The myth of the “new urban poverty”?
Trends in urban poverty in China, 1988-2002
Simon Appleton and Lina Song[1]
Simon Appleton
School of Economics, University of Nottingham, NG7 2RD, U.K., email:
Lina Song
School of Sociology and Social Policy, University of Nottingham, NG7 2RD, U.K., email:
- Introduction
Until the second half of the 1990s, policy-makers and researchers regarded urban poverty in China as either non-existent or a very marginal problem. Government anti-poverty programs focussed on rural areas and, in particular, on selected poor counties. As Khan (1998, p42) commented “China’s official poverty reduction strategy is based on the assumption that poverty is a rural problem.” Within academia, few studies focussed on urban poverty, in contrast to the large literature on income distribution and inequality more generally. This neglect of urban poverty arose partly from using low poverty lines – such as a “dollar a day”. Due to the great urban-rural divide in China (Knight and Song, 1999), a significant proportion of the rural population fell below these poverty lines but only 1% of the urban population were classified poor. With urban poverty in China being defined so as to concern only a very small minority, it is scarcely surprising that the issue was marginalised by government and scholars alike. However, this attitude would be surprising to counter-parts in middle and high income countries, where poverty remains a relevant concept despite only very small proportions of citizens living under a dollar a day. Arguably, the main role of poverty lines is to focus attention on the living standards at the lower end of the income distribution. When set inappropriately low, they can have the opposite effect and help marginalise any discussion of the distributional effect of policy reform. This is what happened in urban China prior to the late 1990s.
Things began to change in the second half of the 1990s with concern over what was seen as a “new urban poverty” caused partly by a wave of rural-urban migration and partly by mass unemployment following a program of retrenchment in state owned enterprises. These new forms of urban poverty differed from the old urban poverty which was often characterised as the “three withouts” – roughly corresponding to the disabled, the sick and the orphaned (Wong, 1998). By the turn of the century, opinion makers began to assert that urban poverty had risen during the 1990s, taking some of the shine of China’s exceptionally high rates of economic growth. For example, The Economist (2001, page 39) declared “And in the cities, absolute poverty is increasing…” while the Chinese government magazine Liaowang (27 June 2002) also argued that urban poverty had increased. Underpinning such commentarywas a concern that rising urban poverty would lead to political unrest, jeopardising the reforms that had enabled China’s rapid economic growth (see Wu, 2004). However, urban poverty in China has continued to be measured using low poverty lines such as “a dollar a day” that, despite perceived adverse developments, still only categorised around 1% of the urban population as poor.
In this chapter, we use a range of higher poverty lines in order to consider more broadly how lower income urban households fared in the 1990s. Despite thus categorising a larger segment of the urban population as poor, we find that concern over adverse poverty trends in the 1990s appear misplaced. There has perhaps been an over-reaction to the previous neglect of urban poverty in China, with unwarranted pessimism about the living standards of poorer households. In particular, we challenge the assertion that urban poverty rose in absolute terms. While this claim makes dramatic headlines, it is does not appear to be supported by the evidence. In particular, we present new evidence on trends in urban poverty from 1988 to 2002 based on our analysis of the best available data-set on income distribution in China. The data come from the surveys in 1988, 1995, 1999 and 2002 conducted as part of the Chinese Academy of Social Sciences (CASS) Household Income Project, which provides a much fuller accounting of household income than the biased measures used in official statistics. While undoubtedly there have been some losers from China’s urban reforms, the CASS surveys imply that this has not translated into a worsening of real income of those at the bottom of the income distribution. Indeed, while real incomes have grown less at the lower end of the distribution than higher up, they nonetheless have grown. As a consequence, whatever fixed point is taken as the poverty line, urban poverty has fallen. Although the existing literature on this question is mixed, we argue this result is consistent with official household statistics, despite the latter’s biased measurement of household incomes. Consequently, while we do not deny that the concept of the “new urban poverty” identifies new forms of relative poverty, the idea that these trends have led to absolute impoverishment appears to be something of a myth.
An important caveat to our argument is that the CASS surveys on which we base our estimates of urban poverty are based on the government’s official sampling frame. While this has the advantage of making the surveys broadly representative of the country as a whole, it has the drawback of excluding the “floating population” of rural-urban migrants. Due to this data limitation, we – like all existing quantitative estimates of rates of urban poverty in China - confine ourselves to the study of poverty levels among Chinese residents with urban registration (hukou). Since rural-urban migrants are no doubt poorer as a group than residents with urban hukou, their exclusion will in some sense lead us to under-estimate urban poverty. Indeed, since rural-urban migration has increased in the 1990s, the exclusion of rural-urban migration will lead us to bias downward poverty rates in urban areas. However, this limitation is perhaps less serious than it seems. Migrants have presumably moved to the cities to make themselves better off – indeed, some estimates put the return to migrant labour at ten times the marginal product of labour in agriculture (Song, 2000). There is no real suggestion that migrants as a group have impoverished themselves by moving to the cities – if anything, the presumption is that migration has provided a means by which they can escape poverty (Park, Du and Wang, 2004). Contrary to the premise of the Harris-Todaro model of migration, what data we have on rural-urban migrants in China (for the year 1999) suggests that unemployment rates among them are negligible (Appleton et al., 2002). Hence even if we could get data on migrants, any rise in urban poverty due to their movement to the cities would surely be more than offset by a contrary decline in rural poverty. By contrast, the second half of the 1990s saw the emergence of mass unemployment among residents with urban hukous. Employees in loss-making State Owned Enterprises found themselves laid-off and enduring long spells of unemployment. It is within this, formerly protected, group that a rise in absolute poverty is most commonly assumed to have occurred.
The structure of the chapter is as follows. Section 2 discusses some of the key issues – both the general debate about growth, poverty and inequality debate; and the specific case of China during its period of urban reforms. Conceptual and practical issues in the measurement of poverty are addressed in Section 3. The small existing literature on trends in urban poverty in China is reviewed in Section 4. Our own findings on household income regarding trends in growth, inequality and poverty are documented in Section 5. Section 6 summarises and concludes.
2.Issues
2.1The International Debate on Poverty Reduction and Growth
The study of the poverty trends during China’s rapid growth is of general interest because it informs a wider international debate. Approaches to poverty reduction differ in the relative emphasis they place on promoting economic growth and on redistributing existing resources in favour of the poor. One extreme would be to rely on economic growth – sometimes characterised as the “trickle down” approach, although this description implies a unjustified passivity in the role of the poor in bettering their own lot. At the other extreme, some argue that it is essential for the government to undertake large-scale redistribution through progressive taxation and welfare policies. How growth is actually distributed clearly has some bearing on this debate. If the poor did not benefit from growth alone, then redistribution would be necessary for poverty reduction. By contrast, if growth benefited the poor disproportionately, then the case for redistribution would be weakened.
The debate on poverty reduction and growth is particularly relevant to China, as its government has shifted its position during the reform period. In the era of planning, the Chinese state played a very strong redistributive role – for example, in urban areas, wage inequalities were compressed, unemployment avoided through implicit job guarantees and a variety of welfare benefits provided by work units. However, during the planning period, it could be some have characterised the Chinese government’s approach to poverty reduction as being closer to “trickle down” in emphasising growth over redistribution – in the words of Deng Xiao-Ping, “let some get rich first”. In the interests of promoting economic efficiency and hence growth, enterprises have been given more freedom in letting worker remuneration reflect productivity, excess workers in stated owned enterprises have been made unemployed and many state transfers have been removed. As Khan (1998) argues, post-reform, the Chinese government has tended to reject a “relief approach” to poverty reduction in favour of efforts to increase income generation. Indeed, Khan suggests that the reluctance of the Chinese government to adopt a redistributive approach to poverty reduction may partly be a backlash against the extreme egalitarianism of the planning period. Clearly, the efficacy of China’s current strategy of emphasising economic growth over redistribution depends partly on the extent to which growth has actually reduced poverty. However, despite the importance of the issue, theory and evidence provide suggest that a priori, we cannot be confident about the distributional effects of economic growth in a particular instance like that of urban China after 1988.
Early theorising about the relation over time between growth, inequality and poverty centred on Kuznets (1955) “inverse U” hypothesis. This postulated that inequality should rise with initial development, but then subsequently fall as development continued. The hypothesis was based on initially rather fragmentary cross-country evidence; Kuznets made the hypothesis after examining data on income distribution from only six countries at different levels of development and only at one point in time. While simple bivariate correlations using cross-country data appeared to corroborate Kuznets’ hypothesis, further work has shown that this hypothesis is not a valid empirical generalisation and specifically is not robust to controlling for a “Latin America dummy”. Middle income countries are found disproportionately in Latin America, where the distribution of income is particularly unequal – in part due to the very uneven division of land. Without regional controls, this gives the appearance that being middle income leads to inequality, when in fact it is simply being a Latin American country that is associated with higher inequality. Analysis of different country experiences in the post-war period generally implies that economic development does not have a systematic effect on inequality, if country-specific effects are controlled for (Fields, 1991). However, this generalisation masks a wide range of experiences, implying the desirability of monitoring the distribution of growth in countries of interest.
Kuznets’ provided some theoretical support for his hypothesis using a simple theoretical model. The model postulated that growth took the form of moving labour from a low return traditional sector to a high return modern sector. This paradigm has some relevance to inequality in China as a whole with the traditional sector perhaps being equated with agriculture and the modern sector with non-agricultural production. However, while relevant to China as a whole, it is not clear to what extent this characterisation is germane to inequality within urban areas[2]. An alternative use of Kuznets’ model might be to characterise the movement of urban labour out of loss-making state owned enterprises into more economically viable companies, whether in the state or private sector.
To the extent that cross-country empirical evidence implies that, on average, there is no effect of development on inequality, it follows that one should expect growth to tend to raise the incomes of the poor on a one-for-one basis. Cross-country evidence appears to support this (Dollar and Kraay, 2000). Simple economic theories of growth are also consistent with a favourable effect of growth on poverty. For example, within the Solow (1956) neoclassical growth model, growth in income per capita may arise in the long-term due to improvements in total factor productivity (TFP) or, in the short term, due to increased rates of accumulation of physical capital. TFP growth may be as a result of improved efficiency – perhaps due to marketisation, state reform or greater international competition – or due to improvements in technology. However, whether growth is due to improvements in TFP or to capital accumulation, the model predicts that wages will rise and, with them, the main income source of the poor. The Solow model can also be augmented to allow human capital accumulation to play a role. In the Chinese case, the gradual improvement in the educational attainment of the urban population is likely to lead to increased returns to labour in general, although perhaps a fall in the return to education ceteris paribus.
However, while cross-country evidence implies that, on average, growth appears to reduce absolute poverty and have no impact on inequality, there is considerable variation in country experiences. Rapid economic growth in other East Asian economies since 1960 appears to have been progressively distributed and to have sharply reduced poverty. Conversely, there are examples of worsening income distribution during growth and indeed of the incomes of the poor falling[3]. China, by virtue of its large population and sustained high economic growth, is arguably the most important recent case for study of the relations between growth, inequality and poverty. Poverty trends in rural areas are, of course, important for a complete treatment of this topic – but consideration of developments in urban areas is a necessary part of the analysis.
2.2The distributional effects of China’s urban reforms
Since international evidence provide no firm predictions on the likely distributional effects of China’s economic growth after 1988, it is important to consider particular features of China’s experience that may be relevant. Perhaps the most important feature is that the period in question has seen a gradual transition from a predominantly planned economy to one that is more market led. Nee’s (1989) “market transition” hypothesis postulates that the move from a command economy to a competitive one will see a fall in returns to certain unproductive characteristics of workers – such as political connections – and a rise in returns to productive characteristics such as education. Conceivably, this may lead to a fall in wages of those with unproductive characteristics – e.g. the relatively unskilled – and ceteris paribus this may contribute to increased poverty. Whether this trend leads to a rise in absolute poverty, or merely relative poverty, will depend on the extent to which marketisation leads to general rises in wages, as well as changes in wage differentials. However, the marketisation hypothesis has been disputed by those who argue that, far from leading to competitive outcomes, reform may at least initially lead to managerialism and the freedom of employers to vary wages according to a variety of factors, not just those affecting productivity. In such a situation, there may be more scope for rent-seeking behaviour, clientelism and other inequalities that cannot be justified on “efficiency” grounds. This may explain why urban wage data shows a rise in the premium for being a Communist Party member (Appleton et al., 2004).
Somewhat contrary to the predictions of the “market transition” hypothesis are those surrounding the impact of globalisation and increased international trade on China. A characteristic of China’s rapid economic growth is that it has been associated with an increase in exports and a general increase in the openness of the Chinese economy to international trade, most clearly signalled by China’s ascension of to the World Trade Organisation. Wood (1995) uses a modified version of the Heckscher-Ohlin trade theory to make predictions about the distributional consequences of increased trade between rich countries and poor countries like China. He uses the theory to show that in poor countries, where unskilled labour is relatively abundant, one should expect a rise in the returns to unskilled labour and fall in the relative returns to skilled labour. This trend may be enhanced by the increase in the supply of human capital noted earlier. Both developments should tend to make the distribution of wages more equal. However, some there are countervailing trends– for example, the introduction of skill-biased technological change (e.g. computerisation).