Rural Industrialisation and Spatial Inequality in China, 1978-2006

Chris Bramall ( ) teaches Chinese Political Economy at the School of East Asian Studies, Sheffield University, Sheffield, UK.

CHINA SINCE 1978december 27, 2008 EPW Economic & Political Weekly44hard to believe that total employment in rural industry (TVEs, SOEs and COEs combined) can have been much less than 100 million in 2006. Official rural employment stood at around 500 million in that year, suggesting an industrial share of about 20% of the rural workforce.4Much of the growth in the early 1980s was driven by small-scale private enterprises, partly because the more inefficient CBEs were closed during the “Readjustment” of the early 1980s. Never-theless, the public sector – rural enterprises owned and often managed by township and village governments – expanded rap-idly after the mid-1980s and dominated the TVE sector until the mid-1990s; the public sector share was 50% of total TVE employ-ment and 65% of value added in 1994 (Bramall 2007: 78-79). Thereafter, privatisation was swift. By 2005, the public share was only 25% of value added and 18% of employment (Nongye bu 2006: 226).5 Even in Jiangsu province, the heartland of public rural industry in the 1980s and 1990s, privatisation was rapid and thoroughgoing. By 2005, the share of the public sector (usu-ally called “collective industry” in Chinese sources) in Jiangsu was down to a meagre 8% of value added (JSB 2006: 202). 2 IndigenousRuralIndustrialisation and PovertyThe most positive consequence of industrialisation has been its contribution to absolute poverty reduction. Official Chinese data show a fall in the absolute poverty headcount from 250 million in 1978 to 14.8 million in 2007 (CPIRC 2008). These magnitudes are not reliable; it is hard to square an absolute poverty figure of 250 million with the high levels of life expectancy recorded in the late 1970s. However, there is little doubt that absolute poverty has fallen sharply over the last three decades.There is equally little doubt that poverty reduction has been rapid in those places where industry has developed most quickly because the new local enterprises have enabled workers to aban-don (low wage) jobs in farming, thus “leaving the land but not the countryside” (litu bu lixiang). Rural industrialisation has been most rapid in eastern China, which entered the 1980s with both a well-developed industrial base (one of the key legacies of Maoism) and favourable economic geography (especially in terms of low transport costs). And many parts of the region have locked in those advantages by attracting large inflows of foreign capital. The most obvious examples are the deltas of the Yangzi (Sunan)6 and Pearl rivers, but rapid industrialisation has been the norm across the coastal region. As a result, 29% of the rural workforce in the eastern region was employed in industry by the end of 2006, compared with only 5% in western China (SSB 2008). No wonder, then, that absolute rural poverty has been virtually eliminated in eastern China: only 0.7% of the eastern population was officially categorised as poor in 2006 compared with 5.1% in western China (SSB 2007a: 8).The significance of industrialisation is apparent from the sources of rural income (Table 1). The gap in per capita income between China’s richest provinces like Jiangsu and Guangdong, and poor provinces such as Gansu and Yunnan was very marked by 2006. Yet differences in income from household production (mainly farming) were small as a comparison between Yunnan and Guangdong demonstrates. The main reason for rural prosperity in the richer provinces was the high wage income from industrial employment. For example, wage income per head in Jiangsu was over seven times larger than in Yunnan. Moreover, wages were the fastest-growing com-ponent of rural income (Gustaffson et al 2008: 64-65). In other words, industrial em-ployment offered a route out of poverty in the 1980s and 1990s. It is the scarcity of such jobs that ex-plains its persistence in western China.The role played by industrial growth in poverty reduction is also apparent from county-level data onGDP. Much of rural China was very poor in the late 1970s, mainly because of dependence on low productivity agriculture, but rapid industrialisation has transformed the fortunes of the people. Henan province, for ex-ample, has comfortably eclipsed its neighbours in terms of per capitaGDP growth (Figure 2, p 47). This is especially interesting for two reasons. First, the auguries for rural development were unpropitious because Henan was one of the poorest parts of China in the late 1970s. Twenty-six of the 221 poor counties identified in 1977-79 were located in Henan alone, a total second only to that recorded in Guizhou (Nongye bu 1981: 120). Second, rural industrialisation has driven growth; rapid GDP growth has occurred in both urban and rural areas, and it has been based on industrial development.Rural industrialisation has not yet transformed Henan’s pros-pects. For one thing, poverty has not been eradicated. According to a list of poor counties used by the State Council’s Office of the Leading Group for Poverty Reduction (OLG) in 2004, 31 of the 592 counties were to be found in Henan (OLG 2004). The main prob-lem, as Table 1 shows, was that wages were still contributing much less in Henan than in Jiangsu or Guangdong,TVE expan-sion notwithstanding. For another thing, industrial growth has brought environmental damage in its wake, with unknown con-sequences for long-term poverty reduction. The main problem is water pollution in the Huaihe river basin, mainly caused by Henan’s paper and pulp industry. As Nygard and Guo (2001: 17) conclude: “Henan [was] the largest water and COD load generator ... [this] explains why Henan Province was singled out as one of the main targets for the closing down interventions applied from 1996.” Some 17,000 industrialTVEs were closed between 1996 and 1997 in Henan (out of a national total of over 64,000 closures), which reduced chemical oxygen demand in the province’s rivers by about 70% (Nygard and Guo 2001: 32, 39). For all that, the problem of pollution persists: the Haihe was still the most pol-luted of China’s river basins in 2005 (SEPA 2006: 19). The impact of rural industrialisation on short-term poverty reduction has been positive across Henan but the province has stored up a range of future problems for itself.The positive impact of rural industrialisation on poverty is per-haps more apparent in China’s coastal provinces. The breakneck Table 1: Sources of Rural Income(2006)(yuan per capita)Province Total Net Wage Income Net Income IncomefromHousehold ProductionJiangsu 5,813 3,105 2,271Guangdong 5,0802,906 1,694Henan 3,261 1,023 2,108Guizhou 1,9857161,113Yunnan 2,251 442 1,632Gansu 2,134 637 1,292China 3,587 1,375 1,931The surveys on which these data are based are not fully reliable because they under-sample both rural rich and rural poor.Source: SSB (2007a: 320).

0 5000 5001 25000 25001 50000 50001 206227 Migrant Numbers Beijing Tianjin Shandong JiangsuAnhui Shanghai Zhejiang Fujian Guangdong

CHINA SINCE 1978december 27, 2008 EPW Economic & Political Weekly46transformed counties into cities during the 1980s and 1990s by attracting migrants.The case of Anhui exemplifies the underlying dynamics of the migration process. Although agricultural growth in the early 1980s had made a dent in rural poverty, the province was still very poor in 2000. Per capitaGDP was only two-thirds of the national average and Anhui was alone among the eastern provinces in that its illiteracy rate (13.5%) exceeded the national figure of 8.7% (SSB 2001). Not surprisingly, therefore, many of Anhui’s rural population were keen to migrate during the late 1980s and the 1990s and – equally unsurprisingly – the main destinations were Shanghai and the neighbouring provinces of Zhejiang and Jiangsu (Figure 1, p 45). Some found their way to Beijing and to Guangdong, but most sought the industrial jobs on offer in the rural industries springing up across the Greater Yangzi delta region.Such migration out of Anhui and elsewhere undoubtedly con-tributed to poverty reduction directly (migrants increased their income by moving to industrial jobs), and indirectly via the pay-ment of remittances by migrants to their natal village. Remit-tances totalled between 15 and 30% of net rural income in prov-inces like Jiangxi by 2006, and virtually all of China’s poor prov-inces (Yunnan was the only exception) were significant recipients (SSB 2007a: 320, 322). Moreover, the 2002 data show households living in nationally designated poor counties receiving a per cap-ita remittance income of 240 yuan. Although less than the na-tional average in absolute terms (which was 298 yuan), this in-come amounted to 18% of net income in poor counties, compared to only 12% in an average county (SSB 2003: 81). Nevertheless, migration was at best an incomplete solution to the poverty problem. For one thing, many of those living in nation-ally designated poor counties were not poor at all (Riskin 1993). And it was these non-poor households which benefited more from remittances than those living below the poverty line. This emerges from a 2002 survey of those living in designated poor counties (Table 3). Of the 52,624 households covered in this sur-vey, 273 had a per capita income of less than 100 yuan. Although this income level was pitifully low – the official poverty line in 2002 was 627 yuan (SSB 2007a: 43) – these very poor households benefited more from remittance income in percentage terms than any other class. However, the most striking feature of Table 3 is that the remittance income of other households living below the poverty line was lower than (both abso-lutely and proportionately) the remit-tances received by non-poor households.9 Tellingly, the income class just below the poverty line (earning between 300 and 500 yuan) earned the least in percentage terms from remittances. Only when per capita income reached 2,500 yuan did re-mittance income become less important in relative terms. The conclusion is clear. Re-mittances generated by outmigration helped all the income groupsinpoorcoun-ties but it was a remarkably inefficient way of reducing absolute poverty because the bulk of remittances went to the non-poor. Poor households gained little from outmigration because most migrants were younger and better educated than the rural norm whereas many poor households were in poverty precisely be-cause they lacked educated young workers who could find out-side work. For example, the 2002 survey of poor counties revealed that only 8% of workers from illiterate households migrated, whereas the figure was 16% for those with a middle school education (SSB 2003: 155). The comprehensive Second Agricultural Census of 2006 also shows that migrants were not representative of the rural population: 53% of migrants were aged 30 or younger, compared with 30% of the rural population as a whole. Only 36% of migrants were women compared with 49% of the rural labour force. And, whereas 6.8% of the rural labour force was illiterate, that was true of only 1.2% of migrants (SSB 2008). This was truer of a province like Guangdong. The proportion of illiterates in the floating population was slightly higher (1.3%) but no less than 71% of the floaters were aged under 30 in 2000 (GSB 2003: 78-79). In short, the typical migrant was male, young and relatively well educated and therefore little remittance income went to the neediest rural households.Moreover, because migrants were more skilled than the aver-age rural worker, the countryside suffered a skills drain which constrained indigenous industrialisation. To be sure, this drain was partly offset by return migration. However, recent research suggests that the contribution of returnees was small. According to Wang and Fan (2006: 951): “Contrary to most studies on China that emphasise success returnees such as entrepreneurs, this study shows that these returnees are rare but that failure return-ees – those who are rejected by the destination and who have difficulties surviving there – are prevalent.” There is not even much evidence that return migration increased rural investment; the accumulated capital of returnees was typically spent on house-building and durable consumer goods in non-poor vil-lages, and on food, clothing and other necessities in poor villages (De Brauw and Rozelle 2008).In short, it is difficult to be sure about the net effect of migra-tion on absolute rural poverty. Most of the literature is positive (Fan 2008: 79) and there can be no denying the positive effects of migration. However, remittances accrued mainly to the relatively well off rather than to the rural poor. More negatively still, coun-ties which were large exporters of labour suffered a skill drain. The extent to which this constrained industrialisation, and thereby offset the remittance effects, re-mains largely unexplored terrain. There is no evidence of declining per capita real incomes even in poor areas, but this may just be because the positive effects of (limited) indigenous development and continuing educational expansion are off-setting the adverse impact of outmigration. As Hare and Zhao (2000: 150) say: “Migra-tion should not be seen as a substitute for locally based efforts to improve economic conditions in the rural areas”.Table 3: Income Received from Migrant Remittances in Poor Counties(2002)Income Class Households Remittance Total Net Remittance (Yuan Per Head) (Number) Income Per Income Per Share (%) Head (Yuan) Head (Yuan) 0-100 273 215935.6100-300 1,273212219.5300-500 3,068 33411 8.0500-800 9,003 69 650 10.6800-1,000 7,137 123 896 13.71,000-1,500 14,380 216 1,230 17.61,500-2,000 8,309 351 1,721 20.42,000-2,500 4,312 5022,220 22.62,500-3,000 2,164 6152,724 22.6> 3,000 2,705 848 4,125 20.6All classes 52,624 240 1,305 18Sources: SSB (2003: 81, 158-59).

Ningxia 0-16 17 20 21-40 GDP Growth 1982-2000 Qinghai Liaoning Nei Menggu Beijing Tianjin Heber Shanxi Shandong Shaanxi Henah Jiangsu Hubei Shanghai Zhejang Anhui Hunan Jiangxi Fujian Guangdong Sichuan Guizhou Yunhan Guangxi Hainan Gansu

CHINA SINCE 1978december 27, 2008 EPW Economic & Political Weekly48overall inequality also fell (from a massive 60% in 1982 to only 38% in 2000).13 By contrast, the contribution of inequality be-tween counties (inter-rural) to overall inequality rose from 14% in 1982 to 42% by 2000. The impact of rural industrialisation on overall spatial inequality was twofold. On the one hand, it played a role in the narrowing of the urban-rural gap shown in Table 4. Established urban centres like Shanghai grew from a relatively high base after 1978, whereas suc-cessful industrialising counties – such as Kunshan in Jiangsu or Dongguanin Guangdong – have enjoyed the advantagesofback-wardness. The impact is apparent from the example of Wenzhou. Wenzhou city proper experiencedGDP per capita growth of only 12% a year between 1978 and 2004, whereas the median growth rate for the counties within Wenzhou municipality was 15% (Wenzhou tongjiju 2005: 42-43, 80-81). As a result of this narrowing of the urban-rural gap, the overall CV for Zhejiang increased only marginally between 1982 and 2000 (from 0.39 to 0.40), and in Jiangsu it actually de-clined (0.73 to 0.67). In both provinces, as elsewhere in China, rural industrialisation has been so rapid that the most successful counties have been reclassified as cities. However, the principal effect of rural industrialisation has been to drive the increase in inequality between counties ob-served in Table 4. While some counties have forged ahead on the back of successful rural industrialisation, others have fallen be-hind because of industrial underdevelopment. Many of these un-successful counties during the 1982-2000 period were in western China: there is a clear correlation between location on the one hand, and both the level and the rate of growth of GDP per head on the other. Nevertheless, many relatively slow-growing coun-ties were located in eastern China. As a result, inequalities inGDP per capita between counties have risen even in the coastal prov-inces. For example, inter-rural inequality also increased over the course of the 1980s and 1990s in Jiangsu, a province with a well-established transport infrastructure and few real geographical obstacles to development. Although the overall CV fell from 0.74 to 0.66, the rural CV increased massively, rising from 0.28 in 1982 to 0.69 in 2000. This reflects the joint impact of rapid rural industrialisation in the southern part of the province and slow industrialisation in its northern counties.The trend in spatial inequality between counties for Zhejiang is especially revealing given both its well-developed infrastruc-ture and its twin growth poles in Hangzhou-Ningbo to the north and in the Wenzhou area in the south-east. In Zhejiang at least, one might therefore have expected powerful spread effects and hence convergence. However, the CV for countyGDP per capita rose from 0.32 to 0.39 between 1982 and 2000. Ye and Wei (2005) arrive at the same conclusion; their data shows an even more dra-matic increase in the ruralCV between 1978 and 1998.14 In fact, there is a clear distinction between the core and the periphery in Zhejiang. Per capitaGDP levels are high around the urban centres of Hangzhou, Ningbo and Wenzhou but they are much lower in the counties on the southern border of the province. Alarmingly, there is evidence of divergence even within a geographical unit as small as Wenzhou municipality itself. As noted earlier, its two poorest counties (Wencheng and Taishun) enjoyed an annual real per capitaGDP growth of 10 and 13% respectively between 1978 and 2004, but this did not lead to convergence because me-dianGDP per capita growth in the municipality was almost 16%. Thus, despite their good growth record, Wencheng and Taishun fell further behind and they remain part of Zhejiang’s poor south-ern border region. The likely explanation is the exodus of skilled labour and capital from the two to faster-growing Wenzhou counties. This evidence of rising inter-rural ine-quality even in provinces like Jiangsu and Zhejiang suggests that rural industrialisa-tion not only failed to generate large posi-tive spillovers but also may have produced backwash effects. Most migrants being young and skilled, their exodus may have denuded poor regions of the human capi-talneeded for indigenous industrialisation. The very fact that neither JiangxinorAnhui experienced rapid per capitaGDP growth in the 1980s and 1990s despite massive outmigration lends support to the conclu-sion that industrialisation, for all its poverty-reducing effects, ex-acerbated spatial inequality between counties.4 ConclusionsRural industrialisation has proved to be a double-edged sword for China. On the one hand, it has played a key role in reducing absolute poverty by job creation. These new jobs have attracted local farmers, hence reducing poverty in the main centres of rural industrialisation such as Zhejiang, Jiangsu and Guangdong. Rural industries have also attracted farm workers from the deep interior, and in the process helped to reduce poverty in histori-cally underdeveloped provinces like Anhui and Guizhou. In this sense at least, the fruits of rural industrialisation in regions fa-voured by history and by geography have trickled down to other parts of the People’s Republic. Moreover, rural industrialisation has been so rapid that the per capitaGDP gap between cities and those areas which were rural in the early 1980s has narrowed. For example, much of Sunan has started to catch up with Shang-hai. This outcome fits better with traditional diminishing returns-based neoclassical theory than the predictions of much of the new economic geography.However, and precisely because the process has been so uneven, rural industrialisation has exacerbated inequality across rural China. At a local level, the main beneficiaries of remittance in-come appear to have been non-poor households, such that the net effect of remittances on the income distribution is disequalising. The impact on spatial inequality has also been adverse. Although some counties have bounded ahead, many have been left behind; as a result, inequalities in per capitaGDP between China’s counties have risen. Migration has brought some benefits to the periphery via remittances, but growth rates have still been much slower Table 4: Coefficient of Variation for County-Level Per Capita GDP(1982 and 2000) All Cities Counties Only Cities Only Urban-Rural andCounties Gap(Ratio)1982 0.74 0.40 0.563.0:12000 0.84 0.75 0.63 2.4:1Counties and (county-level) cities are as designated at the time of the 1982 Population Census for both the 1982 and 2000 calculations. This approach is adopted because we are primarily interested in the fate of areas which were rural in 1982. As many counties have been reclassified as cities over the last two decades, it makes little sense to use the jurisdictional boundaries of 2000. That type of approach makes the notion of a successful county a contradiction in terms: any county which enjoys very rapid growth is reclassified as a city and therefore successful rural industrialisation is definitionally impossible. The population data used here are the census figures on the resident population, which include floaters of six months or more. Note that the coefficients of variation are not sensitive to population weighting.Sources: See Bramall (2007) for GDP sources and methodology. Population data from SSB (1988; 2000).

CHINA SINCE 1978Economic & Political Weekly EPW december 27, 200849than in the core because of the exodus of skilled labour. More-over, there is little to suggest that the process of outmigration has slowed or halted in more recent years. To be sure, there have been reports of labour shortages in the coastal provinces, seem-ingly driven by a stagnation of real wages in rural industry (Shao et al 2007). Further, the very fact that the size of the national floating population in 2005 was only a little higher than it was in 2000 (147 to 144 million) suggests that the process of outmigra-tion is slowing. Nevertheless, and although we must await the results of the next national census to be sure, there is little sign that the flood of inter-provincial migrants into the most dynamic coastal region is abating. Jiangsu’s permanent population (which includes floaters of six months duration or more) increased by 0.6% a year between 2000 and 2007. However, the rate was much faster in the municipalities of Sunan. Nanjing’s population increased annually by 2.8%, Wuxi’s by 2.4% and Suzhou’s by 3.8% (JSB 2008). Similarly for Hunan; the latest data suggest a total of 6.9 million extra-provincial migrants, well up on the 4.3 million of 2000 (HSB 2008). This evidence that rural industrialisation (and the migration flows it engenders) is both continuing, and disequalising in its effects on spatial inequality in the countryside, suggests that policy action is needed to accelerate the pace of industrialisation in backward regions. However, this remedy has found little favour amongst scholars. Instead, the conventional wisdom is that rural industrialisation is not the solution to backwardness, and that poor areas would do better to focus on agriculture or on the develop-ment of infrastructure (Rozelle and Nyberg 1999; Zhang et al 2003). There are two lines of argument at work here. One is outright geographical determinism: industrialisation is simply infeasible in western China. Second, industrial development is frowned upon because it will have to be state-run and state-sub-sidised in the early stages of development for the usual infant in-dustry reasons (private sector banks are unwilling to supply funds to infant industries because of the uncertainty involved). Hostility towards state-led industrial policy feeds into hostility towards rural industrialisation, irrespective of the merits of that industrialisation.Yet neither of these arguments against industrialisation is compelling. Of course geography imposes constraints upon de-velopment. There are remote areas across China like the Hima-layan plateau and the Gobi desert where industrialisation is not feasible. However, for most of central and even western China, geography is only as a minor constraint. Moreover, many of the obstacles to inter-provincial trade result from local protectionism rather than high transport costs. And Henan’s experience sup-ports the conclusion that the disadvantages of being located in the interior are all too easy to exaggerate. Although historically a very poor province and lacking the geographical advantages en-joyed by the coastal provinces, Henan has experienced rapid in-dustrialisation since the early 1980s. This suggests that, for many poor areas, geography is not destiny: the return to investment in manufacturing is higher than the return to investment in educa-tion, agriculture or infrastructure. As for the second argument, the Chinese evidence suggests a positive role for the state. Convergence between China’s cities and parts of the countryside may be occurring: that outcome is very much a neoclassical one. However, the underlying processes driv-ing convergence are anything but neoclassical. Rather, successful rural industrialisation across China has been built by local govern-ments acting as developmental states. The very fact that rapid industrialisation was both rapid and state-led in Jiangsu in parti-cular, and in rural China in general, during the 1980s and 1990s suggests that the local state is the solution to backwardness, not part of the problem. It was not private industry that allowed many parts of rural Sunan to begin to catch up with the well-established cities of the Yangzi delta but the agency of local government.15 All this suggests a clear policy conclusion: China needs more rural industrialisation, rather than less, if poor counties are to catch up. To reduce spatial inequality, China should aim for a high level equilibrium (in which the majority of counties are industrialised) than a low level equilibrium (in which industry is underdevel-oped). Successful rural areas have closed the gap between them-selves and the cities by rapid industrialisation, and logic dictates that poor rural areas can achieve the same outcome if they per-sist with – or in some cases initiate – industrial development. Notes 1 For a variety of perspectives on the causes of rural industrialisation, see Oi (1999), Whiting (2001), and Bramall (2007). 2 The abolition of communes and brigades in 1984 meant that the ownership of CBEs was transferred to the newly formed town, township and village governments. Since then, the TVE epithet has been used to describe the sector. Note, however, that the coverage of the TVE sector is broader than that of the old CBE sector because it includes the private enterprises (in which category I in-clude household, or what the Chinese callgeti en-terprises), which developed quickly in the early 1980s. The Chinese data series are not consistent across the 1984 divide; much of the growth of the mid-1980s (total employment increased by over 60% in 1984) simply reflects redefinition. 3 Output and employment in the TVE sector have been over-reported. Much of it was identified dur-ing the course of the 1995 Industrial Census and the 1997 data were adjusted accordingly. The cov-erage of the sector was also partially redefined in line with a more general re-definition of Chinese industrial sub-sectors; the effect was to reduce total employment in the TVE sector by around 3 million workers. Nevertheless, the “fact” of rapid rural industrialisation after 1978 is not in doubt. 4 As there is some evidence that the size of the farm workforce is exaggerated (a large portion of the notional farm workforce had migrated to the cit-ies; see Rawski and Mead 1998), it may be that rural industry employed close to a quarter of the rural workforce. 5 These figures categorise all enterprises designated as collective, cooperative, joint or limited liability as “public”. Most had a large state-ownership share, but this approach does somewhat over-state the size of the public share. 6 Sunan in this article refers to the counties of southern Jiangsu and northern Zhejiang. 7 For an excellent survey of migration issues, see Fan (2008). I use “migrant” here to refer to the floating population (liudong renkou), meaning (as defined in the 2000 Census) anyone living away from home for six months or more. The official definition of a migrant in China is a person who changes his or her place of household registration (qianyi renkou) but the scale of these flows is small; it is the floating population which is of in-terest to scholars and policymakers. 8 Many migrants were employed in the tertiary sec-tor; the 2006 second agricultural census reveals that 57% of migrants were employed in industry, 41% in the tertiary sector and only 2% in agricul-ture. However, the tertiary sector itself only de-veloped because of rural industrialisation. 9 National surveys of rural income report the same finding: wage income, of which remittance in-come was a key component, had a strongly dise-qualising effect on the distribution of rural in-come (Gustaffson et al 2008: 66-67).10 Almost all the literature considers only inequality between provinces, which is not especially useful given both the level of aggregation (many Chinese provinces are as big as large European countries) and the mix of urban and rural jurisdictions within each province. An exception is Rozelle (1996), but his study uses county-level data on gross output

Sonobe, T, D Y Hu and K Otsuka (2004): “From Inferior to Superior Products: An Inquiry into the Wenzhou Model of Industrial Development in China”, Journal of Comparative Economics, 32 (3), 542-63.