WTO Accession, Rural Labour Migration and Urban Unemployment in China

Urban Studies, Vol. 39, No. 12, 2199 2217, 2002 WTO Accession, Rural Labour Migration and Urban Unemployment in China Fan Zhai and Zhi Wang [Paper received in nal form, May 2002] Summary. This paper evaluates the rural urban migration and urban unemployment effects of China s WTO accession based on the China US bilateral agreement on market accession. The evaluation is conducted by a 53-sector recursive dynamic computable general equilibrium (CGE) model of China benchmarked to a most recent version of the Chinese Social Accounting Matrix (SAM). The simulation results show that there is a great need for co-ordination between China s rural urban migration policy, labour market reform and the implementation of trade liberalisation measures after China s WTO entry. The net bene t from WTO membership will be maximised if China adopts a policy gradually to relax its rural urban migration control in conjunction with its labour market reform. Doing so not only prevents a dramatic worsening of the urban unemployment problem, but also permits enough labour market exibility to create more employment opportunities for rural unskilled labour shifted out of the farming sector. 1. Introduction Since the beginning of economic reform and its opening to the outside world, China s economy has been growing at a rate of nearly 10 per cent annually while its external trade has expanded by more than 15 per cent a year. Thus in the past decade China s trade volume more than quadrupled, compared with an 83 per cent growth of world trade as a whole. In 2000, its trade reached US$474.3 billion, ranking eighth in the world, with export volume reaching US$249.2 billion. Thus China has emerged as an important player in world trade, a role that is likely to increase as the nation continues to grow and industrialise. In order to ful ll its membership requirements at the World Trade Organisation (WTO), China has to adopt broad and deep trade liberalisation measures, resulting in substantial reductions in trade barriers across all economic sectors. A major concern is that this will have a dramatic impact on China s industries and may induce massive unemployment, especially in the inef cient stateowned sectors. Due to the economic slowdown since 1997 and dif culties associated with restructuring state-owned enterprises, the urban unemployment rate has grown rapidly in recent years. It is estimated that urban unemployment was maintained at Fan Zhai is in the Department of Policy and Fiscal Affairs, Ministry of Finance, PRC, Sanlihe, Xicheng District, Beijing 100820, P. R. China. Fax: 86 10 6855 1523. E-mail: fzhai 99@yahoo.com. Zhi Wang is in the School of Computational Sciences, George Mason University, and with the Economic Research Service, US Department of Agriculture, 1800 M Street, NW, Room N2109, Washington, DC 20036, USA. Fax: 202 694 5662. E-mail: zwang@ers.usda.gov. The authors are grateful to the Editor of this journal and anonymous referees for comments and suggestions. Appreciation also extends to Fred Gale at ERS/USDA for his valuable comments on an earlier draft of this paper. The views expressed in the paper are those of the authors and should not be attributed to their af liated institutions. 0042-0980 Print/1360-063X On-line/02/122199-19 Ó 2002 The Editors of Urban Studies DOI: 10.1080/004209802200003382 7

2200 FAN ZHAI AND ZHI WANG 14 million during the period 1995 2000, and that it reached 21.6 million, or 9.3 per cent, at the end of 1999 if laid-off workers (Xia- Gang ZhiGong) are taken into account (Cai et al., 2001). Dealing with this problem is already one of the most pressing policy issues currently facing the Chinese government. There is a hot debate on how much the in ow of rural labour has contributed to urban unemployment. Accompanying the rapid economic growth in the past two decades, the income gap between urban and rural areas widened sharply. As a result of this income divergence and the gradual erosion of mobility restrictions, rural urban migration has increased rapidly, causing the urban population to grow faster than the rural population. In many cities, regulations restricting this in ow of rural labour have been issued in order to protect the employment of local workers. Undoubtedly, China s WTO accession will have important implications for rural urban migration and urban employment. Given the strong political in uence of urban residents, these effects are signi cant for the feasibility of any trade policy reform in China. This paper aims at providing a quantitative assessment of the rural urban migration and urban unemployment effects of China s WTO membership based on the China US bilateral agreement on market accession signed in November 1999. The assessment is conducted by a 53-sector recursive dynamic computable general equilibrium (CGE) model of China benchmarked to a most recent version of the Chinese Social Accounting Matrix (SAM). It estimates quantitatively the sectoral employment shifts likely to result from WTO accession, and the resulting social costs, under different assumptions about labour market rigidity. In this way, it provides some insight into the challenges and opportunities that WTO membership may bring. The paper is organised as follows. The next section outlines the basic structure of the CGE model for China used in this analysis. Section 3 highlights the economic structure and degree of market openness of the current Chinese economy and describes the base-year data used in our model. Section 4 describes the major assumptions used to calibrate a baseline scenario and three alternative scenarios for the years 1998 2005. Section 5 uses simulation results to assess the impact of China s WTO accession on rural urban migration and urban unemployment. Section 6 concludes by drawing policy implications. 2. Model Features The Chinese CGE model constructed for this study is an extension of the CGE model that had been used in China s WTO accession study (Development Research Center, 1998). It is closely related to the applied general equilibrium model already used extensively over the past two decades to analyse the impact of trade policy reform (see Dervis et al., 1982; de Melo, 1988; Shoven and Whalley, 1992; de Melo and Tarr, 1992; Hertel, 1997). The starting-point for our model is the prototype CGE model developed for the Trade and Environment Programme of the OECD Development Centre (Beghin et al., 1994). However, signi cant modi cations were made to capture the major features of the trade and tax system in the current Chinese economy (Wang and Zhai, 1998; Zhai and Li, 2000). A more complete description of the model is provided in the Appendix. The main features of the model are summarised in this section. As pointed out by Naughton (1996), China had established two separate foreign trading regimes by 1986/87. The CGE model explicitly treats them both. One is the processing trade or export promotion regime, which is extremely open and is participated in by most foreign-owned and many domestic rms. Under this regime, producers process and assemble imported goods, turning them into nished products for export; these imported goods are exempt from tariffs and value-added taxes. The other regime consists of ordinary trade carried out under traditional, although increasingly reformed, taxes and regulations. Since 1990, the rst

WTO ACCESSION, MIGRATION AND UNEMPLOYMENT 2201 regime has grown rapidly, now accounting for more than half of all exports. Production is modelled using nested constant-elasticity-of-substitution (CES) production functions, assuming constant returns to scale. Household demand is modelled by the extended linear expenditure system (ELES). The other nal demand accounts are assumed as xed-coef cient expenditure functions. Trade is modelled using the Armington assumption for import demand and a constant elasticity of transformation (CET) for export supply. World import prices are assumed exogenous in US dollars. Exports are demanded according to constantelasticity demand curves. All commodity and factor markets, except that for urban unskilled labour, are assumed to clear through market prices. For urban unskilled labour, we adopt the marketequilibrium speci cation of Harris and Todaro (1970), in which the wage is assumed xed relative to a GDP de ator (which is the numeraire of the model). This re ects the rigidity of urban labour markets and the abundant supply of unskilled labour in the Chinese economy. Urban unemployment is endogenous, adjusting to clear the market for urban unskilled labour. The model has a built-in demographic module which endogenises population growth and labour supply in both urban and rural areas. Migration from rural to urban areas is modelled by assuming imperfect labour mobility, which re ects the policy and institutional factors that limit labour movement as well as the location preferences of residents. The movement of labour between rural and urban areas is determined by the relative expected wage (de ned as the product of wage and employment rate in this model) and a constant elasticity of transformation. 1 Capital is assumed partially mobile across sectors, re ecting difference in the marketability of capital goods across sectors. Rural labour is assumed perfectly mobile across sectors, with 77 per cent of it agricultural in the base year. Urban labour is assumed not employed in agriculture, but perfectly mobile across non-agricultural sectors. Furthermore, within non-agricultural sectors, rural and urban labour can be substituted for each other in production. Thus agricultural labour could become nonagricultural either through migration or through rural industrialisation. The CGE model has a simple recursive dynamic structure. Dynamics originate from the accumulation of changes in productive factors and productivity. The model is benchmarked on 1997 China s Social Accounting Matrix (SAM) and is solved for subsequent years from 1998 to 2005. Life expectancy, the total fertility rate and growth of gross domestic product (GDP) are all assumed exogenous. The growth rate of the capital stock, on the other hand, is determined endogenously by saving and investment relations. 3. Economic Structure and Market Openness in the Current Chinese Economy The 1997 Social Accounting Matrix was developed by the Development Research Centre (DRC) at the State Council from the most recent input output table. This section outlines the basic features of the industrial structure and market openness of the Chinese economy in 1997 based upon the SAM. Table 1 summarises this information in a 12-sector aggregation. 2 For each sector, the base-year data for shares of output, employment, imports, exports, trade dependence and the share of ordinary trade are reported. Columns 10 and 11 give information about the degree of import protection. As may be seen in columns 1 4, the data are notably asymmetric among the shares of output, employment and trade. For example, the grain sectors account for 14 per cent of China s employment but only produce 2.3 per cent of its output and account for 1 per cent of China s total trade. Textile and apparel industries employ 2.1 per cent of China s labour force, but produce 6.6 per cent of its total output and account for more than 21 per cent of the country s total exports. Export dependency is high for textiles and

2202 FAN ZHAI AND ZHI WANG Table 1. Economic structure and market openness in China, 1997 Ordinary Ordinary Import/ exports/ imports/ Domestic Total Total Net export Nominal Collect Output Employment Imports Exports use Export/Outputs exports imports (billion tariff rate tariff rate (percentage) (percentage) (percentage) (percentage) (percentage) (percentage) (percentage) (percentage) yuan) (percentage) (percentage) Grain 2 2 2 2 2 2 2.3 14.3 1.1 0.6 3.1 1.9 100.0 77.2 5.0 1.0 0.2 Cotton 0.3 2.0 0.6 0.0 10.7 0.0 100.0 16.6 7.7 3.0 0.6 Other agriculture 9.7 39.4 1.4 2.1 0.9 1.6 97.7 25.3 17.2 6.5 5.1 Processing food 6.9 1.4 3.6 4.8 3.4 5.3 71.2 50.9 34.3 12.3 5.5 Mining 3.4 2.4 6.1 2.7 10.6 5.7 95.4 58.7 31.8 1.4 0.7 Textiles and apparel 6.6 2.1 7.5 21.3 8.6 23.9 53.1 0.9 260.6 30.2 0.2 Other light industry 5.0 1.7 6.7 13.6 9.4 19.5 35.3 21.7 142.6 9.8 1.9 Manufactured 17.4 5.1 26.6 15.8 9.3 7.1 55.4 29.1 70.1 10.3 2.7 intermediates Automobiles 1.6 0.4 1.1 0.4 4.2 1.9 62.6 72.9 7.6 50.6 32.6 Electronics 2.5 0.4 13.2 11.5 34.7 36.3 8.6 19.9 25.2 11.8 2.1 Other capital-intensive 12.0 4.2 25.8 17.7 13.3 11.7 35.8 24.1 29.2 13.4 2.8 Services 32.3 26.7 6.2 9.5 1.2 2.5 100.0 100.0 80.9 0.1 0.1 Total/Average 100 100 100 100 6.4 7.7 51.0 32.0 409.5 11.2 2.5 Notes: Imports/Domestic use and Exports/output are in domestic prices; the sectoral share of imports and exports are in world prices. The imports of rice, wheat, corn, cotton, grain mill and vegetable oil are averages for the 1993 97 period.. Source: Chinese Social Accounting Matrix, 1997, Development Research Centre of the State Council.

WTO ACCESSION, MIGRATION AND UNEMPLOYMENT 2203 apparel, electronics and other light industry, as more than 20 per cent of their products are sold on foreign markets. The sectors with the largest import shares are manufactured intermediates and capital-intensive manufactures. The electronics sector has both high export and import dependency, re ecting the fact that a large percentage of this sector s production involves processing and assembling products from abroad i.e. the processing trade. The trade balances by industry in column 9 re ect China s comparative advantage. China is a net exporter of labour-intensive manufactures and a net importer of capitalintensive manufactures. The largest share of the trade surplus in China comes from textiles and apparel. In the agricultural sector, China is net importer of grain and cotton, but has a trade surplus of other agricultural products. The processing trade is the most rapidly expanding portion of China s foreign trade. It accounted for 49 per cent China s total exports and 46 per cent of China s total imports in 1997. 3 It was more important for nished goods than primary and intermediates products. For instance, nearly 80 per cent of China s primary iron and steel, electric machinery, electronics and instruments exports, and more than two-thirds of China s leather, social articles and chemical bres exports were processing exports. The high shares of processing exports in these sectors require a large volume of raw materials, components and semi-processed products imported from abroad as their inputs. Column 8 in Table 1 shows that in the sector of textiles and apparel, the ordinary imports were only about 1 per cent of its total imports; the rest were used for the production of processing exports. Most imports of paper products, building materials, chemicals, basic metal and metal products, machinery and electronics were also used as intermediate inputs to produce processing exports. Another notable feature of the base-year data is the signi cant differences between China s nominal tariff rate and the actual collected rate. It is well known that China s tariff collection is signi cantly below its nominal tariff level because of imports for the processed trade, extensive import duty exemptions and widespread smuggling (World Bank, 1994; Bach et al., 1996). The nal two columns of Table 1 provide the nominal tariff rate and actual collected rate. The ratio of nominal tariff rate to actual collections varies dramatically across sectors, ranging from 140 for textiles and apparel to less than 2 for automobiles. In general, the more export-oriented sectors have a higher ratio because of tariff exemptions applied to their imports of intermediate inputs. China s tariff structure is typical of that of developing countries in providing high protection for manufacture, especially capitalintensive manufacturing and the manufacture of nal-consumption goods. But in aggregate, China s actual tariff rate is moderate. Automobiles are subject to the highest nominal tariff rate, over 50 per cent. The actual collected rate for ordinary automobile imports is slightly lower, at 45 per cent. The tariff rates for other manufactures and for textiles and apparel are also relatively high (30 per cent), but effective rates are much lower since such a large share of imports in these sectors are exempt from duties. 4. Base-case Projections and Simulation Design Because China s market accession commitments to the WTO will be phased in over a 5-year transition period, a baseline (non- WTO accession) from 1998 to 2005 is established using our recursive dynamic model. The baseline assumes that China would continue its grain self-suf ciency policy and that the import quota for agricultural goods would grow at 3 per cent annually from 2000 to 2005. This counterfactual benchmark serves as a basis for comparison with simulations conducted in the WTO accession experiments. Major results from the baseline calibration are summarised in Table 2. It uses economy-wide total factor productivity (TFP) as a residual, letting it adjust to be

2204 FAN ZHAI AND ZHI WANG Table 2. Summary of baseline calibration 1997 2000 2001 2002 2003 2004 2005 Exogenous speci ed variables GDP growth rate (percentage) 7.64 7.98 7.96 7.94 7.93 7.62 Life expectancy 70.3 71.2 71.52 71.84 72.16 72.48 72.8 Total fertility rate Urban 1.65 1.37 1.36 1.35 1.34 1.33 1.32 Rural 2.04 2.29 2.31 2.32 2.34 2.36 2.38 Calibrated results Macroeconomic trends Growth rate (percentage) Labour force 1.09 1.68 1.37 1.57 1.45 0.80 Capital stock 9.97 9.81 9.63 9.47 9.33 9.21 Total factor productivity 2.95 2.69 2.84 2.72 2.80 2.83 Fraction of GDP (percentage) Private consumption 46.8 49.0 49.5 49.9 50.3 50.6 50.8 Investment 32.9 34.5 34.5 34.6 34.6 34.6 34.7 Export 21.8 17.4 17.2 17.0 16.8 16.6 16.4 Import 16.4 15.1 15.0 15.0 15.0 14.9 14.9 Urban rural wage ratio 3.64 4.78 4.77 4.75 4.75 4.78 4.79 Urban rural per capital 2.96 3.44 3.36 3.30 3.23 3.17 3.13 income ratio

WTO ACCESSION, MIGRATION AND UNEMPLOYMENT 2205 Population and labour force Population (millions) 1236.3 1269.3 1279.7 1290.2 1300.8 1311.6 1322.7 Urban 369.9 396.3 405.4 414.4 423.4 432.6 442.0 Rural 866.4 873.0 874.3 875.7 877.4 879.0 880.8 Labour force (millions) 636.7 651.9 662.8 671.8 682.3 692.1 697.5 Urban 177.1 184.2 189.7 194.9 200.5 205.9 210.6 Rural 459.6 467.8 473.2 476.9 481.8 486.2 486.9 Grain self-suf ciency rate 98.8 98.4 98.0 97.6 97.4 97.0 97.0 (percentage) Urban unemployment rate 3.00 3.01 3.01 3.02 3.02 3.03 3.03 (percentage) Rural share of labour force 72.2 71.8 71.4 71.0 70.6 70.2 69.8 (percentage) Share of agricultural 77.1 75.2 74.8 74.3 73.9 73.4 72.9 employment in total rural labour force (percentage) Source: authors calculation.

2206 FAN ZHAI AND ZHI WANG consistent with the pre-speci ed values of life expectancy, urban and rural fertility, and GDP growth. We then consider three additional scenarios in reference to the baseline. All the assumptions for the baseline scenario and the three WTO accession scenarios are summarised in Table 3. The rst scenario considers China s WTO accession without labour market reform. In order to re ect trade liberalisation under WTO accession, import tariffs are reduced gradually and the sectoral reduction rates are aggregated from the Harmonised Commodity Description and Coding System (HS) tariff schedules for the period of 2000 05 in the China-US agreement and weighted by 1997 ordinary trade data. The growth rate of the import quota for petroleum re ning and automobiles will be accelerated during 2000 05 and the quantitative restriction will be eliminated in 2005. The tariff rate quota (TRQ) system is introduced for rice, wheat, corn, cotton, wool, vegetable oil and sugar. Moreover, the tariff for other agricultural goods will also be reduced. Quota elimination under the Multi-Fibre Agreement (MFA) is taken into account in this scenario as well. Once China becomes member of the WTO, its exports of textiles and apparel to North America and European markets will be subjected to accelerated MFA quota growth during 2001 04, similar to other developing countries that are WTO members. The remaining quota restriction will be eliminated in the year 2005 according to the Agreement on Textiles and Clothing (ATC). 4 It should be pointed out that China s WTO accession includes a complex package of trade and investment liberalisation. In this paper, however, only parts of the issue are captured. It does not take into account other major aspects of WTO membership, such as reduction of barriers in service trade and foreign investment, protection of intellectual property rights, securing market access, enforcement of commitment and co-operation in dispute settlement. The second scenario looks at the impact of reduced restrictions on labour movement. It assumes that the elasticity of labour mobility between rural and urban areas would be increased gradually from 0.25 to 0.5 during China s WTO accession process while the real wage rate of urban labour is xed. 5 The third scenario assumes WTO accession with a full reform in the labour market. The elasticity of labour mobility between rural and urban areas would increase as above and, in addition, the wage rigidity of urban labour is eliminated. Urban unemployment is xed at the baseline level and the urban wage is fully exible in this scenario. 5. Major Simulation Results 5.1 Aggregate Effects Table 4 reports the main ef ciency and other macroeconomic indicators under the three scenarios of China s WTO accession. They are deviations from baseline in the year of 2005. The results show that China can bene t from its WTO accession (scenario E1) in terms of real GDP and social welfare, the latter measured by Hicksian equivalent variations (EV). In 2005, China s real GDP is 1.1 per cent higher in the E1 scenario than in the baseline. The welfare gains are slightly smaller than this, 0.97 per cent of GDP, because of a 0.7 per cent deterioration in terms of trade. Given the closure rule which we have chosen, the trade balance is xed at the baseline level in foreign currency terms across different simulations, so deterioration in terms of trade would result in more exports and less imports, which in turn decrease the real domestic absorption and EV. Private consumption increases by 0.95 per cent, indicating the bene ts to consumers from WTO accession. Investment increases by 1.27 per cent. Exports and imports are 10.7 per cent and 10.9 per cent higher respectively, in the E1 scenario, indicating signi cant trade expansion. Many factors are interacting to determine these general equilibrium results. First, the large gains in GDP result from the enhanced ef ciency of resource allocation through in-

Experiment Description WTO ACCESSION, MIGRATION AND UNEMPLOYMENT 2207 Table 3. Summary of simulations design Base E1 Base case Real GDP and agricultural output exogenous Sectoral-speci c TFP growth rate endogenous Three per cent growth rate of import quota for goods subjected to quantitative restriction (rice, wheat, corn, cotton, wool, vegetable oil, sugar, petroleum re ning, automobiles) Exogenous export quota growth for textiles and apparel: textiles, 5.0 per cent; apparel, 6.2 per cent (annual average) All tax rates are xed at their base-year level Balance of payment gradually declines to 45 per cent of its base-year level in 2005 WTO accession without labour market reform Tariff reduction and quotas elimination on industrial products An average 55 per cent cut of 2000 tariff level during 2000 08, based on the nominal tariff schedule in China US agreement Phased elimination of import quotas on petroleum re ning and automobiles from 2000 05 Initial quota in 2000: petroleum re ning, 27.6 billion yuan; automobile, 496.8 billion yuan Annual growth rate of quota from 2000 05: petroleum re ning, 15 per cent; automobiles, 15 per cent Agricultural trade liberalisation Quota restrictions Initial quota in 2000 Annual growth rate of quota (10 billion yuan) (percentage) Rice 0.857 18.9 Wheat 1.158 7.2 Corn 0.325 12.5 Cotton 1.046 4.7 Wool 0.635 4.5 Vegetable oil 10.428 14.5 Sugar 1.523 8.0 Tariff cut for other agricultural goods, based on the nominal tariff schedule in China US agreement Phase out of MFA Acceleration of MFA quota growth rate from 2001 04 Zero export tax of textiles and apparel in 2005 E2 E3 WTO accession with high labour mobility E1 Elasticity of labour mobility is doubled gradually from 2001 to 2005 WTO accession with high labour mobility and exible urban labour market E2 Unemployment is xed at baseline level and urban wage is endogenous creased specialisation according to comparative advantage, including the migration of additional agricultural labour to manufacturing activities. The latter increases labour productivity. Secondly, there is more rapid physical capital accumulation and the result-

2208 FAN ZHAI AND ZHI WANG Table 4. The impact of China s WTO accession: major macroeconomic results, 2005 WTO accession with WTO accession with full labour market WTO accession high labour mobility reform (E1) (E2) (E3) Welfare change (equivalent 0.97 2 0.38 2.27 variation) as percentage of GDP Percentage change relative to baseline GDP 1.09 2 0.26 2.45 Consumption 0.95 2 0.10 1.28 Investment 1.27 2 0.94 4.31 Exports 10.73 10.02 13.32 Imports 10.94 10.20 13.56 2 2 Per capita income Urban households 1.11 5.57 4.34 Rural households 0.52 3.42 6.35 Terms of trade 2 0.73 2 0.69 2 0.90 Change relative to baseline Income share of rural households 2 0.003 0.001 0.012 Urban rural income ratio 0.018 2 0.009 2 0.077 Urban rural wage ratio 2 0.005 2 0.237 2 0.607 Absolute values Grain self-suf ciency rate 91.8 91.4 92.7 (percentage) Urban unemployment rate 2.2 8.7 2.9 (percentage) Source: simulation results. ing larger stock of physical capital further enhances productivity. Thirdly, removal of high protection rates induces real currency depreciation, enhancing the international competitiveness of China s exports. Fourthly, elimination of the MFA quota further increases the competitiveness of China s textiles and apparel, leading to export expansion of these sectors. Finally, total factor productivity grows more rapidly due to technology transfer via expansion of capital and intermediate goods imports from advanced industrial countries. As China expands its labour-intensive exports to world markets after its WTO entry, Chinese rms will import more capital- and technology-intensive goods as both investment and intermediate inputs from industrial counties. Those goods are usually embodied with advanced technology from abroad, thus stimulating productivity growth. What is the impact of such ef ciency gains on China s food security? The simulation results suggest that its grain imports would increase signi cantly, by more than 160 per cent relative to the baseline. Nevertheless, the grain self-suf ciency rate remains very high at 92 per cent. Thus China would still have low dependence on world grain markets at the point when its grain market would be opened in 2005. The urban rural income disparity would be increased by China s WTO accession under the assumption of low labour mobility. The results in Table 4 show that in 2005 the per capita income of urban households is 1.1 per cent higher in the E1 scenario than in the baseline, whereas that for rural households is only 0.5 per cent higher. Agricultural trade liberalisation would decrease the domestic price of agricultural products and decrease the return of production factors in the agri-

WTO ACCESSION, MIGRATION AND UNEMPLOYMENT 2209 cultural sector. Low labour mobility restricts the migration of rural labour; as a result, the gains of rural households resulting from expansion of manufacturing exports and urban employment are smaller than for urban households. However, when a high level of labour mobility is assumed (scenario E2), which implies a further relaxation of restrictions on rural urban migration, more rural labour will migrate to the urban areas looking for jobs. In this scenario, the rigid urban labour market cannot absorb all the unskilled labour available (laid-off workers in contracted sectors plus new rural immigrants). The urban unemployment rate thus increases from 2.2 per cent to 8.7 per cent, lowering urban household income by 5.6 per cent from the baseline. Because urban household income constitutes 60 per cent of China s total household income, and urban savings constitute 70 per cent of total household savings, the result is that the decline in urban household income will reduce national consumption and investment, thus lowering real GDP growth. This result has important policy implications. It indicates that rural urban migration is a crucial issue to realise the potential welfare gain from China s WTO accession. If the government does not handle the rural labour migration problem properly, it may result in very high unemployment in Chinese cities, thus having a signi cant and detrimental effect on the net welfare gains that China hopes to obtain from its WTO accession. Now consider the experiment of China s WTO accession with full labour market reform (scenario E3). Different from simulation E2, in this case, the higher level of rural labour mobility leads to a higher level of welfare gain. Both welfare gain and GDP growth rate are more than doubled comparing with simulation E1. Rural households are dramatically better off, their per capita income increasing by 6.5 per cent. The income of Chinese households is 1.21 per cent higher, thus producing a higher level of national consumption and investment, although the per capita income of Chinese urban households still declines by 4.3 per cent from the baseline. The underlying reason is quite straightforward. When the labour market is fully exible, the wage rate of urban unskilled labour will decline as the supply increases with immigration from rural areas. It will further increase the competitiveness of Chinese exports in the world market and is one of the major forces driving manufacturing production and export expansion in China under scenario E3. The lower urban wage will decrease urban households income (relative to the baseline) but will dramatically increase rural households income because urban incomes are still much higher than rural incomes and there is more migration in this scenario. Although the new urban wage for unskilled labour is lower than non-wto levels, it is substantially higher than the rural wage. Thus, rural migrants are much better off. The downward pressure on urban unskilled labour keeps Chinese exports competitive and creates additional jobs through raising exports. 5.2 The Effect on Rural Labour Migration and Urban Unemployment Given the importance of rural labour migration and urban unemployment to the actual realisation of the ef ciency gain from China s WTO accession, Table 5 presents more detail related to these phenomena. As the economy grows and industrialises, rural labour will continually move from farm sectors to non-farming activities, and migrate from rural to urban areas. WTO accession will accelerate this process. The simulation results show that agricultural employment will decrease and rural urban migration will increase in each of the three WTO accession scenarios relative to the baseline. Obviously, increased labour mobility will speed up China s industrialisation and urbanisation process signi cantly. As the results indicate, there are more rural labourers leaving agriculture and migrating to cities in scenarios E2 and E3 than in scenario E1. However, the simulation results also indicate that there is a possible con ict between removing restric-

2210 FAN ZHAI AND ZHI WANG Table 5. The impact of China s WTO accession: rural labour migration and urban unemployment, 2005 WTO accession WTO accession with full with high labour market WTO accession labour mobility reform Base (E1) (E2) (E3) (million persons) 2 2 2 2 2 2 2 2 2 2 2 2 2 2 (change from baseline, million persons) Urban unemployment 6.59 1.84 13.54 0.00 Labour in agriculture 354.76 2.76 11.99 11.39 Non-agricultural 132.15 1.94 1.50 2.63 employment in rural areas Migration to urban areas 30.65 0.81 13.24 8.57 (ratio) (change in ratio from baseline) Urban rural wage ratio 4.786 0.005 0.237 0.607 (percentage change relative to baseline) Urban unemployment 27.95 205.65 4.95 (urban labour wage) Labour in agriculture 0.78 3.38 3.21 Non-agricultural 1.47 1.14 1.99 employment in rural areas Migration to urban areas 2.64 43.20 27.96

WTO ACCESSION, MIGRATION AND UNEMPLOYMENT 2211 tions on rural urban labour mobility and reduced urban unemployment. Results from scenario E1 show that WTO accession will reduce urban unemployment as a result of more ef cient resource allocation and expansion of manufacturing exports to world markets if the controls on rural urban migration remain in place (the urban unemployment rate decreases by 28 per cent in 2005). However, scenario E2 indicates that without a full reform of the urban labour market, reduced control on migration may signi cantly worsen the employment situation in urban areas (urban unemployment would more than double in 2005). This result shows that there is a great need for co-ordination between China s rural urban migration policy, labour market reform and the implementation of trade liberalisation measures after China s WTO entry. The net bene t from WTO membership will be maximised if China adopts a policy gradually to relax its rural urban migration control in conjunction with its labour market reform. Dramatically to relax rural urban migration controls without corresponding reform measures in the urban labour market may create more problems than it solves, as our simulation results show (scenario E2). 5.3 Sectoral Adjustments A careful examination of the pattern of employment shifts among industries may help us to understand better such potential problems. Table 6 reports changes in gross output, employment and trade due to China s WTO accession under different labour market assumption in a 12-sector disaggregation. 6 Generally speaking, when restrictions on the labour market are reduced, the output response and shift of employment between sectors become stronger. There is even a sign switch in some sectors. There are two contracting sectors (grains and automobiles) and one expanding sector (textiles) regardless of assumptions on labour market conditions. This is consistent with China s factor endowment structure: abundant labour with scarce arable land and physical capital. Although the increase in agricultural imports is dramatic, ranging from 14 per cent (other agriculture, E1) to nearly 200 per cent (grains, E3), domestic production is only affected moderately. The output of the agricultural sectors falls from less than 1 per cent (other agriculture, E3) to 5 per cent (grains). This is because the amount of grain imports is very small relative to the size of production in the baseline. The signi cant increase in cotton imports (more than 50 per cent, all scenarios) is driven by the expansion of textiles and apparel due to the elimination of MFA quotas. Those sectors with high protection now, such as the automobile industry, would also experience a sharp increase in imports (about 150 per cent!). The lower import prices induce consumers to substitute imports for domestic products, resulting in dramatic decline in output and employment in those sectors. The removal of tariff and non-tariff barriers is only one factor that contributes to the signi cant surge in imports after joining the WTO. Export growth, caused by further realisation of China s comparative advantage in labour-intensive products, also contributes to the increase of imports. This is because the expansion of labour-intensive production drives up the demand for capital, especially technology-intensive equipment, as well as semi-processed products and intermediate inputs. As noted earlier, there is a large proportion of processing exports in China s total exports, which means that export growth is particularly important for China s import growth. Its effects are shown clearly from the sharp increase of imports of textiles and apparel, electronics, manufactured intermediates and other manufactured products. Processing imports account for more than 90 per cent of total imports in those industries. The contraction of agricultural production will divert agricultural labour and capital to non-agricultural sectors. Given the large share of agricultural employment in China, it is quite obvious that the most important adjustment cost will be related to the relocation of agricultural labour. It will bring challenges

2212 FAN ZHAI AND ZHI WANG Table 6. The impact of China s WTO accession: structural adjustment in production, employment and trade, 2005 Output Exports Imports Employment Tariff cut (percentage Base Base Base Base Tariff change (10 E1 E2 E3 (10 E1 E2 E3 (10 E1 E2 E3 (10 E1 E2 E3 rate from billion (percentage change billion (percentage change billion (percentage change billion (percentage change (percentage) baseline yuan) from baseline) yuan) from baseline) yuan) from baseline) yuan) from baseline) Grain 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 0.3 0.3 55.0 3.6 4.9 5.0 0.1 4.8 1.2 10.3 1.8 166.5 174.7 197.9 95.2 3.9 6.2 6.4 Cotton 3.6 0.0 9.6 0.2 1.4 0.8 0.0 9.0 2.6 6.4 1.0 50.4 52.6 63.0 16.6 0.0 2.6 2.0 Other agriculture 26.6 28.8 323.2 1.0 1.0 0.7 2.9 8.1 0.4 8.4 5.0 13.7 22.7 35.2 242.9 0.4 2.3 2.0 Processing food 15.6 18.4 227.1 0.8 0.4 0.1 8.9 11.7 7.8 3.7 8.4 20.1 21.2 27.7 8.9 0.8 0.8 0.7 Mining 1.9 1.2 102.1 0.5 1.0 3.0 3.8 4.1 3.4 5.9 13.9 1.1 2.5 1.7 19.3 0.7 1.3 2.4 Textiles and clothing 40.4 51.8 217.9 10.8 9.5 12.1 48.6 34.4 32.7 36.3 14.8 33.4 31.9 36.4 14.2 8.0 6.1 9.4 Other light industry 13.0 43.4 171.0 1.4 0.0 2.6 25.2 6.2 4.3 7.2 14.6 7.4 6.5 9.0 11.6 1.3 0.6 2.4 Manufactured 15.9 43.7 639.7 0.2 1.3 2.5 39.3 3.9 3.1 6.9 56.7 8.4 7.4 10.3 39.7 0.6 1.5 2.6 intermediates Automobiles 70.7 69.7 61.9 8.5 9.8 6.0 1.0 0.6 0.3 5.4 1.9 151.9 147.1 151.9 2.6 6.2 7.9 3.9 Electronics 15.2 75.9 99.8 2.0 2.1 1.3 32.9 1.8 3.5 6.2 31.7 8.4 8.6 11.3 3.0 1.4 2.5 1.9 Other capital-intensive 20.3 47.1 440.0 0.1 1.3 3.2 41.0 3.7 3.6 9.0 64.9 5.8 4.2 6.9 30.2 0.5 1.4 3.3 Service 0.1 0.7 1286.6 1.1 0.2 3.2 32.0 3.4 3.6 7.3 18.5 1.7 3.1 1.8 213.3 1.4 0.4 3.4 Total 16.8 47.4 3633.9 1.0 2 0.3 2.7 235.7 10.4 9.7 13.0 233.3 10.9 10.2 13.6 697.5 0.3 2 1.9 0.0 Source: simulation results.

WTO ACCESSION, MIGRATION AND UNEMPLOYMENT 2213 to both central and local governments to make such relocation possible. The dramatic expansion of the textiles and apparel sector will create new job opportunities, but its contribution to reducing unemployment is modest because its share of employment in the whole economy is small. Our simulations suggest that employment opportunities in sectors such as other agriculture, food processing, manufactures other than textiles, and services are closely related to conditions in the urban labour market. In scenario E2, where rural urban migration control is eliminated without reform in the urban labour market, much labour moves to urban areas but most sectors do not have the capacity to absorb it. This is a scenario that the Chinese government should avoid. The ideal situation is the combination of scenarios E1 and E3: actively implement reform measures in the urban labour market to keep the speed of liberalising rural urban migration consistent with the employment capacity building in urban areas. Simultaneous development of labour-intensive agricultural and non-agricultural production in rural areas would help to slow the rural urban migration ow. Such a set of policies will not only prevent a dramatic worsening of the urban unemployment problem, but will also increase the degree of exibility in the labour market. A more exible labour market would allow creation of more employment opportunities in both rural and urban areas to absorb rural unskilled labour shifted out from grain production due to trade liberalisation. 6. Conclusions and Limitations Some important lessons can be learned from our simulation results. First, labour market rigidity is an important factor that determines the impacts of trade liberalisation the degree of mobility of rural unskilled labour matters in general equilibrium analysis. Different assumptions on labour market rigidity affect the overall level of welfare gain as well as its distribution. Secondly, structural unemployment may rise following China s accession to the WTO as millions of farmers transfer to non-agricultural sectors. If the government does not handle rural urban migration properly, high urban unemployment could have signi cant and detrimental effects. Thirdly, there is a great need for coordination among migration policy, labour market reform and the implementation of trade liberalisation measures. Bene ts from WTO membership will be greatest if China adopts a policy that maintains some degree of rural urban migration control and actively implements reform measures in the urban labour market in order to maintain a balance between the urban labour force and urban employment capacity. Relative scarcity of arable land and physical capital, and abundant unskilled labour, are the basic national conditions in China. China s trade, production and employment patterns will be determined largely by its resource endowment. It should be a major consideration in making economic policy decisions, including labour and urbanisation policies. Our analysis illustrates how CGE models can be valuable tools for evaluating trade and labour market policies. However, several caveats should be mentioned. First, the model used is a stylised simpli cation of the Chinese economy and does not differentiate explicitly between skilled and unskilled labour. Secondly, the model does not take job-search and migration costs into account. Finally, the model is calibrated according to the standard method widely used in CGE modelling: many key parameters are based on only one year of data or on a literature search rather than on econometric estimations. Therefore, the results obtained must be interpreted with caution and may be best understood as indicative of the real effects. Notes 1. The combination of the Harris-Todaro labour market equilibrium condition and CET speci cation of imperfect rural urban labour migration re ects the double transition simultaneously going on in the Chinese economy: the transition from a backward rural economy to a modern industrialised econ-

2214 FAN ZHAI AND ZHI WANG omy (the HT model); and the transition from a planned economy to a market economy (the institutional constraints for rural labour migration in the labour market). 2. Details on the more disaggregate 53 sectors are available from the authors. 3. The other 22 per cent of total imports are investment goods imported by foreigninvested enterprises. 4. On 1 January 1995, the ATC entered into force and replaced the old Multi Fibre Arrangements (MFA). The ATC provides for the elimination of the quotas and the complete integration of textiles and apparel into the WTO regime over a 10-year transition period ending on 1 January 2005. All WTO countries are subject to ATC disciplines and only WTO members are eligible for ATC bene ts. 5. A back-of-the-envelope calculation was conducted to determine the value of elasticity of transformation in the labour market based on the data for three recent years. We use the urban/rural per capita income as a proxy of the urban/rural wage and assume the natural growth rates of the labour force in urban and rural areas to be the same. The calculation shows that the value of elasticity is between 0.12 and 0.34 during 1997 2000. A mid value of 0.25 is simply chosen in the baseline. 6. Results in 53-sector detail are available from the authors. References ARMINGTON, P. S. (1969) A theory of demand for products distinguished by place of production, IMF Staff Papers, 16, pp. 159 176. BACH, C., MARTIN, W. and STEVENS, J. (1996) China and the WTO: tariff offers, exemptions and welfare implications, Weltwirtschaftliches Archiv, 132(3), pp. 409 431. BEGHIN, J., DESSUS, S., ROLAND-HOLST, D. and MENSBRUGGHE, D. VAN DER (1994) Prototype CGE model for the trade and the environment programme: technical speci cation. Paris: OECD Development Centre. CAI, F., WANG, M. and WANG, D. (2001) The current situation of employment and its prospect: analysis of labor market in shaping, in: G. LIU ET AL. (Eds.) Analysis of Chinese Economic Prospect Spring Report of 2001, pp. 145 162. Beijing: Social Science Literature Press. DERVIS, K., DE MELO, J. and ROBINSON, S. (1982) General Equilibrium Models for Development Policy. Cambridge: Cambridge University Press. DEVELOPMENT RESEARCH CENTRE (1998), The global and domestic impact of China joining the World Trade Organization. Research Paper, Development Research Centre of the State Council, PRC, Beijing. GILBERT, J. and WAHL, T. (2001) Imperfect labor mobility, urban unemployment and agricultural trade reform in China. Paper presented at the World Bank Institutional Development Project Policy Seminar, January, Beijing. HARRIS, J. R. and TODARO, M. P. (1970) Migration, unemployment and development: a two-sector analysis, American Economic Review, 60, pp. 126 142. HERTEL, T. W. (Eds.) (1997) Global Trade Analysis: Modelling and Applications. Cambridge: Cambridge University Press. LESLIE, P. H. (1945) On the use of matrix in certain population mathematics, Biometrika, 33, pp. 183 212. MELO, J. DE (1988), Computable general equilibrium models for trade policy analysis in developing countries: a survey, Journal of Policy Modeling, 10(4), pp. 469 503. MELO, J. DE and TARR, D. (1992) A General Equilibrium Analysis of US Foreign Trade Policy. Cambridge, MA: The MIT Press. NAUGHTON, B. (1996) China s emergence and prospects as a trading nation, Brookings Paper on Economic Activity, 1996(2), pp. 273 344. SHOVEN, J. B. and WHALLEY, J. (1992) Applied General Equilibrium Analysis. Cambridge, MA: Cambridge University Press. WANG, Z. and ZHAI, F. (1998), Tariff reduction, tax replacement and implication for income distribution in China, Journal of Comparative Economics, 26, pp. 358 387. WORLD BANK (1994) China foreign trade reform. A World Bank Country Study, The World Bank, Washington, DC. WORLD BANK (1997a) Sharing rising incomes: disparities in China. A World Bank Country Study, The World Bank, Washington, DC. WORLD BANK (1997b) Old age security: pension reform in China. A World Bank Country Study, The World Bank, Washington, DC. ZHAI, F. and LI, S. (2000) The implications of accession to WTO on China s economy. Paper presented to the Third Annual Conference on Global Economic Analysis, Melbourne, Australia, June. Appendix. The Recursive Dynamic Chinese CGE Model 1. Model Dimension The model has 53 industries, 4 production factors and 2 representative households (urban and rural). Among the factors, labour and capital are used by all sectors, while land is used only by agricultural activities. labour is disaggregated by urban and

WTO ACCESSION, MIGRATION AND UNEMPLOYMENT 2215 rural. Within the 53 sectors, there are 10 agricultural sectors, 5 mining sectors, 29 manufacturing sectors, 1 utility sector and 8 services sectors. The detailed disaggregation of the agricultural sector makes it possible to model explicitly the quantitative restriction on agricultural products. 2. Production and Factor Markets The model assumes that there are two types of competitive rm ordinary rms and export processing rms that produce the same product in the same industry. The products of ordinary rms are assumed to be sold on the domestic market or to be exported to the rest of the world by a constant elasticity of transformation (CET) function, while products of the export processing rms are for exports only. All sectors are assumed to operate under constant returns to scale and cost optimisation. Production technology is represented by a nested constant elasticity of substitution (CES) functions. At the rst level, output results from two composite inputs: a composite of primary factors plus energy inputs and an aggregate non-energy intermediate input. At the second level, the split of the non-energy intermediate bundle into intermediate demand is assumed to follow the Leontief speci cation i.e. there is no substitution among non-energy intermediate inputs. At the same level, the value-added plus energy component is decomposed into an aggregate labour and energycapital bundle. At the third level, aggregate labour is further split into two types of labour force. And the energy-capital bundle is decomposed into an energy and capital land bundles. Finally, the energy bundle is made up of three types of base fuel component, and capital land is split into capital and land in the agricultural sector. The model distinguishes between old and new capital goods. This assumption of vintage capital allows the substitution elasticity in the production function to differ according to the vintage of capital. The model also includes adjustment rigidities in the capital market. It is assumed that new capital goods are homogeneous and that old capital goods are supplied through second-hand markets. The installed old capital in a sector can disinvest when this sector is in decline. The supply curve of old capital is a simple constant elasticity function of the relative rental rates. The higher the rental rates on old capital, the greater the supply of old capital. But the rental rate on old capital is not allowed to exceed the rental rate on new capital. Within each sector, the capital is fully mobile among ordinary rms and export processing rms. Each type of labour is assumed to be fully mobile across sectors and across the two types of rm. Rural labour is employed in agricultural sectors and non-agricultural sectors; urban labour is employed only in non-agricultural sectors. The model utilises a CET function to describe imperfect labour mobility among urban and rural areas. The movement of labour is determined by the relative expected wage and the constant elasticity of transformation. The expected wage is de ned as the product of wage and employment rate, one of which is endogenous. 3. Foreign Trade The rest of the world supplies imports and demands exports. Given China s small trade share in the world, import prices are exogenous in foreign currency (an in nite price elasticity). Exports are demanded according to constant-elasticity demand curves, the price elasticities of which are assumed to be 15. The ordinary rms allocate their output between exports and domestic sales to maximise pro ts, subject to imperfect transformation between the two alternatives. All the output of export processing rms is sold to the overseas market. We assume that exports by ordinary rms and export processing rms are heterogeneous. A CES aggregation function with relatively high substitution elasticities is employed to form the composite exports. In other words, we assume that buyers in the rest of the world choose a mix between the two types of export to minimise their cost. Three types of import are differentiated in the model. The rst one is ordinary imports, which are subjected to import tariff and non-tariff barriers. The second category is duty-free imports of raw materials and components used primarily for processing exports, but some may be transferred to the domestic market. The third category is duty-free imports of investment goods for foreign-invested and export processing enterprises. Products are assumed to be differentiated by region of origin i.e. the Armington assumption (Armington, 1969). A two-level nested CES aggregation function is speci ed for each Armington composite commodity. At the top level, agents choose an optimal combination of the aggregate domestic good and an imports aggregate, which is determined by a set of relative prices and the degree of substitutability. At the second level of the nest, the imports aggregate is further split into ordinary imports, duty-free imports of investment and the imports for the processing trade that are transferred into the domestic market. The split again is a function of the relative import prices and the degree of substitution across different import types. Note that the import prices differ by import type since the last two types are dutyfree. We establish the difference between domestic