International trade and urban-rural income inequality in China

ABSTRACT The issue of income inequality has persisted for decades in China, rendering it one of the world’s most unequal countries. Given the increasing significance of international trade in China, it is important to understand whether it widens urban-rural income inequality. This paper aims to investigate the causality between international trade and the urban-rural income gap in China through the Granger causality test. We found that variables including foreign direct investment, dependence on exports, and GDP can cause urban-rural income inequality, while urban-rural income inequality causes rural income growth. Our results contribute to the controversial literature on income inequality with a focus on the urban-rural difference.


I. Introduction
What drives urban-rural income inequality?Several studies have attempted to answer this question through urbanization, migration, education, and other factors (Wang, Shao, and Li 2019;Ma 2017;Marginson 2018).Interestingly, existing research on the relationship between international trade and urban-rural income inequality remains controversial.Some research asserted that international trade can directly lead to an increase in the income level of the population by growing economic development without affecting or even reducing income inequality (Lin et al. 2010;Cerdeiro and Komaromi 2020).However, other scholars argued that international trade will widen income disparities and the mechanisms behind it differ between developed and developing countries.Reichelt, Malik, and Suesse (2020) found that in developed countries, particularly Germany, only rising exports will favour highskilled occupations, and thus affect their wages, widening wage dispersion.However, in developing countries, Meschi and Vivarelli (2009) argued that both imports and exports with developed countries will worsen income distribution through skilledbiased technical change.
The urban-rural wage inequality has become one of the biggest challenges in China.For the past few decades, China's rapid economic growth has not equally benefited all parts of the population, with income disparities growing up.The urban-rural gap is the most important determinant of income inequality, which was estimated to contribute 44% of China's total income inequality in 1995 and increased further till 2007, then declined in 2013 to 34% (Jain-Chandra et al. 2018).Figure 1 summarizes the ratio of urban-rural disposable income from 1978 to 2021 in China.It suggests that the urban-rural income gap, which is measured by the ratio of urban-rural disposable income per capita, has risen since 1985 but began to fall in 2007.
Persistent urban-rural inequality will hinder the sustainable development, dampening social welfare and justice (Liu et al. 2019).Given the increasing significance of international trade in China, it is important to understand how it can adversely affect the economy and whether it widens urbanrural income inequality.This paper aims to examine the causal relationship between international trade and the urban-rural income gap in China through the Granger causality test.Policymakers and researchers, especially from developing countries, may find the results of the study useful in terms of improving social welfare and alleviating income inequality.

II. Data and methodology
To study the causal relationship between trade and urban-rural income inequality, the author employed the Granger causality test using annual data over the period 2002-2021 from the National Bureau of Statistics of China ( 2022).The study also transformed all data into their natural logarithm to mitigate heteroskedasticity issues.All variables are summarized in Table 1, where lnINE is the log of the urban-rural income inequality measured by the urban-rural disposable income ratio, lnX is the log of the share of exports in GDP, lnIM is the share of imports in GDP, lnFDI is the log of the actual amount of foreign direct investment (million US dollars), lnG is the log of the growth rate of per capita disposable income, lnGUR is the log of the growth rate of urban per capita disposable income, lnGRU is the log of the growth rate of rural per capita disposable income, lnGDP is the log of China's gross domestic product per capita.
This study employed the augmented Dicky-Fuller test (ADF) test and the Phillips -Perron (PP) tests to ensure stationarity (see Table 1).All variables are stationary at the first difference at the 10%, 5%, or 1% level of significance in both ADF and PP tests.Thus, the Granger causality test can be applied.Also, based on the result of AIC, HQIC, SBIC, LR, LL, the lag length of 3 is selected for the models.

III. Empirical analysis
The results of the Granger causality test are represented in Table 2.If the probability is at significant levels (e.g.5%), then the null hypothesis -variables do not have a causal relationship can be rejected.We find that the results reject the null hypothesis that 1) foreign direct investment, GDP and the share of export in GDP do not Granger cause the urban-rural income inequality; 2) the urban-rural income inequality does not Granger cause the growth rate of rural per capita disposable income.Other variables were not found to have a causal relationship with urban-rural income inequality.
The results can be summarized as follows: (1) Foreign direct investment Granger causes urban-rural inequality.This result is consistent with several previous research on the FDI-inequality relationship and supports the inequality-widening hypothesis (Wang and Lee 2021).As FDI tends to invest in capital-intensive (distinguished from labour-intensive) industries, rural labour thus has less employment than urban labour, leading to more unbalanced income distribution (Girling 1973).This result calls for policies to balance the impact of foreign direct investment on the unbalanced industry development and income distribution in the urban and rural areas.
The result shows that China fails to balance rural and urban developments over the past decades.The unbalanced development might cause social problems including rural migrants' welfare issues, high migration costs, and decreasing labour force in the rural, suggesting a trade-off between inequality and development is essential for the country.
Further research that distinguishes the urbanrural difference is needed when considering the potential impact of GDP growth in China.
(3) Dependence on exports Granger causes urbanrural inequality.The result supports the theoretical prediction that trade openness tends to restructure the labour market, shifting towards skilled workers while leaving low-skilled labours low-paid (Le et al. 2020).Our result brings evidence and concerns about the opposite effects of trade on the labour market structure and the urban-rural income unbalance.Diversifying exports can be considered to reduce the impact of trade on income inequality (Le et al. 2020).(4) The urban-rural inequality Granger causes the growth of rural per capita disposable income.It represents a positive sign that the awareness of urban-rural income inequality might prompt development in lagging rural areas.However, income growth does not necessarily close the income inequality.

IV. Conclusion
Our results contribute to the controversial literature on income inequality with a focus on the urbanrural difference.We found that urban-rural income inequality can be caused by foreign direct investment, GDP, and the dependence on exports.Also, urban-rural income inequality can cause the growth of rural income, which does not necessarily address the inequality.Policies that encourage rural development should be considered to narrow the urban- rural gap.No significant causality from imports, overall and urban income growth to urban-rural income inequality was found.Future research that explores other factors related to income inequality and strategies to balance economic growth and equality are warranted.

Figure 1 .
Figure 1.Urban-rural inequality in China.Data were collected from the National Bureau of Statistics of China (2022).

Table 1 .
Summary statistics and stationarity.
Symbols * represents a 10% level of statistical significance, ** represents a 5% level of statistical significance and *** represents a 1% level of statistical significance.

Table 2 .
Results of granger causality test.10% level of statistical significance, ** represents a 5% level of statistical significance and *** represents a 1% level of statistical significance.