 5.3 percent.64 in fact looks to be much larger than the “India effect” In 2010, China’s enumerated population was or the “Indonesia effect,” even though these other 1.33 billion. Our simple estimates for under- two populous and densely populated countries also urbanization in China that year would imply that look distinctly under-urbanized in such analyses.63 China should have 70.1–103.6 million more urban

45 URBANIZATION WITH CHINESE CHARACTERISTICS

Figure 49. Predicted Urbanization Rate Figure 50. Predicted Urbanization Rate (UN Population Division) by Life Expectancy (UN Population Division) by Life Expectancy (World Development Indicators), Mean Years (World Development Indicators), Mean Years of Schooling 15+ (Wittgenstein), and World of Schooling 15+ (Barro-Lee), and World Development Indicators GDP PPP, 2010 Development Indicators GDP PPP, 2010

100 100

90 90

80 80

70 70

60 60

50 50

40 40 Actual Urbanization Rate

Actual Urbanization Rate 30 30 20 20 10

10 0 0102030405060708090100 0 0102030405060708090100 Predicted Urbanization Rate Predicted Urbanization Rate Rest of World China Linear (Rest of World)

Rest of World China Linear (Rest of World)

Source: World Bank, World Development Indicators, accessed Source: World Bank, World Development Indicators, accessed October 2017, https://datacatalog.worldbank.org/dataset/ October 2017, https://datacatalog.worldbank.org/dataset/ world-development-indicators; UN Population Division, World world-development-indicators; UN Population Division, World Urbanization Prospects: The 2018 Revision, 2018; and Wittgen- Urbanization Prospects: The 2018 Revision, 2018; and Robert stein Centre for Demography and Global Human Capital, Witt- Barro and Jong-Wha Lee, “A New Data Set of Educational Attain- genstein Centre Data Explorer, http://www.wittgensteincentre. ment in the World, 1950–2010,” Journal of Development Eco- org/dataexplorer. nomics 104 (2013): 184–98.

inhabitants if it were a “typical” country, given its has been—an under-urbanized society. And inter- income and educational levels. Rounding off to avoid estingly enough, our simple calculations are keeping false precision, we can call this range 70–100 million. with the Chinese government’s own assessments. In We are hardly the first to suggest that contem- 2014, when Beijing unveiled its “National New-Type porary China’s development has been marked by Urbanization Plan (2014–2020)” (i.e., urbanization under-urbanization. To the contrary, Chinese schol- drive), authorities declared that China’s urbanization ars have been discussing this topic for years. Professor ratio in 2013 was 6.3 percentage points below those of Zhang Li of Fudan University, for example, published other countries with similar income levels (53.7 per- an entire book in English in 2003 on what he termed cent versus 60 percent).66 That official Chinese esti- “China’s limited urbanization,” which he described as mate is almost exactly the midpoint of our calculated the paradox of “high industrial growth without corre- range (5.3–7.8 percent). sponding growth of urban population.”65 The subject If our range of 70–100 million “missing” urban- is a matter of not only academic studies but also pol- ites in China in 2010 proved to be roughly correct, icy research. migrant family separation alone could not account Indeed, Beijing’s current urbanization drive is for that discrepancy. Of China’s officially counted predicated on the understanding that China is—or migrant population in the 2010 census, 223 million

46 NICHOLAS EBERSTADT AND ALEX COBLIN

Figure 51. Predicted Urbanization Rate Figure 52. Predicted Urbanization Rate (UN Population Division) by Life Expectancy (UN Population Division) by Life Expectancy (World Development Indicators), Mean Years (World Development Indicators), Mean Years of Schooling 15+ (Wittgenstein), World of Schooling 15+ (Wittgenstein), World Development Indicators GDP PPP, and Wall Development Indicators GDP PPP, and Fraser, Street Journal/Heritage, 2010 2010

100 100

90 90

80 80

70 70

60 60

50 50

40 40 Actual Urbanization Rate

30 Actual Urbanization Rate 30

20 20

10 10

0 0102030405060708090100 0 0102030405060708090100 Predicted Urbanization Rate Predicted Urbanization Rate Rest of World China Linear (Rest of World) Rest of World China Linear (Rest of World) Source: World Bank, World Development Indicators, accessed October 2017, https://datacatalog.worldbank.org/dataset/ Source: World Bank, World Development Indicators, accessed world-development-indicators; UN Population Division, World October 2017, https://datacatalog.worldbank.org/dataset/ Urbanization Prospects: The 2018 Revision, 2018; Wittgenstein world-development-indicators; UN Population Division, World Centre for Demography and Global Human Capital, Wittgen- Urbanization Prospects: The 2018 Revision, 2018; Wittgen- stein Centre Data Explorer, http://www.wittgensteincentre.org/ stein Centre for Demography and Global Human Capital, Witt- dataexplorer; and Heritage Foundation and Wall Street Journal, genstein Centre Data Explorer, http://www.wittgensteincentre. Index of Economic Freedom, http://www.heritage.org/index. org/dataexplorer; and Fraser Institute, Economic Freedom Net- work, http://www.freetheworld.com/. migrants were reportedly of working age (15–65). If 70–100 million range—all the more when we recall the dependency ratio in China in 2010 were the same that many left-behind children are from urban areas for these migrants as for the nation as a whole, Chi- and therefore are already counted in urban popula- na’s total migrant population in 2010 would have been tion totals, whereas the 70–100 million figure would 301 million—that is, about 40 million higher than the refer to the prospective net urban population deficit. migrant total reported. But that gap falls far short of We can tentatively surmise that hukou policy was our 70–100 million number. a main factor responsible for suppressing urban pop- As is well-known, contemporary China contends ulation growth in China in 2010—through not only with a “left-behind children” problem because mov- separating migrant families but also discouraging or ing entire families is difficult under the strictures of preventing tens of millions of additional would-be the current hukou regimen. Working off the 2010 PRC migrants from moving out of the countryside. Clearly, census, UNICEF has estimated there were as many reconstituting migrant families in their new locales, as 61 million left-behind children in China in 2010, of which is difficult under currenthukou rules, would which 29 million were living with neither father nor increase China’s urban population. But that alone mother.67 But such numbers also fall far short of our would not make up the 70–100 million population

47 URBANIZATION WITH CHINESE CHARACTERISTICS

Figure 53. Predicting per Capita GDP PPP with Life Expectancy, Urbanization, Education, and Index of Economic Freedom, 1970–2010

100,000

10,000

1,000 ctual GDP per Capita, PPP A Constant 2005 International $) ( 100 100 1,000 10,000 100,000 Predicted GDP per Capita, PPP (Constant 2005 International $)

ln(GDP per Cap PPP) = 0.047 (Life Expectancy) + 0.022 (Percent Urban) + 0.080 (IIASA 15+ MYS) + 0.070 (Fraser EFI) + 3.343 (20.68) (23.77) (9.66) (4.60) (29.00) Adj R-Squared: 0.8616 Number of Observation: 1,202

Note: Variables have a five-year lag. Source: World Bank, World Development Indicators, accessed July 31, 2013, http://data.worldbank.org/data-catalog/world- development-indicators; UN Population Division, World Urbanization Prospects, http://esa.un.org/unpd/wup/CD-ROM/Default. aspx; International Institute for Applied Systems Analysis, accessed July 31, 2012, http://www.nationsencyclopedia.com/WorldStats/ Edu-iiasa-mean-years-schooling-15.html; Fraser Institute, Economic Freedom Network, http://www.freetheworld.com/; and Heritage Foundation and Wall Street Journal, Index of Economic Freedom, http://www.heritage.org/index. increase necessary to bring China’s 2010 urban popu- Elsewhere we have delved into the interplay lation up to the level one expects of a country of com- between per capita output on the one hand and parable population size and economic performance. human resources and business climate on the other.68 Yet the costs of under-urbanization in contempo- Indeed, we believe we have developed a compact and rary China are not measured only in human terms, surprisingly powerful model for explaining and esti- such as the suffering of involuntarily separated fam- mating postwar differences in per capita GDP between ilies or the thwarted aspirations of would-be rural countries at any point in time—and within countries migrants to urban areas. Under-urbanization entails over time. Our model uses health (represented by life a loss of national economic potential for China as expectancy at birth), education (represented by mean well—and a considerable one. years of schooling for the 15+ population), social com- Estimating the economic implications of Chi- plexity (represented by urbanization rates), and busi- na’s apparent under-urbanization is the final aspect ness climate or quality of institutions and policies in this report of placing Chinese urbanization in an (represented by the aforementioned indexes of eco- international perspective. One way to do this is by nomic freedom). This simple model can be seen in calculating the international “value” of urbanization Figures 53 and 54. as revealed by postwar patterns of economic develop- Using just these four variables, we can explain ment around the world. roughly 85 percent of the differences in per capita

48 NICHOLAS EBERSTADT AND ALEX COBLIN

Figure 54. Predicting Global per Capita GDP PPP with Life Expectancy, Urbanization, Education, and Index of Economic Freedom, 1995–2012 100,000

10,000

1,000 ctual GDP per Capita, PPP A Constant 2005 International $) (

100 100 1,000 10,000 100,000 Predicted GDP per Capita, PPP (Constant 2005 International $)

ln(GDP per cap PPP) = 0.047 (Life Expectancy) + 0.019 (Percent Urban) + 0.100 (IIASA 15+ MYS) + 0.016 (Heritage EFI) + 2.783 (24.20) (21.50) (15.20) (10.57)(24.81) Adj R-Squared: 0.8712 Number of Observation: 1,434

Note: Variables have a five-year lag. Source: World Bank, World Development Indicators, July 31, 2013, http://data.worldbank.org/data-catalog/world-development- indicators; UN Population Division, World Urbanization Prospects, http://esa.un.org/unpd/wup/CD-ROM/Default.aspx; International Institute for Applied Systems Analysis, accessed July 31, 2013, http://www.nationsencyclopedia.com/WorldStats/Edu-iiasa-mean- years-schooling-15.html; Fraser Institute, Economic Freedom Network, http://www.freetheworld.com/; and Heritage Foundation and Wall Street Journal, Index of Economic Freedom, http://www.heritage.org/index.

GDP within and between countries for the decades education, and business climate into account. In fact, since 1970.69 Note further that the predictive power the independent contribution of a 1 percentage point of the model is attained using lagged variables (that increase in urbanization looks to be roughly a 2 per- is, using our independent variables for a given year to centage point increase in overall GDP per capita. estimate the per capita output for a country five years This should be no surprise: There are many rea- later). This lagged-variable effect means we are not sons to expect urbanization inherently stimulates just dealing with simultaneous correlations. Although development. What matters here is that our analy- econometric analysis can never prove causation, sis can quantify this statistically. If China’s degree of we nevertheless have insight into causation in the under-urbanization in 2010 were accurately bounded dynamic connecting human resources, business cli- by the range in our estimates, our simple models mate, and economic outcome in our modern world.70 would suggest that under-urbanization was costing The simple regression models presented in Figures China around 10–15 percent in forgone GDP around 53 and 54 demonstrate one particular and important 2010—not an inconsiderable sum. These same mod- facet of the modern development dynamic: This is the els would suggest that under-urbanization in 2010 independent contribution of urbanization to income would continue to exact a cost in terms of reduced growth. Our urbanization variable strongly tracks national productivity for some years into the future with higher GDP per capita, even after taking health, as well.

49 URBANIZATION WITH CHINESE CHARACTERISTICS

Beijing’s urbanization drive is informed by this So-called second-tier cities have more relaxed migra- economic reality. And eliciting untapped growth tion rules compared to first-tier cities (such as Bei- potential from migration and urbanization looks jing and Shanghai), where the hukou system is strictly increasingly important as the tempo of growth con- enforced. tinues to slow (albeit from stratospheric heights) Migration from the countryside, in turn, looks to be and generating growth from other sources becomes the main source of urban population growth for China increasingly challenging. today—and virtually the only source in the decades For example, ghost cities can be seen as a prod- ahead. Given the realities of the hukou system and of uct of under-urbanization, as measured by the the tremendous existing differences in education and movement of people because of hukou regime, and skills between city and country in China today, this over-urbanization, as measured by urban housing. will make for political and economic complications. Therefore, we can see ghost cities as a manifestation Some of these complications have likely been envi- of the dangerous overstock of real estate, which prob- sioned in advance by Chinese authorities. But others ably is one of the motivations behind state encour- may not have been. In the following sections, we will agement of rural-urban migration. The urbanization cast light on some of these prospective problems. drive is not uniformly applied across Chinese cities.

50 VI. The Arithmetic (and Politics) of Migration and Urbanization with Chinese Characteristics

n this section, we examine China’s population population of urban and rural China (Figures 56 and Istructure, as reported in the 2010 census, by resi- 57). China’s total urban population is composed of dential and migration status. Our analysis results in “cities”—that is, major urban agglomerations—and two significant findings. First, it would appear that “towns”—that is, smaller urban centers. We pres- urban China is no longer capable of maintaining, let ent the reported residential population structure for alone increasing, its future manpower numbers on these separate types of urban concentrations as well the basis of “town” and “city” residents legally living (Figures 58 and 59). We then present the reported in their own hukou. To sustain—much less increase— population structure for China’s total migrant pop- urban manpower in the years immediately ahead, a ulation (Figure 60) and, finally, the population continued influx of migrants is therefore absolutely structures reported for the nonmigrant residents of essential. But such new migrants, under current urban China, China’s cities, China’s towns, and rural arrangements in China, are not in general lawful res- China (Figures 61–64). (We also refer to rural China idents of their new urban abodes—and this brings us as “left-behind rural China.”71) to our second finding. The data from Figure 55 show some curiosities. Migrants already account for a sizable fraction of One of these is the surprisingly low sex ratio for the China’s urban manpower. Indeed, in the big cities, 25–34 group (103 males per 100 females)—an oddly migrants already constituted a majority of youthful low ratio, especially considering that men and women manpower in 2010—and the share of such migrants in in their mid and late 20s in the 2010 census count had urban centers was likely higher in 2015 than it was five been born under the shadow of the One-Child Pol- years earlier. Urban China has thus been transformed icy, which ushered in a sharp increase in sex ratios at into a two-tier society—legal hukou residents and birth. (We commented on these discrepancies in Sec- provisional or de jure illegal migrants. And the cur- tion II.) Despite such curiosities, we take these data as rent second-class residents may constitute a majority “given” in our analysis. of urban cohorts in many urban venues, irrespective We should remember that Figures 61–64 present of the urbanization drive and its announced plan to China’s population according to their 2010 residential/ grant new local hukou to many millions of out-of- migration status. They do not indicate the residential hukou migrants in towns and cities every year. or hukou origins of those who have migrated. We do Figures 55–64 display the main aspects of China’s not have access to those data through the materials urban and rural population structure, as well as the available for our analysis. population structure for China’s migrant popula- Fortunately, Zheng Zhenzhen and Yang Ge of the tion, from the 2010 PRC census. In these successive Chinese Academy of Social Sciences Institute of Pop- snapshots, we see the reported structure for Chi- ulation and Labor Economics have done this.72 Their na’s total population (Figure 55) and the residential work underscores an important point about the

51 URBANIZATION WITH CHINESE CHARACTERISTICS

Figure 55. Population Structure for China: Nationwide, 2010

100+ Population (Millions) 1,332.8 96 Median Age35 92 88 Percentage Age 65+8.9% 84 Percentage Age 15–6474.5% 80 76 Percentage Age 0–14 16.6% 72 Ratio Age 55–64 to Age 15–2462:100 68 Male-to-Female Sex Ratio105:100 64 60 Male-to-Female Ratio Age 25–34103:100 56 52

Age 48 44 40 36 32 28 24 20 16 12 8 4 0 15 10 50510 15 Population (Millions) Male Female

Figure 56. Population Structure for China: Urban (Cities and Towns), 2010

100+ Population (Millions) 670.0 95 90 Median Age 35 85 Percentage Age 65+ 7.8% 80 Percentage Age 15–64 78.1% 75 Percentage Age 0–14 14.1% 70 Ratio Age 55–64 to Age 15–24 54:100 65 60 Male-to-Female Sex Ratio 105:100 55 Male-to-Female Ratio Age 25–34 102:100 50 Age 45 40 35 30 25 20 15 10 5 0 864202468 Population (Millions) Male Female

52 NICHOLAS EBERSTADT AND ALEX COBLIN

Figure 57. Population Structure for China: Rural (“Counties”), 2010

100+ Population (Millions) 662.8 95 90 Median Age 36 85 Percentage Age 65+ 10.1% 80 Percentage Age 15–64 70.8% 75 Percentage Age 0–14 19.2% 70 Ratio Age 55–64 to Age 15–24 75:100 65 60 Male-to-Female Sex Ratio 105:100 55 Male-to-Female Ratio Age 25–34 104:100 50

Age 45 40 35 30 25 20 15 10 5 0 864202468 Population (Millions) Male Female

Figure 58. Population Structure for China: Cities, 2010

100+ Population (Millions) 403.8 95 90 Median Age 35 85 Percentage Age 65+ 7.7% 80 Percentage Age 15–64 80.1% 75 Percentage Age 0–14 12.2% 70 Ratio Age 55–64 to Age 15–24 65 48:100 60 Male-to-Female Sex Ratio 105:100 55 Male-to-Female Ratio Age 25–34 103:100 50

Age 45 40 35 30 25 20 15 10 5 0 6420246 Population (Millions) Male Female

53 URBANIZATION WITH CHINESE CHARACTERISTICS

Figure 59. Population Structure for China: Towns, 2010

100+ Population (Millions) 266.2 95 90 Median Age 35 85 Percentage Age 65+ 8.0% 80 Percentage Age 15–64 75.1% 75 Percentage Age 0–14 16.9% 70 65 Ratio Age 55–64 to Age 15–24 55:100 60 Male-to-Female Sex Ratio 105:100 55 Male-to-Female Ratio Age 25–34 100:100 50 Age 45 40 35 30 25 20 15 10 5 0 432101234 Population (Millions) Male Female

Figure 60. Population Structure for China: National Migrant Population, 2010

100+ 95 Population (Millions) 260.9 90 Median Age 30 85 Percentage Age 65+ 3.8% 80 Percentage Age 15–64 85.5% 75 Percentage Age 0–14 10.7% 70 65 Ratio Age 55–64 to Age 15–24 22:100 60 Male-to-Female Sex Ratio 110:100 55 Male-to-Female Ratio Age 25–34 106:100 50 Age 45 40 35 30 25 20 15 10 5 0 864202468 Population (Millions) Male Female

54 NICHOLAS EBERSTADT AND ALEX COBLIN

Figure 61. Chinese Urban Population (Without Migrants), 2010

100+ Population (Millions) 444.0 95 90 Median Age 38 85 Percentage Age 65+ 9.85% 80 Percentage Age 15–64 74.4% 75 Percentage Age 0–14 15.74% 70 Ratio Age 55–64 to Age 15–24 80:100 65 60 Male-to-Female Sex Ratio 102:100 55 Male-to-Female Ratio Age 25–34 98:100 50

Age 45 40 35 30 25 20 15 10 5 0 864202468 Population (Millions) Male Female

Figure 62. Chinese City Population (Without Migrants), 2010

100+ Population (Millions) 233.3 95 Median Age 39 90 85 Percentage Age 65+ 10.5% 80 Percentage Age 15–64 75.6% 75 Percentage Age 0–14 13.8% 70 Ratio Age 55–64 to Age 15–24 86:100 65 60 Male-to-Female Sex Ratio 101:100 55 Male-to-Female Ratio Age 25–34 98:100 50 Age 45 40 35 30 25 20 15 10 5 0 6420246 Population (Millions) Male Female

55 URBANIZATION WITH CHINESE CHARACTERISTICS

Figure 63. Chinese Town Population (Without Migrants), 2010

100+ 95 Population (Millions) 210.7 90 Median Age 37 85 Percentage Age 65+ 9.11% 80 Percentage Age 15–64 73.04% 75 Percentage Age 0–14 17.85% 70 65 Ratio Age 55–64 to Age 15–24 73:100 60 Male-to-Female Sex Ratio 104:100 55 Male-to-Female Ratio Age 25–34 98:100 50 Age 45 40 35 30 25 20 15 10 5 0 432101234 Population (Millions) Male Female

Figure 64. Chinese Rural Population (Without Migrants), 2010

100+ Population (Millions) 627.8 95 90 Median Age 37 85 Percentage Age 65+ 10.4% 80 Percentage Age 15–64 69.9% 75 Percentage Age 0–14 19.7% 70 Ratio Age 55–64 to Age 15–24 79:100 65 60 Male-to-Female Sex Ratio 104:100 55 Male-to-Female Ratio Age 25–34 104:100 50 Age 45 40 35 30 25 20 15 10 5 0 864202468 Population (Millions) Male Female

56 NICHOLAS EBERSTADT AND ALEX COBLIN

260-million-plus persons counted as migrants in the 20 percent of the total reported residential urban pop- 2010 census: Not all of them are peasants coming to ulation—even though de jure unauthorized migrants the cities.73 accounted for a much larger share of urban popula- By their estimates, over a third (34.4 percent) of tion in general, and of big-city population in particu- all reported migration in 2010 was urban to urban, lar, as we shall see below. and about a tenth (10.3 percent) was rural to rural. A cursory visual review of Figures 65–74 underscores There was even some reported migration from cit- how different the demographic structures of these ies to the countryside (accounting for 2.6 percent 10 subpopulations were as of 2010. The most distinct of the reported total). But most migration was rural population structure, not surprisingly, is for migrants: to urban (52.7 percent), meaning that net rural-to- This group is composed overwhelmingly (over urban migration accounted for just over half 85 percent) of working-age (15–64) men and women. (50.1 percent) of the total flow of reported migration The median age for this group is 30 years—five years in the 2010 census. That would amount to just over lower than for China as a whole—and among migrants 130 million people. there are only 22 prospective workers age 55–64 for In other words, if China’s 2010 population had been every 100 men and women age 15–24, as opposed to forced to live in their officially designatedhukou loca- a 62:100 ratio for the country as a whole. Migrants tion, 261 million people would have been displaced furthermore tend to skew even more male than from the locales where they were counted—but Chi- gender-imbalanced China does as a whole; the sex na’s rural population would have been augmented by ratio for migrants is 110 males for every 100 females, 131 million, and China’s urban areas reduced by that versus an all-China sex ratio of 105:100. same amount. Net population transfer to urban areas Because China’s 2010 migrants overwhelmingly via migration as of 2010 thus accounted for about (87 percent) are located in urban areas, urban China

Figure 65. Legal Resident vs. Illegal Migrant: Nationwide, 2010

100+ 95 90 85 80 75 70 65 60 55 50 Age 45 40 35 30 25 20 15 10 5 0 15 10 5051015 Population (Millions) Resident Migrant Male Migrant Female

57 URBANIZATION WITH CHINESE CHARACTERISTICS

Figure 66. Legal Resident vs. Illegal Migrant: Reported Composition of Working-Age (15–64) Manpower, Nationwide, 2010

40 Age 35

15 15 10 50510 15 Population (Millions) Resident Migrant Male Migrant Female

Figure 67. Legal Resident vs. Illegal Migrant: China (Villages), 2010

100+ 95 90 85 80 75 70 65 60 55 50

Age 45 40 35 30 25 20 15 10 5 0 864202468 Population (Millions) Resident Migrant Male Migrant Female

58 NICHOLAS EBERSTADT AND ALEX COBLIN

Figure 68. Legal Resident vs. Illegal Migrant: Reported Composition of Working-Age (15–64) Manpower, China (Villages), 2010

40 Age 35

15 864202468 Population (Millions) Resident Migrant Male Migrant Female

Figure 69. Legal Resident vs. Illegal Migrant: Urban China (Cities and Towns), 2010

100+ 95 90 85 80 75 70 65 60 55 50 Age 45 40 35 30 25 20 15 10 5 0 8327 Population (Millions) Resident Migrant Male Migrant Female

59 URBANIZATION WITH CHINESE CHARACTERISTICS

Figure 70. Legal Resident vs. Illegal Migrant: Reported Composition of Working-Age (15–64) Manpower, Urban China (Cities and Towns), 2010

40 Age 35

15 8327 Population (Millions) Resident Migrant Male Migrant Female

Figure 71. Legal Resident vs. Illegal Migrant: China (Towns), 2010

100 95 90 85 80 75 70 65 60 55 50 Age 45 40 35 30 25 20 15 10 5 0 42024 Population (Millions) Resident Migrant Male Migrant Female

60 NICHOLAS EBERSTADT AND ALEX COBLIN

Figure 72. Legal Resident vs. Illegal Migrant: Reported Composition of Working-Age (15–64) Manpower, China (Towns), 2010

40 Age 35

15 42024 Population (Millions) Resident Migrant Male Migrant Female

Figure 73. Legal Resident vs. Illegal Migrant: China (Cities), 2010

100+ 95 90 85 80 75 70 65 60 55 50

Age 45 40 35 30 25 20 15 10 5 0 42024 Population (Millions) Resident Migrant Male Migrant Female

61 URBANIZATION WITH CHINESE CHARACTERISTICS

Figure 74. Legal Resident vs. Illegal Migrant: Reported Composition of Working-Age (15–64) Manpower, China (Cities), 2010

40 Age 35

15 42024 Population (Millions) Resident Migrant Male Migrant Female

in 2010 was, perhaps surprisingly, more youthful momentum for population decline is revealed. In 2010, than rural China—and in some important respects, among big-city legal residents, no age group was as considerably more youthful. Whereas rural China numerous as men and women born in 1963—that is, counted 75 persons of older working age (55–64) for persons then age 47 (babies born immediately after the every 100 persons of younger working age (15–24), Three Lean Years famine brought on by the disastrous the corresponding ratio in urban China was 51:100. Great Leap Forward). Among legal residents of the big Despite longer lives and markedly lower fertility, cities, for every 100 persons age 47, there were only urban China’s proportion of people age 65 and over 58 persons 20 years younger (age 27), and for every was distinctly lower than rural China’s as of 2010. 100 persons age 27, there were only 56 persons 20 years The effect is particularly notable in China’s biggest younger than them (age 7). Similar tendencies, although cities (“cities”), where 80 percent of the total pop- not quite as acute, are likewise apparent among legal ulation (including migrants) were of prospective residents of the smaller cities (“towns”). working age (15–64) and where there were more than Are China’s urban areas, absent migration, facing two persons age 15–24 for each person age 55–64. a severe depopulation? Is the legal in-hukou urban But the picture changes radically when migrants population about to become a “net mortality” soci- are taken out of China’s “city” tableau. If only legal ety? These are important questions, yet answering in-hukou residents in the big cities are counted, the them simply and conclusively is impossible with any median age for that subgroup jumps to 39 years (up readily available Chinese demographic data. China’s from 35 years for big-city population overall); the ratio annual birth and death data are still problematic for of 55- to 64-year-olds to 15- to 24-year-olds nearly dou- the country as a whole and, perhaps surprisingly, for bles (86:100); and a society with tremendous built-in urban areas.

62 NICHOLAS EBERSTADT AND ALEX COBLIN

Figure 75. Projection of China’s Total Population Assuming Zero Net Urban Migration, 2010–40

1.4 Total China

1.2

1.0

0.8 Rural China

Billions Urban China 0.6

0.4 Urban China (Without Migrants)

0.2

0 2012 2015 2018 2021 2024 2027 2030 2033 2036 2039

Note: For methodology and assumptions in these projections, see accompanying text. Source: National Bureau of Statistics, Department of Population and Employment Statistics, “China Population Census: Tabulation of the 2010 Population Census of the People’s Republic of China,” 2012; and UN Population Division, World Population Prospects 2015, http://esa.un.org/unpd/wpp/index.htm.

Figure 76. Projection of China’s Working-Age (15–64) Population Assuming Zero Net Urban Migration, 2010–40

1.4 Total China 0.9

0.8

0.7

0.6 Rural China 0.5

Billions Urban China 0.4

0.3 Urban China (Without Migrants) 0.2

0.1

0 2012 2015 2018 2021 2024 2027 2030 2033 2036 2039

Note: For details of methodology and assumptions, see accompanying text. Source: National Bureau of Statistics, Department of Population and Employment Statistics, “China Population Census: Tabulation of the 2010 Population Census of the People’s Republic of China,” 2012; and UN Population Division, World Population Prospects 2015, http://esa.un.org/unpd/wpp/index.htm.

63 URBANIZATION WITH CHINESE CHARACTERISTICS

Lacking these data, we can still offer a formalized Under these projections, China’s total popula- and internally consistent assessment of the urban tion peaks around 2028. However, urban population demographic prospect by conducting a simple set begins to decline in 2021, and urban nonmigrant of population projections for the nonmigrant pop- population begins to decline around 2018—that is, ulations of urban China based on the 2010 census at this writing. results. We do so using the demographic software More portentous is the intrinsic change of the age package DemProj contained in Spectrum 5, a version 15–64 population—that is, the population of those of the Futures Group’s widely used population pro- conventionally defined as working age. Under these jection software.74 Figures 75 and 76 fill out and refine projections, in-hukou, urban, working-age man- estimates from the previous quarterly report con- power would have already begun to decline; nonmi- cerning the inherent momentum in urban and rural grant, urban manpower would begin to fall in 2013; populations in China as of 2010. and rural, working-age manpower would peak around Using Spectrum demographic software, these pro- 2025 before starting to decline around 2030. In other jections illustrate what urban and rural Chinese pop- words, by these projections, the only thing keeping ulation trends would look like in 2010–40 if: China’s urban, working-age manpower from shrink- ing at the moment may be migration. (1) Official China 2010 data on urban and rural Note that declining manpower totals do not nec- population structures were correct; essarily mean declining workforce totals. Workforce can increase while manpower declines if labor force (2) UN Population Division (UNPD) estimates participation rates increase. But they cannot do so of total fertility rates for China from 2010 to forever. Urban China now appears to be an “eater 2040 from WPP 2017 (medium fertility vari- of men”; it cannot maintain its manpower without ant) were correct, with a positive differential of steady inflows from outside. Soon this will be true for 0.25 births per woman separating rural China the total population as well. from the national average, a negative differ- Thus, to repeat: We may have already reached the ential for urban China, and sex ratios at birth juncture where urban and big-city China are entirely (SRBs) for urban and rural areas comported to dependent on the influx of migrant labor for any fur- the national SRBs that the UNPD projects in ther increases in their total working-age manpower. WPP 2017; Urban China has been the engine of China’s phenomenal and sustained burst of economic growth (3) These fertility differentials were held constant over the past generation and more. But the tempo of until 2040; urban-driven economic growth depended on fueling the urban engine with a growing supply of would-be (4) WPP 2017 projections for life expectancy to workers from the countryside. That supply is tighten- 2040 were accurate—except life expectancy at ing up in purely numerical terms. birth was a year lower than the national average We address education and skills of prospective in rural areas for both males and females, and rural migrants in the following section. But now conversely a year higher in urban areas; and we need to examine the sociopolitical implications of the migration that has helped enable what many (5) China experienced zero rural-urban migration have called modern China’s development miracle. and zero international migration after 2010— While this tremendous movement of people has so that population change in urban and rural brought both national and personal economic ben- China in 2010–40 would be due only to natural efits, it is also directly accompanied by social prob- increase (births versus deaths). lems (and potentially political tensions). This is so, quite simply, because the migration that has helped

64 NICHOLAS EBERSTADT AND ALEX COBLIN

sustain China’s remarkable economic transformation Up to 2010, the influx of migrants to urban locali- is almost entirely outside the law.75 ties (including migrants from other urban localities) In effect, the overwhelming majority of migrants shifted the overall balance between in-hukou urban tabulated in the 2010 census were, in a sense, illegal residents and out-of-hukou newcomers quite rap- aliens in their own land. The hundreds of millions idly. In 2005, for example, migrants accounted for of people in question were as a rule not included 39 percent of the urban population for ages 20–39; in social services in their new locales: health, edu- by 2010, just five years later, the share had risen to cation, housing, pensions, and the like.76 They typi- 44 percent. cally could not bring their families with them to their With the promulgation of the official urbaniza- new places of residence, resulting in a growing army tion drive in 2014, and its promise to reassign hukou of left-behind children and elderly relatives.77 They for tens of millions of migrants to that of their actual were quite typically exploited in the labor market— place of urban employment and residence, we may compensated less than legal residents with the same have already seen the high-water mark for the ratio skills and education levels—though the degree of of “outside” workers in urban China. However, we such exploitation varies and may have lessened over cannot actually calculate those trends for the period time.78 They were, and still are, discriminated against since 2010 with any precision, given our lack of access in other ways as well, so that the situation in their to new census or mini-census data. unauthorized places of residence remains generally We can get some sense of the trends underway, contingent and precarious. though, from the crude metric of the floating popula- With this background in mind, we contrast the tion to total urban population—an obviously imper- legally authorized in-hukou population structures fect measure, but an informative approximation for with the out-of-hukou migrant population struc- our purposes nonetheless.79 In both 2000 and 2005, tures from the 2010 census for men and women of that ratio was about 26 percent. It rose to 33 per- all ages, men and women of conventional working cent in 2010. It reached 34 percent in 2014, the year ages (15–64), China as a whole, China’s rural areas the urbanization drive was launched, and by 2016— (villages), China’s urban areas (towns and cities), the most recent year available at this writing—it was China’s smaller cities (towns), and China’s big cit- about 31 percent. ies (cities). (See Figures 68, 70, 72, and 74.) We can If Beijing’s 2014 urbanization policy proceeds see from these figures that cities, particularly big cit- according to plan, the proportion of national urban ies, have been magnets for China’s migrants. We may residents without local hukou would fall between further note that migrant populations—which is to 2014 and 2020—from 16.9 percent to 15 percent. In say, de jure illegal residents—have come to account 2010, roughly one in three urban residents was out of for absolute majorities across urban China for some hukou; under the new plan, the corresponding frac- age groups. tion in 2020 would perhaps be something like one As of 2010, migrants outnumbered legal residents in four. Implicitly, the plan envisions that more than of China’s big cities among males age 16–35 and for 200 million Chinese will still be living out of hukou in every year of age within this grouping. Among females, 2020—and this is if the plan meets its currently envi- out-of-hukou residents predominate for each year of sioned targets. Suffice it to say that vast populations age throughout the 16–31 age groups. For urban China of de jure illegal migrants—some of them reaching as a whole in 2010, out-of-hukou residents outnum- near majorities or actual majorities of key age groups bered in-hukou residents at each year of age for the in many urban centers—look to be very much part of 18–21 age group among males and for the 16–21 age Beijing’s vision of the country’s future for as far as the group among females. urban planner’s eye can see.

65 VII. Educational Attainment: The Urban-Rural Divide, Migration, and Development

his section addresses educational attainment Recall that these data are based on actual place of Tin relation to migration, urbanization, and eco- residence, not officially designatedhukou residence; nomic development. the reported profiles for urban China include nonle- Skills and education are crucial elements of human gal rural migrants already residing in those locales. In capabilities for economic performance. The prospec- other words, these figures include the130-million -plus tive contribution of migration to economic develop- people who were born (and in large part, educated) in ment in China depends on many considerations and the countryside but now reside in China’s urban areas factors, some of them devilishly difficult to measure as a component in urban China’s educational profile. objectively, such as business climate or the quality of Table 7 quantifies some of the trends in Fig- policies and institutions. Likewise, skills imparted by ures 77–81. Note, for example, that the proportion education per se may not be measured easily or well. of working-age (15–64) Chinese villagers with some But 2010 PRC census data do avail us of information tertiary education or more in 2010 (2.6 percent) was on educational attainment, and these afford import- just about the same as the national average for China ant insights, not least into differences in educational in 1990, 20 years earlier. And it was barely an eighth profiles within China. the level of their urban counterparts and one-tenth Figures 77–81 display population “pyramids” by the level of their big-city counterparts. By the same age, sex, and educational attainment for the pop- token, rural men age 20–39 with no more than a pri- ulation age 6–85+ for China as a whole, rural China mary education accounted for one-sixth of all such (villages), urban China (towns and cities), and Chi- men in China in 2010—three times the share for their na’s towns and cities separately, as reported in urban counterparts and four times the share for male the 2010 PRC census. The “no schooling” group is 20- to 39-year-olds in big cities. In 2010, just 4.4 per- self-explanatory. “Primary schooling” refers to peo- cent of China’s rural women in their 20s and 30s had ple with some schooling, but no more than a primary any tertiary education; the corresponding propor- education. “Secondary schooling” includes those with tion was over 28 percent for women that same age in more than a primary education but no tertiary educa- urban areas and nearly 35 percent in the big cities. tion, and “tertiary education” refers to people with at The patterns of educational attainment revealed in least some higher education, though not necessarily a the 2010 PRC census could also be quantified using college degree or equivalent. mean years of schooling (MYS) for various subgroups These figures show the tremendous differences according to age and residence. MYS is a useful sum- in reported educational attainment between rural mary metric for comparing educational attainment and urban China in 2010. They also underscore the between countries and over time, although it can- differences in reported educational profiles between not take account of differentials in educational qual- smaller and larger cities in China. ity. The 2010 PRC census did not provide its own

66 NICHOLAS EBERSTADT AND ALEX COBLIN

Figure 77. China Population by Age, Sex, and Education Attainment, 2010

85+ 80 75 70 Male Female 65 60 55 50 45 Age 40 35 30 25 20 15 10 5 15 10 5051015 Population (Millions) No Schooling Primary Secondary Tertiary

Figure 78. China Population by Age, Sex, and Education Attainment in Towns, 2010

85+ 80 75 70 Male Female 65 60 55 50 45

Age 40 35 30 25 20 15 10 5 864202468 Population (Millions)

No Schooling Primary Secondary Tertiary

67 URBANIZATION WITH CHINESE CHARACTERISTICS

Figure 79. China Population by Age, Sex, and Education Attainment in Cities, 2010

85+ 80 75 70 Male Female 65 60 55 50

Age 45 40 35 30 25 20 15 10 5 864202468 Population (Millions)

No Education Primary Secondary Tertiary

Figure 80. China Population by Age, Sex, and Education Attainment in Urban Areas, 2010

85+ 81 77 73 69 65 Male Female 61 57 53 49 45 Age 41 37 33 29 25 21 17 13 9 5 864202468 Population (Millions) No Schooling Primary Secondary Tertiary

68 NICHOLAS EBERSTADT AND ALEX COBLIN

Figure 81. Population by Age, Sex, and Education Attainment in Villages/Rural Areas, 2010

85+ 80 75 70 65 Male Female 60 55 50

Age 45 40 35 30 25 20 15 10 5 10 50510 Population (Millions) No schooling Primary Secondary Tertiary

estimates on MYS for the Chinese population, so population at 7.95 years: 8.35 for males and 7.54 years we attempted to reconstruct them from publicly for females. By contrast, we calculated 7.84 years for available information, following the approach of the China overall, 8.20 years for males, and 7.47 years for widely cited Barro-Lee Educational Attainment Data- females. Thus, there is slight downward bias in our set.80 Our calculations are provided in Table 8. calculation, and it is a bit greater for males than for The Barro-Lee approach is not as straightforward females. as it might seem, unfortunately. The problem is deter- That noted, by our reckoning, the urban-rural gap mining the rate of school completion and how much in MYS for China’s adult (15+) population in 2010 was schooling to attribute to men and women who only about 1.6 years: 8.6 versus 7.0.81 The calculated gap partly completed their primary, secondary, or tertiary was greater for women than for men (1.8 years ver- education. Patterns in this regard shift between and sus 1.4 years). And the gap between big cities and the within countries over time, and Barro-Lee publica- countryside, not surprisingly, was even wider than the tions provide no insight into precisely how they dealt urban-rural gap (2.0 years for the overall adult popula- with this matter for China in 2010, or for any other tion, 1.8 years for males, and 2.2 years for females). For specific country and calendar year. the working-age (15–64) China 2010 population, we cal- Our independent calculations for MYS for China culated an urban-rural gap of nearly 1.5 years for males in 2010 resulted in nationwide estimates close to and females combined (8.9 years versus 7.5 years) and the Barro-Lee estimates. But they were not exactly an overall gap of nearly 1.8 years between big cities and the same. For 2010, based on the 2010 PRC census, the countryside (9.2 years versus 7.5 years). Barro-Lee estimates placed MYS for the overall 15+

69 URBANIZATION WITH CHINESE CHARACTERISTICS

Table 7. Indicators of Educational Attainment from PRC Censuses

2010 2010 2010 2010 Census Year 1990 2000 2010 (City) (Town) (Village) (Urban) Percentage of Total (15–64) with N/A 4.9% 2.6% 0.9% 2.0% 4.1% 1.3% No Schooling Percentage of Total (15–64) with 43.6% 31.7% 20.8% 9.1% 17.3% 30.3% 12.2% Primary Education Percentage of Total (15–64) with 53.9% 58.5% 65.1% 65.5% 69.4% 62.9% 67.0% Secondary Education Percentage of Total (15–64) with 2.5% 4.9% 11.5% 24.6% 11.3% 2.6% 19.5% Tertiary Education Percentage of Men (15–64) with N/A 2.5% 1.5% 0.5% 1.1% 2.3% 0.7% No Schooling Percentage of Men (15–64) with 39.2% 27.3% 17.4% 7.6% 14.0% 25.6% 10.1% Primary Education Percentage of Men (15–64) with 57.7% 64.3% 69.0% 66.5% 72.4% 69.2% 68.7% Secondary Education Percentage of Men (15–64) with 3.0% 5.8% 12.2% 25.4% 12.5% 2.9% 20.5% Tertiary Education Percentage of Women (15–64) N/A 7.5% 3.8% 1.3% 2.9% 6.0% 1.9% with No Schooling Percentage of Women (15–64) 49.3% 36.3% 24.3% 10.5% 20.6% 35.2% 14.4% with Primary Education Percentage of Women (15–64) 48.9% 52.3% 61.1% 64.5% 66.3% 56.5% 65.2% with Secondary Education Percentage of Women (15–64) 1.8% 3.9% 10.9% 23.7% 10.2% 2.3% 18.5% with Tertiary Education Percentage of Men (20–39) with N/A 1.1% 0.7% 0.2% 0.5% 1.2% 0.3% No Schooling Percentage of Men (20–39) with 30.9% 20.2% 9.7% 3.9% 7.5% 15.6% 5.2% Primary Education Percentage of Men (20–39) with 65.7% 71.4% 71.1% 61.5% 73.8% 78.0% 65.9% Secondary Education Percentage of Men (20–39) with 3.4% 7.4% 18.4% 34.4% 18.1% 5.3% 28.6% Tertiary Education Percentage of Women (20–39) N/A 2.8% 1.2% 0.3% 0.8% 2.1% 0.5% with No Schooling Percentage of Women (20–39) 44.6% 30.0% 13.4% 4.8% 10.6% 21.8% 6.9% with Primary Education Percentage of Women (20–39) 53.4% 61.7% 67.6% 60.2% 71.9% 71.7% 64.4% with Secondary Education Percentage of Women (20–39) 2.0% 5.4% 17.8% 34.8% 16.6% 4.4% 28.3% with Tertiary Education

Source: National Bureau of Statistics, Department of Population and Employment Statistics, “China Population Census: Tabulation of the 1990 Population Census of the People’s Republic of China,” 1992; National Bureau of Statistics, Department of Population and Employ- ment Statistics, “China Population Census: Tabulation of the 2000 Population Census of the People’s Republic of China,” 2002; and National Bureau of Statistics, Department of Population and Employment Statistics, “China Population Census: Tabulation of the 2010 Population Census of the People’s Republic of China,” 2012.

70 NICHOLAS EBERSTADT AND ALEX COBLIN

Table 8. Preliminary Estimate: Mean Years of Schooling by Region in China, 2010

Age 15+ Age 15–64

Region Total Male Female Total Male Female

National 7.84 8.20 7.47 8.22 8.49 7.95

City 9.00 9.23 8.77 9.25 9.39 9.11

Town 8.02 8.35 7.67 8.36 8.60 8.11

Village (Rural) 7.00 7.45 6.53 7.46 7.81 7.09

Urban 8.62 8.89 8.35 8.91 9.09 8.73

Source: Author’s calculations derived from National Bureau of Statistics, Department of Population and Employment Statistics, “China Population Census: Tabulation of the 2010 Population Census of the People’s Republic of China,” 2012.

These are rather large differences for a nation This pronounced gap in educational attainment where the level of adult educational attainment aver- begs the question of exactly how much Beijing’s ages just eight years. Remember as well that the qual- planned urbanization drive can accelerate economic ity of education in rural areas may not be as high as in growth through rural-to-urban migration (even urban areas at any given level of matriculation. assuming social tensions can be managed in the pro- We can try to put these China 2010 levels and differ- cess). We have a sense of the overall gap in educa- ences in adult educational attainment into historical tional attainment between rural and urban areas. But and international perspective thanks to the Barro-Lee are educational levels for the next wave of prospec- data set. The two-year difference between rural and tive rural migrants comparable to those of the rural big-city China’s census-reported 2010 adult MYS, for migrants now in China’s cities and towns? example, is roughly equivalent to the absolute MYS Unfortunately for our purposes, the 2010 PRC cen- improvement that rapidly developing China achieved sus does not break down the educational attainment over the two decades 1990–2010. By our calculations, of the urban population by migration status; the edu- furthermore, China’s rural adult MYS level in 2010 cational profiles of migrants and legal residents are would have been roughly the same as less-developed aggregated in Figures 77–81 and cannot be disaggre- Vietnam’s overall level that same year, while China’s gated. So far as we can determine, no hukou-specific urban level in 2010 would have been comparable with data (i.e., characteristics by original place of res- the nationwide level for more highly developed Hong idence) on education, employment, or other key Kong in the late 1980s. For its part, big-city China in demographic and socioeconomic indicators are pub- 2010 would have had roughly the same MYS level that licly available from the 2010 PRC census—or for that highly educated South Korea attained nationwide in matter, virtually any other data set from the China 1985, a generation earlier. We already know the edu- National Bureau of Statistics. cational gap between rural and urban China today is We must therefore work with what we have. And a big one, but these sorts of benchmarks may help us the 2010 census does provide a summary of overall better appreciate its dimensions. educational attainment for the population age 6 and

71 URBANIZATION WITH CHINESE CHARACTERISTICS

over for migrants in villages, towns, and cities—so or legal urban residents. All other things being equal, those data can be compared with corresponding data this distinction should bias reported educational for legal city, town, and village residents as of 2010. attainment for adult migrants upward, since edu- We offer these comparisons in Figures 82–88. In cational attainment is higher for young adults than addition to the standard education groups used in older ones. Bearing this important caveat in mind, the Figures 77–81 (no schooling, primary, secondary, and data are nonetheless revealing. tertiary), we further break down the data back into To begin, Figures 82 and 83, which record the dis- the original seven education groupings listed in the tribution of educational attainment among total num- 2010 census so we might better appreciate the differ- bers of migrants to cities, towns, and villages, seem to ences in education profiles for China’s villages, towns, indicate a strong tendency toward higher educational and cities. These visual depictions of the distribution attainment among migrants in the big cities and lower of educational attainment among China’s villages, educational attainment among migrants in villages towns, and cities highlight a number of noteworthy (rural areas). The tendency is further emphasized differences, but one striking distinction is the deficit in Figures 84 and 85, which present the distribution of higher education in rural areas. of educational attainment among migrants in pro- We must caution: These data are not standardized portional terms. Virtually no migrants in villages or by age, and we know the age structure of the popula- towns, for example, reported any graduate schooling. tions under consideration in these figures differ quite But the proportion of migrants in big cities with some considerably. Thus, we are not comparing apples to graduate education was roughly comparable with the apples. China’s migrant population is significantly proportion for legal big-city residents (0.9 percent younger than the populations of legal rural residents versus 1.1 percent).

Figure 82. Migrant Population by Detailed Educational Attainment and Current Location, 2010

30 Population (Millions)

0 City Town Village No Schooling Elementary Middle High Junior College Undergraduate Graduate

72 NICHOLAS EBERSTADT AND ALEX COBLIN

Figure 83. Migrant Population by Educational Attainment and Current Location, 2010

120

100

Population (Millions) 40

0 City TownVillage No Schooling Primary Secondary Tertiary

Figure 84. Distribution of Adult Detailed Educational Attainment by Place of Residence and Migration Status, 2010

60%

50%

40%

30% centage r Pe

20%

10%

0% Migrant Migrant Migrant Legal Legal Legal City Town Village City Town Village Educational Attainment by Location

No Schooling Elementary Middle High Junior College Undergraduate Graduate

73 URBANIZATION WITH CHINESE CHARACTERISTICS

Figure 85. Distribution of Adult Educational Attainment by Place of Residence and Migration Status, 2010

70%

60%

50%

40% centage r 30% Pe

20%

10%

0% Migrant Migrant Migrant Legal Legal Legal City Town Village City Town Village Educational Attainment by Location No Schooling Primary Secondary Tertiary

Figure 86. Distribution of Detailed Educational Attainment in Population Age 6 and over, 2010

70%

60%

50%

40% centage r

Pe 30%

20%

10%

0% No SchoolingPrimary SecondaryTertiary

Educational Attainment by Location Migrant City Migrant Town Migrant Village Legal City Legal Town Legal Village

74 NICHOLAS EBERSTADT AND ALEX COBLIN

Figure 87. Educational Attainment by Location and Residence Status in Absolute Numbers, 2010

300

250

200

150

100 Population (Millions)

0 No Schooling Primary Middle High Junior Under- Graduate School College graduate Educational Attainment Migrant City Migrant Town Migrant Village Legal City Legal Town Legal Village

Figure 88. Distribution of Educational Attainment in Population Age 6 and over, 2010

60%

50%

40%

30% Percentage

20%

10%

0% No Schooling Primary Middle High Junior Under- Graduate School College graduate Educational Attainment by Location

Migrant City Migrant Town Migrant Village Legal City Legal Town Legal Village Source: National Bureau of Statistics, Department of Population and Employment Statistics, “China Population Census: Tabulation of the 2010 Population Census of the People’s Republic of China,” 2012.

75 URBANIZATION WITH CHINESE CHARACTERISTICS

By the same token, 10 percent of legal residents in know precise details until the requisite National big cities had some or more college education (col- Bureau of Statistics data on educational attainment lege plus graduate school). By contrast, slightly over by age, current residence, and hukou location are 10 percent of migrants in big cities, under 5 percent accessible to international researchers. But available of migrants in towns, and 3 percent of migrants in data strongly suggest the process of urban migration villages had some or more college education. Among in China has already cherry-picked the younger peo- legal rural residents, the share was reportedly 0.4 per- ple with the highest educational profiles from the cent. As of 2010, as we see in Figure 83, the number of countryside and brought them to China’s big cities. legal rural residents with some or more tertiary educa- Migrant per migrant, the coming phase of rural-to- tion ( junior college plus college and graduate school) urban migration may therefore be less economically totaled fewer than 10 million—but the corresponding promising than it was over the past several decades. number for big-city migrants was 36 million. Yet, while we cannot rigorously test this surmise, In all of legal rural China, furthermore, fewer than we can at least expose it to the only detailed data on 50 million persons had some high school education hukou-specific educational attainment patterns we in 2010. As we see in Figure 84, that amounts to just could obtain: the survey findings from the Chinese 7.4 percent of the legal rural resident population. This Household Income Project (CHIP), a joint collabora- means that the proportion of legal rural residents tion between Chinese and Western economists that with a high school education or better as of 2010 was has been underway for nearly three decades.83 CHIP under 9 percent. In the big cities, the corresponding has a large sample size—its latest wave includes share for legal residents was about 46 percent. It was 160,000 households—and for our purposes it has the 22.5 percent for migrants in rural areas, 36.8 percent virtue of offering a specific household sample sur- for migrants in towns, and about 45.8 percent for vey for migrants, in addition to its surveys for urban migrants in big cities. households and rural households. We should remember that many of the young We used the 2013 wave of the CHIP survey to exam- migrants in big cities were themselves raised and edu- ine educational attainment patterns reported for Chi- cated in urban areas or big cities. And given the age na’s working-age rural and urban residents by hukou structures in question, we can be fairly confident that background. Our analysis was facilitated by Professor the educational profile of migrants in big cities would Terry Sicular of the University of Western Ontario, a not be quite as high as legal big-city residents. principal in the CHIP effort. The weighting schema But it should also be fairly clear that there is only for the analysis of the 2013 wave was developed by her a limited pool of people in rural areas with relatively and Yue Ximing (Renmin University of China).84 high levels of educational attainment. Indeed, among The CHIP data are not directly comparable to rural China’s resident population (including migrants 2010 PRC census data and should not be expected to to rural regions from urban areas), the best educated match up precisely with the educational attainment cohort is the 20–24 age group. Yet just 5.2 million of numbers we have derived from it. (And they do not.) the 56.3 million men and women in that cohort report Nonetheless, they can offer meaningful insights into exposure to higher education of any sort, for any educational differentials byhukou background. We duration. This works out to 9.3 percent with some ter- present those results in Table 9. tiary education among China’s rising rural men and We will not try to reconcile differences between the women in their early 20s. But that is just half the aver- 2010 PRC census and the CHIP 2013 survey regarding age for men and women of all conventional working educational attainment.85 In general (and ordinal) ages (15–64) in urban China in 2010 (19.5 percent).82 terms, though, CHIP 2013 indicates that educational An urbanization drive of any magnitude, in other attainment for both the 15–64 working-age popula- words, would perforce reduce overall levels of education and the 20–39 working-age population was dis- tional attainment in urban China today. We cannot tinctly higher for those with urban hukou than for the

76 NICHOLAS EBERSTADT AND ALEX COBLIN

Table 9. Education Endowment of Migration Status for Working-Age Population

Working-Age Population Young Adults (15- to 64-Year-Olds) (20- to 39-Year-Olds)

Proportion Proportion of High Proportion of High Proportion Mean School– of Tertiary- Mean School– of Tertiary- Years of Educated Educated Years of Educated Educated Education People People Education People People

Rural Residents in 7.84 8.20 7.47 8.22 8.49 7.95 Their Own Hukou

Urban Residents in Their Own Local 9.00 9.23 8.77 9.25 9.39 9.11 Urban Hukou

Urban Migrants 8.02 8.35 7.67 8.36 8.60 8.11 with Urban Hukou

Urban Migrants 7.00 7.45 6.53 7.46 7.81 7.09 with Rural Hukou

Note: We used the 2013 wave of the CHIP survey to examine educational attainment patterns reported for China’s working-age rural and urban residents by hukou background. Our analysis was facilitated by Professor Terry Sicular of the University of Western Ontario, a principal in the CHIP effort. She and Yue Ximing (Renmin University of China) developed the weighting schema for the 2013 analysis. Source: Chinese Household Income Project survey, 2013.

rest of the population—regardless of whether those One last issue meriting mention is the matter of working-age men and women were in hukou them- the so-called left-behind children—the boys and girls selves or instead migrants working in other urban living apart from their long-term migrant parents. localities. But CHIP also finds that urban migrants of Based on the 2010 PRC census, UNICEF estimated working age from rural areas have significantly higher that nationwide nearly one in four Chinese children educational attainment profiles than their in-hukou under age 18 qualified asleft-behind children (defined rural counterparts—substantially higher MYS, much as living apart from at least one parent).87 The propor- more secondary education, and more tertiary education of left-behind children is higher in rural China. tion as well (although for all adults from rural hukou, The All-China Women’s Federation estimates that the proportion with tertiary education is small). 38 percent of all rural children were left-behind chil- Although we need to be careful in interpreting the dren and nearly half of left-behind children are report- data in Table 9, these results seem entirely consistent edly living with neither parent.88 with the proposition that a major movement of rural The long-term implications for education and workers to urban centers in present-day China would skills from such arrangements are as yet unclear. lower the mean level of educational attainment for Some researchers are concerned that these arrange- the migrant population in urban China and for urban ments will seriously impair the educational and China as a whole.86 cognitive development of this rising cohort.89 The

77 URBANIZATION WITH CHINESE CHARACTERISTICS

counterargument would be that left-behind children China’s rising generation. We may simply observe for typically benefit from remittances that their migrant now that, should such an effect ultimately be proved, parents send and consequently face less material this particular consequence of the current hukou sys- hardship than they would otherwise, possibly even if tem would have many adverse implications, of which both parents had remained in situ. the impact on prospective migrants’ economic poten- Available test score evidence on the question of tial would be only one. left-behind children and systematic disadvantage In light of the data assembled in this section, (that is, disadvantage above and beyond the pov- however, we do not need to presume such an effect erty that prompted parents to move in search of to suggest that the economic contribution of migra- work in the first place) appears to be inconclusive, tion to overall growth in China in the decade imme- at least to date. We do not have enough information diately ahead could prove disappointing to many as yet to take a position on whether there will be a observers—not least those in policymaking posi- left-behind-children effect on education and skills for tions in Beijing.

78 VIII. Concluding Observations

n this report, we analyzed the dynamics of China’s influx of additional manpower from outside. Urban I tightly interlinked phenomena of urbanization and China is the engine of economic growth for China migration—under the shadow of the hukou system, a today; this demographic dilemma would mean that distinctive policy regime in which the Chinese govern- continuing migration from out-of-hukou men and ment both encourages unauthorized mass movements women is crucial to the arithmetic of China’s future of people to urban centers and reserves the preroga- economic growth. tive to deport those unauthorized migrants whenever But the out-of-hukou migrants to urban areas that it so chooses. Chinese economic growth depends on the Chinese growth machine badly needs are mostly continuing flows of workers to cities, but for now the treated as de jure illegal aliens in the cities and towns Chinese Communist Party (CCP) insists on maintain- where they seek work. And this brings us to the third ing social control over those migrants through the major finding of our report:Out-of -hukou migrants— hukou system. The reference year for our study was commonly understood in China to be second-class 2010, the year of the most recent all-China census. citizens due to the discrimination they face where Our report offers a number of findings pertinent they reside and toil—have come to compose a siz- to a better understanding of China’s economic, social, able share of total manpower today for urban China and security realities. First, despite its remarkable eco- as a whole. And they actually composed an abso- nomic record over the past generation, China in 2010 lute majority of working-age men and women in a looked to be a significantlyunder-urbanized society. still-increasing number of younger cohorts in China’s As a rough approximation, in 2010 China appeared big cities as of 2010. to have 70–100 million fewer urban residents than Even though Beijing’s urbanization drive envi- one would have predicted for a country at this level sions granting local residence to scores of millions of of development (also noting the questions and prob- urban migrants, the plan also implicitly accepts that lems in the definitions Beijing has used for classifying hundreds of millions of migrants to cities and villages “urban” and “rural” areas). We suggest the lost eco- will be consigned to second-class out-of-hukou sta- nomic potential from Chinese under-urbanization tus for the foreseeable future—meaning that young was considerable as of 2010, amounting to perhaps second-class citizens will compose a majority of many 10–15 percent of current GDP. age groups in many cities for as far as the official Chi- Second, we show that Chinese population struc- nese planner’s eye can see. This has established an ture and fertility patterns in 2010 were such that the all-but-permanent social tension, based on vivid in-hukou urban population was no longer capable inequality recognized by all involved, at the heart of of generating sustained population growth or man- every Chinese urban center. power growth. To the contrary, our projections sug- Fourth, governmental ambitions to stimulate Chi- gest that, without in-migration, China’s overall urban na’s economic growth through induced urbaniza- in-hukou population would be on course to decline tion notwithstanding, an examination of the huge around 2018—that is, more or less at this writing— rural-urban gap in educational attainment for China’s and that the 15–64 working-age population would working-age population suggests that rural China may be set to peak in 2013, meaning that it would already already have sent most all of its highest-skilled young be well into long-term decline today without steady men and women to urban areas. So any substantial

79 URBANIZATION WITH CHINESE CHARACTERISTICS

flow ofrural-to -urban migration in the years immedi- on socioeconomic disparities, one might also expect ately ahead would lower overall levels of educational the human capital effect here to function as a con- attainment in urban China—begging the question of straint against rapid macroeconomic growth. whether the next wave of urban migration will be as To be sure, the story of urban development glob- much of an engine for growth as past waves. ally has, in broad terms, been a saga of less-educated Our analysis points to both opportunities and rural migrants moving into urban areas and catalyzing problems. The under-urbanization we identified in economic development in the process. But the Chi- China in 2010, for example, can be regarded as an nese case differs from these “stylized facts” in two economic opportunity—and a big one. “Normalizing” crucial respects. China’s urbanization profile for a society at its level of First, China today is concluding its epoch of what economic development would offer substantial eco- might be called “heroic economic growth.” While nomic benefits. And by all indication, Beijing has been rural-to-urban migration still offers China the prom- attempting to reap these benefits with its urbaniza- ise of stimulating economic growth in absolute terms, tion drive. it is part and parcel of a process bringing China down According to official Chinese statistics, China’s toward a relative tempo of growth more familiar in urban population surged by more than 120 million other contemporary settings. between 2010 (the reference year for our study) and Second, unlike elsewhere, China has a hukou sys- 201690 and was reportedly more than 140 million tem that automatically marks migrants into a sepa- higher at year-end 2017 than in 2010.91 Since the offi- rate, and inferior, urban caste. This inferior caste is cial definitions of “urban” and “rural” in China remain not a fringe group in China’s big cities; it is a sub- opaque, we cannot tell how much of this upsurge is stantial minority of the working-age adult population due simply to reclassification or redefinition of for- overall. For younger age groups in many urban cen- merly “rural” areas. ters, this inferior caste is an outright majority of the All other things being equal, an increase in urban overall population. Stark, unanswered social ques- numbers of this magnitude could be expected to cor- tions and even more sensitive issues regarding social rect for much of the under-urbanization that, we stability attend this peculiar arrangement. argue, characterized China’s society and economy in 2010. The corollary of this correction, however, would be that China has already addressed a structural inef- Demographics with Chinese ficiency—and must now look for other, perhaps more Characteristics and the Hukou System challenging, structural changes or reforms for additional sources of growth for the future. Our report highlighted important qualitative differ- On the other hand, the demographic and politi- ences that distinguish the processes of migration and cal arithmetic of urban growth looks to be problem- urbanization in China from those in virtually the rest atic for China. If we are correct that in-hukou urban of the contemporary world. But there is a broader and demographic growth is now essentially exhausted, the deeper point here. Such qualitative distinctions for task of “fueling” China’s urban growth engines with China are not simply limited to the demographics of even more manpower will essentially entail mass domestic migration and urbanization. movement of rural workers to urban jobs. Given the Plainly put, the Chinese state exercises excep- wide gaps in educational attainment between urban tionally intrusive claims over the ordinary, everyday and rural areas in China today—and the possibly even demographic life of its subjects in a variety of realms. wider gaps in skills—such inflows stand to reduce It is in fact tempting to speak of “Demographics with mean educational and skill levels in urban areas and Chinese Characteristics”—the almost reflexive ten- increase the dispersion of their distribution, in both dency for the Chinese government to use the coercive cases perhaps appreciably. In addition to the impact instruments at its disposal to shape and (by so doing)

80 NICHOLAS EBERSTADT AND ALEX COBLIN

distort the demographic rhythms of the society under unintended; some were not. But all were due to the its command, altering demographic outcomes in the considered policies and actions of China’s overarch- name of augmenting state power, enhancing state ing party and state. security, and preserving or perfecting social control And then there is migration. With respect to migra- for the regime. tion, the epitome of Demographics with Chinese To be sure, such unapologetically dirigiste impulses Characteristics is of course the hukou system, the de have a long and storied history in China. But the mech- facto internal passport issued to all Chinese subjects anisms of social control that make Demographics with by the Ministry of Public Security, which we have dis- Chinese Characteristics so distinctive today are, in cussed extensively already. important ways, new. The most striking are features devised under the CCP: innovations from the era of Maoist totalitarianism and, more ironically, the new The Logic of Migration with Chinese social controls unveiled and implemented during the Characteristics post-Mao era of “reforms.” Regarding births and fertility, for example, Demo- Ambitious—and merciless—policy interventions into graphics with Chinese Characteristics means that ordinary people’s personal lives inevitably result in the contemporary Chinese state is a partner with adverse and unintended consequences. The adverse would-be parents in deciding when or whether chil- unintended consequences of China’s coercive birth dren should be conceived and born—and the state control policies are today widely recognized and dis- has the deciding vote. For three and a half decades— cussed. To date, however, there has been rather less from the early years of the Deng Xiaoping period until attention to, and discussion about, the corresponding 2015—Beijing enforced a One-Child Policy through reverberations of the hukou system—but not for want a nationwide system of quotas, population agents, of adverse or unintended consequences. and threats of severe penalties, including criminal As we have noted, hukou powers are very much a penalties.92 part of Beijing’s vision of the future. So as well, as we Although Beijing announced in late 2015 that have noted, is the prospect that hundreds of millions it would be abandoning the “One-Child Norm,” of migrants will continue to suffer discrimination and authorities explicitly emphasized that they would social exclusion as out-of-hukou residents in booming not be dismantling its apparatus for forcible popu- urban China. lation control. Far from repudiating its program of Contemplating this vision of the Chinese future, forcible birth control, Beijing was refining it: reca- the outsider may be excused for asking, “But why?” librating the lifetime birth ration it would grant its Why would the Chinese government deem acceptable subjects to a “Two-Child Norm.” In 2018, Beijing the indefinite nationwide continuation of an order announced it would be rescinding the two-child that might uncharitably be described as “Soweto norm as well—but this did not mean relinquishing with Chinese characteristics”?95 What will it mean this instrument of population control. To this writ- for China to have a caste of second-class citizens as ing, indeed, Beijing insists that “births are a state the majority population for key working-age groups affair”—even as there are signs the government is throughout the nation’s urban areas? How will social toying with reversing course and implementing a tensions be mitigated or managed with the highly vis- pro-natal population policy.93 ible inequalities such distinctions currently presage In the realm of mortality, Demographics with for such densely concentrated population centers? It Chinese Characteristics might arguably refer to the is hard to imagine a conspicuous two-tier society will recurrent spasms of unnatural “excess deaths” that foster much social trust on either side of the divide. punctuated China’s demographic patterns under As best can be told, the answers to these ques- Maoist rule.94 Some of these mortality spikes were tions are that there is a logic to this most unnatural

81 URBANIZATION WITH CHINESE CHARACTERISTICS arrangement for the real, existing Chinese government. Migration and Control in an Increasingly Indeed, the human insecurity that migrants today suf- Affluent and Urbanized China fer should be seen as serving purposes of state. For one thing, financial exploitation of migrants On its current trajectory, if continued, the China of directly serves the interests of municipalities and tomorrow looks to be an increasingly urban and afflu- localities that benefit for migrant labor without being ent society supervised by an authoritarian leadership obliged to provide them with social benefits. In effect, abidingly and deeply invested in social control of its migrants are “cash cows” for public finance in cit- subjects. Political and security calculations figure cen- ies and towns all around the country. By not extend- trally in the development visions the Chinese govern- ing local hukou to working migrants (much less their ment will, and will not, accept for the future. dependents), the central government is protecting Although Beijing’s current plans envision immense the public finances of the locales in which unautho- numbers of out-of-hukou workers in Chinese urban rized migrants reside from the fiscal burdens that centers, they also imply that the share of the urban migrants’ demands for services might impose. Local population made up of such officially disadvantaged authorities are often vocal about the budgetary “cri- men and women should be reduced over time. Chi- ses” they would face if migrants were to receive access nese policy may of course offer thelong-suggested to local services. Of course this is a problem the cen- arrangements for social inclusion of the country’s tral government could solve by assuming respon- still-extralegal migrant population in the communi- sibility for these costs—but that would be a major ties to which they have moved. But no such arrange- expense, and to date Beijing has declined to accept ments have been determined at this writing—and such a commitment. de jure “illegal” migrants continue to move to urban The hukou system also provides a measure of polit- areas.97 ical security for the regime in the face of potential Multiple considerations are behind this determi- instabilities. This is not merely a theoretical option nation, many of which we have already mentioned or or a potential instrument of control always held reviewed. But there are perhaps other reasons that in reserve. In the wake of the 2008 global financial Beijing may be reconciled to a decline in the share of crash, tens of millions of Chinese migrants were sent out-of-hukou population in urban areas as well, even back to their homes by authorities who did not wish though this means granting local residency rights to to risk migrants congregating in cities as jobless dis- an increasingly large population of former farmers contents. They were only permitted to return later, as and peasants. It may be that the relative importance out-of-hukou work became available again. Although of hukou as a tool of social control is lessening as other the total number of migrants so affected has never new, technologically sophisticated tools of social con- been officially announced, and has not been defin- trol are perfected. itively estimated, the lower bound of unofficial esti- Much has been made of the surveillance potentiali- mates is that 20 million people were “un-migrated” ties of emerging new techniques of facial recognition, in 2008–09; some academic estimates range as high for example98; likewise for the social credit system as 48 million.96 that China is planning to unveil in 2020.99 The poten- Preventing unrest in urban areas in the wake of the tial social control via big data (police state use of arti- 2008 global economic crisis is surely regarded by Chi- ficial intelligence and information from e-commerce, nese authorities as a hukou policy “success.” No doubt fintech, and more) are just beginning to be discussed the same is true of subsequent episodic use of hukou abroad.100 to reduce migrant population in Beijing and key urban Still other innovations and advances may likewise centers. Whatever else may be said, these “successes” reduce the relative reliance Beijing vests in hukou as are unlikely to reduce the regime’s appetite for hukou a population-control tool. Just how new technologies perpetuation. match with regime objectives, of course, remains to

82 NICHOLAS EBERSTADT AND ALEX COBLIN

be seen. Suffice it to say, the contradictions of a strat- structured migration and “illegal” labor will affect egy for economic development predicated largely life and society—and possibly politics—in China for on building an urban workforce on a foundation of decades to come.

83 URBANIZATION WITH CHINESE CHARACTERISTICS

Notes

1. Jim Oeppen and James W. Vaupel, “Broken Limits to Life Expectancy,” Science 296, no. 5570 (May 2002): 1029–31, http://science. sciencemag.org/content/296/5570/1029; and UN Population Division, “World Population Prospects 2017,” June 21, 2017, https:// population.un.org/wpp/. 2. UN Population Division, “World Population Prospects 2017.” 3. UN Population Division, “World Urbanization Prospects: The 2018 Revision,” May 16, 2018, https://esa.un.org/unpd/wup/ Publications/Files/WUP2018-KeyFacts.pdf; and UN Population Division, “World Urbanization Prospects 2018,” http://esa.un.org/ unpd/wup/. 4. University of Groningen, Groningen Growth and Development Centre, “Maddison Historical Statistics,” http://www.ggdc.net/ maddison/maddison-project/data.htm#. 5. Simon Kuznets, Modern Economic Growth: Rate, Structure, and Spread (New Haven, CT: Yale University Press, 1967); Hollis Chenery and Moises Syrquin, Patterns of Development, 1950–1970 (New York: Oxford University Press, 1975), http://documents. worldbank.org/curated/en/482491468328816108/Patterns-of-development-1950-1970; Allen C. Kelley and Jeffrey G. Williamson, “Population Growth, Industrial Revolutions, and the Urban Transition,” Population and Development Review 10, no. 3 (September 1984): 419–41, https://www.jstor.org/stable/1973513?seq=1#page_scan_tab_contents; and C. Peter Timmer, A World Without Agriculture: The Structural Transformation in Historical Perspective (Washington, DC: AEI Press, 2009), http://www.aei.org/publication/a- world-without-agriculture-2/. 6. Edward L. Glaeser, Triumph of the City: How Our Greatest Invention Makes Us Richer, Smarter, Greener, Healthier, and Happier (New York: Penguin Books, 2012). 7. National Bureau of Statistics of China, China Statistical Yearbook 2017 (Beijing: China Statistics Press, 2018), Table 2-1, http:// www.stats.gov.cn/tjsj/ndsj/2017/indexeh.htm. 8. National Bureau of Statistics of China, China Statistical Yearbook 2017, Table 2-1. 9. State Council of the People’s Republic of China, “Accelerating Urbanization Ushers in New Economic Driver in China,” February 5, 2018, http://english.gov.cn/news/top_news/2018/02/05/content_281476037070464.htm. 10. Officially, the program is termed the “National New-Type Urbanization Plan (2014–2020).” See, for example, China Daily, “China Unveils Landmark Urbanization Plan,” March 17, 2014, http://europe.chinadaily.com.cn/business/2014-03/17/content_17352583.htm. For further background, see World Bank and People’s Republic of China, Development Research Center of the State Council, Urban China: Toward Efficient, Inclusive, and Sustainable Urbanization, July 1, 2014, https://openknowledge.worldbank.org/handle/ 10986/18865. 11. UN Population Division, “World Urbanization Prospects: The 2018 Revision.” 12. China’s National Bureau of Statistics conducted its fourth 1% National Population Sample Survey in November 2015 and released a bulletin summarizing its results in April 2016. See National Bureau of Statistics, April 2016, http://www.stats.gov.cn/tjsj/zxfb/201604/ t20160420_1346151.html. To date, the 2016 and 2017 editions of the China Statistical Yearbook make reference to data from the 2015 mini-census, and several papers published in Chinese by researchers based in Beijing review and analyze findings from this same survey. See Guo Zhigang, “中国低生育进程的主要特征 :2015 年 1%人口抽样调查结果的启示,” Chinese Journal of Population Science, 2017, http://iple.cssn.cn/rkxzt/yjyts/sy/201801/W020180123406271729734.pdf; Duan Chengrong, Qin Min, and Lv Lidan, “我国农村留守妻子的分布与生存发展现状—基于2015年1%人口抽样调查数据的分析,” South China Population, 2017, http://www. wanfangdata.com.cn/details/detail.do?_type=perio&id=nfrk201702004; Zou Xiangjiang and Wu Dan, “我国流动人口减少了吗?— 基于2015年1%人口抽样调查数据的分析,” Jiangxi Social Sciences, 2017, http://www.wanfangdata.com.cn/details/detail.do?_type= perio&id=672493120; and Wang Xiangjun, “中国老年人口特征和变迁研究—基于2015年1％人口抽样调查数据的分析,” Future and Development, 2018, http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=wlyfz201802008. However, unlike returns from the 2010 PRC census, detailed information from the 2015 mini-census is not electronically available on the China NBS website. Nor is the 2015 mini-census readily accessible in the US. For example, it is not yet catalogued in the Library of Congress; it does not yet

84 NICHOLAS EBERSTADT AND ALEX COBLIN

appear on the website of the University of Michigan’s China Data Center, arguably the foremost such institute in the US; and it is not yet available for sale on Amazon. At Harvard, America’s largest university collection, a search locates several provincial volumes of the 2015 mini-census, but curiously, not the all-China volume. This is not to say that the China 2015 1% National Population Sample Survey might not be found in the US at present. But obtaining it proved impracticable for this project—and judging from the lack of English-language studies citing the source at this juncture, other Western researchers may have had the same experience. 13. Kang Chao, Man and Land in Chinese History: An Economic Analysis (Palo Alto, CA: Stanford University Press, 1986). Chao uses the 2,000-person threshold for urban designations for comparability with the 1953 PRC census, which used the same cutoff. 14. Gilbert Rozman, Urban Networks in Ch’ing China and Tokugawa Japan (Princeton, NJ: Princeton University Press, 1974), 23. 15. Alfred Schinz, Cities in China (Berlin: Gerbruder Borntraeger, 1989), 13–14; and Chao, Man and Land in Chinese History, 48. 16. Chao, Man and Land in Chinese History. 17. Chao, Man and Land in Chinese History, 49–56. 18. Among the noteworthy studies, see Mark Elvin, The Pattern of the Chinese Past: A Social and Economic Interpretation (Stanford, CA: Stanford University Press, 1973). 19. Chao, Man and Land in Chinese History, 46. 20. Chao, Man and Land in Chinese History. 21. For more on Pao-Chia (baojia), see Kung-Chuan Hsiao, Rural China: Imperial Control in the Nineteenth Century (Seattle, WA: University of Washington Press, 1967). 22. Fei-Ling Wang, Organizing Through Division and Exclusion: China’s Hukou System (Stanford CA: Stanford University Press, 2005). 23. See Reeitsu Kojima, “Urbanization in China,” Developing Economies 33, no. 2 (June 1995): Table V, http://www.ide.go.jp/library/ English/Publish/Periodicals/De/pdf/95_02_01.pdf. 24. Without getting too far ahead of ourselves, we may note that Kang Chao’s estimates for historical urbanization ratios comport closely with definitions in the First (1953) PRC population census—so that something like an apples-to-apples comparison indicates a long-term historical decline in urbanization in China up to the period of Communist rule. Definitions for “urban” were tightened between the First and the Second (1964) Census but then loosened again between the Second and Third (1982) Census—the impact of the tightening and loosening perhaps largely offsetting each other. If so, urbanization ratios in China at the start of the 1980s maynot have been historically unprecedented—unlike virtually any other place on earth at the time, with the obvious and understandable exception of Cambodia. 25. Dwight H. Perkins, Agricultural Development in China: 1368–1968 (Chicago: Aldine, 1969). 26. Those interested in technical details should be aware that there is a distinction between “floating population” and “migrant population” in the official Chinese taxonomy. From Beijing’s standpoint, “registration-residence inconsistency” marks a person as a migrant—but that could mean a person living in the same city as their hukou but residing outside their officially designated residence. “Floating population” refers to that portion of the migrant pool who reside outside the village, town, or city in which their official hukou papers stipulate. 27. For clarity, the data we examine in this study—from China’s National Bureau of Statistics, the United Nations Population Divi- sion, and the US Bureau of the Census—all treat “China” as the mainland territories administered directly by Beijing—that is, not including Macau, Hong Kong, or the Republic of China on Taiwan, irrespective of legal or jurisdictional claims. 28. Strictly speaking, other demographic data sources do exist for contemporary China; these can and are used as “correctives” in reconstructing all-China trends. In addition to the hukou registry, there are figures on educational enrollment for children and youth, public health data, and more. But it is the census and mini-census data that are central to all estimates of contemporary Chinese demographic trends. See Chen Wei, “Evaluation of the Completeness of Birth Registration in China Using Analytical Methods and Multiple Sources of Data,” November 2016, http://www.un.org/en/development/desa/population/events/pdf/expert/26/notes/Chen_2016_ Birth%20Registration%20in%20China.pdf. 29. Shi Yaojiang and John James Kennedy, “Delayed Registration and Identifying the ‘Missing Girls’ in China,” China Quarterly 228

85 URBANIZATION WITH CHINESE CHARACTERISTICS

(December 2016): 1018–38, https://www.cambridge.org/core/journals/china-quarterly/article/delayed-registration-and-identifying-the- missing-girls-in-china/0759987A48A37E3D2CFE157778747E33. 30. National Bureau of Statistics of China, “Communiqué of the National Bureau of Statistics of People’s Republic of China on Major Figures of the 2010 Population Census[1] (No. 1),” April 28, 2011, http://web.archive.org/web/20131108022004/http://www.stats.gov.cn/ english/newsandcomingevents/t20110428_402722244.htm. 31. US Census Bureau, “Census Bureau Releases Estimates of Undercount and Overcount in the 2010 Census,” press release, May 22, 2012, https://www.census.gov/newsroom/releases/archives/2010_census/cb12-95.html. 32. Cai Yong, “China’s New Demographic Reality: Learning from the 2010 Census,” Population and Development Review 39, no. 3 (September 2013): 371–96, https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1728-4457.2013.00608.x. 33. UN Population Division, “World Urbanization Prospects 2018,” https://population.un.org/wup/. 34. As the WUP 2018 website says, “The 2018 Revision of World Urbanization Prospects . . . are consistent with the size of the total population of each country as estimated or projected in the 2017 Revision of World Population Prospects” [emphasis added]. UN Popu- lation Division, “World Urbanization Prospects 2018.” “Consistent with,” not “identical”—the population estimates in question are not exactly the same, but they are close. 35. Note that WUP 2014 and 2018 followed the 2012 and 2017 revisions, respectively, of World Population Prospects. If age-sex breakdowns for rural and urban population for China in 2010 were available from the 2018 WUP revision, these would presumably differ from those of the 2014 revision in some respects. This explains the apparent inconsistencies between Tables 1 and 2 and Tables 4 and 5. 36. Kam Wing Chan, “What Is the True Urban Population of China? Which Is the Largest City in China?,” January 2009, http:// faculty.washington.edu/kwchan/Chan-urban.pdf. 37. World Bank and People’s Republic of China, Development Research Center of the State Council, Urban China: Toward Efficient, Inclusive, and Sustainable Urbanization. 38. World Bank and People’s Republic of China, Development Research Center of the State Council, Urban China: Toward Efficient, Inclusive, and Sustainable Urbanization. 39. National Bureau of Statistics of China, “Classification & Methods: Population,” April 24, 2002,http://www.stats.gov.cn/english/ ClassificationsMethods/Definitions/200204/t20020424_72391.html. 40. Districts with population densities larger than 1,500 persons per square kilometer were considered urban locales. 41. Zhou Yixing and Laurence J. C. Ma, “China’s Urbanization Levels: Reconstructing a Baseline from the Fifth Population Census,” China Quarterly 173 (March 2003): 176–96, https://www.cambridge.org/core/journals/china-quarterly/article/chinas-urbanization-levels- reconstructing-a-baseline-from-the-fifth-population-census/6EE598E6BE4B8F64CA1F3D75510F4B7F. 42. Kojima, “Urbanization in China,” Table V. 43. Qin Bo and Zhang Yu, “Note on Urbanization in China: Urban Definitions and Census Data,”China Economic Review 30 (Sep- tember 2014): 495–502, http://www.sciencedirect.com/science/article/pii/S1043951X1400087X. 44. Even such an effort, however, would still encounter difficulties in adjusting for seemingly arbitrary political decisions embedded in Beijing’s constantly shifting definitions of “urban” and “rural” areas, as these make international standardization especially challenging. 45. UN Population Division, “World Urbanization Prospects 2018.” 46. UN Population Division, “World Urbanization Prospects 2018.” 47. Hirotsugu Uchida and Andrew Nelson, “Agglomeration Index: Towards a New Measure of Urban Concentration,” February 15, 2008, https://openknowledge.worldbank.org/handle/10986/9039?locale-attribute=fr. 48. Note these net interprovincial migration tallies represent inflows minus outflows. For some places in China, such as Shanghai and Guangdong, there is essentially no reported outflow at all—that is, gross provincial out-migration of the population of 1 percent or less. 49. Of course, men and women younger than 15 or older than 64 may in fact have been working, but this big fact about the 15–64 population provides an important first take on migration and manpower in China. In fact, the Chinese census does not appear to collect

86 NICHOLAS EBERSTADT AND ALEX COBLIN

data on work for migrants under the age of 15 because of child labor concerns. As for most migration around the world, migration in China has predominantly been fueled by the desire to find better working conditions than currently exist in rural areas. 50. The figures for this portion of the report were prepared by Grace Finley, whose work we gratefully acknowledge. 51. UN Population Division, “World Urbanization Prospects 2018.” 52. Wittgenstein Centre for Demography and Global Human Capital, Wittgenstein Centre Human Capital Data Explorer Version 1.2, 2015, http://www.wittgensteincentre.org/dataexplorer. 53. Robert J. Barro and Jong-Wha Lee, “A New Data Set of Educational Attainment in the World, 1950–2010,” Journal of Development Economics 104 (2013): 184–98. 54. World Bank, World Development Indicators, June 11, 2010, http://data.worldbank.org/data-catalog/world-development- indicators. 55. University of Groningen, Groningen Growth and Development Centre, “Maddison Historical Statistics.” 56. The WDI figures appear on the UNPD’s biennial World Population Prospects effort—but those life expectancy estimates are only offered for five-year periods. WDI interpolates these to derive annual soundings. 57. Fraser Institute, Economic Freedom, http://www.freetheworld.com/. 58. Heritage Foundation, 2018 Index of Economic Freedom, 2018, http://www.heritage.org/index/. 59. Some of these data sets offer observations extending back to 1970. 60. Depending on the source and whether the estimates are PPP-adjusted or based on foreign exchange rates, current World Bank WDI and Maddison Project data file estimates for 2010 place per capita GDP in China between 13 percent and 49 percent above that of Indonesia. 61. The figures for Taiwan are more difficult to parse since Taiwan has adopted a different set of thresholds for defining urban settlements from those widely used elsewhere. Since the Taiwanese threshold for “urban” is much higher than in most other parts of the world, it would appear that China was also considerably less urbanized in 2010 than Taiwan had been when it reached a comparable income level. 62. Uchida and Nelson, “Agglomeration Index: Towards a New Measure of Urban Concentration.” 63. In other words, the value of the “dummy variables” for China in the regressions in question were not only strongly negative but also consistently greater in absolute value than those for India or Indonesia—often twice the magnitude of the “India effect” or the “Indonesia effect.” 64. We are well aware that our methodological approach here is “simplistic” and that a much more sophisticated econometric examination of this relationship could be undertaken. We prefer the “simplistic” path precisely because the approach is so conceptually straightforward. We are also wary of the risks of false precision inherent in a more sophisticated analysis of China’s time-series data on urbanization, which we regard as embedded with major and typically undiscussed distortions and biases. 65. Zhang Li, China’s Limited Urbanization: Under Socialism and Beyond (New York: Nova Science, 2003). 66. China Daily, “China Unveils Landmark Urbanization Plan.” 67. National Bureau of Statistics of China, “What Census Data Can Tell Us About Children in China: Facts and Figures 2013,” http:// www.unicef.cn/en/uploadfile/2013/1216/20131216111141945.pdf. 68. Nicholas Eberstadt, Understanding North Korea’s “Epic Economic Fail” in International Perspective, Asan Institute for Public Pol- icy, November 2015, http://www.nkeconwatch.com/nk-uploads/Eberstadt-Asan-Report_North-Koreas-Epic-Economic-Fail.pdf. 69. We should note that the estimative power of our model is hardly affected by choosing one measure of per capita GDP over another—that is, PPP-adjusted or non-PPP-adjusted—or one index of business climate over another—that is, Fraser versus Heritage/ Wall Street Journal. 70. Here again, we should explicitly recognize the shortcomings of such “simplistic” modeling. As more sophisticated readers will immediately understand, the approach embraced here is subject to a number of potentially important pitfalls: among these, the problems of collinearity, omitted variable bias, and endogeneity. Even so, our results appear to be sufficiently stable and robust, so we are comfortable presenting them, if only as an illustrative first approximation of the magnitude of the effects at play here. 71. We recognize that “left behind” has a more particular connotation in China today, being used to refer to children and older

87 URBANIZATION WITH CHINESE CHARACTERISTICS

relatives of working-age migrants who do not accompany them to their new places of residence. There is considerable discussion in China today of left-behind children in particular. We are using “left behind” in a different sense here, applying it to the entire population not included in the great trek to the cities. But we will also directly address the matter of left-behind children later in this report. 72. Zheng Zhenzhen and Yang Ge, “Internal Migration in China: Changes and Trends,” in Contemporary Demographic Transforma- tions in China, India and Indonesia, ed. Christophe Z. Guilmoto and Gavin W. Jones (Heidelberg, Germany: Springer, 2016), 223–37, https://www.researchgate.net/profile/Nandita_Saikia2/publication/301242418_Trends_in_Mortality_Differentials_in_India/links/ 5a0b48f7458515e482748fcd/Trends-in-Mortality-Differentials-in-India.pdf. 73. There is an official distinction, previously noted, between the “migrant population” and the “floating population”—the latter being the subset of the former who are both out of hukou and residing outside the village, town, or city for which their hukou papers are designated. In the 2010 PRC census, Beijing counted 261 million migrants, of which 221 million, or about 85 percent, constituted floating population. The 2010 PRC census data we use in this report provide information on the total migrant population—not the floating population per se. Since the floating population makes up the overwhelming majority of migrants, we tend to use these two terms syn- onymously. But strictly speaking, there is a distinction here, and tens of millions of migrants in China today are not considered mem- bers of the floating population. 74. Health Policy Project, DemProj Spectrum 5, http://www.healthpolicyproject.com/index.cfm?id=software&get=Spectrum. 75. Although the actual numbers entailed are unknown outside Chinese officialdom, it is apparent that only a tiny minority of migrants in China over the past four decades have as yet seen their hukou officially changed to their new place of residence. 76. While the central government has dictated that migrant children are entitled to nine years of compulsory education in their local school districts, access to schooling is still not completely universal due to considerable barriers and lack of enforcement by local governments. Some locals have extended health services to migrants, but this remains exceptional. 77. China’s total population structure is composed of 74.5 percent working-age people (age 15–64). If one assumed that China’s migrant population had a similar composition, that the 223 million people constituted 74.5 percent of the total migrant population, then about 78 million older persons age 65+ and children under 15 would have been found with China’s 223 million working-age migrants. However, the reported migrant population for older and younger persons totaled about 38 million—a discrepancy of 40 million. A “normal China” population containing 223 million persons of working age would be expected to include 49 million children— whereas the enumerated China migrant population counted just 28 million. This is a large gap—amounting to 21 million children under age 15 in 2010. Even so, that gap is much smaller than some of the numbers often cited for total left-behind children in contemporary China, general estimates for which today often fall in the vicinity of 60 million or even more. We will return to the question of left-behind children in the next section. 78. The two components of total compensation to bear in mind are (1) wages or salaries and (2) social benefits, including health insurance, pension, and so forth. Even if migrants are remunerated at the same wage rates as in-hukou workers of comparable skills and experience, their overall compensation will be lower if social benefits are not granted—thus, the distinctive (and to foreign observers, sometimes surprising) current phenomenon in China of worker strikes for social benefits rather than higher pay. Tom Mitchell, “Chi- nese Factory Workers Strike over Benefits,”Financial Times, March 18, 2015, https://www.ft.com/content/c46252d0-cd48-11e4-9144- 00144feab7de. 79. National Bureau of Statistics of China, China Statistical Yearbook 2017 (Beijing: China Statistics Press, 2018), Tables 2-1 and 2-3, http://www.stats.gov.cn/tjsj/ndsj/2017/indexeh.htm. 80. Barro and Lee, “A New Data Set of Educational Attainment in the World, 1950–2010.” 81. Any apparent discrepancies here are due to rounding. 82. For their part, 45 percent of urban men and women in their early 20s reported at least some tertiary education in 2010—and this includes the majority who were unauthorized migrants. 83. China Institute for Income Distribution, CHIP Dataset, http://ciid.bnu.edu.cn/chip/index.asp?lang=EN. 84. John Knight, Li Shi, and Wan Haiyuan, “The Increasing Inequality of Wealth in China, 2002–2013,” University of Oxford, Decem- ber 20, 2016, https://ideas.repec.org/p/oxf/wpaper/816.html. Our thanks to Professors Sicular and Yue for their generous cooperation. 85. The most important of these is the high level of urban educational attainment reported. The 2013 CHIP survey attributes

88 NICHOLAS EBERSTADT AND ALEX COBLIN

11.5 mean years of schooling to China’s working-age (15–64), in-hukou, urban population—whereas we estimate MYS for all urban adults 15 and older in 2010 to be about nine years. To go by CHIP 2013 data and the Barro-Lee data set, moreover, the level of MYS for China’s urban, in-hukou, working-age population would have been almost the same as Hong Kong in 2010. This does not seem plausible for urban China as a whole. We suspect this curious result to be an artifact of sampling bias. 86. We note that MYS for urban migrants from rural areas in Table 9 is only marginally higher for the 20–39 group than for the 15–64 group—unlike their urban hukou counterparts and in-hukou rural counterparts. One interpretation of this distinction could be that recent influxes have already slowed the improvement of educational attainment for rural migrants in urban areas. Another possibility could be sample bias or measurement error. 87. UNICEF, National Bureau of Statistics of China, and National Working Committee on Children and Women Under the State Council, “Children in China: An Atlas of Social Indicators,” 2014, http://www.unicef.cn/en/uploadfile/2015/0114/20150114094309619.pdf. 88. Zhao Chenyue et al., “Care for Left-Behind Children in Rural China: A Realist Evaluation of a Community-Based Intervention,” Children and Youth Services Review 82 (November 2017): 239–45, https://www.sciencedirect.com/science/article/pii/S0190740917306424 #bbb0005. 89. Dennis Normile, “One in Three Chinese Children Faces an Education Apocalypse. An Ambitious Experiment Hopes to Save Them,” Science, September 21, 2017, https://www.sciencemag.org/news/2017/09/one-three-chinese-children-faces-education- apocalypse-ambitious-experiment-hopes-save. 90. National Bureau of Statistics of China, China Statistical Yearbook 2017 (Beijing: China Statistics Press, 2018), Table 2-1, http:// www.stats.gov.cn/tjsj/ndsj/2017/indexeh.htm. 91. Xinhua, “Accelerating Urbanization Ushers in New Economic Driver in China,” February 5, 2018, http://english.gov.cn/news/top_ news/2018/02/05/content_281476037070464.htm. 92. John S. Aird, Slaughter of the Innocents: Coercive Birth Control in China (Washington, DC: AEI Press, 1990); Susan Greenhalgh, Just One Child: Science and Policy in Deng’s China (Berkeley, CA: University of California Press, 2008); and Wang Feng, Gu Baochang, and Cai Yong, “The End of China’s One-Child Policy,” Studies in Family Planning 47, no. 1 (March 2016): 83–86, https://onlinelibrary. wiley.com/doi/abs/10.1111/j.1728-4465.2016.00052.x. 93. See, for example, Kangyu Mark Wang and Cecilia Joy-Pérez, “China Finally Wants More Kids. But There Are Problems,” AEIdeas, September 24, 2018, http://www.aei.org/publication/china-finally-wants-more-kids-but-there-are-problems/. 94. These would include the mass deaths during land reform and “socialist transformation” in the Chinese countryside in the years following the Liberation and the death toll from the violence and upheaval during the Great Proletarian Cultural Revolution in the 1960s—both arguably intended consequences of Mao’s political designs, each thought to have claimed millions of lives. The most cat- astrophic mortality paroxysm was the terrible famine of 1959–61, which was a direct but unintended consequence of the disastrous Great Leap Forward, a madly ambitious campaign for super-rapid modernization that brought the entire economy to the verge of col- lapse and caused a breakdown of the nation’s food system. Thus “mortality with Chinese characteristics” might be understood to describe the tens of millions of deaths brought about by “socialist transformation,” “socialist construction,” and the fateful political campaigns of the Maoist regime, exacted from a society ostensibly at peace. To be sure, regime-made famines and deadly campaigns of terror were not unique to China in the 20th century—Communist China shared this “family resemblance” with other totalitarian states of the period—but China’s absolute head count from death-by-government actions was unparalleled. 95. Uncharitably—but not altogether inaccurately—recall that the apartheid regime of South Africa referred to the policy by which it shaped and governed Soweto and other similar settlements as “influx control.” 96. For a more detailed discussion, see Jeremy L. Wallace, Cities and Stability: Urbanization, Redistribution, and Regime Survival in China (Oxford: Oxford University Press, 2014), chap. 6. 97. Hukou reform is not a new topic in Chinese political dialogue. The central government, which is aware of the potential political instability the system may cause, has mandated specific local governments to implement experimental hukou reforms. However, such reforms have proved ineffective. An important reason for their ineffectiveness is pushback from local governments. The announcement of national hukou reform following China’s November 2013 Third Plenum meeting was the first sign of potentially meaningful hukou reform. Although much fanfare was created around the announcement, little real action has been seen to date. While the State

89 URBANIZATION WITH CHINESE CHARACTERISTICS

Council released a draft measure in December 2014 for the relaxation of hukou restrictions and access to social services in small cities and towns, where only a small fraction of migrants reside, big cities such as Beijing and Shanghai will continue to strictly enforce the current system. These restrictions signify that half, if not more, of China’s migrants will not benefit from these reforms because they currently live in urban (big-city) locations. 98. Amy B. Wang, “A Suspect Tried to Blend in with 60,000 Concertgoers. China’s Facial-Recognition Cameras Caught Him.,” Wash- ington Post, April 13, 2018, https://www.washingtonpost.com/news/worldviews/wp/2018/04/13/china-crime-facial-recognition-cameras- catch-suspect-at-concert-with-60000-people. 99. Martin Chorzempa, Paul Triolo, and Samm Sacks, “China’s Social Credit System: A Mark of Progress or a Threat to Privacy?,” Peterson Institute for International Economics, June 2018, https://piie.com/publications/policy-briefs/chinas-social-credit-system- mark-progress-or-threat-privacy. 100. Fred Hiatt, “If the Chinese Look to the West for a Democratic Model, What Are We Showing Them?,” Washington Post, Novem- ber 4, 2018, https://www.washingtonpost.com/opinions/global-opinions/if-the-chinese-look-to-the-west-for-a-democratic-model- what-are-we-showing-them/2018/11/04/8fa457d6-ddbc-11e8-b732-3c72cbf131f2_story.html.

90 Appendix A. China’s Population Policy U-Turn: A Tougher New Battle Ahead

Kangyu Mark Wang

hina’s mass fertility-control policy, infamously Newest Wave of Policy Changes C known for its ruthless one-child restriction, is ending after four decades of controversy. Although The relaxation of family planning policy follows China’s fertility rate began to fall even before the similar patterns of policy change in China.3 First, implementation of the One-Child Policy in 1979,1 the scholars and National People’s Congress (NPC) Chinese Communist Party (CCP) leadership seems representatives advocate for policy change. Second, to have awakened to this problem only recently. Dis- CCP media propose ideas without explicitly endors- mantlement of the fertility-control regime accelerated ing them, allowing some public discussion to test after Chinese President Xi Jinping took power in 2012, the waters. Third, CCP Congresses and Plenums and all Chinese couples were allowed to have two chil- approve the new policy framework, and then the dren starting in 2016. NPC session the following year changes the corre- Ahead of the Fourth Plenum of the 19th Party sponding laws. Congress, which is yet to take place and will report- Two major policy changes took place after Xi Jin- edly focus on economic issues,2 the government ping took power in 2012: seems to be taking the temperature by allowing some open discussion on pro-natalist measures and prob- (1) Since 2013, families could have two children if ing people’s reactions. However, public opinion on one parent was an only child.4 the new direction of population policy seems far from positive. Some of the government’s proposals (2) Universal two-child policy was implemented in triggered fierce criticism of state coercion in fertil- 2016.5 ity promotion. Backlash may also come from ethnic minorities if, as the new policy line suggests, birth Also, starting in 2017, the term “family planning” restrictions are not lifted on them as they are on the ceased to appear in President Xi’s Party Congress ethnic Han majority. Also, a nationwide fiscal incen- speech and Premier Li Keqiang’s Annual Report tive program to promote fertility would divert mas- of the Work of the Government.6 Along with pol- sive public resources and thus is difficult to sustain icy changes, a new wave of institutional reform in without any taxation or fine on unmarried people 2018 removed “family planning” from the name of and childless couples. Therefore, whatever radical the Ministry of Health.7 This new round of reform policy change we see from the forthcoming Fourth radically changed the mandate of the new ministry Plenum of the 19th Party Congress, China’s new bat- and its departments regarding population policy and tle on human reproduction will only be tougher, eco- decisively weakened the family planning apparatus nomically and politically. at the national level.8

91 URBANIZATION WITH CHINESE CHARACTERISTICS

Party-State Policy Statements. Party Congress Draft of the Civil Code: Family Planning–Related Con- Reports are paramount policy statements in China. tent Eliminated. To keep pace with the CCP line, a new They are signed by party general secretaries and are draft of the Civil Code of China submitted to the NPC supposed to be followed by the whole party, govern- Standing Committee in August 2018 also dropped all ment, and nation until the next Party Congress is held. mentions of “family planning.”19 The new Civil Code is scheduled to be approved by 2020, and by then Party Congress Reports on Family Planning. In the marriage and adoption will no longer be subject to 17th and 18th National Party Congress reports, for- family planning limits. mer President Hu Jintao emphasized family planning. However, this is far from a complete purge of “fam- However, from the 17th to 18th Congress, the goal of ily planning” in China’s legal vocabulary; most impor- “stabilizing low fertility level” was replaced by the tantly, the term and the policy are enshrined in the need to “optimize policy” with the aim being “facili- constitution.20 Also, the National Population and tating long-term balanced population development,” Family Planning Law is not incorporated into the clearly suggesting policy change.9 Civil Code, and thus the universal two-child policy Furthermore, in the 19th Party Congress Report, will technically remain until the National Population President Xi did not mention family planning.10 The and Family Planning Law is revised, despite imple- overall goal of population policy became “promoting mentation of the new Civil Code. The Fourth Plenum the link between fertility policy and related socioeco- of the 19th Party Congress and the NPC session in nomic policies, and strengthening research on popu- spring 2019 will be the earliest time window to legally lation development strategy,” again emphasizing the end family planning policy. need for research.11 This leaves room for unspecified, probably dramatic, policy change in the future. Institutional Reform at the National Level. Along the lines of population policy change, two waves of Reports on the Work of the Government: Parallel with Party institutional reforms were implemented. The merger Line. The Reports on the Work of the Government of the Ministry of Health and the National Commis- given by the premier reflect the principles of Party sion of Population and Family Planning in 2013 failed to Congress Reports and direct work of the government. make a dent in the family planning apparatus. By stark After 2014, Reports on the Work of the Government contrast, the newest overhaul of the Ministry of Health stopped describing family planning as a “basic national in 2018 is more than a name change: The new com- policy,” but the substance of family planning policies mission completely purged the term “family planning” remained.12 After the 19th Party Congress, however, Li and decisively weakened its fertility policy apparatus in stopped mentioning population policy at all.13 terms of internal structure and personnel. In 2014, Premier Li reiterated that China should “unwaveringly adhere to the basic state policy of fam- 2013 Merger of Health and Family Planning Minis- ily planning” and “implement the policy that allows tries. In 2013, the Commission of Population and married couples to have two children if one parent is Family Planning merged with the Ministry of Health a single child.”14 In 2015, by contrast, his comments to become the Commission of Health and Family on population policy were shortened to “advance the Planning. However, the two former ministries were reform of family planning services and the way they coalesced rather mechanically without much substan- are managed.”15 In 2016, he further emphasized the tial integration. need to support “policies to complement the decision First, the official duty of the former Commis- to allow all couples to have two children.”16 In 2017, sion of Population and Family Planning remained he emphasized the two-child policy again, but with- unchanged.21 Second, in terms of subdivisions, all out mentioning the term “family planning” itself.17 In the former departments of the Commission of Popu- 2018, he did not mention population policy at all.18 lation and Family Planning were kept under the new

92 KANGYU MARK WANG

commission. Third, in terms of personnel, the for- implement fertility policies, and establish and perfect mer Ministry of Health had 387 personnel, the former aid policy to special planned families.”28 Commission of Population and Family Planning had More importantly, the new round of institutional 143 personnel, and the new commission had 545 per- reform in 2018 downsized the commission’s person- sonnel.22 In other words, the new commission after nel, in sharp contrast to the 2013 merger. The former the merger was even larger than the sum of its two Health and Family Planning Commission had 545 per- parts. Therefore, little real integration could have sonnel, while the new Health Commission has only taken place. The family planning apparatus was hardly 525. However, this 20-person reduction is still too weakened in 2013. small to cause all the 143 personnel of the former Commission of Population and Family Planning to 2018 Renaming of the Health Commission. On March 23, lose their jobs. 2018, the former National Commission of Health and In addition to removing family planning subdi- Family Planning was renamed the National Commis- visions, the new National Health Commission also sion of Health. According to a State Council document, promises to correct for some of the unwanted con- “family planning policy making” is replaced by “policy sequences of almost four decades of draconian birth making coping with aging population” in its mandate.23 control. It still shows no regret over a wide range Except for this general State Council document, of much criticized practices, including forced abor- official documents on the specific duty, subdivision, tion, forced sterilization, confiscation of belongings, and personnel arrangements of the new commission and dismissal from work of people who fail to “plan” were not published as scheduled by June 20, 2018.24 their families. However, it does vow to help “specially In other words, the new National Commission of planned families” or “family planning special fam- Health missed the official deadline to specify its new ilies.” These are special types of families caused by organizational structure. This unusual delay contrasts family planning policies, including rural one-daughter sharply with other newly established or overhauled families and families with a deceased only child.29 ministry-level agencies, the majority of which made They usually face big social and economic difficulties. the new subdivisions public in time.25 Given that this In short, the renaming of the Health Commission round of overhaul almost eliminated the former fam- in 2018 is far more substantial than the 2013 merger of ily planning departments and thus must have import- the Health and Family Planning Ministries. The gov- ant implications for vested interests, the delay is likely ernment still preserves the power to make fertility pol- a result of bureaucratic infighting. icy, but no longer in the name of family planning. Local On July 30, the Health Commission’s detailed health and family planning bureaus were also sched- institutional reform plan was finally made public uled to follow suit and reconstruct.30 after being more than one month overdue.26 The plan clearly signals radical policy change. First, the Provincial-Level Variation of Population Pol- term “family planning” is completely purged from the icy: Revolution from Below? Under the princi- names and duties of the new commission and its sub- ples specified by the Fifth Plenum of the 18th Central divisions. More substantially, the number of subdivi- Party Committee, the Law of Population and Family sions in charge of fertility policy is reduced from three Planning was edited in 2015. Compared to the 2001 to one.27 All former departments in charge of imple- version, it dropped stipulation of “encouraging late menting family planning policy are removed, and the marriage and late childbirth.”31 It also authorizes only subdivision responsible for population policy, provinces to make policies on children beyond the the Population Monitoring and Family Development universally allowed two. Department, is designed to “monitor population and Since 2016, a new wave of provincial lawmaking report warning, propose policy suggestions related on family planning has been seen across the country. to population and family development, perfect and In particular, the depressed northeast rust belt began

93 URBANIZATION WITH CHINESE CHARACTERISTICS

Table A1. Birth Rate of Northeastern Rust Belt 2016 Birth Rate 2017 Birth Rate Province (Per Thousand) (Per Thousand) Third-Child Policy Liaoning 6.6 6.49 • Parents with disabled children • Border county residents Jilin 5.55 6.76 • Parents with disabled children • Foreign Chinese Heilongjiang 6.12 6.22 • Border county residents • Parents with disabled children National 12.95 12.43

Source: Liaoning Statistical Information Net, “Statistical Statements,” http://www.ln.stats.gov.cn/tjsj/tjgb/; Statistical Bureau of Jilin, “Statistical Statements,” http://tjj.jl.gov.cn/tjsj/tjgb/ndgb/; and Heilongjiang Bureau of Statistics, “Statistical Statements,” http:// www.hlj.stats.gov.cn/tjgb/shgb/. to encourage reproduction. As shown in Table A1, all while lowering the sex ratio and improving education of the three northeast provinces have birth rates far (Table A2).36 lower than the national level, and they have the most lenient third-child policy. Both the Jilin and Heilong- The First Year of the Two-Child Policy. The jiang provinces allow border county residents to have universal two-child policy was first implemented in a third child; this policy is not seen in any other bor- 2016. The national birth rate increased from 12.07 der province.32 An extreme case is Liaoning, where to 12.95, and among the 32 provinces, 24 saw a birth the government plans to “reward” families with a sec- rate increase. However, in 2017 the national birth ond child through “taxation, education, social secu- rate dropped to 12.43, and 10 provinces also saw a rity and housing” policies to lighten the burden of drop in the birth rate. (See Table A3.) Directed by the giving birth to and bringing up children.33 National Bureau of Statistics, multiple local govern- Also worthy of attention is the change in fertil- ments conducted surveys on new births in the first ity restrictions on ethnic minorities: They used to be half of 2018, which again showed decreases in the allowed to give birth to more children than Han people number of newborns as compared to the first half were, but no longer. For example, in Xinjiang, urban eth- of 2017.37 nic minorities used to be allowed to have two children At the end of 2015, Deputy Director of National but urban Han people only one. Rural ethnic minorities Health and Family Planning Wang Pei’an declared that used to be allowed to have three, as opposed to only the universal two-child policy would result in more two for rural Han couples.34 After 2016, all urban cou- than three million more births per year.38 However, ples are allowed to have two children, and all rural cou- even though 2016 saw an increase in the birth rate, the ples are allowed to have three children.35 number of newborn babies increased by only 1.21 million, as calculated by annual birth figures provided by National Bureau of Statistics.39 The increase further Goals Versus Reality: Limited Effects of shrank in 2017 to 0.58 million. Hence, relaxation of the Universal Two-Child Policy family planning policy to the universal two-child limit failed to meet the authorities’ expectation. Probably As specified by the State Council National Population alarmed by people’s lackluster response to the univer- Development Plan (2016–30), the government wants sal two-child policy, the National Bureau of Statistics to boost China’s total fertility rate back to 1.8 in 2020 initiated another nationwide survey on “population

94 KANGYU MARK WANG

Table A2. Fertility Targets as Set by National Population Development Plan (2016–30) Indicator 2015 2020 2030 Population 1.375 billion 1.42 billion 1.45 billion Total Population Total Fertility Rate 1.5–1.6 1.8 1.8 Population Structure Sex Ratio 113.5 <=112 107 Life Expectancy 76.3 77.3 79 (Years) Population Quality Labor Population 10.23 10.8 11.8 Years of Education Population Urbanization Rate 56.1% 60% 70% Distribution Source: Central People’s Government, “Main Duty, Internal Agencies and Personnel Arrangement of Ministry of Health,” http://www. gov.cn/zhengce/content/2016-06/12/content_5081151.htm.

change” in September 2018 to “strengthen population As often occurs with policymaking in China, development strategy research.”40 the government is now testing the waters by yield- ing proposals on fertility promotion policies and allowing some open discussion to probe people’s Next Step to Avoiding a Population reactions. Yet, there are concerns remaining if Crisis: “Independent Fertility” or Fertility birth control measures are completely scrapped: Promotion? rapid increase of ethnic minority population, overcrowded megacities, and environmental pressures. Talk of “independent fertility”—that is, “couples Therefore, China’s population policy after the era deciding the number of children and time interval of of family planning is likely to involve both fertil- fertility”—began in late 2016.41 Director of National ity promotion for the Han majority in ecologically Development and Reform Commission Development stronger eastern China and fertility restriction in Research Institute Yang Yiyong said that “after 2020 economically backward, ecologically fragile western China may realize independent fertility very soon.”42 China, where ethnic groups historically have had On May 21, 2018, Bloomberg News reported that “an higher fertility rates. initial feasibility study was submitted to Chinese Pre- mier Li Keqiang in April,” and Premier Li “requested Promoting Fertility for the Han Majority. By more research on the social impact of scrapping the now, both the government and the public seem to policy altogether.” According to Bloomberg, “the have reached the consensus that there should be no decision could be made as soon as the fourth quar- more birth restrictions, at least for the Han ethnic ter,” and “announcement might also be pushed into majority. The debate has shifted to the types of incen- 2019.”43 Chinese news media then reposted this news. tives and how much state coercion should be involved However, despite a U-turn in policy statements in fertility promotion. and important institutional reforms, Chinese cou- The government seems to be considering both ples are unlikely to be granted full independence in positive and negative incentives, contemplating sub- their fertility. The government does not intend to sidizing extra children and fining unmarried people yield control over human reproduction, insisting that and married couples without children. For exam- childbearing is a state affair and that people should ple, on August 14, 2018, the Jiangsu Province Party have children for the sake of the nation.44 Committee newspaper advocated for establishing a

95 URBANIZATION WITH CHINESE CHARACTERISTICS

Table A3. Birth Rate as Calculated by Annual Statistical Reports

2015 (Per Thousand) 2016 (Per Thousand) 2017 (Per Thousand) National 12.07 12.95 12.43 Beijing 7.96 9.32 9.06 Tianjin 5.84 7.37 7.65 Hebei 11.35 12.42 13.20 Shanxi 9.98 10.29 11.06 Inner Mongol 7.72 9.00 9.47 Liaoning 6.17 6.60 6.49 Jilin 5.87 5.55 6.76 Heilongjiang 6.00 6.12 6.22 Shanghai 7.52 9.00 8.10 Jiangsu 9.05 9.76 9.71 Zhejiang 10.52 11.22 11.92 Anhui 12.92 13.02 14.07 Fujian 13.90 14.50 15.00 Jiangxi 13.20 13.45 13.79 Shandong 12.55 17.89 17.54 Henan 12.70 13.26 12.95 Hubei 10.74 12.04 12.60 Hunan 13.58 13.57 13.27 Guangdong 11.12 11.85 13.68 Guangxi 14.05 13.82 15.14 Hainan 14.57 14.57 14.73 Chongqing 11.05 11.77 11.18 Sichuan 10.30 10.48 11.26 Guizhou 13.00 13.43 13.98 Yunnan 12.88 13.16 13.53 Tibet (Xizang) 15.75 15.79 16.00 Shaanxi 10.10 10.64 11.11 Gansu 12.36 12.18 12.54 Qinghai 14.72 14.70 14.42 Ningxia 12.62 13.69 13.44 Xinjiang 15.59 15.34 15.88

Source: China Economic Net, http://district.ce.cn/zg/201702/24/t20170224_20510455.shtml; and Henan Province Bureau of Sta- tistics, National and Provincial Statements, http://www.ha.stats.gov.cn/sitesources/hntj/page_pc/tjfw/tjgb/gjhgsgb/list1.html.

96 KANGYU MARK WANG

Table A4. Population of Ethnic Minorities Has Been Growing Faster Than Majority Han Chinese

Growth Rate from Proportion of 0- to Ethnic Group 2000 Population 2010 Population 2000 to 2010 14-Year-Olds in 2010

Total 1,242,612,226 1,332,810,869 7% 17% Han 1,137,386,112 1,220,844,520 7% 16% Tibetan 5,416,021 6,282,187 16% 26% Mongolian 5,813,947 5,981,840 3% 20% Manchurian 10,682,262 10,387,958 –3% 17% Uighur (Muslim) 8,399,393 10,069,346 20% 26% Kazakh (Muslim) 1,250,458 1,462,588 17% 24% Hui (Muslim) 9,816,805 10,586,087 8% 21%

Source: National Bureau of Statistics, Department of Population and Employment Statistics, “China Population Census: Tabulation of the 2010 Population Census of the People’s Republic of China,” 2012, Table 2-1; and National Bureau of Statistics, Department of Population and Employment Statistics, “China Population Census: Tabulation of the 2000 Population Census of the People’s Republic of China,” 2002, Table 2-1.

compulsory “birth fund” that subsidizes families with incentives would cover taxation, education, social two or more kids.45 security, and housing. The idea, to tax childless couples and then cre- However, these new pro-natalist policies would ate a “fund” to help with raising children, triggered inevitably divert massive public resources. The author fierce criticism. People not only complained about of the birth fund proposal argues that the government the high costs of raising children but also criticized should use previous “social payment of upbringing,” the potential for human rights violations involved or fine parents with more children than allowed, to in fertility promotion. Some have even drawn paral- finance newpro-natalist programs. Commentators lels with Communist Romania’s pro-natalist policy immediately mocked this suggestion as naive because under Nicolae Ceauşescu, which banned abortion “social payment of upbringing,” which was supposed and severely limited contraception options. Conse- to be turned over to the State Treasury, has been in quently, the government seems to have backed down reality spent locally “in mysterious ways.”48 Hence, a little, with an op-ed on the official CCTV website without any form of tax or fine on unmarried or child- dismissing the birth fund proposal as “ridiculous,” less people, positive financial incentives seem diffi- arguing that the government should focus more on cult to sustain. positive incentives.46 Regarding positive incentives, monetary rewards Dealing with High Fertility Rates Among Eth- for families with more than one child are clearly on the nic Minorities. During four decades of family plan- table. According to Chinese news sources, since 2018, ning, China has always imposed more lenient birth the National Health Commission has been “organiz- restrictions on ethnic minorities than on the majority ing experts” to discuss the possibility of “rewarding Han people.49 As a result, the ethnic population grew fertility” and estimate the likely effects of these mon- faster than the Han majority, which seems to alarm etary rewarding programs.47 If the pro-natalist Liaon- the government. This pattern is clearly illustrated in ing model is implemented nationwide, the positive Table A4: The populations of Muslim minorities and

97 URBANIZATION WITH CHINESE CHARACTERISTICS

Tibetans in poorer western China grew rather rapidly (and thus not explicitly ethnically) differentiated compared to Han people.50 Also, as of 2010, all major population policy that still holds on to restrictions for ethnic minority groups had higher proportions of areas with relatively high fertility rates while encour- 0- to 14-year-olds than the Han majority did, and aging childbearing in low-fertility areas. Tibetans, Uighurs, and Kazakhs again topped the list. Although the State Council National Population Development Plan (2016–30) mostly warns of the low Conclusion: A Tougher New Battle in the fertility rate and the aging population, it nonetheless War on Human Reproduction mentions “higher fertility rates among ethnic minorities than national average,” the “increase of ethnic The leadership seems to fully endorse the plan to minority population,” and “unbalanced population reverse four decades of fertility control policy since development among ethnic groups in some ethnic neither President Xi nor Premier Li mentioned the regions.” Clearly, the government is worried about term “family planning” after 2017. Moreover, insti- fast ethnic, especially Muslim, population growth and tutionally, the latest reform of the newly rebranded lagged Han population growth. National Commission of Health is substantial, com- The government is particularly worried about pletely removing the term “family planning” and “unbalanced population development among eth- weakening the fertility policy apparatus. In addition, nic groups in ethnic regions.”51 The State Council provincial-level legislation is demonstrating some Plan does not clarify which “ethnic regions” these flexibility in terms of ethnic minority population con- are. However, if we take into account that Tibetans, trol and policy on the third child. This is especially the Uighurs, and Kazakhs have particularly high popu- case for the northeast rust belt provinces that vow to lation growth rates, then politically restive Xinjiang boost fertility. and Tibet must be among the worst headaches for the Notwithstanding, the effect of the universal two- central government. It is not far-fetched to assume child policy since 2016 seems limited, and birth rates that some people believe “unbalanced population continued to drop in 2017 and the first half of 2018. development” will result in ethnic minority, espe- Worried about unbalanced population development cially Muslim, “takeover” of those border regions. among ethnic minorities, the government is unlikely The central government seems to be more inter- to allow truly independent fertility. Instead, it will ested in positive fiscal incentives to convince ethnic actively intervene to provide positive incentives for minorities to give birth to fewer children. Explic- Han majority couples to have more children and hold itly capping the number of births for ethnic minori- on to monetary incentive schemes for ethnic minorities while encouraging fertility for the Han majority ties to have fewer children. would be politically provocative and thus less plau- Low fertility rates have proved notoriously diffi- sible. Therefore, the central government intends to cult to raise worldwide, and state-sponsored financial carry on the “Fewer Births, Faster Wealth” program, incentives for parents can be expensive.53 Also, nega- a monetary incentive scheme for households in west- tive incentives that punish single people or childless ern China—where the population is largely composed couples and fertility-control programs targeted at of ethnic minorities—to limit the number of children ethnic minorities can meet strong resistance and thus they have.52 are politically costly. China is ready to continue the If the central government is really paranoid about war on fertility, and we are likely to hear the war horn the “explosion” of population of certain ethnic from the Fourth Plenum of the 19th Party Congress. minority groups, especially Muslims in Xinjiang, we But this new battle will be tougher, economically and cannot rule out the possibility of a more regionally politically.

98 KANGYU MARK WANG

Notes

1. Justin Fox, “Ending China’s Birth Limits Won’t Bring a Baby Boom,” Bloomberg Opinion, May 21, 2018, https://www.bloomberg. com/view/articles/2018-05-21/ending-china-s-birth-limits-won-t-bring-a-baby-boom. 2. Association of Industry and Commerce, “Time and Theme of 2018 Fourth Plenum of 19th Party Congress,” August 20, 2018, http://www.shjfzx.gov.cn/a/jingjifuwu/20180810/4774.html. 3. Andrew Mertha, “‘Fragmented Authoritarianism 2.0’: Political Pluralization in the Chinese Policy Process,” China Quarterly 200 (December 2009): 995–1012, https://www.cambridge.org/core/journals/china-quarterly/article/fragmented-authoritarianism-20- political-pluralization-in-the-chinese-policy-process/EA5E4FE9316DA47EB53C777C879DCA29; and Yik Chan Chin, “Policy Process, Policy Learning, and the Role of the Provincial Media in China,” Media, Culture & Society 33, no. 2 (March 2011): 193–210, https:// journals.sagepub.com/doi/abs/10.1177/0163443710393381. 4. State Council, “The Decision on Major Issues Concerning Comprehensively Deepening Reforms,” November 15, 2013, http:// www.gov.cn/jrzg/2013-11/15/content_2528179.htm. 5. Xinhua Net, “Communiqué of the Fifth Plenary Session of the 18th Central Committee of the CPC,” October 29, 2015, http:// www.xinhuanet.com//politics/2015-10/29/c_1116983078.htm. 6. Xinhua Net, “Full Text of Xi Jinping’s Report at 19th CPC National Congress,” November 13, 2017, http://www.xinhuanet.com/ english/special/2017-11/03/c_136725942.htm; and Li Keqiang, “Report on the Work of the Government,” People’s Daily Online, April 3, 2018, http://en.people.cn/n3/2018/0403/c90000-9445262.html. 7. Central People’s Government, “Plan of State Council Institutional Reform,” March 17, 2018, http://www.gov.cn/xinwen/2018-03/17/ content_5275116.htm. 8. Central People’s Government, “Main Duty, Internal Agencies and Personal Arrangement of National Health Commission,” Sep- tember 10, 2018, http://www.gov.cn/zhengce/2018-09/10/content_5320817.htm. 9. Hu Jintao, “Hold High the Great Banner of Socialism with Chinese Characteristics and Strive for New Victories in Building a Moderately Prosperous Society in All Respects,” September 17, 2018, http://www.china.org.cn/english/congress/229611.htm; and Hu Jintao, “Firmly March on the Path of Socialism with Chinese Characteristics and Strive to Complete the Building of a Moderately Pros- perous Society in All Respects,” November 8, 2012, http://www.china.org.cn/china/18th_cpc_congress/2012-11/16/content_27137540. htm. 10. Xinhua Net, “Full Text of Xi Jinping’s Report at 19th CPC National Congress.” 11. Xinhua Net, “Full Text of Xi Jinping’s Report at 19th CPC National Congress.” 12. State Council, “Full Text: Report on the Work of the Government (2014),” March 14, 2014, http://english.gov.cn/archive/ publications/2014/08/23/content_281474982987826.htm. 13. Keqiang, “Report on the Work of the Government.” 14. State Council, “Full Text: Report on the Work of the Government (2014).” 15. State Council, “Full Text: Report on the Work of the Government (2015),” March 16, 2015, http://english.gov.cn/archive/ publications/2015/03/05/content_281475066179954.htm. 16. State Council, “Full Text: Report on the Work of the Government (2016),” March 17, 2016, http://english.gov.cn/premier/ news/2016/03/17/content_281475309417987.htm. 17. Li Keqiang, “Report on the Work of the Government,” Xinhua Net, March 5, 2017, http://www.xinhuanet.com/english/ china/2017-03/16/c_136134017.htm. 18. Li Keqiang, “Report on the Work of the Government,” People’s Daily Online, April 3, 2018, http://en.people.cn/n3/2018/0403/ c90000-9445262.html. 19. Bloomberg News, “China Signals End to Child Birth Limits by 2020 at Latest,” August 27, 2018, https://www.bloomberg.com/ news/articles/2018-08-28/china-signals-end-to-limits-on-child-births-by-2020-at-latest. 20. Pkulaw, “Constitution of the People’s Republic of China (2018 Amendment),” September 15, 2018, http://en.pkulaw.cn/Display.

99 URBANIZATION WITH CHINESE CHARACTERISTICS

aspx?lib=law&Cgid=311950. 21. Central People’s Government, “Main Duty, Internal Agencies and Personal Arrangement of National Population and Family Planning,” http://www.gov.cn/banshi/qy/rlzy/2012-11/05/content_2257876.htm. 22. Central People’s Government, “Plan of State Council Institutional Reform.” 23. Central People’s Government, “Main Duty, Internal Agencies and Personnel Arrangement of Ministry of Health,” http://www. gov.cn/zhengce/content/2016-06/12/content_5081151.htm. 24. Sina News, “Institutional Reform Due Times,” http://k.sina.com.cn/article_6143128258_16e28b2c20190076dx.html. 25. China Urban-Townization Promotion Council, “Ministry of Natural Resources Will Set Up 6 Divisions and 16 Departments,” http://www.cupc.org.cn/index.php?m=content&c=index&a=show&catid=182&id=1352; and Paper, “Divisions of Ministry of Veteran Affairs,”https://www.thepaper.cn/newsDetail_forward_2181035 . 26. Central People’s Government, “Main Duty, Internal Agencies and Personal Arrangement of National Health Commission.” 27. The three subdivisions are the Local Family Planning Direction Subdivision, the Family Planning Development Subdivision, and the Floating Population Family Planning Service and Management Subdivision. 28. Reuters, “China Does Away with Family Planning Offices,” September 10, 2018,https://ca.reuters.com/article/topNews/ idCAKCN1LR06Q-OCATP; and Central People’s Government, “Main Duty, Internal Agencies and Personal Arrangement of National Health Commission.” 29. Huaxia Times, “Three Decision on National Health Commission Published,” iFeng, September 12, 2018, http://finance.ifeng. com/a/20180912/16498577_0.shtml. 30. Sina News, “Institutional Reform Due Times.” 31. National People’s Congress of People’s Republic of China, “Law of Population and Family Planning,” http://www.npc.gov.cn/npc/ xinwen/2015-12/28/content_1957360.htm; and Central People’s Government, “Law of Population and Family Planning,” http://www. gov.cn/banshi/2005-08/21/content_25059.htm. 32. Xinjiang Health Inspection, “Xinjiang Uyghur Autonomous Area Population and Family Planning Rules,” http://www.xjwsjd.gov. cn/flfg/jgsyfg/xjwwezcgbb/14509.htm; Yunnan Human Resource and Social Security Net, “Yunnan Province Population and Family Planning Rules,” http://www.ynhrss.gov.cn/NewsView.aspx?NewsID=18417&ClassID=682; and Guangxi Zhuang Autonomous Region Health and Family Planning Commission, “Guangxi Zhuang Autonomous Region Population and Family Planning Rules,” http://www. gxhfpc.gov.cn/gzdt/bt/2016/0117/17291.html. 33. Liaoning Province People’s Government, “Liaoning Province Population Development Plan,” http://www.ln.gov.cn/zfxx/zfwj/ szfwj/zfwj2011_125195/201807/t20180703_3273280.html. 34. Xinjiang Health Inspection, “Xinjiang Uyghur Autonomous Area Population and Family Planning Rules,” http://www.xjwsjd.gov. cn/flfg/jgsyfg/xjwwezcgbb/14509.htm. 35. Tianshang Net, “Xinjiang Uyghur Autonomous Area Population and Family Planning Rules,” http://www.ts.cn/special/ content/2007-06/05/content_1960020.htm. 36. Central People’s Government, “National Population Development Plan,” http://www.gov.cn/zhengce/content/2017-01/25/ content_5163309.htm. 37. 21st Century Economic Report, “Local Statistic Bureaus Survey Effects of Two Child Policy: Weak Fertility Desire,” August 24, 2018, http://3g.163.com/news/article/DPVFPIQC029797TC.html. 38. Xinhua News, “Health and Family Planning Commission Introduces Universal Two-Child Policy,” http://www.xinhuanet.com/ live/20151110z/index.htm. 39. National Bureau of Statistics, “Statistical Communiqué on National Economic and Social Development of People’s Republic of China in 2015,” February 29, 2016, http://www.stats.gov.cn/tjsj/zxfb/201602/t20160229_1323991.html; National Bureau of Statistics, “Statistical Communiqué on National Economic and Social Development of People’s Republic of China in 2016,” February 28, 2017, http://www.stats.gov.cn/tjsj/zxfb/201702/t20170228_1467424.html; and National Bureau of Statistics, “Statistical Communiqué on National Economic and Social Development of People’s Republic of China in 2017,” February 28, 2018, http://www.stats.gov.cn/tjsj/ zxfb/201802/t20180228_1585631.html.

100 KANGYU MARK WANG

40. National Bureau of Statistics, “National Bureau of Statistics Notice on Population Change Survey,” September 5, 2018, http:// www.stats.gov.cn/tjgz/tzgb/201809/t20180905_1621099.html. 41. Wang Hongru, “Our Country Might Be Able to Enjoy Independent Fertility After 2020,” July 19, 2016, http://epaper.gmw.cn/wzb/ html/2016-07/19/nw.D110000wzb_20160719_4-01.htm. 42. 163 News, “New Trend After 2020 and China Might Realize Autonomous Fertility?,” http://news.163.com/16/0715/14/ BS189VI300014AED.html. 43. Bloomberg, “China Considers Ending Birth Limits as Soon as This Year,” May 21, 2018, https://www.bloomberg.com/news/ articles/2018-05-21/china-said-to-consider-ending-birth-limits-as-soon-as-this-year. 44. Global Times, “Party Newspaper Calls Giving Birth a ‘State Affair,’” August 6, 2018,http://www.globaltimes.cn/content/1114047. shtml. 45. Charlotte Gao, “To Encourage More Births, Chinese Specialists Propose Birth Fund, Childless Tax,” Diplomat, August 17, 2018, https://thediplomat.com/2018/08/to-encourage-more-births-chinese-specialists-propose-birth-fund-childless-tax/. 46. CCTV News, “‘Setting Up Birth Fund System’ Is an Absurd Proposal,” August 17, 2018, http://news.cctv.com/2018/08/17/ ARTI9rKqMejVTmx4PQvsUe8Z180817.shtml. 47. Sina News, “National Health Commission Is Investigating the Possibility of Rewarding Fertility,” Paper, July 13, 2018, http://baby. sina.com.cn/news/2018-07-13/doc-ihfhfwmu4155002.shtml. 48. For example, see Huoshen Bahao, “‘Fertility Fund’? You Are a Fucking Genius,” 163 News, August 18, 2018, http://3g.163.com/dy/ article/DPG3J5OB05219T0D.html; and China Youth News, “How Is Social Payment of Upbringing Used?,” September 16, 2018, http:// zqb.cyol.com/html/2013-09/16/nw.D110000zgqnb_20130916_1-08.htm. 49. Economist, “China Allows All Couples to Have Two Children,” October 29, 2015, https://www.economist.com/china/2015/10/29/ china-allows-all-couples-to-have-two-children. 50. The slow population growth among Mongols and Manchus from 2000 to 2010 looks rather mysterious. Notably, many Mongol people and most Manchu people live in northeastern rust belt provinces, where fertility rates have always been low. 51. Central People’s Government, “National Population Development Plan.” 52. General Office of State Council, “Division of Fiscal Responsibilities Between Central and Local Governments in the Field of Medicine and Health,” August 13, 2018, http://www.gov.cn/zhengce/content/2018-08/13/content_5313489.htm. 53. Nicholas Eberstadt, “The Demographic Future: What Population Growth—and Decline—Means for the Global Economy,” For- eign Affairs 89, no. 6 (November/December 2010): 54–64, https://www.jstor.org/stable/20788716?seq=1#page_scan_tab_contents.

101 Appendix B. Changes in the Classification of Urban and Rural Population in China’s National Statistics

Kangyu Mark Wang

he People’s Republic of China’s (PRC) census residents’ committees and villagers’ committees as a Tclassifies human settlements into cities (cheng- base unit for classifying the population. shi), towns (chengzhen), and rural areas (xiangcun). Second, I compare the current rules for classifi- To compare urbanization statistics in China over time cation of cities, towns, and rural areas in China with and with other countries, we need to understand the the UNPD’s population-based classifications of big criteria used to classify the urban and rural population cities (300,000+), small cities (50,000–300,000), in China’s censuses. and urban areas. While the UNPD strictly follows Unfortunately, the six censuses that the PRC con- population criteria, the classification of cities and ducted used differenturban-rural distinction rules. towns in China is largely administrative, depending The current rule made for the 2010 Sixth National on the rules of reclassification of counties xian( ) as Population Census is based on idiosyncratic and county-level cities (xian ji shi) and townships (xiang) politicized administrative divisions. Therefore, we as towns (zhen). China’s population-based rules are should be cautious when comparing urbanization usually much lower than the UNPD’s thresholds, but indicators among different rounds of national cen- they should be seen as minimum requirements. Other suses and between China’s statistics and the United economic and political factors usually play more Nations Population Division (UNPD) figures. important roles, with significant regional variation. First, I compare the rules used by China’s six One of the challenges in classifying urban and rural national censuses and examine whether they residential areas is that the censuses classify areas into under- or overestimated urbanization rates. From the units called “cities” and “towns,” which exist within Second National Population Census in 1964 to the larger administrative cities. For example, a large city Sixth National Population Census in 2010, all cen- such as Hangzhou may have areas designated as urban suses have combined geographical and administrative towns and cities and areas that are considered rural. standards to classify urban and rural populations. In general, administrative rules identify urban residential areas as those with a larger population (munici- Comparisons Among Six National palities and administrative cities), while geographical Censuses rules classify urban and rural residents based on the size of the permanent population and on popula- There have been three official regulations for tion density in suburbs and smaller residential zones. urban-rural distinction: (1) the 1955 State Council The current rule made in 2008 incorporates rural Rules About Urban-Rural Distinction, (2) the 1999

102 KANGYU MARK WANG

Figure B1. Conglomerate Classification Rules Set by the 1955 State Council Rules About Urban- Rural Distinction

Residential Areas

Urban Cities and Residential Rural Areas Towns Areas

Cities

Towns

Source: State Council, “Rules on Statistical Urban-Rural Distinction,” http://www.jxsmz.gov.cn/New/MainShow.aspx?id=302.

Rules About Statistical Urban-Rural Distinction (for There were two definitions of “urban residential trial implementation) by the State Bureau of Statis- areas” (chengzhen xing juminqu). The first included tics, and (3) the 2008 Rules on Statistical Urban-Rural industrial and mining enterprises, railway stations, Distinction by the State Bureau of Statistics. industrial and commercial centers, transportation hubs, schools above the middle school level, loca- The 1955 State Council Rules About Urban-Rural tions of scientific research institutions, and residen- Distinction.1 In the 1955 rules, “cities and towns” tial areas for employees. These had to have more than (chengzhen) were defined as (1) prefectural seats and 1,000 permanent residents, and the nonagricultural localities of People’s Committees (the predecessor of population had to exceed 75 percent. The second was governments) above the county level and (2) residen- convalescent centers, where the convalescence pop- tial areas with more than 2,000 permanent residents, ulation exceed permanent residents by more than more than 50 percent of whom were not nonagricul- 50 percent. tural. “Cities and towns” were further divided into If a suburban area was adjacent to cities, it was “cities” (chengshi) and “towns” (jizhen). (See Figure considered “cities and towns,” regardless of the pro- B1.) The following were considered “cities” and the portion of the nonagricultural population. Other resi- rest considered “towns”: dential areas were considered rural areas.

• Municipalities (e.g., Beijing and Shanghai), The 1999 Rules About Statistical Urban-Rural • Prefectural-level cities, and Distinction (for Trial Implementation).2 In • County seats and “industrial and commercial the 1999 rules, the definition of “cities and towns” areas” with a permanent population of more (chengzhen) was divided into “cities” and “towns.” than 20,000 people. “Cities” were further divided into “downtown areas

103 URBANIZATION WITH CHINESE CHARACTERISTICS

Figure B2. Conglomerate Classification Rules Set by the 1999 Rules About Statistical Urban- Rural Distinction (for Trial Implementation)

Residential Areas

Cities and Towns Rural Areas

Cities Towns

Downtown Areas Administrative Towns of Cities with That Are Seats of Districts Aﬃliated Governments at and Above County Level Downtown Areas of Cities Without Other Districts Aﬃliated Administrative Towns

Source: State Council, “Rules on Statistical Urban-Rural Distinction (for Trial Implementation),” http://www.stats.gov.cn/tjsj/ndsj/ renkoupucha/2000pucha/html/append7.htm.

of administrative cities with districts affiliated” and to adjacent administrative towns and townships “downtown areas of cities without districts affili- (xiang), the entire administrative areas of those ated.” (See Figure B2.) towns and townships would be considered “down- For administrative districts of administrative cities, town area.” if their population density was at or above 1,500/km², “Towns” were “downtown areas of towns.” They their entire administrative districts would be con- included administrative town seats and other urban sidered “downtown area.” If it was below 1,500/km², residential committees (jumin weiyuanhui) that only the district seat and other administrative subdis- belonged to the administrative town. If the “con- tricts were considered “downtown area.” If the “constructed area” of town seats had already expanded structed area” of district seats had already expanded to adjacent villagers’ committees (cunmin weiyuan- to adjacent administrative towns and townships, the hui), entire administrative areas of those villagers’ entire administrative areas of those towns and town- committees would be considered “downtown area.” ships would be considered “downtown area.” Some industrial and mining areas, development In the definition of “downtown areas of admin- zones, tourist areas, scientific research institutes, uni- istrative cities without districts affiliated,” only the versities, and colleges that were located outside towns city seat and other administrative subdistricts were and cities were included in the “towns and cities” des- considered “downtown area.” However, if the “con- ignation. If they had more than 3,000 permanent res- structed area” of city seats had already expanded idents, they would be considered “towns”; otherwise,

104 KANGYU MARK WANG

Figure B3. Conglomerate Classification Rules Set by the 2008 Rules on Statistical Urban-Rural Distinction by State Bureau of Statistics

Residential Areas

Cities and Towns Rural Areas

Downtown Downtown Area of Cities Area of Towns

Urban Residential Committees Urban Residential Committees That Belong to Subdistricts of That Belong to Administrative Cities (Regardless Administrative Towns of Whether They Have Districts Aﬃliated) Areas of Other Urban Residential Committees and Villagers’ Areas of Other Urban Residential Committees That “Urban Committees and Villagers’ Facilities” Expand to Committees That “Urban Facilities” Expand to Special Areas

Source: State Council, “Rules on Statistical Urban-Rural Distinction,” http://www.stats.gov.cn/tjsj/tjbz/200610/t20061018_8666.html. they were considered rural areas. Other residential “Downtown area of towns” include: areas were considered rural areas. • Urban residential committees that belong to The 2008 Rules on Statistical Urban-Rural Distinc- administrative towns, tion by the State Bureau of Statistics.3 In the 2008 • Areas of other urban residential committees rules, “cities and towns” (chengzhen) include “down- and villagers’ committees that “urban facilities” town area of cities” and “downtown area of towns.” (See expand to, and Figure B3.) “Downtown area of cities” include: • Special areas, including industrial and mining areas, development zones, research institutes, • Urban residential committees that belong to universities, farms, and forest farms, with more subdistricts of administrative cities (regardless than 3,000 permanent residents. of whether they have districts affiliated) and • Areas of other urban residential committees Other residential areas are considered rural areas. and villagers’ committees that “urban facilities” expand to.

105 URBANIZATION WITH CHINESE CHARACTERISTICS

Observations on How Urban-Rural Populations urban-rural distinction. The population of “cities and Are Qualified in Six Censuses. The rules above towns” was considered urban population, and every- were applied unevenly in the different census sur- thing else was rural population. veys. Below are some comments on how each census adjusted the rules. Changing Bias of Urbanization Statistics. As outlined above, all six censuses followed different First Census (1953). No unified code was available back rules to classify urban and rural population, caus- then, and the criteria were inconsistent among prov- ing inconsistency in urbanization statistics. A range inces. Generally, only administrative cities, county of Chinese literature analyzed the directions of his- seats, “towns with prosperous commerce and indus- torical bias. In summary, the First, Third, and Sixth try and more than two or three thousand population,” Census results are most accurate; the Second Census industrial and mining enterprises, and forestry oper- underestimated the urban population; and the Fourth ating areas were considered urban areas.4 and Fifth Censuses slightly overestimated the urban population. Second Census (1964). The definition of “urban” was narrowed to include only the nonagricultural popula- First Census and 1953–62: Mostly Accurate. Statistics tion in cities and towns, as defined by the 1955 State during this time relatively strictly followed the 1955 Council rules about urban-rural distinction. State Council rules about urban-rural distinction, and downtown population size and nonagricultural pop- Third Census (1982). The definition of “urban” was ulation proportion were used to identify administra- expanded to include the agricultural population, tive city and town areas.7 Therefore, most scholars again in cities and towns, as defined by the 1955 State believe these figures accurately reflected the real sizes Council rules about urban-rural distinction. of the rural and urban population.8

Fourth Census (1990).5 The population-based crite- Second Census and 1963–75: Underestimate. Because of ria stipulated by the 1955 State Council rules about the economic disaster caused by the Great Leap For- urban-rural distinction were completely abandoned ward and the Cultural Revolution, the state during due to the rapid establishment of cities and towns that time cut the number of cities and towns to reduce and expansion of city and town administrative bor- the urban population. As a result, the criteria for city ders. Therefore, “city and town population” (shizhen and town establishment were raised. In particular, renkou) was divided into “city population” and “town only the nonagricultural population was classified population.” Administrative borders at lower lev- as urban population. Therefore, the second census els (districts, subdistricts, rural residential commit- underestimated the urban population and urbaniza- tees, and villagers’ committees) were used to identify tion ratio.9 them.6 Third Census and 1976–83: Mostly Accurate. The Third Fifth Census (2000). The distinction between urban Census in 1982 again used the administrative criteria and rural population was made in accordance with set in 1955, and it was only after 1983 that new admin- the 1999 rules about statistical urban-rural distinction istrative towns and cities with mostly rural population (for trial implementation). The population of “cities were established. Therefore, the urbanization ratio in and towns” was considered urban population, and the 1982 census is believed to be mostly accurate.10 everything else was rural population. Fourth Census and 1984–99: Slight Overestimate. With Sixth Census (2010). The Sixth National Popula- administrative cities and towns rapidly expanding, tion Census followed the 2008 rules on statistical they could no longer be the basis of the rural-urban

106 KANGYU MARK WANG

distinction. As a result, the 1990 census began to and other economic and political factors usually play use districts, subdistricts, and residential commit- more important roles, with significant regional varia- tees to identify urban areas, giving more or less accu- tions. In general, the minimal population criteria are rate results. However, it slightly overestimated city higher in the eastern provinces than in the western population and underestimated town population provinces and mountainous areas. and, in total, still slightly overestimated total urban population.11 Rules of Settlement Classification. As shown in Table B1, the UNPD strictly adheres to population size Fifth Census and 1999–2008: Slight Overestimate. The standards in classifying settlements. However, admin- 1999 rules about statistical urban-rural distinction istrative classification is the basis for settlement clas- (for trial implementation) made only slight changes sification in China. Whether an urban settlement is to the rules for the Fourth Census in 1990.12 It tight- considered a city largely depends on whether it is an ened the rules for the city population but loosened administrative county seat or administrative city seat. the rules for the town population and, in total, still Similarly, whether a settlement is considered a town slightly overestimated total urban population. or rural depends on whether it is an administrative town seat. Since 2008: Mostly Accurate. The 2008 rules on statistical urban-rural distinction corrected the overesti- Establishment of Administrative Cities in mation of the urban population in the 1999 rules by China. Two recent pieces of national-level legis- using “constructed area” standards instead of popula- lations regulate the establishment of administration density standards.13 As a result, statistical figures tive cities in China: the 1993 Report About Adjusting on urban population have decelerated since 2005.14 City Establishment Criteria and the 2016 Criteria of County-Level City Establishment.

Comparison Between Chinese and 1993 Report About Adjusting City Establishment Cri- UNPD Rules of the Classification of teria.15 The 1993 Report About Adjusting City Estab- Conglomerates lishment Criteria set four types of county-level city establishment criteria: (1) population density of the The UNPD estimate of China’s big-city population whole county area, (2) the population size of a county in 2010 (414.8 million) is close to China’s 2010 cen- seat, (3) the economic power of the county, and (4) its sus city population (403.8 million), and the UNPD military-political importance. They are summarized estimate of small-city population (254.6 million) in Table B2. For a county to become a country-level is close to China’s 2010 census town population city, it needs to satisfy the four types of criteria in any (266.2 million). While the UNPD strictly follows one of the three rows in the table. Similarly, the estab- population criteria, the classification of cities and lishment of prefectural-level cities follows both pop- towns in China is largely administrative, depend- ulation and economic power criteria, as summarized ing on the rules of reclassification of counties as in Table B3. county-level cities and the reclassification of town- After 1993, China saw a big surge in the number of ships as towns. county-level cities, many of which were established China’s quantified, population-based standards based on “bribery and falsified materials,” resulting in are much lower than the UNPD’s thresholds: The “false urbanization.”16 Therefore, the 1993 rules were minimum population requirement for county-level suspended in 1997, and administrative cities began to cities is around 150,000, while the minimum popu- be established on purely a case-by-case basis.17 From lation requirement for towns is around 6,000. How- 1997 to 2016, new county-level cities were mostly in ever, these should be seen as minimum requirements, border and ethnic minority areas.

107 URBANIZATION WITH CHINESE CHARACTERISTICS

Table B1. UNPD and 2010 PRC Census Rules of Settlement Classification

Residential Type 1 2 3

UNPD Name Big Cities Small Cities Rural Estimation Definition Urban agglomerations of Urban areas with pop- Settlements smaller than 300,000 or more ulations of fewer than 50,000 persons 300,000 but more than 50,000 2010 PRC Name Cities Towns Rural Census Definition Urban residential com- Urban residential com- Settlements smaller than Estimation mittees that belong to mittees that belong to 50,000 persons subdistricts of adminis- county seats and admin- trative cities (regardless istrative towns; areas of of whether they have other urban residential districts affiliated); areas committees and villagers’ of other urban residential committees that “urban committees and villagers’ facilities” expand to; committees that “urban special areas facilities” expand to

Source: UN Population Division, World Urbanization Prospects: The 2014 Revision, 2014, http://esa.un.org/unpd/wup/; and National Bureau of Statistics, “Rules on Statistical Urban-Rural Distinction,” http://www.stats.gov.cn/tjsj/tjbz/200610/t20061018_8666.html.

Table B2. County-Level City Establishment Criteria Set by the 1993 Report About Adjusting City Establishment Criteria

Population Density of Population Size of the Military-Political Economic Power the Whole County Area County Seat Importance

Industrial output >= 80 percent Criteria can be >= 120,000; with urban >= 400 per km2 of total industrial and agricultural relaxed for coun- hukou population >= 80,000 output; GDP >= 1 billion yuan ties with special political and mili- Industrial output >= 70 percent tary importance. >= 100 per km2 and <= >= 100,000; with urban of total industrial and agricultural 400 per km2 hukou population >= 70,000 output; GDP >= 800 million yuan

Industrial output >= 60 percent >= 80,000; with urban <= 100 per km2 of total industrial and agricultural hukou population >= 60,000 output; GDP >= 600 million yuan

Source: State Council, “Report About Adjusting City Establishment Criteria,” http://www.gzsmzt.gov.cn/xxgk/xxgkml/zcwj/fgwj/ 201702/t20170228_1961090.html.

108 KANGYU MARK WANG

Table B3. Prefectural-Level City Establishment Criteria Set by the 1993 Report About Adjusting City Establishment Criteria

Population Size of City Seat Economic Power

>= 250,000; with urban hukou population Industrial output >= 80 percent of total industrial and agricultural >= 200,000 output; GDP >= 2.5 billion yuan

Source: State Council, “Report About Adjusting City Establishment Criteria,” http://www.gzsmzt.gov.cn/xxgk/xxgkml/zcwj/ fgwj/201702/t20170228_1961090.html.

2016 Criteria of County-Level City Establishment. A new Establishment of Administrative Towns in wave of county-level city establishment began in 2016 China: Localized Rules. There is currently no under the new Criteria of County-Level City Estab- national rule regulating the establishment of towns lishment.18 Population, economic, and public service and abolishment of townships in China, and provincial criteria remain. Most importantly, the permanent rules differ. For example, in Hebei, based on township population of a county seat should be no fewer than total population density, the number of permanent 150,000, still lower than the UNPD line of 300,000. urban residents of new towns should be no fewer than However, these quantified criteria are said to be 4,500–8,000 people.19 By contrast, in Henan, where only in principle and would be tightened for east- population density and total population are also crite- ern China and relaxed for western China. Also, the ria, the number of permanent urban residents of new actual approval would still follow a case-by-case pro- towns should be no fewer than 5,000–8,000 people.20 cess. In short, although hard and quantified criteria of All these urban population criteria should be seen as city establishment have always been lower than the minimal requirements because economic and politi- UNPD’s 300,000 line, other criteria have been rather cal factors are also taken into account. opaque and politicized, and increasingly so.

109 URBANIZATION WITH CHINESE CHARACTERISTICS

Notes

1. State Council, “State Council Rules About Urban-Rural Distinction,” December 31, 2017, http://xzqh.mca.gov.cn/statistics/2017. html. 2. National Bureau of Statistics, “Statistical Urban-Rural Distinction (for Trial Implementation),” http://www.stats.gov.cn/tjsj/ndsj/ renkoupucha/2000pucha/html/append7.htm. 3. National Bureau of Statistics, “Rules on Statistical Urban-Rural Distinction,” http://www.stats.gov.cn/tjsj/tjbz/200610/ t20061018_8666.html. 4. State Council, “State Council Rules About Urban-Rural Distinction.” 5. Zhou Yixing and Sun Ying, “Dui Woguo Shizhen Renkou Bizhong de Fenxi,” Zhongguo Tongji 5 (1993): 20–25, http://www.cnki. com.cn/Article/CJFDTotal-ZGTJ199305008.htm. 6. City residents: For cities at and above prefectural level, only residents in “districts” (administratively, qu) are considered city residents. For cities without districts affiliated, only residents in “subdistricts” (administratively,jiedao banshichu) are considered city residents. Town residents: For cities without districts affiliated, only residents in urban residential committees (administratively, chengshi jumin weiyuanhui) are considered town residents. For towns affiliated to counties, only residents in urban residential committees (administratively, chengshi jumin weiyuanhui) are considered town residents. 7. State Council, “State Council Rules About Urban-Rural Distinction.” 8. Zhang Qingwu, “Zhongguo Chengxiang Huafen yu Chengzhen Renkou Tongji Wenti,” Renkou yu Jingji 3 (1989): 3–7, http://rkyjj. cueb.edu.cn/dzqk/62134.htm. 9. Zhang, “Zhongguo Chengxiang Huafen yu Chengzhen Renkou Tongji Wenti.” 10. Zhou Yixing and Sun Ying, “Dui Woguo Disici Renkou Pucha Shizhen Renkou Bizhong de Fenxi,” Renkou yu Jingji 1 (1992): 21–27, http://www.cnki.com.cn/Article/CJFD1992-RKJJ199201003.htm. 11. Wang Fang, “Shizhen Shezhi Biaozhun ji Chengzhen Renkou Tongji Koujing Dui Zhongguo Chengshihua Fazhan de Yingxiang,” Renkou yu Fazhan 17, no. 2 (2011): 82–87, http://www.oaj.pku.edu.cn/rkyfz/CN/article/downloadArticleFile.do?attachType= PDF&id=29814; and Chen Jinyong, “Dangqian Zhongguo de Chengzhen Renkou Tongji Wenti ji qi dui Jingji Fenxi de Yingxiang,” Renkou yu Laodong Lvpishu, 2010, https://wenku.baidu.com/view/818a89f90242a8956bece4ad.html. 12. National Bureau of Statistics, “Statistical Urban-Rural Distinction (for Trial Implementation).” 13. National Bureau of Statistics, “Rules on Statistical Urban-Rural Distinction”; National Bureau of Statistics, http://www.stats.gov. cn/tjsj/tjbz/200610/t20061018_8666.html; and Tao Li, Yuan Jing, and Fan Chunke, “Jiyu Liupu Tongji Kouji dui 1982-2012 Nian Chengzhenhua Shuiping de Xiuzheng,” Zhongguo Waizi 10 (2014), http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id= zgwz-s201410130. 14. Wang Fang, “Shizhen Shezhi Biaozhun ji Chengzhen Renkou Tongji Koujing Dui Zhongguo Chengshihua Fazhan de Yingxiang.” 15. Xia Hai and Zhongguo Zhengfu Jiagou, Qinghua Daxue Chubanshe, 2001, https://books.google.com/books?id= qkfXM428fQ4C&pg=PA7&lpg=PA7&dq=%E5%85%B3%E4%BA%8E%E8%B0%83%E6%95%B4%E8%AE%BE%E5%B8%82%E6%A0%8 7%E5%87%86%E7%9A%84%E6%8A%A5%E5%91%8A+1993&source=bl&ots=QgthxoYbqf&sig=DDchl8A8jYI7KvMfZbkbFWcNF vs&hl=en&sa=X&ved=2ahUKEwiP5sGqhZXdAhWGwFkKHQ4xBPU4ChDoATADegQICBAB#v=onepage&q=%E5%85%B3%E4%BA% 8E%E8%B0%83%E6%95%B4%E8%AE%BE%E5%B8%82%E6%A0%87%E5%87%86%E7%9A%84%E6%8A%A5%E5%91%8A%20 1993&f=false. 16. Gao Yanjun, “New Round of County Abolishment and City Establishment in 2017,” Guancha, April 18, 2017, https://www.guancha. cn/kengqiangxia/2017_04_18_404103.shtml. 17. Central Committee of Chinese Communist Party and State Council, “Notice on Strengthening Land Management and Protect- ing Arable Land,” News of the Communist Party of China, April 15, 1997, http://cpc.people.com.cn/GB/64162/71380/71382/71481/4854245. html. 18. Lin Xiaozhao, “Without Criteria Published, Why Can These Six Counties Become Cities,” Yi Caijing, April 18, 2018, http://m.yicai.

110 KANGYU MARK WANG

com/news/5268796.html. 19. Department of Civil Affairs of Hebei Province, “Notes on Abolishment of Townships and Establishment of Towns and Abolish- ment of Towns and Establishment of Sub-Districts,” December 14, 2015, http://www.hebmz.gov.cn/dmpcw/zcfg/201512/t20151214_45868. html. 20. Lankao County Office of Administration, “Criteria of Town Establishment,” August 8, 2016,http://www.lkxdmw.com/fwzn/561. htm.

111 Appendix C. The Political Economy of China’s Evolving Hukou Regime

Kangyu Mark Wang

he hukou regime that restricts the Chinese peo- cherry-picking talented individuals and excluding the Tple’s free movement within the country has pro- majority of low-skill migrants. Therefore, this appen- found implications for various aspects of China’s dix argues that the newest round of reform does not economic and social life. In particular, it obstructs address a major problem of the hukou system: unequal people’s rural-urban movement and denies migrants access to social security and public services between city social welfare and public services. The relaxation low-skill migrant workers and local residents in big of the hukou system is now considered the major drive cities in eastern China. for China’s “new modernization” plan. At the national level, the Third Plenum of the 18th Party Congress in 2013 outlined the principles of the Principles of the Newest Round of Hukou newest round of hukou reform, which were further Reform detailed in the “National Plan on New Urbanization (2014–2020)” (hereafter “New Urbanization Plan”) The process to grant rural migrants urban hukou, or and “Opinions of the State Council on Further Pro- the “citizenization” of rural migrants, is not intended motion of Reform of the Household Registration for the majority of rural migrants. In 2017, 291 million System” (hereafter “Opinions on Household Regis- Chinese lived outside their designated hukou, among tration”) in 2014.1 The government promised to grant whom 244 million are considered the floating pop- city hukou for more rural migrants, but the hukou sys- ulation.3 However, according to the New Urbaniza- tem is still largely seen as a tool for population control tion Plan, the citizenization of rural migrants will be in super- and megacities. In addition, the government at best slow-paced. The government intends to close intends to further expand the coverage of social wel- the gap between the permanent residents’ urban- fare and public services for the floating population ization rate and the hukou urbanization rate by only without hukou, which is likely to meet resistance 2 percentage points. According to the New Urbaniza- from local governments due to concerns over public tion Plan, migrants in urban areas without city hukou financing. will still account for 15 percent of the Chinese popula- At the local level, worried about real estate over- tion by 2020 (Table C1). supply, some big and supercities in central and west- The expansion of non-hukou-based public services ern China have relaxed the requirements for city is still underdeveloped, and local governments lack hukou registration since 2016.2 However, even the an incentive to provide anything more than what the recently lowered standards are too high for the major- State Council requires. In addition, the move to abol- ity of migrant workers to meet, and the hukou regimes ish the nominal rural-urban hukou division is actu- of super- and megacities remain strictly meritocratic, ally hollow and does not shake the hierarchy of city

112 KANGYU MARK WANG

Table C1. The Timeline of Hukou Reform: New Urbanization Plan

Index 2012 2020

Permanent Residents Urbanization Rate (%) 52.6 About 60

Hukou Urbanization Rate (%) 35.3 About 45 Coverage of Compulsory Education Service Among Children of Rural Immigrant N/A >= 99 Workers (%) Coverage of Basic Endowment Insurance Among Urban Permanent Residents (%) 66.9 >= 90

Coverage of Basic Health Insurance Among Urban Permanent Residents (%) 95 98

Coverage of Basic Housing Service Among Urban Permanent Residents (%) 12.5 >= 23

Source: State Council, “Opinions of the State Council on Further Promotion of Reform of the Household Registration System,” July 24, 2014, http://www.gov.cn/zhengce/content/2014-07/30/content_8944.htm; 任远, “中国户籍制度改革: 现实困境和机制重构,” 南京社会科学 8 (2016): 46–52, http://www.cssn.cn/shx/shx_fzshx/201611/t20161101_3259162.shtml; and State Council, “National Plan on New Urbanization (2014–2020),” August 1, 2018, http://www.mohrss.gov.cn/gkml/xxgk/201509/ t20150925_221486.html.

welfare and public services provision or result in the Residence Permit System and Expansion of privatization of agricultural land. Non-Hukou Social Welfare and Public Services Coverage. Besides vowing to relax city hukou registra- Citizenization of Rural Migrants and Control tion rules, the government also intends to institution- over City Size. To date, the Chinese Communist alize the residence permit system as a gateway to a city Party (CCP) has never shown any intention to dis- hukou. Anyone who lives outside their original hukou mantle the hukou regime completely. Instead, hukou can apply for a residence permit, which gives them reform is seen as part of an urbanization drive, with “basic” access to labor, education, and health services. a focus on increasing the proportion of urban resi- However, the local governments determine the dents with city hukou.4 The CCP considers the hukou scale of public services provided for residence permit regime a curb on the population size of big and super- holders, and they do not have any incentive to expand cities. The Third Plenum of 18th Party Central Com- it. Firstly, the population registered under the local mittee articulated these principles: hukou still acts as the basis for calculating many key performance indicators of the government, and thus We will help the eligible population to move away officials in big cities are incentivized to favor the local from agriculture and become urban residents. We population in welfare spending and public services. will introduce new population management meth- People in China who have a local hukou usually oppose ods, accelerate the reform of household registration relaxing the hukou regime and expanding the city’s system, completely lift restrictions on new residence public services to include immigrants.6 Secondly, the registration in administrative townships and small State Council has yet to come up with a nationwide cities, relax restrictions on new residence registra- transfer payment mechanism based on the actual per- tion in medium-sized cities in an orderly manner, lay manent population. Therefore, providing social secu- down appropriate conditions for new residence reg- rity and public services for rural migrants could incur istration in large cities, and strictly control the popu- heavy financial burdens on migration destinations lation size of megacities.5 without compensation.

113 URBANIZATION WITH CHINESE CHARACTERISTICS

Therefore, rural migrants still enjoy few, if any, evolution of hukou regime; it is still city-centrist but public services. Most importantly, children of rural has become more relaxed, elitist, and hierarchical. migrants, who do not have city hukou even if they are born there, are thus largely unable to attend public Social Control. The hukou regime was initially high school or take part in the college entrance exam- established as the People’s Republic of China govern- ination. Hence, they either go back home for educa- ment began to consolidate itself in the early 1950s, tion and become left-behind children or get employed with the purpose of “surveilling and controlling in low-paying jobs.7 counter-revolutionary and suspicious elements.”10 The function of surveillance and control was largely Abolishment of the Rural-Urban Divide in replaced by the resident identity card system adopted Hukou Regime. Opinions on Household Registra- in 1985, on which the current state surveillance regime, tion also directed local governments to nominally using the internet and artificial intelligence, is based.11 abolish the division between rural and urban hukou However, the dissociation between hukou registration and merge them into “residential hukou.” The name and the urban population’s actual geographical loca- change lacks substance. tion still worries some policymakers because they On the one hand, the de facto rural-urban divide in “make urban social governance more difficult.”12 access to a wide range of public services never disap- peared.8 Quite the contrary: As market reforms exac- Economic Management. The hukou regime is the erbated regional inequality, it morphed into regional pillar of the rural-urban dual economic structure that divides based on hukou origin. The residential hukou allows city industries to exploit the agricultural pop- of downtown Shanghai and the residential hukou of ulation, in both the Maoist and post-Maoist years.13 Tibetan grasslands still endow people with very dif- Since the mid-1950s, the Communist command econ- ferent access to public services. omy and overindustrialization began to bind the rural On the other hand, “Rural Land Contracted population to low-returning agricultural activities, Management Right,” or pseudo-ownership of nom- instead pouring resources into urban industries. This inally collectively owned rural land, remains locked pattern, however, was later replaced by the rapidly in with the hukou system. Instead of automati- developing coastal labor-intensive industries exploit- cally granting pseudo-ownership of land to peo- ing rural migrants without hukou, who were paid far ple with rural hukou, the new land regime grants less than urban state-owned enterprise employees land pseudo-ownership based on village commit- and had little access to basic social welfare and pub- tee membership. Currently, rural migrants can still lic services. If migrant workers were to get city hukou, reregister hukou in city areas without losing access city governments would need to pay for their welfare to land, and thus the dismantling of the rural-urban and public services. This is still considered the largest divide did not result in the full-scale privatization of cost of further hukou reform. agricultural land.9 Maoist economic policy, featuring command economy and state-led heavy industrialization, began to take shape in the early 1950s. To accelerate the accu- Understanding the Political-Economic mulation of industrial capital, the state tried to mini- Dynamic of Hukou Reform mize wages and thus needed to minimize food prices. Therefore, in 1953, the Communist state began to Social control and economic management are con- monopolize the trade of food: City residents were sidered the two key functions of the hukou regime. provided food for low prices, and the relative indus- Furthermore, the current economic model based on trial wage was set higher than the agricultural wage. mass real estate development and industrial upgrad- This naturally resulted in rural-urban migration, ing provides new dynamics at the local level to the which was contained only by the hukou regime.

114 KANGYU MARK WANG

The hukou regime’s original economic function is At the early stage of market reform, local differ- well illustrated by the fact that migration permission entiation of the city hukou regime did not change was given by the destination, in most cases cities. The the number of migrant workers drawn to cities, sug- city-centrism of hukou reregistration remained until gesting that the hukou regime had little impact on today. In short, Maoist China was both overindus- rural-urban migration itself. Migrant workers were trialized and over-urbanized compared to its level of going to leave for cities anyway, whether they could economic development, but it was under-urbanized enjoy public services or not.18 compared to its level of heavy industrialization. The As found by Ren Yuan, this pattern began to change hukou regime facilitated this development model by in the early 2000s, when the hukou regime began to severely limiting rural-urban migration. impair labor market liquidity; in other words, the Maoist economic policy failed to provide sufficient Lewis turning point had come to China.19 However, food for city residents. Because of the famine in the the heavy burden on city welfare and public services late 1950s, 20 million urban residents were forced expenditures is still seen as an important economic to go back to the countryside. Again during the Cul- hurdle to scrapping the hukou regime. tural Revolution, 40 million young urban residents were sent to the countryside. At the same time, con- Intragovernmental Competition over “Talent” trol on rural-urban migration was further tightened. and Real Estate Buying Power. An aversion to the The state created a nationwide annual rural-urban possibility of diverting local financial resources to migration quota: Urban hukou registration for rural rural migrants discourages local governments from migrants could not exceed 1.5 percent of the nonagri- pursuing more liberal hukou reforms. However, in cultural population.14 the newest round of hukou reform, big and supercit- The quota system of city hukou registration still ies began to relax hukou registration restrictions for exists today in some big cities. In 1975, the constitu- “talents” and people with real estate buying power. tion finally dropped the “migration freedom” clause. Local governments, which are financially depen- As shown by the quantitative research of Cai Fang et dent on mass real estate development and increas- al., the extent of food shortage was a strong predic- ingly under pressure to “upgrade” industry to sustain tor for the actual scale of rural-urban migration in the growth, have begun to see hukou policy as a tool to Maoist era.15 attract more desirable immigrants who boost the The food shortage was significantly eased and local economy. the state monopoly on food trade dismantled in the The current hukou registration rules implemented 1980s and 1990s, and food provision ceased to be sta- in most big, super-, and megacities are essentially tistically related to the scale of migration. Despite the competitive and meritocratic. Despite significant hukou regime, mass rural-urban migration took place regional variations, the qualifying requirements for at unprecedented rates and resulted in a large number big-city hukou put a similar emphasis on invest- of migrants working informally in coastal industries.16 ment, real estate ownership, education, and skills. However, the hukou regime was never eased. For example, top graduates from universities, central The new economic motivation behind the hukou and Beijing government personnel, returning over- system is to exploit cheap migrant labor by maintain- seas Chinese students, and successful entrepreneurs ing the unequal access to social security and public enjoy premium access to a Beijing hukou.20 Hukou services. As calculated by Peng Xizhe et al., from 1992 candidates are given different points according to to 2004, the hukou regime enabled the extraction of their career, educational, and financial backgrounds, around one trillion renminbi from migrant labor.17 and the cutoff line is determined yearly by the migra- Migrant workers were paid far less than workers with tion quota.21 urban hukou and had access to essentially no welfare This point-based meritocratic system in effect or public services. excludes all low-skilled labor. For megacities such as

115 URBANIZATION WITH CHINESE CHARACTERISTICS

Table C2. China’s City Hierarchy and DifferentHukou Regimes

The Level in the Urban Urban Area Hierarchy Permanent Population Hukou Regime Example Town

Small City Largely inclusive, open to all with <= 20,000 social endowment insurance Medium City 50,000–1 million enrollment Big City (Type II) 1–30 million Nanning, Luoyang Big City (Type I) 3–5 million Exclusive, combination of quota- Hefei, Xiamen Supercity 5–10 million and point-based systems; open Tianjin, Wuhan to locally employed university Guangzhou 10–15 million and associate college degree and Shenzhen holders Megacity Beijing and Exclusive; not open to most >= 15 million Shanghai university graduates

Source: Xinhua Net, “Li Keqiang Held State Council Executive Meeting, Planning In-Depth Promotion of People-Oriented New Urban- ization,” January 25, 2016, http://www.xinhuanet.com/mrdx/2016-01/25/c_135041446.htm; Guangdong Province Online Adminis- tration Hall Guangzhou Branch, “Hukou Registration in Guangzhou,” July 31, 2018, http://wsbs.gz.gov.cn/gz/rhzt/wdr.jsp; Shenzhen City People’s Government, “Notes on Hukou Registration in Shenzhen,” June 1, 2017, http://www.baoan.gov.cn/zxbs/zdyw/hjbl/ zcfg_114460/201712/t20171212_10472604.htm; Shanghai Bureau of Public Security, “Rules on Permanent Hukou Management,” March 8, 2018, http://www.shanghai.gov.cn/nw2/nw2314/nw2319/nw41149/userobject83aw882.html; and Beijing People’s Government, “Notes on Further Implementing Reforms on Household Registration,” September 8, 2016, http://zfxxgk.beijing.gov. cn/110001/szfwj/2016-09/10/content_11334cb8b12849b1915da0937f8d1a8e.shtml.

Shanghai, people holding master’s or even doctoral Beijing and Shanghai still insist on the tightest degrees sometimes fail to reach the minimal point to hukou policies in China. Other super- and megac- register hukou.22 Ironically, in the post-Mao era, the ities have significantly loosenedhukou control for hukou regime does not really curb the rural-urban “talents,” allowing university graduates and even migration of low-skill laborers, who are unable to set- associate degree holders to register, dropping the tle down in cities without a long-term employment requirement of local employment.25 However, the contract anyway. By contrast, it is more likely to con- majority of rural migrant workers still cannot meet strain “high-end talents” who want to enjoy city wel- even those lowered criteria with education and social fare and public services. insurance enrollment requirements. Therefore, the A new round of relaxation of big and megac- hukou regimes of mega- and supercities are still elitist ity hukou registration rules began in 2016. They are and exclusive. implemented under the banner of “talent attrac- Other big, medium, and small cities are supposed tion,” though critics believe that local governments to open hukou registration for all, according to the are more interested in stimulating the dangerously National Development and Reform Commission.26 overcrowded real estate market, on which they rely As argued by Zhang Guosheng and Chen Mingming27 financially.23 At the national level, Premier Li Keqiang and Ren Yuan,28 hukou of small and medium cities in 2016 encouraged the “complete liberalization” of does not endow its holders with many more wel- hukou registration for university graduates, skilled fare and public services than rural hukou, and thus laborers, and returning overseas Chinese students, rural migrants do not necessarily want to reregister. except in “some few megacities.”24 Nonetheless, to register hukou even in towns and

116 KANGYU MARK WANG

small cities, migrants are still required to be “legally with city hukou, or even equal access to city public and stably” employed, own or rent “legal and stable” services. First, the nationwide goal of hukou reregis- housing, and be enrolled in urban social endowment tration for rural immigrants is conservative, and big, insurance programs, presumably inviting bureau- super-, and megacities are still required to control cratic obstruction. the population size using the hukou policy. Second, despite State Council directions, local officials are dis- couraged politically and economically from expand- The Current Hukou Regime ing public services to include rural migrants. Third, the phaseout of the rural-urban divide in hukou does As shown above, the logic of social control and eco- not make a dent in the city hukou-based welfare and nomic management remains today, and historical public services hierarchy. institutions have left their imprints on the current The logic of social control and agricultural popu- hukou regime. Hukou registration at birth is jus san- lation exploitation has been in place since the estab- guinis; newborns are registered under the same Hu lishment of the hukou system. At present, hukou is no (household) as their parents. As a result, city-born longer the main unit of state surveillance, but the eco- children of rural migrant workers still cannot get city nomic benefits of denying migrant workers welfare hukou and thus cannot enjoy health care and educa- and public services still discourage city governments tion services in the city. from pursuing more liberal hukou reforms. In addi- After the abolishment of the nominal rural-urban tion, the latest wave of relaxing hukou registration divide of hukou types, the hierarchical structure of among most big cities, except Beijing and Shanghai, is different cityhukou is the most important hurdle a result of intragovernmental competition for human to rural-urban migration. As shown in Table C2, at resource and real estate buying power. As a result, the present, hukou registration in towns and small and hukou regime in big cities continues to be exclusive medium cities has largely been liberalized, open and elitist. By contrast, the almost-complete liberal- to all residents who have been enrolled in local ization of town and small-city hukou registration has social endowment insurance programs for some only minimal effect because it grants residents little time. Hukou regimes in some big cities and most privilege compared to rural hukou. super- and megacities are still largely closed to ordi- In short, the key features of the hukou regime— nary low-skilled labor but have become more inclu- namely, jus sanguinis registration policy and sive to university and associate college degree holders. city-centered, elitist, and hierarchical migration pol- The only exceptions are Beijing and Shanghai, where icy—remain intact despite rapid economic and social most university graduates or even advanced degree changes. The central government has shown little holders do not have the hope to get local hukou.29 intention to end the hukou system’s main issue— unequal access to social security and public services in big cities—and local governments clearly lack the Conclusion motivation to do so. As a result, despite recent registration liberalization for urban “talents,” the majority The newest round of hukou reform sees the relaxation of China’s engine of growth, low-skill migrant work- of urban hukou registration as an important dynamic ers, have little hope that they get hukou in big, super-, of China’s “new urbanization.” However, current pol- and megacities, as the state wants them to continue to icies cannot provide the majority of rural migrants be the engine of growth.

117 URBANIZATION WITH CHINESE CHARACTERISTICS

Notes

1. State Council, “National Plan on New Urbanization (2014–2020),” http://www.mohrss.gov.cn/gkml/xxgk/201509/t20150925_ 221486.html; and State Council, “Opinions of the State Council on Further Promotion of Reform of the Household Registration System,” July 24, 2014, http://www.gov.cn/zhengce/content/2014-07/30/content_8944.htm. 2. People’s Daily, “Results of ‘Competition for Talents’ in Round One: Multiple Localities See Hukou Population Burgeoning,” May 14, 2018, http://www.gqb.gov.cn/news/2018/0514/44870.shtml. 3. National Bureau of Statistics, “Statistical Communiqué of the People’s Republic of China on the 2017 National Economic and Social Development,” February 28, 2018, http://www.stats.gov.cn/tjsj/zxfb/201802/t20180228_1585631.html. 4. According to Deputy Director of National Development and Reform Commission Xu Xianping, the “most important progress” made by National Plan on New Urbanization (2014–2020) is the use of the index of the hukou urbanization rate (the proportion of urban residents with urban hukou) and its difference with the permanent residents urbanization rate (the proportion of de facto urban residents). State Council Information Office of the People’s Republic of China, “Press Conference on New Urbanization Plan,” March 19, 2014. 5. 18th Central Committee of the Communist Party of China, “Decision of the Central Committee of the Communist Party of China on Some Major Issues Concerning Comprehensively Deepening the Reform,” January 15, 2014, http://www.gov.cn/jrzg/2013- 11/15/content_2528179.htm. For the English translation, see https://china.usc.edu/decision-central-committee-communist-party-china- some-major-issues-concerning-comprehensively. 6. Peng Xizhe, Zhao Deyu, and Guo Xiuyun, “Huji Zhidu Gaige de Zhengzhijingjixue Sukao,” Fudan Xuebao: Shehui Kexue Ban 3 (2009): 1–11, http://www.fdwkxb.fudan.edu.cn/CN/Y2009/V0/I3/1. 7. Ren Yuan, “Zhongguo Huji Zhidu Gaige: Xianshi Kunjing yu Jizhi Chonggou,” Nanjing Shehui Kexue 8 (2016): 46–52, http://www. cssn.cn/shx/shx_fzshx/201611/t20161101_3259162.shtml. 8. Li Yu, “Woguo Tongchou Chengxiang Huji Zhidu Gaige de Lixing Sikao,” Renkou yu Jingji 5 (2011): 63–69, https://www.ixueshu. com/document/10afbf0724d7b734318947a18e7f9386.html. 9. State Council, “Opinions of the State Council on Further Promotion of Reform of the Household Registration System.” 10. China Legal Information, “Hukou Reform from a Legal Perspective,” April 27, 2016, http://www.legalinfo.gov.cn/index/ content/2016-04/27/content_6605258.htm. 11. Shannon Bond, “Inside China’s Surveillance State,” Financial Times, https://www.ft.com/content/1b8c193c-8e67-11e8-bb8f- a6a2f7bca546; and Li Tao, “A Look at China’s Push for Digital National ID Cards,” South China Morning Post, January 23, 2018, https:// www.scmp.com/tech/article/2129957/look-chinas-push-national-digital-id-cards. 12. People.cn, “Be Wary of the Negative Spillover Effect of ‘Professional War,’” July 10, 2018,http://leaders.people.com.cn/ n1/2018/0710/c58278-30137323.html. 13. Peng, Zhao, and Guo, “Huji Zhidu Gaige de Zhengzhijingjixue Sukao.” 14. State Council, “State Council Notice on Endorsing Ministry of Public Security Regulations on Migration of Hukou,” May 4, 2015, http://gaj.changchun.gov.cn/xxgk/flfg/201505/t20150504_1280733.html. 15. Cai Fang, Du Yang and Wang Meiyan, “Huji Zhidu yu Laodongli Shichang Baohu,” Jingji Yanjiu 12 (2001): 41–49. 16. Peng, Zhao, and Guo, “Huji Zhidu Gaige de Zhengzhijingjixue Sukao.” 17. Peng, Zhao, and Guo, “Huji Zhidu Gaige de Zhengzhijingjixue Sukao.” 18. Ren Yuan, “Zhongguo Huji Zhidu Gaige: Xianshi Kunjing yu Jizhi Chonggou,” Nanjing Shehui Kexue 8 (2016): 46–52, http://www. cssn.cn/shx/shx_fzshx/201611/t20161101_3259162.shtml. 19. Yuan, “Zhongguo Huji Zhidu Gaige: Xianshi Kunjing yu Jizhi Chonggou.” 20. Haidian People’s Government, “A Graphic Demonstration of Beijing Municipality Point-Based Hukou Registration Management Rule (for Trial Implementation),” April 11, 2018, http://www.bjhd.gov.cn/xinxigongkai/zcfg/zcjd/201804/t20180411_1502683.htm. 21. Xinhua Net, “Beijing Launches Point-Based Hukou System,” April 11, 2018, http://www.xinhuanet.com/english/2018-04/11/

118 KANGYU MARK WANG

c_137103450.htm. 22. Yuan, “Zhongguo Huji Zhidu Gaige: Xianshi Kunjing yu Jizhi Chonggou.” 23. Li Yan, “Talent Policies Boost House Prices,” Global Times, May 24, 2018, http://www.ecns.cn/news/economy/2018-05-24/detail-if-� yuqkxh5545300.shtml. 24. Xinhua Net, “Li Keqiang Held State Council Executive Meeting, Planning In-Depth Promotion of People-Oriented New Urban- ization,” January 25, 2016, http://www.xinhuanet.com/mrdx/2016-01/25/c_135041446.htm. 25. EZJSR, “Up-Grade of Hangzhou Hukou Registration Policy,” May 26, 2017, http://biz.zjol.com.cn/zjjjbd/zjxw/201705/ t20170526_4050845.shtml; Guangdong Province Online Administration Hall Guangzhou Branch, “Hukou Registration in Guangzhou,” July 31, 2018, http://wsbs.gz.gov.cn/gz/rhzt/wdr.jsp; Bendibao, “2018 Relaxation of Nanjing Hukou Registration,” July 12, 2018, http:// nj.bendibao.com/news/201815/66851.shtm; and Bendibao, “Chengdu Talent Recruit and Education Hukou Registration Criterion,” July 18, 2018, http://cd.bendibao.com/live/20141017/63129.shtm. 26. National Development and Reform Commission, “National Development and Reform Commission Note on Implementing Key Tasks of New Urbanization Promotion,” http://www.ndrc.gov.cn/gzdt/201803/t20180313_879342.html. 27. Zhang Guosheng and Chen Mingming, “Woguo Xin Yilun Huji Zhidu Gaige de Jiazhi Quxiang, Zhengce Pinggu yu Dingceng Sheji,” Jingji Xuejia 7 (2016): 58–65, http://jjxj.swufe.edu.cn/CN/article/downloadArticleFile.do?attachType=PDF&id=2536. 28. Ren Yuan, “Dangqiang Zhongguo Huji Zhidu Gaige de Mubiao, Yuanze he Lujing,” Nanjing Shehui Kexue 2 (2016): 63–70, http:// www.cssn.cn/shx/201607/t20160722_3131965_3.shtml. 29. State Council, “State Council Note on Adjusting the Classification Standards of City Sizes,” November 20, 2014, http://www.gov. cn/zhengce/content/2014-11/20/content_9225.htm.

119 Appendix D. Explaining the Migration Patterns of Tertiary- Educated Migrants: Higher Education Overcapacity and the Economics of Labor

Kangyu Mark Wang

he migration pattern of highly educated man- those provinces are tertiary degree holders than are Tpower in China cannot be explained solely by each interprovincial migrants for municipalities such as province’s economic development level. Along with Beijing and Shanghai. Beijing, inland provinces such as Shaanxi and Henan This seems surprising at first, but the level of local see a high proportion of tertiary-educated migrants. labor market development and higher education I will disentangle this puzzling pattern in two steps. overcapacity can explain more than 60 percent of this First, I break up the total migration of each province tertiary-education proportion variance in interprovin- into intra- and interprovincial migration, which fol- cial migration. Higher education overcapacity simply low distinct patterns. Second, I use the level of labor means there are more seats at colleges and universi- market development and higher education overca- ties than there are eligible students in that province. pacity to explain the dynamics of intra- and interpro- In Heilongjiang and Liaoning, for example, this is the vincial migration. case because cities such as Harbin and Shenyang are Beijing and Shanghai, highly economically devel- so important in China’s modern history. oped eastern provinces, and Shaanxi, Shandong, and This appendix uses ordinary least squared (OLS) Henan, less developed inland provinces, draw the regressions to better explain the migratory trends largest portion of intra- and interprovincial migrants of tertiary-educated people. I argue that the stage of with tertiary education or above in China. Two labor market development has a positive impact on very different forces shape this pattern. First, Bei- the proportion of tertiary-educated intra-provincial jing, Shanghai, and Tianjin draw more interprovin- migration and a U-shaped impact on the proportion cial migrants than intra-provincial migrants. A large of tertiary-educated interprovincial migration. Higher proportion of intra-provincial migrants there have education overcapacity has a positive impact on the pro- particularly high education levels, while the pool of portion of tertiary-educated interprovincial migration. interprovincial migrants has proportionally fewer This appendix intends to provide the following tertiary degree holders. Second, for inland prov- insights into the report. First, it argues that coastal inces such as Shaanxi, Hunan, and Hubei, the major- cities and provinces are still the most attractive for ity of migrants are intra-provincial migrants, many high-skilled workers. Second, it highlights the edu- of whom are not tertiary degree holders. However, cational dimension of migration, which can be as a higher proportion of interprovincial migrants into important as the economic dimension. Many rural,

120 KANGYU MARK WANG

uneducated, young people migrate Figure D1. Percentage of Intra- and Interprovincial Migrants to cities to become educated, and Combined: Tertiary Schooling, 2010 college and graduate students account for significant proportions of high-skill labor migration. Third, it proposes a new way of comparing what draws skilled labor across provinces. 0–0.13 0.13–0.15 The Breakup of Migration 0.15–0.17 0.17–0.19 with Tertiary Education into 0.19–0.21 Inter- and Intra-Provincials 0.21–0.24 0.24–1 Among the provinces of the Peo- ple’s Republic of China, Bei- jing, Shaanxi, Shandong, Henan, Note: This figure appears as Figure 24 in the main text. Source: National Bureau of Statistics, “China Population Census: Tabulation of the 2010 Hunan, Hubei, and Shanghai Population Census of the People’s Republic of China,” 2012, Table 7-6, Table 7-7.

Figure D2. Educational Endowment of Intra- and Interprovincial Migration

25 0.6

0.5 20

0.4 15

0.3

Population (Millions) 0.2 Tertiary Proportion

5 0.1

0 0 Jilin Fujian Anhui Hubei Hebei Tianjin Gansu Jiangxi Shanxi Hunan Henan Beijing Hainan Jiangsu Yunnan Shaanxi Ningxia Sichuan Xinjian g Qinghai Liaoning Guangxi Guizhou Zhejian g Shanghai Shandong Chongqing Guangdon g Heilongjiang InnerMongol Tibet (Xizang ) Interprovincial Migration with Tertiary Education Intra-Provincial Migration Without Tertiary Education Interprovincial Migration Without Tertiary Education Tertiary Interprovincial Migration Proportion (Right Axis) Intra-Provincial Migration with Tertiary Education Tertiary Intra-Provincial Migration Proportion (Right Axis)

121 URBANIZATION WITH CHINESE CHARACTERISTICS

Figure D3. Educational Endowment of Intra-Provincial Migration

16 0.6

14 0.5 12 0.4 10

8 0.3

6 0.2

4 Tertiary Proportion

Migrant Population (Millions) 0.1 2

Figure D4. Educational Endowment of Interprovincial Migration

25 0.4

0.35 20 0.3

0.25 15

0.2

10 0.15 Tertiary Proportion 0.1

Migrant Population (Millions) 5 0.05

0 0 Jilin Fujian Anhui Hubei Hebei Tianjin Gansu Jiangxi Shanxi Hunan Henan Beijin g Hainan Jiangsu Yunnan Shaanxi Ningxia Sichuan Xinjian g Qinghai Liaoning Guangxi Guizhou Zhejiang Shanghai Shandong Chongqing Guangdon g Heilongjiang Tibet (Xizang ) Inner Mongol Non-Tertiary Interprovincial Migration Tertiary Interprovincial Migration Interprovincial Tertiary Proportion Source: National Bureau of Statistics, Department of Population and Employment Statistics, “China Population Census: Tabulation of the 2010 Population Census of the People’s Republic of China,” 2012, Table 7-7.

122 KANGYU MARK WANG

have the highest proportion of migrants with ter- Hypotheses: The Economics of Labor tiary education, from within and outside the prov- Skills and Higher Education Capacity ince. (See Figure D1, which is identical to Figure 24 in the report.) While Beijing and Shanghai are meg- I hypothesize that two main factors drive the higher acities with many opportunities for educated people, education endowment of migrants: higher education Shaanxi and Henan are economically backward. overcapacity and the economics of labor. The propor- By contrast, Tibet, Guangdong, Zhejiang, and tion of intra-provincial migrants with tertiary edu- Guizhou have the lowest proportion of migrants cation largely reflects the stage of provincial labor with tertiary education. Zhejiang and Guangdong are market development. The proportion of interprovin- coastal manufacturing powerhouses, while Guizhou cial migrants with tertiary education is determined and Tibet are the poorest provinces in China. These by both the provincial labor market development seemingly confusing patterns get clearer when we stage and the level of higher education institute break down migrants into inter- and intra-provincial overcapacity. groups. Provinces with a high proportion of tertiary- The Economics of Labor. The stage of local eco- educated migrants have very different inter- and nomic development is related to the pattern of intra-provincial migration patterns. (See Figures D2– intra-provincial migration in a linear manner and D4.) First, Beijing and Shanghai receive more inter- to interprovincial migration in a U-shaped manner. provincial migrants than intra-provincial migrants. I hypothesize that provincial labor market devel- An extraordinarily high proportion of intra-provincial opment in China progresses through three distinct migrants have tertiary education. Given hukou resi- stages, which are summarized in Table D1. I also dents’ wealth and educational endowments in these hypothesize that manpower within a given province two cities, this is not surprising. However, interpro- will initially fill that province’s labor market and out- vincial migrants to these two cities are not usually side labor will then fill the vacancies. Another factor highly educated. that affects these trends is that thehigh-skill labor Second, Shandong and Henan see fewer interpro- market is more national than the low-skill labor mar- vincial than intra-provincial migrants, and the former ket because it is more difficult to find people locally group has a lower proportion of tertiary-educated who can perform highly skilled jobs. people than the latter group does. Third, Shaanxi, At the low stage of economic development, both Hunan, and Hubei also have fewer interprovincial low-skill manufacturing and low-skill service indus- than intra-provincial migrants, but the former group tries do not provide enough employment opportuni- has a higher proportion of tertiary-educated people ties to meet the demand for low-skill jobs. Low-skill than the latter group does. workers who are unsuccessful in securing employ- Among provinces with low proportions of ment need to look for jobs outside their own prov- tertiary-educated migrants, Guangdong and Zhejiang inces. At the same time, a few high-skill jobs attract have more inter- than intra-provincial migration, and high-skill interprovincial migrants. the proportion of tertiary-educated people is lower At the medium stage of economic development, for interprovincial migrants than for intra-provincial low-end manufacturing and services begin to develop migrants. This is understandable, taking into rapidly, which creates a demand for low-skill labor account the large scale of labor-intensive manufac- that the provincial labor supply cannot meet. Thus, turing industries in these two provinces. By con- low-skill labor from other provinces floods in. trast, Guizhou, a poor southwest province with little High-end jobs are still few, filled bybetter-educated labor-intensive industry, has more intra- than inter- intra-provincial migrants and locals. Therefore, these provincial migrants. provinces see a mass interprovincial migration of low-skill labor. The proportion of tertiary-educated

123 URBANIZATION WITH CHINESE CHARACTERISTICS

Table D1. Three Stages of Labor Market Development in Chinese Provinces: Demand and Supply of Manpower Stage of Economic Low Medium High Development

Skill of Within-Provincial Low Medium High Labor

Low-Skill Manufacturing Few, filled by low-skill Many, filled by low-skill Few, filled by low-skill Jobs locals and intra-provincial interprovincial migrants, intra-provincial migrants migrants intra-provincial migrants, and locals and locals

Low-Skill Service Jobs Few, filled by low-skill Many, filled by low-skill Many, filled by low-skill intra-provincial migrants interprovincial migrants, intra-provincial migrants, and locals intra-provincial migrants, interprovincial migrants, and locals and locals

High-Skill Jobs Few, filled by high-skill Some, filled by high-skill Many, filled by high-skill intra-provincial migrants, intra-provincial migrants intra-provincial migrants, interprovincial migrants, and locals interprovincial migrants, and locals and locals

Examples Shaanxi and Jiangxi Zhejiang and Guangdong Beijing and Shanghai

Source: Author.

interprovincial migrants is lowest when a province is • Hypothesis 2: There is a positive relationship at the medium stage of development. between the stage of economic development At the high stage of economic development, costs and the proportion of tertiary-educated intra- are too high for low-end manufacturing jobs to stay, provincial migration. but low-end service and construction jobs for megacity residents remain, which still attract mass From the hypotheses above, the proportion of the low-skill interprovincial migrants. On the other hand, total inter- and intra-migration of the three types of high-end jobs begin to increase, resulting in both provinces are presented in Table D2. intra-provincial and interprovincial high-skill labor migration. Therefore, although highly developed eco- Higher Education Capacity. If there is higher edu- nomic provinces still see mass low-skill interpro- cation overcapacity in a province, there are more seats vincial migration, intra-provincial migration is high at colleges and universities than college entrance skilled in character, and there is significanthigh-skill examination (CEE, gaokao) participants can fill. In interprovincial migration as well. fact, college and graduate students account for a large I present two hypotheses on economic develop- share of tertiary-educated migrants in both intra- and ment and tertiary-educated migration. interprovincial groups. For example, in 2018 there were 1.13 million higher education students in Shaanxi • Hypothesis 1: There is a U-shaped relationship province, who accounted for about 20 percent of between the stage of economic development 5.65 million total migrants. and the proportion of tertiary-educated inter- College and graduate students are classified as provincial migration. having a tertiary education. To continue to higher

124 KANGYU MARK WANG

Table D2. Three Stages of Labor Market Development in Chinese Provinces: Migration Patterns

Development Stage Low Medium High

Proportion of Tertiary-Educated Medium to High Low Medium to High Migrants in Total Migration

Proportion of Tertiary-Educated High Low Medium Migrants in Interprovincial Migration

Proportion of Tertiary-Educated Low Medium High Migrants in Intra-Provincial Migration

Source: Author.

education, most of them need to move outside of y1 refers to the proportion of interprovincial migra- their assigned hukou towns, townships, and subdis- tion with tertiary education or more. x1 stands for the tricts to study because most Chinese towns, town- overcapacity rate of higher education per capita, as ships, and subdistricts do not have higher education measured by the annual higher education institution institutions. About 25 percent of total migrants in capacity1 minus the number of CEE candidates,2 and Shaanxi have tertiary education. Therefore, a sim- then divided by provincial population. ple estimation shows that college and graduate students account for a stunning 80 percent of total Annual Higher Education Capacity–Number of CEE Candidates x₁= tertiary-educated migrants. Total Population I hypothesize that higher education overcapacity of a province also affects the proportion of x₂ stands for the proportion of net migrants in tertiary-educated interprovincial migrants. If a prov- the total number of permanent residents of a province can provide more seats in higher education than ince, which is used as a measure for the stage of what the local population needs, it is likely to attract labor market development.3 If a province absorbs interprovincial students, and thus it is likely to have a high number of net migrants per local resident, it a higher proportion of tertiary-educated interprovin- is considered to be at a high level of labor market cial migrants. development. Figure D5 shows a strong positive correlation • Hypothesis 3: There is a positive relationship (correlation coefficient = 0.844) between x₂ and between higher education overcapacity and the GDP per capita. However, I consider x₂ to be a bet- proportion of tertiary-educated interprovincial ter measurement of the “real” level of labor market migration. development than is per capita GDP, which favors natural-resource-reliant provinces such as Inner Mon-

golia. If my hypothesis holds, β₁ should be positive, β2 Empirical Analysis: OLS Regressions should be negative, and β3 should be positive. Regres- sion results are reported in Section (2) of Table D3. The main regression function for interprovincial Section (1) gives a linear “robust check” regression: migration is as follows: The quadratic model has higher adjusted R-squared than the linear model and thus is more desirable.

y1 = β0 + β₁ ∙ x₁ + β₂ ∙ x₂ + β₃ ∙ x2²

125 URBANIZATION WITH CHINESE CHARACTERISTICS

Figure D5. Net Migration per Capita and GDP per Capita

0.5

0.4 Shanghai Beijing

0.3

Guangdong Tianjin 0.2 Zhejiang

0.1 Xinjiang Fujian Jiangsu Xizang Hainan Ningxia Net Migration per Capita Qinghai LiaoningInner Mongol 0 Yunnan Shanxi 0102030405Shandong060708090 ShaanxHebeJilini i Gansu Heilongjiang Guangxi HenanChongqingHubei –0.1 Guizhou SichuanHunan Jiangxi Anhui –0.2 GDP per Capita (Thousand RMB) Source: National Bureau of Statistics, Department of Population and Employment Statistics, “China Population Census: Tabulation of the 2010 Population Census of the People’s Republic of China,” 2012, Table 7-6, Table 7-7; and National Bureau of Statistics of China, “National Data,” accessed September 5, 2018, http://data.stats.gov.cn/english/easyquery.htm?cn=E0103.

As shown in Table D3 and Figure D6, both Hypoth- of permanent residents of a province, which is used as esis 1 and Hypothesis 3 hold. A 1 percentage point a measure for the stage of labor market development. increase in the higher education overcapacity rate If my hypothesis holds, d₁ should be positive. Regres- results in around a 15 percentage point increase in the sion results are reported in Table D4. tertiary-educated proportion in interprovincial migra- As shown in Table D4, Hypothesis 2 also holds. A tion. Also, there is a U-shaped relationship between 0.01 increase in net migration per capita results in a the proportion of tertiary-educated interprovincial 0.4 percentage point increase in the tertiary-educated migrants and the level of labor market development. proportion of intra-provincial migration. This model explains more than 60 percent of this tertiary-educated proportion variance in interprovin- Models for Alternative Measurements. As argued cial migration. above, the proportion of tertiary-educated migrants The main regression function for intra-provincial to the number of total provincial permanent resi- migration is as follows: dents is a better measurement than the proportion of tertiary-educated migrants to the number of total

y2 = d0 + d₁ ∙ x₂ migrants. The higher a province’s economic development, the more attractive it is for tertiary-educated

y2 stands for the proportion of interprovincial migrants. If that is the case, the U-shaped relationship migration with tertiary education or more. x₂ stands in Hypothesis 1 is no longer necessary. Thus, Hypoth- for the proportion of net migrants in the total number esis 1 shall be replaced by Hypothesis 4.

126 KANGYU MARK WANG

Table D3. Regression Results of Interprovincial Migration on Higher-Education Overcapacity and Labor Market Development Dependent Variable Proportion of Interprovincial Migration with Tertiary Education Linear Model Quadratic Model (1) (2) Higher Education Overcapacity per Capita 20.270*** 14.953*** (4.000) (4.242) Net Migration per Capita –0.717*** –0.444*** –0.118 –0.086 Net Migration per Capita Squared 0.156*** (0.062) Constant 0.214*** 0.192*** (0.014) (0.012) Observations 31 31 R2 0.576 0.655 Adjusted R2 0.545 0.617 Residual Standard Error 0.059 (df = 28) 0.054 (df = 27) F Statistic 18.998*** (df = 2; 28) 17.122*** (df = 3; 27) Note: *p < 0.1; **p < 0.05; ***p < 0.01. Standard errors are reported in parentheses. Source: National Bureau of Statistics, Department of Population and Employment Statistics, “China Population Census: Tabulation of the 2010 Population Census of the People’s Republic of China,” 2012, Table 7-6, Table 7-7; and Ministry of Education of China, “Pro- vincial Number of Students at Different Levels of School per 100,000 Residents,”http://www.moe.gov.cn/s78/A03/moe_560/ s6200/201201/t20120117_129521.html.

• Hypothesis 4: There is a positive relationship • Hypothesis 6: There is a positive relationship between the stage of labor market development between higher education overcapacity and and the ratio of tertiary-educated interprovin- the ratio of tertiary-educated interprovincial cial migrants to total provincial permanent migrants to total provincial permanent residents. residents. A new model for “real” attractiveness to tertiary- Hypotheses 2 and 3 should be rewritten as Hypoth- educated interprovincial migrants is as follows: eses 5 and 6.

y3 = f0 + f₁ ∙ x₁ + f₂ ∙ x₂ • Hypothesis 5: There is a positive relationship

between the level of provincial labor market y3 refers to the ratio of tertiary-educated interpro- development and the ratio of tertiary-educated vincial migration to the number of total provincial intra-provincial migrants to total provincial per- permanent residents. x₁ stands for the overcapacity manent residents. rate of higher education per capita. x₂ stands for the proportion of net migrants to the total number of

127 URBANIZATION WITH CHINESE CHARACTERISTICS

Figure D6. Residuals vs. Net Migration per Capita

0.25 Jiangxi 0.2

0.15 Shaanxi HunanChongqingHubei Sichuan Jilin 0.1

HenaHeilongjiangn Beijing Anhui Gansu0.05 Hainan

HebeiShandong Guangxi 0 –0.2 –0.1 00Liaoning .1 0.20.3 0.40.5 ShanxiNingxia Shanghai Guizhou –0.05Qinghai YunnanInner Mongol XizangJiangsu Tianjin Xinjiang Guangdong –0.1 Fujian Proportion of Tertiary-Educated Interprovincial Migrants Zhejiang

–0.15 Net Migration Per Capita

Note: *After controlling for higher education overcapacity. Source: National Bureau of Statistics, Department of Population and Employment Statistics, “China Population Census: Tabulation of the 2010 Population Census of the People’s Republic of China,” 2012, Table 7-6, Table 7-7; and Ministry of Education of China, “Pro- vincial Number of Students at Different Levels of School per 100,000 Residents,”http://www.moe.gov.cn/s78/A03/moe_560/ s6200/201201/t20120117_129521.html.

Table D4. Regression Results of Intra-Provincial Migration on Labor Market Development Dependent Variable Proportion of Intra-Provincial Migration with Tertiary Education Linear Model Net Migration per Capita 0.391*** (0.075) Constant 0.207*** (0.010) Observations 31 R2 0.484 Adjusted R2 0.466 Residual Standard Error 0.055 (df = 29) F Statistic 27.217*** (df = 1; 29) Note: *p < 0.1; **p < 0.05; ***p < 0.01. Standard errors are reported in parentheses. Source: National Bureau of Statistics, Department of Population and Employment Statistics, “China Population Census: Tabulation of the 2010 Population Census of the People’s Republic of China,” 2012, Table 7-6, Table 7-7.

128 KANGYU MARK WANG

Table D5. Regression Results of Real Attractiveness to Tertiary Interprovincial Migration on Higher Education Overcapacity and Labor Market Development Dependent Variable Ratio of Tertiary-Educated Interprovincial Migration to the Number of Total Provincial Permanent Residents Linear Model Higher Education Overcapacity per Capita 2.503*** (0.597) Net Migration per Capita 0.048** (0.018) Constant 0.014*** (0.002) Observations 31 R2 0.766 Adjusted R2 0.749 df = 28 0.009 (df = 28) F Statistic 45.749*** (df = 2; 28) Note: *p < 0.1; **p < 0.05; ***p < 0.01. Standard errors are reported in parentheses. Source: National Bureau of Statistics, Department of Population and Employment Statistics, “China Population Census: Tabulation of the 2010 Population Census of the People’s Republic of China,” 2012, Table 7-6, Table 7-7; and Ministry of Education of China, “Pro- vincial Number of Students at Different Levels of School per 100,000 Residents,”http://www.moe.gov.cn/s78/A03/moe_560/ s6200/201201/t20120117_129521.html. permanent residents in a province. If my hypothesis mid-2010s, this model seems to have some predic- holds, both f₁ and f2 should be positive. Regression tive power. results are reported in Table D5. A new model for “real” attractiveness to tertiary- As shown in Table D5 and Figure D7, both Hypothe- educated intra-provincial migrants is as follows: sis 4 and Hypothesis 6 hold for the new measurement of attractiveness to tertiary-educated interprovincial y4 = y0 + y₁ ∙ x₂ migrants. A 1 percentage point increase in the higher education overcapacity rate results in around a 2.5 per- y4 refers to the ratio of tertiary-educated intra- centage point increase in the ratio of tertiary-educated provincial migrants to the number of total provincial interprovincial migration to the number of total pro- permanent residents. x₂ stands for the proportion of vincial permanent residents. This model explains net migrants in the total number of permanent resi- more than 70 percent of the total variance. dents in a province. If my hypothesis holds, y₁ should A stunning result is that the most “unattractive” be positive. Regression results are reported in Table D6. provinces are exactly the rust belt provinces that are As shown in Table D6, Hypothesis 5 also holds now under severe economic depression4—namely, for the new measurement of attractiveness to Heilongjiang, Jilin, Liaoning, and Tianjin—although intra-provincial migration. A 0.01 increase in net the unattractiveness of coastal powerhouses such migrants per capita results in a 0.09 percentage as Jiangsu, Zhejiang, and Shandong is more surpris- point increase in the ratio of tertiary-educated intra- ing. Given that I based my analysis on 2010 data provincial migration to the number of total provincial and the provinces only began to face a crisis in the permanent residents.

129 URBANIZATION WITH CHINESE CHARACTERISTICS

Figure D7. Residual vs. Net Migration per Capita (New Independent Variable)

0.04

Beijing 0.03

0.02 Gansu

Shanghai Henan 0.01 Ningxia Anhui ShanxiInnerQinghai Mongol HainanXinjiang Guizhou Guangdong Shaanxi Fujian Xizang GuangxiHebei JiangxiChongqing 0 –0.2 Sichuan-0.1 00.1 0.20.3 0.40.5 Hubei Shandong Hunan Zhejiang Yunnan Jiangsu –0.01 Jilin Liaoning Tianjin

Proportion of Tertiary-Educated Interprovincial Migrants Heilongjiang

–0.02 Net Migration per Capita

Table D6. Regression Results of Real Attractiveness to Tertiary Intra-Provincial Migration on Labor Market Development Dependent Variable Ratio of Tertiary-Educated Intra-Provincial Migration to the Number of Total Provincial Permanent Residents Linear Model Net Migration per Capita 0.084*** (0.014) Constant 0.029*** (0.002) Observations 31 R2 0.553 Adjusted R2 0.537 Residual Standard Error 0.010 (df = 29) F Statistic 35.854*** (df = 1; 29) Note: *p < 0.1; **p < 0.05; ***p < 0.01. Standard errors are reported in parentheses. Source: National Bureau of Statistics, Department of Population and Employment Statistics, “China Population Census: Tabulation of the 2010 Population Census of the People’s Republic of China,” 2012, Table 7-6, Table 7-7.

130 KANGYU MARK WANG

Implications for the Migration Pattern of and bringing them to big cities. Higher education Tertiary-Educated Manpower in China (associate, college, and graduate) is an important driving force of migration, and rural people migrate As shown in the report, Beijing is disproportionately to cities to get educated first and then find jobs. As the venue for migrants with at least some tertiary edu- in the case of Shaanxi, 80 percent of total migrants cation, along with inland Shaanxi and Henan. How- with tertiary education are college and graduate stu- ever, as analyzed above, that does not mean that all dents. The same tentative calculation can be applied of them are attractive for tertiary-educated migrants. to other provinces, and results are reported in Fig- Shaanxi and Henan stand out not because the number ure D11.5 of high-skill migrants is large but because the num- Beijing and Shanghai conspicuously have the low- ber of low-skill migrants is small. If measured by the est proportion of students in their tertiary-educated number of tertiary-educated migrants per local resi- migrants. In other words, despite the large number of dent, coastal provinces such as Beijing, Shanghai, and students pouring into those two megacities, most of Tianjin come out on top (Figures D8–D10). the migrants in Beijing and Shanghai are there to find work; they are purely economically motivated. Other The Educational Function of Migration. As coastal provinces such as Guangdong and Fujian see argued in the report, the proportion of tertiary- a similar trend. educated men and women is far lower in rural areas The low proportion of students in tertiary-educated than in towns and cities. Also, the process of urban migrants in northwestern provinces such as Inner migration has been absorbing younger people with Mongolia, Ningxia, Qinghai, and Xinjiang are an indi- the highest educational profiles from the countryside cator of the low higher education capacity in those

Figure D8. Attractiveness to Intra-Provincial Migration with Tertiary Education

0.04

Beijing 0.03

0.02 Gansu

Proportion of Tertiary-Educated Interprovincial Migrants Heilongjiang

–0.02 Source: National Bureau of Statistics, Department of PopulationNet Migration and Employment per Capita Statistics, “China Population Census: Tabulation of the 2010 Population Census of the People’s Republic of China,” 2012, Table 7-6.

131 URBANIZATION WITH CHINESE CHARACTERISTICS

Figure D9. Attractiveness to Interprovincial Migration with Tertiary Education

1.8 0.10

1.6 0.09

0.08 1.4 0.07 1.2 0.06 1.0 0.05 0.8 0.04 0.6 0.03 Migrants per Local Resident 0.4 0.02 Tertiary-Educated Interprovincial

0.2 0.01 Tertiary-Educated Interprovincial Migrants (Millions) 0 0 Jilin Fujian Anhui Hubei Hebei Tianjin Gansu Jiangxi Shanxi Hunan Henan Beijing Hainan Jiangsu Yunnan Shaanxi Ningxia Sichuan Xinjiang Qinghai Liaoning Guangxi Guizhou Zhejiang Shanghai Shandong Chongqing Guangdong Heilongjiang Tibet (Xizang) Inner Mongol Number of Migrants Ratio of Tertiary-Educated Migrants to Local Residents Source: National Bureau of Statistics, Department of Population and Employment Statistics, “China Population Census: Tabulation of the 2010 Population Census of the People’s Republic of China,” 2012, Table 7-7.

Figure D10. Attractiveness to Combined Inter- and Intra-Provincial Migration with Tertiary Education

4.5 0.20

4.0 0.18

3.5 0.16 0.14 3.0 0.12 2.5 0.10 2.0 0.08 1.5 0.06 per Local Resident 1.0 0.04 Tertiary-Educated Migrants 0.5 0.02 Tertiary-Educated Migrants (Millions) 0 0 Jilin Fujian Anhui Hubei Hebei Tianjin Gansu Jiangxi Shanxi Hunan Henan Beijing Hainan Jiangsu Yunnan Shaanxi Ningxia Sichuan Xinjiang Qinghai Liaoning Guangxi Guizhou Zhejiang Shanghai Shandong Chongqing Guangdong Heilongjiang Tibet (Xizang) Inner Mongol

Number of Migrants Ratio of Tertiary-Educated Migrants to Local Residents

132 KANGYU MARK WANG

Figure D11. The Educational Factor in Migration of Tertiary-Educated Manpower

1.0 0.9

0.8

0.7

0.6

0.5 0.4

0.3

0.2 Total Tertiary-Educated Migrants

Ratio of Higher Education Students to 0.1

0 i i i i i i i a u u u n n n n n g g xi ai ol li n ji an an ng ng ng ng ng ng nx xi hu gx ha gx be be Ji ns jia gs na na ua on ij in an in ia ni ia gh ng an jia nn zho aa do an an i ng ng gqi An an Fu ej n gj Hu He ch Ga Ti Sh Ji Hu He Be ao gd an Ji Ha Yu Sh Ni Xi an Mo Si Qi Li on Gu Gu Zh on an r Sh il Sh Ch Gu He nne I

Source: National Bureau of Statistics, Department of Population and Employment Statistics, “China Population Census: Tabulation of the 2010 Population Census of the People’s Republic of China,” 2012, Table 7-6, Table 7-7; and Ministry of Education of China, “Pro- vincial Number of Students at Different Levels of School per 100,000 Residents,”http://www.moe.gov.cn/s78/A03/moe_560/ s6200/201201/t20120117_129521.html.

provinces. By contrast, some central and northeast- initially classified asless-educated rural residents ern provinces such as Heilongjiang, Hubei, Jiangxi, but later as more educated urban migrants. There- Jilin, and Shaanxi see the largest proportion of stu- fore, education should be considered another import- dents in tertiary-educated migrants, indicating weak ant driving force of migration, at least for the 15–24 economic draw for educated migrants or overdevel- age group. With adequate higher education capacity, opment of higher education as compared to their provinces can attract degree seekers and effectively labor market development level. This is not surpris- increase the educational level of migrants. ing, given that most higher education institutions were established before the reform era in 1978. It is The “Real” Attractiveness of Employment only after the 1980s and 1990s that China’s economic Opportunities. Educational overcapacity plays an activities further shifted to the coast, and northeast important role in attracting interprovincial high-skill and central provinces began to lag behind. migrants, which does not necessarily reflect the “real” It is reasonable to believe that the same thing hap- absorbing power of job opportunities. I suggest that pens to high school education: High school students the residuals of the first regression would be better to usually move away from their hukou towns, townships, compare provinces. or subdistricts to study, becoming intra-provincial We can measure the draw of the provinces for migrants. It is tricky because they become migrants tertiary-educated interprovincial migrants by looking to start school and join the ranks of the “tertiary edu- at the proportion of tertiary-educated people among cated” at the same time. In other words, they are interprovincial migrants. After controlling for higher

133 URBANIZATION WITH CHINESE CHARACTERISTICS

educational overcapacity and the stage of labor mar- educated interprovincial migration to the number of ket development, Beijing performs better than Shang- total provincial permanent residents. As shown in hai does in attracting tertiary-educated migrants; Figure D6, Beijing performs much better than Shang- Guangdong performs better than Zhejiang does; and hai, while Guangdong performs better than Zhejiang. Shaanxi, Jilin, and Hainan perform better than Yun- Rustbelt provinces and Tianjin are absolute under- nan and Guangxi do (Figure D6). performing outliers. Hainan and some northwestern As an alternative, we can measure a province’s provinces including Xinjiang, Qinghai, and Gansu also attractiveness by examining the ratio of tertiary- outperform.

134 KANGYU MARK WANG

Notes

1. The Ministry of Education of China only provides data on the total number of higher education students per 100,000 local residents of each province in 2010, so I multiply this number by the 2010 provincial total population to get the total higher education capacity of each province. Given that higher education students include three-year associate degree students, four-year bachelor’s degree students, and graduate students, I hypothesize that on average the length of higher education is four years. Thus, dividing total higher education capacity by four would get annual higher education capacity. See Ministry of Education of China, “Provincial Number of Students at Different Levels of School per 100,000 Residents,”http://www.moe.gov.cn/s78/A03/moe_560/s6200/201201/t20120117_ 129521.html. 2. China Education Online, “2010 CEE Registration Number and Local Admission Rates,” http://www.eol.cn/html/g/bmsj/index_1. shtml. 3. Department of Population and Employment Statistics National Bureau of Statistics, “China Population Census: Tabulation of the 2010 Population Census of the People’s Republic of China,” 2012, Table 1-8. 4. South China Morning Post, “Liaoning Worst Performer as China’s Northeast Lags Behind Country’s Economic Growth,” June 21, 2018, https://www.scmp.com/news/china/economy/article/2104789/liaoning-worst-performer-chinas-northeast-lags-behind-countrys. 5. Here I make the same assumption I did for Shaanxi: All higher education students are classified as migrants.

135

About the Authors

Nicholas Eberstadt holds the Henry Wendt Chair economics PhD student at the University of Califor- in Political Economy at the American Enterprise nia, Los Angeles. Institute, where he researches and writes extensively on demographics and economic development gen- Cecilia Joy-Pérez was a research associate at the erally, and more specifically on international secu- American Enterprise Institute and is currently an ana- rity in the Korean peninsula and Asia. Domestically, lyst at Pointe Bello. he focuses on poverty and social well-being. Eber- stadt is also a senior adviser to the National Bureau Kangyu Mark Wang is a master’s student at the of Asian Research. University of California, San Diego, where he studies international economics with a focus on China Alex Coblin was a senior research associate at the and India. American Enterprise Institute and is currently an

137