Causal Effect of Urbanization on Income Inequality

We measure income inequality using the popular Gini index, which is mathematically based on the Lorenz curve [14]. Suppose the cumulative income-to-population distribution function of a country isF(t), wheret is a random variable of personal (or household) income andF(t) is the share of population (or households) earning less than or equal tot. The Lorenz curve is depicted inFig 1, where the vertical axisL denotes the cumulative income of the populationF(t). The Gini index is the area between the Lorenz curve and the straight lineL = F(t) (the shaded gray area inFig 1) multiplied by 2.

First, we follow steps illustrated in previous studies [15–17] to derive the function from the income variablet to the Gini indexG. Based on the definition of the Lorenz curve, we have

and

whereV denotes the average income andf(t) denotes the income probability density function. LetB denote the area between the Lorenz curve and the horizontal axis such that we have

By integrating by parts and replacing the integral bounds, we have

After substituting (2) into (4), changing the integral bounds to those oft, and integrating by parts, we have

To further simplify (5), it is reasonable to assume that in reality the income variablet is within the range [0,T], whereT denotes the highest income. Then, we can rewrite (5) as

Based on the definition ofF(t), we haveF(0) = N₀/N, whereN denotes the total population size andN₀ is the number of people (households) with zero income. AsN₀ is usually small relative toN, we can further assume thatF(0) → 0 whenN is large. Similarly, we can assume thatF(T) approximates 1 whenN is large. Then, (6) becomes

Under the same assumptions, we have

LetG denote the Gini index and it is straightforward that

Next, we derive the relation between urbanization and the Gini index. Suppose the cumulative population-to-income distribution functions in the urban and rural sectors can be represented byF_u(t) andF_r(t), respectively. Letx ∈ [0,1] denote the urbanization rate (i.e., the percentage of urban population), and lety = 1- x denote the percentage of rural population. We have

LetU denote the average income of the urban population,R denote the average income of the rural population,G_u denote the urban Gini index, andG_r denote the rural Gini index. Based on (8) and (9), we have

Moreover, we define

By substituting (10)–(15) into (9) and lettingG' = G_u—G_r denote the difference between the urban and rural Gini indexes, we have

To study the monotonicity of (16), we consider the first derivative ofG:

As {(U—R)x + R}² > 0, setting (17) to 0 requires the nominator to be 0. By solving the quadratic equation

we have

Note that by definition0 ≤ x ≤ 1 and we assumeU > R. Therefore,$\frac{-\sqrt{UR\{1+\frac{G^{\prime }}{D}(U-R)\}}-R}{U-R}<0$ and the turning point of (16) should be at

As the opening of the quadratic curve is downward andX is the larger root of (18), we can see that when0 ≤ x < X,dG/dx > 0 and the Gini index increases with urbanization. Similarly, whenX < x ≤ 1,dG/dx < 0 and the Gini index decreases with urbanization. This is consistent with the inverse-U-shaped curve depicted by Kuznets. It is noteworthy thatX is not always on [0, 1]. IfG' > D/R, i.e., if the urban Gini index exceeds the rural Gini index plusD/R, thenX > 1 and urbanization will always causeG to increase. On the other hand, ifG' < -D/U, i.e., if the urban Gini index is smaller than the rural Gini index minusD/U, thenX < 0 and urbanization will always causeG to decrease.

Predictive Causal Model

The advantage of (20) is that it only depends on the rural and urban income distributions, asU,R,G_u, andG_r are all functions ofF_u(t) andF_r(t). A main assumption is that bothF_u(t) andF_r(t) remain the same after the population shift. In reality, however, this condition is unlikely to hold. Newly arriving rural migrants are concentrated in the lower-income strata and/or informal sectors of the urban economy, which tends to downwardly skewF_u(t). It takes time forF_u(t) to recover and the extent to whichF_u(t) can recover depends on a range of factors, including the structure of the urban economy, the extent of social mobility, and access to opportunities for the migrant groups. Moreover, some of the impact of urbanization could be long-lasting and/or lag over time. For example, adding needed labor to a fast-growing urban sector can accelerate economic expansion, whereas barriers to opportunities can trap rural migrants in persistent poverty [18,19]. In both situations, income distribution will shift and income inequality may change over time.

For these reasons, in reality the impact of urbanization on income inequality is a complex process that may span multiple years. Under the traditional causal inference framework based onceteris paribus (i.e., all other factors equal), addressing such dynamics requires highly complex structural models (or systems of models) with a large number of parameters. In most situations, however, such models are difficult to correctly specify and/or efficiently estimate due to limitations in data availability, measurement quality, and model statistical power. Moreover, such models are highly sensitive to misspecification. It is not uncommon that a dynamic structural model with many parameters is not immediately identifiable and additional assumptions are needed to restrict the parameter space. As argued by C. Sims, these requirements not only obfuscate the logic of causal inference but they can also produce highly misleading results, as the numerous assumptions imposed by the structural model are not necessarily warranted in real-world analysis [20].

The concept of predictive causality, largely pioneered by Clive Granger [21], offers an alternative approach to causal inference in dynamic domains. In this approach, causality between two processes is defined by whether observations of one can help predict the other. Compared withceteris paribus this is a weaker causality framework, as it merely describes the temporal correlation between two time-series without controlling for other factors. In other words, the predictive causality model focuses on prediction but may not offer the same explanatory power as a traditional dynamic structural model would. However, it provides a simple yet powerful tool in practice for domains such as macroeconomic analysis, where data are often limited and detailed structural relationships are usually too complex to correctly specify and/or identify. In this study, we hypothesize that predictive causal relationships exist between urbanization and income inequality. We first use vector autoregression (VAR) to explore such relationships. Since we have a relatively small number of time points, the results of the VAR model are not intended to provide inferential evidence; instead they are interpreted exploratorily to guide the specification of a more parsimonious autoregressive model. The final inference is based on a simple autoregressive model, which requires fewer parameters and thus can be more convincingly supported by the short time series.

From the perspective of traditional structural regression, the omission of variables could be a serious problem for our analyses. For example, one may argue that any relationships between urbanization and Gini index are spurious because the omission of economic development, which affects both urbanization and income inequality. This is indeed an issue if we take the model at face value in policy-making and attempt to regulate inequality by boosting or controlling urbanization without checking the conditions of other relevant factors such as economic growth [20]. However, it is less a concern if we stick with the logic of predictive causality and treat the results only as evidence of two temporally correlated processes observed in the study time frame. In other words, our goal is to reveal the temporal relationships between urbanization and income inequality as suggested by the time series, which should not be interpreted explanatorily but could offer an informative baseline to elicit more detailed explanatory studies.

Table 1. Results of unit tests.

Table 2. Summary of VAR estimates.

Table 3. Summary of AR(1) estimates.

Potential Aggravating Factors of Inequality

First, urbanization may have worsened regional inequality. Regional development in China is known to follow a laddered geographic pattern, where the fast-growing coastal areas and the less-developed inland provinces (or the eastern, central, and western regions in a tri-partite view) form distinct growth “clubs” [46]. It has been shown that economic development has converged within each club, which helps suppress regional inequality at the national level [47]. But the inter-club gap seems to be persistent despite the central government’s recent efforts to accelerate inland development [48]. As migrants from the inland rural areas have been an important force for both the urbanization and economic development of the coastal regions, it is reasonable to hypothesize that urbanization may have causally increased the income gap between the origin and destination of the migration due to the widening difference in economic growth. Note that this process is probably more subtle than coarse scale statistics (e.g., average GDP and/or income at the provincial level) can indicate, because it is highly dependent on the structure of local urbanization, the conditions of the local labor market, and specific migration flows.

Second, the lagged positive response in the Gini index may reflect a growing difference in productivity between the rural and urban sectors. Despite the success of the rural decollectivization in the early stage of China’s economic reform, the incentives for farming have gradually languished due in part to high taxes and fees and relatively low returns [49]. Moreover, cropland in rural China is tenured through the household responsibility system (HRS), under which land is owned collectively but controlled by individual households. Although HRS boosts household-level productivity and protects land security for peasants [50], it limits technological advancements and economies of scale due to ambiguous property rights [51], the under-developed land rental market [52], and the lack of adequate financial services for individual farmers [53]. As the urban economy continues to outpace its rural counterpart, the opportunity cost for farming rises and the most productive group of the rural population (e.g., young, healthy, and educated adults) continue to leave for the cities. Consequently, a structural shortage of human resources has emerged in some rural areas [54] and the rural sector lags further behind the booming urban sector in terms of productivity. Again, this process may be difficult to identify using aggregated data given the variation in population flows and local labor markets.

Third, urbanization’s lagged effect on income inequality could be the result of the persistent marginalization of the rural migrants. Rural migrants in China are known to face additional hurdles set up by discriminatory institutions, especially those related to the Hukou household registration system [55], which ties people’s access to services to their residential status. Due to the institutionalized measures whereby migrants are deprived of job opportunities, educational rights, and social welfare benefits, the marginalization of a rural migrant usually lasts for an extended period of time [56]. Consequently, urbanization can lead to a ballooning low-income group that further widens urban income inequality, and the lack of integration with the mainstream urban society can trap migrant groups in segregated spaces and long-term poverty [19,57]. Although a well-recognized problem in the related literature, the persistent poverty of rural migrants is poorly documented in many official and unofficial data sources, as most general-purpose surveys tend to undersample the migrant population [58]. Future research on this subject, therefore, must draw extensively on more representative and preferably firsthand data sources.

The Importance of Social Reform

It is noteworthy that some of the above problems have been addressed to some extent in recent years. The relocation of the manufacturing industry from the coastal areas to the inland provinces [37] has the potential to reduce urbanization’s aggravating effects on regional inequality, as many rural migrants can now find jobs in nearby cities. The abolition of the agricultural tax [38] and the continued development of the cropland rental market help boost agricultural productivity [52]. There are also advances in health care and social welfare reforms, which have greatly expanded the population protected by the emerging social safety net [39,42,59]. A main challenge at this moment, however, is probably the Hukou reform. Although restrictions on employment and residential mobility have gradually withered as economic marketization has deepened, the role of Hukou as a social divider continues to prevail. The ongoing effort to extend social and welfare services to non-local Hukou holders has been beleaguered by the lack of an affordable and accountable solution. Currently, local governments at the receiving end of urbanization are asked to carry out the reform and absorb its enormous cost, which has created significant budgetary pressure on the cities. As the recent slowdown in China’s real estate sector has worsened widespread local debt situations [60], real progress in regard to the Hukou reform is at a standstill in most parts of China.

From a broader policy perspective, the problems of assimilation and integration caused by long-standing discriminatory institutions have resulted in a permanent dichotomy in the social and economic system, which have already obstructed China’s progress toward the second stage of the Lewis model. For example, many rural migrants have to leave dependents, including women, children, and aging parents, in the countryside because of the cost and restrictions imposed by the Hukou system on housing, education, and retirement arrangements. The resulting separation of families, which is estimated as affecting more than 60 million children, is creating profound social, psychological, and human development problems that may haunt Chinese society for generations to come [61]. Due to the high stakes nature of this issue, it is important for both central and local governments to design and effect a more comprehensive, cross-region, and cross-administrative-level reform strategy, as problems such as those related to Hukou cannot be solved by the central or local government alone.

In summary, a broad message from this study is the importance of comprehensive social reforms. Currently, the discussion on China’s urbanization is preoccupied by two dominant theses: the pro-growth thesis that considers urbanization an economic engine, and the pro-sustainability thesis fueled by reactions to problems created or exposed by urbanization. In our opinion, both theses, in their current forms, lack a social dimension, and as a result, the ongoing policy debate is often out of touch with the people directly affected by the urban transition. Pro-growth urbanization policies embraced by GDP-chasing local governments, for example, have led to land-oriented rather than people-centered urban expansion, creating problems ranging from urban sprawl to environmental degradation [62]. On the other hand, the pro-sustainability argument often mistakenly places the blame for environmental woes on the growing urban population, ignoring the fact that expanding land development and increasing consumption among the existing urban residents are probably more culpable than any effect wrought by newcomers. Addressing the income-inequality problem helps reintroduce social justice into the discussion, which constitutes an important first step toward a more socially balanced urbanization policy.

• 1.National Bureau of Statistics of China. China Statistical YearbookBeijing, China: China Statistics Press; 2015. [Google Scholar]
• 2.Ho P, Eyferth JJK, Vermeer EB. Rural development in transitional China: the new agricultureLondon; Portland, OR: Frank Cass; 2004. 309 p. p. [Google Scholar]
• 3.Yao SJ. Economic development and poverty reduction in China over 20 years of reforms. Economic Development and Cultural Change. 2000;48(3):447–74. 10.1086/452606 . [DOI] [Google Scholar]
• 4.Ravallion M, Chen S. China's (uneven) progress against poverty. Journal of development economics. 2007;82(1):1–42. [Google Scholar]
• 5.Yang DT. Urban-Biased Policies and Rising Income Inequality in China. The American Economic Review. 1999;89(2):306–10. 10.2307/117126 [DOI] [Google Scholar]
• 6.Bao SM, Chang GH, Sachs JD, Woo WT. Geographic factors and China's regional development under market reforms, 1978–1998. China Economic Review. 2002;13(1):89–111. 10.1016/s1043-951x(02)00055-x . [DOI] [Google Scholar]
• 7.Fan CC. Uneven development and beyond: Regional development theory in post-Mao China. International Journal of Urban and Regional Research. 1997;21(4):620-+. 10.1111/1468-2427.00105 . [DOI] [Google Scholar]
• 8.Huang Y. Capitalism with Chinese characteristics: entrepreneurship and the stateCambridge; New York: Cambridge University Press; 2008. xviii, 348 p. p. [Google Scholar]
• 9.Xie Y, Zhang X, Li J, Yu X, Ren Q. China Minsheng Development Report 2014Beijing, China: Peking University Press; 2014. [Google Scholar]
• 10.Kuznets S. Economic growth and income inequality. The American economic review. 1955:1–28. [Google Scholar]
• 11.Kanbur SMR. Inequality in Asia and the Pacific: trends, drivers, and policy implications2014. xxiii, 412 pages p.
• 12.Xie Y, Zhou X. Income inequality in today’s China. Proceedings of the National Academy of Sciences. 2014;111(19):6928–33. 10.1073/pnas.1403158111 [DOI] [PMC free article] [PubMed] [Google Scholar]
• 13.Lin J, Chen B. Urbanization and urban-rural inequality in china: a new perspective from the government’s development strategy. Front Econ China. 2011;6(1):1–21. 10.1007/s11459-011-0119-3 [DOI] [Google Scholar]
• 14.Gini C. Concentration and Dependency Ratios. Rivista di politica economica. 1997;87(8–9):769–90. [Google Scholar]
• 15.Cheng Y. Calculation and decomposition of the overall gini coefficient in dual economies. Economic Research Journal. 2006;1(1):109–20. [Google Scholar]
• 16.Brown JAC, Mazzarino G. PRACTITIONER'S CORNER*: Drawing the Lorenz Curve and Calculating the Gini Concentration Index from Grouped Data by Computer. Oxford Bulletin of Economics and Statistics. 1984;46(3):273–8. 10.1111/j.1468-0084.1984.mp46003005.x [DOI] [Google Scholar]
• 17.Dorfman R. A Formula for the Gini Coefficient. The Review of Economics and Statistics. 1979;61(1):146–9. 10.2307/1924845 [DOI] [PubMed] [Google Scholar]
• 18.Chen G, Gu C, Wu F. Urban poverty in the transitional economy: A case of Nanjing, China. Habitat International. 2006;30(1):1–26. [Google Scholar]
• 19.Liu Y, W F., H S.. The making of the new urban poor in transitional China: Market versus institutionally based exclusion. Urban Geography. 2008;29(8):811–34. [Google Scholar]
• 20.Sims CA. Macroeconomics and Reality. Econometrica. 1980;48(1):1–48. 10.2307/1912017 [DOI] [Google Scholar]
• 21.Granger CWJ. Investigating Causal Relations by Econometric Models and Cross-spectral Methods. Econometrica. 1969;37(3):424–38. 10.2307/1912791 [DOI] [Google Scholar]
• 22.Chan KW, Hu Y. Urbanization in China in the 1990s: New Definition, Different Series, and Revised Trends. China Review. 2003;3(2):49–71. 10.2307/23461904 [DOI] [Google Scholar]
• 23.Shi L. Chinese Household Income Project, 2002. Inter-university Consortium for Political and Social Research (ICPSR) [distributor]; 2009.
• 24.Xie Y, Hu J. An Introduction to the China Family Panel Studies (CFPS). Chinese Sociological Review. 2014;47(1):3–29. 10.2753/CSA2162-0555470101 [DOI] [Google Scholar]
• 25.Chen J, Dai D, Pu M, Hou W, Feng Q. The trend of the Gini coefficient of China. BWPI, The University of Manchester, 2010. [Google Scholar]
• 26.tkepohl H. New introduction to multiple time series analysis. Berlin;: New York: Springer; 2005. xxi, 764 p. p. [Google Scholar]
• 27.McKenzie CR, Takaoka S. EViews 5.1. Journal of Applied Econometrics. 2007;22(6):1145–52. 10.1002/jae.947 [DOI] [Google Scholar]
• 28.R Core Team. R: A language and environment for statistical computing.2013. Available from:http://www.R-project.org/.
• 29.Baiocchi G, Distaso W. GRETL: Econometric software for the GNU generation. Journal of Applied Econometrics. 2003;18(1):105–10. 10.1002/jae.704 [DOI] [Google Scholar]
• 30.Knight J, Song L. The rural-urban divide: Economic disparities and interactions in ChinaOUP Catalogue; 1999. [Google Scholar]
• 31.Du R. Many people, little land: China's rural economic reform: Foreign Languages Press; 1989. [Google Scholar]
• 32.Knight J, Song L. China’s emerging urban wage structure, 1995–2002. Inequality and public policy in China. 2008:221–42. [Google Scholar]
• 33.Yueh LY. Wage Reforms in China During the 1990s. Asian Economic Journal. 2004;18(2):149–64. 10.1111/j.1467-8381.2004.00187.x [DOI] [Google Scholar]
• 34.Appleton S, Song L, Xia Q. Has China crossed the river? The evolution of wage structure in urban China during reform and retrenchment. Journal of Comparative Economics. 2005;33(4):644–63. 10.1016/j.jce.2005.08.005. [DOI] [Google Scholar]
• 35.Garnaut R, Song L, Yao Y. Impact and Significance of State-Owned Enterprise Restructuring in China. The China Journal. 2006;(55):35–63. 10.2307/20066119 [DOI] [Google Scholar]
• 36.Lin JY, Cai F, Li Z. State-owned enterprise reform in China: Chinese University Press; 2001. [Google Scholar]
• 37.Lemoine F, Poncet S, Ünal D. Spatial rebalancing and industrial convergence in China. China Economic Review. 2015;34:39–63. 10.1016/j.chieco.2015.03.007. [DOI] [Google Scholar]
• 38.Chen S, Wang J. Rural Tax Reform: Agricultural Productivity and Local Public Finance. Available at SSRN 2374923. 2014.
• 39.Cai Y, Cheng Y. Pension reform in China: challenges and opportunities. Journal of Economic Surveys. 2014;28(4):636–51. [Google Scholar]
• 40.Yip W, Hsiao WC. What Drove the Cycles of Chinese Health System Reforms?Health Systems & Reform. 2015;1(1):52–61. [DOI] [PubMed] [Google Scholar]
• 41.Ravallion M. An emerging new form of social protection in twenty-first-century China. The Oxford Companion to the Economics of China. 2014:441. [Google Scholar]
• 42.Gao Q, Yang S, Li S. Welfare, targeting, and anti-poverty effectiveness: The case of urban China. The Quarterly Review of Economics and Finance. 2015;56:30–42. [Google Scholar]
• 43.Wu FL. China's recent urban development in the process of land and housing marketisation and economic globalisation. Habitat International. 2001;25(3):273–89. . [Google Scholar]
• 44.Cheng Y. Evolution of overall national Gini coefficient and its urban-rural decomposition (in Chinese). Social Sciences in China. 2007;(4). [Google Scholar]
• 45.Wu X, Perloff JM. China's income distribution, 1985–2001. Review of Economics and Statistics. 2005;87(4):763–75. [Google Scholar]
• 46.Yao S, Zhang Z. On regional inequality and diverging clubs: a case study of contemporary China. Journal of Comparative Economics. 2001;29(3):466–84. [Google Scholar]
• 47.Zhang Q, Zou H-f. Regional inequality in contemporary China. Annals of Economics and Finance. 2012;13(1):113–37. [Google Scholar]
• 48.Candelaria C, Daly M, Hale G. Persistence of regional inequality in China. Citeseer, 2013.
• 49.Yep R. Can “Tax-for-Fee” reform reduce rural tension in China? The process, progress and limitations. The China Quarterly. 2004;177:42–70. [Google Scholar]
• 50.Zhang QF, Donaldson JA. China's Agrarian Reform and the Privatization of Land: a contrarian view. Journal of contemporary China. 2013;22(80):255–72. [Google Scholar]
• 51.Ho P. Who owns China's land? Policies, property rights and deliberate institutional ambiguity. The China Quarterly. 2001;166:394–421. [Google Scholar]
• 52.Jin S, Deininger K. Land rental markets in the process of rural structural transformation: Productivity and equity impacts from China. Journal of Comparative Economics. 2009;37(4):629–46. [Google Scholar]
• 53.Zhao Y. China's rural financial system: households' demand for credit and recent reforms1st ed.London; New York, NY: Routledge; 2011. xviii, 168 p. p. [Google Scholar]
• 54.Xie H, Wang P, Yao G. Exploring the Dynamic Mechanisms of Farmland Abandonment Based on a Spatially Explicit Economic Model for Environmental Sustainability: A Case Study in Jiangxi Province, China. Sustainability. 2014;6(3):1260–82. [Google Scholar]
• 55.Chan KW. Cities with invisible walls: reinterpreting urbanization in post-1949 ChinaHong Kong; New York: Oxford University Press; 1994. ix, 193 p. p. [Google Scholar]
• 56.Fan CC. The elite, the natives, and the outsiders: Migration and labor market segmentation in urban China. Annals of the Association of American Geographers. 2002;92(1):103–24. [Google Scholar]
• 57.Chen G. Privatization, marketization, and deprivation: Interpreting the homeownership paradox in postreform urban China. Environ Plann A. 2011;43(5):1135–53. . [Google Scholar]
• 58.Chen G. The heterogeneity of housing-tenure choice in urban China: A case study based in Guangzhou. Urban Studies. 2015:0042098015571822.
• 59.Yip WC-M, Hsiao WC, Chen W, Hu S, Ma J, Maynard A. Early appraisal of China's huge and complex health-care reforms. The Lancet. 2012;379(9818):833–42. [DOI] [PubMed] [Google Scholar]
• 60.Ambrose BW, Deng Y, Wu J. Understanding the Risk of China's Local Government Debts and its Linkage with Property Markets. Available at SSRN 2557031. 2015.
• 61.Ye J, He C, Wu H. Left-behind population in rural ChinaBeijing: Social Sciences Academic Press; 2010. [Google Scholar]
• 62.Bai X, Chen J, Shi P. Landscape Urbanization and Economic Growth in China: Positive Feedbacks and Sustainability Dilemmas. Environmental Science & Technology. 2012;46(1):132–9. 10.1021/es202329f [DOI] [PMC free article] [PubMed] [Google Scholar]

Supplementary Materials

Data Availability Statement

All relevant data are within the paper and its Supporting Information files. Data are also accessible through the National Bureau of Statistics of China (http://www.stats.gov.cn/english/) and the China Data Center (http://chinadatacenter.org).