Labour market distortions and the impacts of further trade liberalisation in China
Preliminary draft
Haiyan Wang Alan Matthews Department of Economics Trinity College Dublin Email address:
[email protected] 15th April 2008
1
I. Introduction In November 2001, the fourth WTO Ministerial Meeting in Doha, Qatar, launched a new round of trade negotiations. Because of its focus on the interests of the developing countries and development issues, the new round is also known as the Doha Development Agenda (DDA).This ambitious round aims to lower trade barriers in a wide range of areas including agriculture, manufacture and services, and in addition to strengthen the rules governing international trade. An important element is to rebalance trade rules in favour of developing countries and providing additional market access for their exports. Formally stalled since last June due to the wide gap in views on agricultural liberalisation among members, the Doha Round is now struggling to reach a deal by June 2007 before “fast track” authority expires in the United States. A successful Doha Round would have significant implications for China. China joined the WTO in 2001 and made significant concessions in reducing its tariff and non-tariff trade barriers on accession. The overall bound tariff level was reduced from 42% to 15% (Long, 2001). China also committed to reform its trade regime and domestic legislation. Encouraged by the greater openness and deeper integration with the world economy after WTO membership, China’s trade and exports continue to grow. Growth in exports accelerated to 29.6% per annum between 2002 and 2006, compared to 12.4% p.a. during the 5 years preceding accession (Calculation from MOFCOM). It is not difficult to predict that China will benefit further from the current Doha Round negotiations, through improved market access for its exports, especially in labour-intensive sectors. However, how these gains will be distributed in China has become a hot political topic, particularly as between rural and urban areas. The gap between rural and urban income in China is particularly high compared to other countries studied by Yang and Zhou (1996). They find that in most countries rural incomes are 66 per cent or more of urban incomes, while in China this ratio is less than 40 per cent. As we can see from Figure 1.1, the gap in net income between the rural and urban residents has grown at an increasing rate since the mid 1990s. In fact, such disparity has been the biggest contributor to the whole problem of equity in China followed by inter-regional disparity. China has had significant success in lifting people out of poverty in the last 25 years. “Rural poverty declined significantly from 2
250 million to 32 million persons over the 1978-2000 period.” (OECD 2005:157). However, tens of millions of Chinese farmers still live in poverty. The Chinese government is determined to protect farmers from being adversely affected by global market liberalisation. Figure 1.1: Urban to rural per capita income and living expenditures ratios (1978-2004)
Source: OECD edition of China Statistical Year Book (2005)
Further trade liberalisation also raises a challenge for China’s grain self-sufficiency policy. Despite its impressive economic performance, China’s agriculture remains relatively inefficient and unproductive. China’s WTO agricultural commitments are substantial. Apart from tariff rate reduction, a tariff-rate-quota system is applied to “sensitive” products such as wheat, rice, corn, cotton, vegetable oils, sugar, etc. to replace the previous planned quotas. Part of the quotas are allocated to non-STEs to make sure that the quotas are filled. The overall implications of further trade liberalisation depend crucially on whether those farmers predicted to be disadvantaged by further trade liberalisation are able to take advantage of the boom in other industries and whether these expanding industries can avail of the abundant cheap labour from rural China. Traditionally, however, the household registration (hukou) system has prevented rural people from migrating into the urban areas and helped to create a wide rural-urban income gap. Although there have been a series of positive reforms in labour market policies since the 1980s, mobility barriers still exist. Labour markets act as a mechanism for distributing the national welfare gain to individual
3
households and have drawn increasing attentions from researchers (Meng, 2000; Sicular and Zhao, 2004; Hertel and Zhai, 2004; Martin and Ianchovichina 2004)). Previous studies agree on the aggregate welfare gains for China from further trade liberalisation (Martin and Van der Mensbrugge, 2005; Polaski, 2006; Hertel and Keeney, 2005; Diao, Diaz, Bonnila, Robinson and Orden, 2005). This paper also assesses the effects of possible Doha outcomes for China in a computable general equilibrium (CGE) setting. Our particular contribution is to examine the impact of reform in China’s restrictions on rural-urban migration for poverty and the rural-urban income gap in China. In the next section, we provide a description of methodology used in this paper. Section III presents the initial GTAP analysis with perfect labour mobility assumption. Section IV estimates labour market distortions in China and Section V uses this estimation to conduct further CGE analysis. Section VI concludes.
II. Methodology Computable General Equilibrium models are widely used to quantify the economic effects of trade reforms. We begin our analysis by using the standard GTAP model Version 6 to examine the welfare implications of further trade liberalisation, modified to take into account China’s accession agreements in 2001. In particular, we modify the basic GTAP model to take account of the important Tariff Rate Quotas (TRQs) for imports of wheat, rice, sugar and plant-based fibre in China. We explicitly model the TRQ reform in China as a complementarity problem in GEMPACK. We use this modified model to perform two experiments: Agricultural trade liberalisation simulation (including TRQ expansion) and merchandise trade liberalisation simulation. However, this conventional model makes rather strong assumptions about factor markets skilled and unskilled labour is assumed perfectly mobile between sectors within a region. Although there has been a surge of rural-urban migration since the 1990s, the household registration (hukou) system still remains a restrictive barrier in China’s labour market. It creates difficulties for rural residents to get employment in urban areas. Even if they find jobs, they are not entitled to social welfare and other benefits which are provided only to
4
urban residents. They also face substantial institutional burdens introduced by urban local governments. To take account of this feature of China’s labour market, we propose a two-step approach in our modelling. In the first stage, we estimate rural labour supply equations for China and derive the labour mobility elasticity between rural and urban areas using household survey data. We follow the approach of Jacoby (1993), Meng (2000) and Sicular and Zhao (2004). First, the shadow wages or the marginal products of labour are obtained by estimating the agricultural household production function. Labour supply is then specified as a function of shadow wages, off-farm wages and other relevant variables. In addition to total labour supply, we also model labour allocation between family agricultural production and wage employment. We are able to derive the labour mobility elasticity based on these labour supply equations. The data used in this study are from a survey conducted by Rozelle et al on 1,200 rural households across six provinces in China in 2000. In the second stage, we incorporate this elasticity into the GTAP model and re-examine the impact of the Doha Round on China. The mobility of unskilled workers between agricultural and non-agricultural sectors is represented by a constant elasticity of transformation function. The unskilled workers make their decisions on labour distribution according to the wage ratios of these two sectors.
III. Initial GTAP analysis with perfect labour mobility within China 3.1 Background - The structure of Chinese economy As a large producer and consumer, changes in China’s economy can have a significant impact on the rest of the world. Table 3.1 below presents the structure of the Chinese economy following the update of the GTAP 6 baseline to include China’s WTO entry and EU enlargement. China consumes 20% of world rice, 15% of world cotton, 10% of world wheat and 10% of live animals and meat. China has been successful in achieving a high self sufficiency rate in these products, with very low trade flows with the rest of the world. Imports of oilseeds and sugar are large compared to domestic production, with self sufficiency rates of 55% and 81% respectively.
5
China has a comparative advantage in most manufacturing sectors, especially apparel, leather products and light manufacturing. China’s exports in apparel and leather products amount to more than 20% and nearly 30% of total world trade, respectively. On the other hand, the automobile sector and chemical & petroleum products have lower market shares. The last two columns of the table show the levels of trade protection applied to Chinese imports and exports. On average China’s exports face a much higher tariff (21.4%) than imports into China (8.3%). Due to the application of TRQs, the applied tariff rates faced by agricultural imports into China are extremely low, providing TRQ quotas are not binding. The tariffs on rice and wheat are only 0.3% and 0.4%, but tariffs imposed on China’s exports of rice and wheat are 142% and 15%. China’s protection on manufacturing products is slightly higher, but still only 9.2%. 3.2 Simulation Design In this section we use the Global Trade Analysis Project (GTAP) model and database version 6. GTAP is a comparative static, multi-regional, computable general equilibrium model of the world economy, with 87 regions and 57 sectors separately distinguished. The standard closure is applied here. This is full employment in each region. Prices of goods and factors are endogenous and they adjust until all markets are clear. Labour and capital are perfectly mobile between sectors, while land and natural resources are imperfectly mobile between sectors. All factors are region specific; there is no mobility between different regions. Regions and economies are linked through trade. The regional balance of trade is determined by the relationship of regional investment and savings. International capital mobility is determined by equalising rates of return across regions.
6
Table 3.1: Structure of the Chinese Economy in 2001 (2001 US$ Millions) Value of Sector Output Rice 43,993 Wheat 9,749 Oilseeds 5,331 Sugar 1,522 Plant based fibre 6,610 Live animals and meat 114,305 Other Agricultural Products 146,701 Other Primary Products 121,216 Processed Food 70,260 Beverage and Tobacco 48,175 Textile 146,833 Apparel 66,044 Leather Products 57,014 Light Manufacturing 93,222 Chemical & Petroleum Products 276,625 Automobiles and parts 42,400 Electronic machinery 392,808 Metals 215,131 Other Manufacturing 268,661 Services 977,142 Source: GTAP Database and own calculations, see text.
1
Value of Exports 676 55 596 34 100 3,219 4,767 5,577 6,221 990 26,496 39,463 32,969 16,336 28,588 2,942 138,560 19,123 58,917 22,698
Value of Imports 121 220 4,889 384 236 3,128 4,733 14,306 3,475 663 24,238 3,594 2,711 10,795 47,887 9,568 122,009 23,506 14,232 39,087
Consumption of Domestic Goods 43,316 9,694 4,735 1,488 6,510 111,085 141,934 115,639 64,039 47,186 121,930 30,469 24,046 76,886 227,527 95,736 70,220 164,974 39,442 1,013,455
Self Sufficiency is calculated as Production divided by the sum of Imports and Consumption of domestic goods
7
Self Sufficiency1 101% 98% 63% 81% 98% 100% 100% 93% 104% 101% 100% 191% 212% 106% 94% 88% 104% 98% 120% 99%
World Output
Total World Trade
227,527 95,736 70,220 164,974 39,442 1,013,455 1,073,291 1,172,163 1,461,072 648,279 764,076 505,724 211,237 1,833,626 3,450,514 1,672,998 4,311,770 2,321,570 1,921,360 35,560,416
12,606 19,986 20,184 13,442 8,853 89,492 149,887 401,808 188,148 60,458 227,121 183,828 102,174 299,084 869,346 560,190 1,933,959 466,145 544,821 1,247,406
China’s Applied Tariff Rates 0.3 0.4 10.9 15.1 2.4 10.1 15.0 1.5 12.8 18.5 10.1 15.7 8.6 3.0 8.2 13.7 5.4 4.5 9.8 0.0
Tariff Rates Imposed on China 142.8 14.7 34.1 43.6 1.5 11.2 29.8 3.2 14.8 33.0 13.8 18.5 15.6 8.2 7.2 9.2 6.4 8.7 10.8 0.0
To keep the sizes of the table of results reasonable, we aggregate the database to 20 regions and 20 sectors. The base year of data is 2001. Since then substantial trade reforms occurred, including China’s accession to the WTO in November 2001 and EU enlargement in 2004. These reforms are simulated as a pre-experiment, against which the Doha Development Agenda simulations are implemented. Unfortunately, there remains much uncertainty as to the outcome of the Round. Agricultural trade liberalisation Scenario For agriculture, the June/July 2006 Modalities do not include specific targets for the cuts in market access and domestic support. On market access, bound custom duties shall be reduced in equal annual instalments using a tiered formula such that those members with higher tariff levels will implement the largest cuts. The depth of the reductions, the thresholds for the tiers and the treatment of sensitive products have yet to be negotiated. In this paper we take the first scenario estimated by Jean, Laborde, and Martin (2005), in which a tiered formula is applied with transition points at 15 and 90 percent and marginal tariff cuts of 45, 70, and 75 percent for industrialized countries. The transition points for developing countries are placed at 20, 60 and 10 percent, and the marginal cuts at 35, 40, 50 and 60 per cent. Least developed countries are exempted from any commitments. Manufacturing trade liberalisation scenario The June/July Geneva Framework text on non-agricultural market access (NAMA) also contains few details. WTO members have agreed that improvements in non-agricultural market access in the Doha Round are to be based on a formula approach, with the precise details, and other issues such as the treatment of tariff escalation and tariff peaks, still the subject of negotiation. As a net exporter in manufacturing products, China will welcome the reduction in non-agricultural tariffs. In this study, linear cuts of 50 per cent and 34 per cent across all 11 manufacturing sectors are imposed for industrialized countries and developing countries respectively. Consistent with the agricultural simulation, least developed countries are exempted from further trade liberalization. 8
Tariff Rate Quotas Simulation Scenario In the WTO accession agreement, China committed to use tariff-rate quotas (TRQs) to replace the previous planned quotas for products considered to be “sensitive”. There are three possible scenarios under the TRQ regime. If the import volume is below the quota, the in-quota tariff is effective. The TRQ behaves just like a tariff and no rent accrues. If the import volume is above the quota, the out-of quota tariff is effective. Rents are collected by whoever holds the right to import at a lower in-quota tariff level. In this paper we assume that all the rents accrue to importers. If the import volume is at quota, the out-of-quota tariff is prohibitive. China agreed to increase quotas for wheat, corn and rice to 9.636 million, 7.2 million and 5.3 million tonnes. Quotas for cotton and sugar are raised to 894 000 tonnes and 1.945 million tonnes. The out-of-quota tariff was also reduced. Any further increase of quota volume and the out-of-quota tariff cut depends on the outcome of the agricultural negotiations in the Doha Round. Table 3.2 below presents China’s TRQ performance in 2002 and 2003. Since China joined the WTO, there has been a large increase in the value of imports of agri-food products. China has experienced an agricultural trade deficit in recent years. However, China’s TRQ system has generally been under-filled, especially for grain products. This is partly due to administrative problems in implementing the system. Particularly in 2002, the allocation of import quotas was delayed. Further data will be needed to properly assess China’s TRQ performance. In the simulation we increase the quota volumes for rice, wheat, sugar and plant-based fibres by 60, 20, 15 and 15 per cent respectively based on China’s WTO commitments. Table 3.2: China’s TRQ Performance Commodity Wheat Corn Rice Soybeans Palm oil Sugar Cotton
2002 Quota (000 tonnes) 8,468 5,850 3,990 2,518 2,400 1,764 818.5
Fill-rate (%) 7 0 6 35 71 67 22
Source: AMAD, OECD, 2004
9
2003 Quota (000 tonnes) 9,052 6,525 4,655 2,818 2,600 1,852 856.25
Fill-rate (%) 5 0 6 67 90 42 102
The three Doha simulation scenarios chosen and the pre-experiment are summarised in Box 1 below. Box 1: Summary of the Trade Liberalisation Scenarios Pre-experiment: China’s accession into the WTO and EU enlargement Agricultural liberalisation: cut the applied agricultural protection rates corresponding to 45, 70, or 75 per cent bound rate cuts for developed countries and 35, 40, 50, or 60 per cent cuts for developing countries. The transition points for developing countries are placed at 20, 60 and 10 percent, and the marginal cuts at 35, 40, 50 and 60 per cent. Least developed countries are exempted from any commitments. (CEPII Doha Scenario 12) Manufacturing liberalisation: applied import tariffs are reduced by 50, 34 and 0 per cent for developed, developing and lease developed countries receptively. Modelling TRQs: Increase tariff rate quotas for rice, wheat, sugar and plant-based fibre by 60, 20, 15 and 15 per cent respectively. 3.3 Results Analysis 3.3.1 Welfare effects In GTAP, welfare changes are measured using Equivalent Variation (EV), which is equal to the difference between the expenditure required to obtain the new (post-simulation) level of utility at initial prices and that available initially (Huff and Hertel, 2000). The welfare changes reported in GTAP arise principally from the reallocation of resources within an economy, which is called the allocative efficiency effect. When import tariffs are reduced or eliminated, the market can move closer to its competitive equilibrium and reduce the efficiency losses associated with any tax or subsidy. Welfare change may also result from terms of trade effects or other effects.
See the 10 Doha agricultural scenarios in Jean, Laborde and Martin (2005). This study takes their first scenario. 2
10
Welfare Changes from all three simulations The global welfare changes from the Doha trade liberalisation scenarios are presented in table 3.3. There is a welfare gain to the world economy, which is equivalent to 0.12 per cent of total world GDP. China’s welfare also increases as a result of further trade liberalisation, also by 0.12 per cent total GDP. All other countries, except the US and New EU members, gain from the Doha liberalisation scenarios simulated here as well. Table 3.3 Total Welfare Effects (Measured as Equivalent Variation in 2001 US$ Millions) WELFARE EU USA Japan China Hongkong ASEAN XASIA Taiwan Brazil Korea AUandNZ NewEU India SSA Canada ROW Total Source: GTAP simulation results
Total Welfare Change 4,003 -1,501 16,331 3,031 653.94 2,436 400.81 726.59 2,028 2,886 942.97 -252.39 867.84 281.64 -103.11 3,384 37,369
% of GDP 0.05 -0.01 0.39 0.27 0.4 0.45 0.19 0.26 0.4 0.67 0.23 -0.07 0.18 0.09 -0.01 0.17 0.12
The decomposition of the welfare changes for China is presented in Table 3.4. Allocative efficiency effects and terms of trade effects drive the positive welfare result. The former is due to the 34 per cent tariff reduction in manufacturing sectors, and the latter is due to a large demand in developed countries for manufacturing goods from China. The investment and savings price effect is a terms of trade effect for the capital account.
11
The small effect from agricultural liberalisation can be explained by the small tariff cut. Due to the application of TRQs, the applied rates of rice, wheat, sugar and plant-based fibre in China are in-quota tariffs, which are only 1% in most cases. There is no reduction in protection for these products under normal GTAP simulations. So it is important to explicitly model the quota expansion. When TRQs are incorporated into the simulation, there will be a further loss of 14 million US dollars from agricultural liberalisation. In both agricultural and industrial scenarios developed countries’ trade liberalisation has a much bigger impact on China’s welfare change. In the latter case China benefits by nearly 3 billion US dollars, which is slightly over 95% of the total gain from manufacturing. Table 3.4: Welfare Effects for China (Measured as Equivalent Variation in 2001 US$ Millions) Welfare Change US$ Million Total 3,031 Allocative Efficiency Effect 1,624 Terms of Trade Effect 1,774 Investment-Saving Price Effect -353 Agriculture -20 Developed countries liberalisation -18.36 Developing countries liberalisation -1.69 Manufacturing Developed countries liberalisation Developing countries liberalisation
3,065 2,919 146
TRQ Simulation Source: GTAP Model simulation results.
-14
3.3.2 Sectoral effects - output, exports and imports The changes at a sectoral level in China arising from the trade liberalisation simulation (no TRQ modelling) can be seen in Table 3. 5. The effects on agricultural sectors in China are mixed. There are small drops in production of wheat, oilseeds, sugar and other agricultural products, while the output of rice and plant based fibre increases. Corresponding to the output change, exports of rice and plant based fibre both increase and other exports decrease. Imports of all the agriculture sectors increase especially live animals and meat. There is a shift from grain to meat when income rises.
12
Overall, the changes in manufacturing sectors are greater than those in agricultural sectors, but there is also a great variation within manufacturing sectors. The net exporting sectors such as textile, apparel and leather products see an increase in both production and trade volume. However, sectors such as electronic machinery, automobile industry, chemical and petroleum products, metals, etc. will contract. Table 3.5 Resource Allocation Effects for China (Changes shown as 2001 US$ Millions) Sector Rice Wheat Oilseeds Sugar Plant based fibre Live animals and meat Other Agricultural Products Other Primary Products Processed Food Beverage and Tobacco Textile Apparel Leather Products Light Manufacturing Chemical & Petroleum Products Automobiles and parts Electronic machinery Metals Other Manufacturing Services Source: GTAP Model Simulation Results.
Output 840 -9 -33 -0.1 173 24 -150 -813 -50 47 3,931 5,004 3,540 -993 -3,992 -1,473 -5,777 -2,825 -576
Exports 609 -4 -8 3 3 -698 -234 -79 244 41 2,754 5,464 3,066 -258 742 -34 580 566 1,234
Imports 0.2 5 37 1 9 113 50 -66 318 37 2,892 548 404 375 2,710 791 4,557 1,207 1,360
674
-670
398
Table 3.6 below provides self-sufficiency rates in China after further trade liberalisation. There is not much change compared to the rates before the simulation (see table 3.1). China remains self-sufficient in most agricultural products, except for oilseeds and sugar. China continues to import a great amount of these two products.
13
Table 3.6 Food self-sufficiency in China after further trade liberalisation Product Rice Wheat Oilseeds Sugar Plant based fibre Live animals and meat Other Agricultural Products
Sufficiency rate 102.67% 98.25% 55.16% 81.39% 97.96% 99.39% 99.83%
Source: GTAP model Simulation results
3.3.3 Poverty effects - wages and prices Table 3.7 gives percentage changes in wages and domestic market prices in the combined simulations. In the standard GTAP model, a world average price of primary factors is set as the numeraire. All the other price changes are relative to this. From the table we can see that most prices of agricultural and manufacturing goods are increasing. The results indicate a positive poverty and equality effect from trade liberalisation. As the majority of the labour employed in agriculture is unskilled, a relative increase in unskilled labour wages means farmers are getting paid relatively more. Returns to unskilled labour rise by more than skilled labour, which means equality is also improving. Secondly, the rents farmers can get from the land increase by almost 1%. Table 3.7 Percentage changes in wages and prices Agricultural Sector Rice Wheat Oilseeds Sugar Plant based fibre Live animals and meat Other Agricultural Products Other Primary Products
% change Manufacturing Sector 0.95 0.79 0.9 0.85 0.98 0.91 0.92 0.26
Processed Food Beverage and Tobacco Textile Apparel Leather Products Light Manufacturing Chemical & Petroleum Products Automobiles and parts Electronic machinery Metals Other Manufacturing
Source: GTAP model Simulation results
14
% Change 0.81 0.77 0.63 0.64 0.69 0.64 0.53 0.53 0.51 0.57
0.63
Production Factor Unskilled Labor Skilled Labor Land Capital Natural resources
% Change 0.21 0.12 0.98 0.11 -2.63
We can also look at the consequences for agricultural value added (net farm incomes) for poverty impact analysis. Table 3.8 below summarises the percentage change in agriculture and nonagriculture’s value added in China. The net farm incomes from agricultural production increase by 0.53% and the value added for non-agriculture is increased by 0.63%. Overall there is a positive effect on poverty reduction. Table 3.8: Changes in sectoral value added from Doha Development Agenda Sector Agriculture Non-agriculture Source: GTAP model Simulation results
% change 0.53 0.63
IV. Estimation of labour mobility elasticity in China 4.1 Background: Labour mobility in China The analysis above is based on a strong assumption that labour is perfectly mobile within China. Substantial restrictions on migration have been a feature of China’s labour market for decades. The household registration (hukou) system was first introduced in the mid 1950s, when labour mobility between rural and urban areas was initially prohibited. Accompanying the take-off of the economy since the 1980s, the Chinese government has gradually relaxed legal barriers to allow for labour migration between agricultural and non-agricultural sectors and migration between regions. Since then, rural-urban migration has accelerated considerably. Between 1989 and 2000, rural migrants accounted for 13% of the total rural population in China and 20-30% in some areas in west China. In those fast growing regions (e.g. Guangdong), most unskilled positions are filled by rural migrants (Chinese Agricultural Policy Survey Report, 2004). Under the current regime, rural people can change their rural residency status in a limited number of ways, such as finding employment after leaving colleges and universities or the army, purchasing houses in cities on the commercial market, or obtaining urban residency due to confiscation of cultivated land. To accelerate the process of urbanisation, in 1997 the Chinese government began an experiment concerning 382 counties to allow small cities and towns to accommodate rural migrants. Rural workers who have lived and worked constantly in such cities or towns can apply for urban residency status. 15
However, by 2001 only 1.3 million rural residents had changed their residency status (MOA 2002, OECD 2005). The system favours those migrants endowed with human capital and financial resources. Most of the prospective rural migrants still face major obstacles in the urban areas. Without an urban permit, they find it difficult to get permanent positions. They are not entitled to most social welfare benefits and services. A survey found the biggest concern of rural migrants in Beijing is their children’s education (Cui and Pan 2002). They have to pay extra fees for their kids to go to local schools, which they are hardly able to afford. They don’t have access to social health services and housing services. They are also required to pay various administrative fees to local urban governments, such as a temporary residency fee, a family planning fee and an urban size expansion fee (Bai and Song, 2002: OECD, 2005). The barriers to labour mobility are further exacerbated by the land tenure system in China. Land is the most important asset and a key source of income for farm households. Land also acts as a sort of insurance policy in the sense that rural migrants face uncertainties and possible fluctuations in urban labour markets. However, under the current system rural residents run the risk of losing their entitlement to land by migrating permanently to urban areas. This risk might deter their decision to migrate. 4.2 Model Most of the rural households in China are self-employed farmers. If all markets work, and if family members work in both the agricultural and non-agricultural sectors, the exogenous market wage rate can be taken as the wage for agricultural self-employment. Labour supply equations can then be easily estimated using a sub-sample of households whose members work in both sectors. However, such a sample might be very unrepresentative in China. More importantly, markets might fail due to various reasons. Poor infrastructure or long distance from the market could easily make the transaction costs prohibitively high so that farmers find it difficult to supply a small amount of their labour to the market. Some goods market might be missing, for example, if the price margins between selling prices and purchasing prices are too wide.
16
When a market fails, the price of the good is no longer determined by market forces. In the case of labour, the price becomes the shadow wage of household workers. It is determined simultaneously when the household makes a decision on production and consumption. Using a household approach, Jacoby (1993) first develops a methodology to explicitly estimate the shadow wages of household workers from an agricultural production function. A farm household which consumes part of its production maximises utility subject to time, income and production constraints. At equilibrium, the household’s shadow price of labour equals the value of the marginal product of labour in household production. Labour supply, then, can be modelled as a function of the shadow wages, observed market wage rates, as well as other relevant variables. 4.3 Data and description of variables The data used in this study come from a survey of 1,200 rural households across six provinces in China in November and December 2000. We thank Professor Scott Rozelle for kindly providing the data to us. The survey team was led by Loren Brandt of the University of Toronto, Scott Rozelle of the University of California at Davis, and Linxiu Zhang of the Center for Chinese Agricultural Policy at the Chinese Academy of Sciences. The six provinces are: Hebei, Liaoning, Shanxi, Zhejiang, Hubei, Sichuan. The data provides detailed information on households’ demographic characteristics, farm and non-farm activities, as well as expenditure patterns and net income transfers. Table 4.1 summarizes the key characteristics of the households. The value of total agricultural production is used instead of the output level since there are multiple outputs for most of the households. It is calculated as the aggregate value of all crops, fruits and vegetables, forestry products, animals and livestock. The value of each product is computed using the selling price if there is market activity or the village level price otherwise. Land is measured as the land area used in production, including own land and rented land. Aggregate family labour hours are used on the strong assumption that male and female members of the household have the same productivity. The majority of the surveyed households (88%) don’t hire any workers. For fertilizer, pesticide and seeds the expenditure
17
level data are used given the variation in quality and price. The value of equipment is a proxy of capital input, mainly tractor, harvester, cow and horse, and thresher. Education might increase management and productivity of the household. It is used as an indicator of the potential productivity of the individual. The average years of schooling of household heads are about 6.5 years. Age is a measure of experience. It is expected to have a positive effect on output. Provincial dummies are also included as explanatory variables to account for different natural endowments and economic environments. Table 4.1: Description of the variables used in the estimation of the production function (RMB in 2000 prices for monetary variables) Variable Household size Value of output Land Labour hours Equipment Fertilizer Pesticide Seeds Head’s age Head’s education phone Enterprise Geographic Hebei Shanxi Liaoning Zhejiang Sichuan Hubei
Description Mean Number of household members 4.1 Value of all crops and livestock 3963.8 Land area in mu, owned or rented 8.5 Hours of family farm work 2862.2 Value of farm equipment 1018.1 Expenditures on fertilizer 493.9 Expenditures on pesticide 124.9 Expenditures on pesticide 124.4 Age of household head 45.2 Education level of household head 6.5 Dummy: 1 if phone present in village 0.9 Dummy: 1 if town and village enterprise present in 0.8 village Dummy: 1 if plain 0.4 Dummy: 1 if live in Hebei Province 0.2 Dummy: 1 if live in Shanxi Province 0.2 Dummy: 1 if live in Liaoning Province 0.2 Dummy: 1 if live in Zhejiang Province 0.2 Dummy: if live in Sichuan Province 0.2 Dummy: if live in Hubei Provine 0.2
Std. Dev. 1.3 4687.2 10.4 2517.4 1986.5 455.8 224.2 194.3 11.0 3.4 0.3 0.4 0.5 0.4 0.4 0.4 0.4 0.4 0.4
4.4 Estimation of the production function The Cobb-Douglas functional form is applied to estimate the agricultural production function, specified as follows: n
n
i =1
i =1
ln Yi = ∑ α i ln M i + ∑ β i Z i + ε i
18
Where Yi is the total value of output produced by the household, including crops, vegetables, fruits, hogs and animals. M i is a vector of production inputs, including labour hours, land, value of equipment, expenditure on fertilizer, seeds and pesticide. Z i is a vector of control variables, including years of schooling of household head, age of head and provincial dummies. The error term in the production function might contain unobservable variables such as managerial ability or anticipation of shocks which are correlated with the variable inputs. Instrumental variables can be used to deal with this sort of endogeneity or simultaneity issue. Examples of instruments for labour input are household composition variables, such as number of working age adults, number of elderly, number of children and their quadratic terms. Community level variables such as village size dummy, average distance to land, average village income and main crop price at village level are also used. Table.4.2 presents the OLS and Instrument Variable estimation of the Cobb-Douglas production function. The OLS results indicate that all the input variables used in the analysis are statistically significant. As a variable input, fertilizer contributes most to the output. Land input is the second contributor. Family labour hours have an elasticity of 0.156. The coefficients of seeds, pesticide and equipment are significant too. The education of household head has a statistically positive impact on output, while the age of head is not significant. Compared to Hebei, province dummies such as Liaoning, Zhejiang, Sichuan and Hubei all have a positive impact on output. The instrumental variable estimates are very similar to the OLS estimates. The Durbin-WuHausman test indicates that the OLS estimates are consistent. Therefore in the next section we use the OLS results to calculate and predict shadow wages.
19
Table 4.2: Estimation of Household Agricultural Production Function Dependent variable: log value of agricultural output Log family hours Log land area Log value of equipment Log expenditure on fertilizer Log expenditure on pesticide Log expenditure on seeds Head age Head age squared Head education level Phone present in village Geographic dummy (1 if plain, 0 otherwise) Shanxi Liaoning Zhejiang Sichuan Hubei Constant Observations R-squared Absolute value of t statistics in parentheses * significant at 5%; ** significant at 1%
20
OLS 0.156 (3.22)** 0.196 (4.44)** 0.039 (4.18)** 0.247 (6.04)** 0.067 (4.11)** 0.133 (6.55)** 0.031 (1.71) 0 (1.56) 0.026 (3.31)** 0.173 (2.07)* 0.243 (4.58)** 0.177 (1.92) 0.376 (3.74)** 0.342 (3.08)** 0.44 (4.77)** 0.181 (2.11)* 2.73 (5.67)** 989 0.52
IV 0.177 (1.45) 0.191 (3.50)** 0.037 (3.93)** 0.246 (5.32)** 0.064 (3.20)** 0.135 (6.62)** 0.028 (1.37) 0 (1.23) 0.027 (3.18)** 0.172 (2.11)* 0.255 (4.34)** 0.171 (1.71) 0.364 (3.14)** 0.366 (2.66)** 0.438 (4.18)** 0.178 (1.86) 2.639 (4.04)** 970 0.51
4.5 Estimation of the shadow wages The advantage of the Cobb-Douglas function is that it is easy to estimate and interpret. The coefficient of an input is the production elasticity of that input. The shadow wage rates of family labour hours can be easily calculated as: ∧ ∧
∧
Wi =
αi Y Fi ∧
∧
where Y is the predicted value of output derived from the estimated coefficient α i . Fi is the total hours of family labour. Households make their labour supply decisions between sectors based on their expected wage differentials. When the decision is made, we can only observe the wages in the actually selected sectors. We need to estimate the counterfactual wage rates for households who do not participate in labour markets and estimate the shadow wage rates for households who do not participate in agricultural production. We do this by regressing the estimated shadow wages and market wage rates on a series of exogenous variables as shown in tables below. The results of these regressions are used to predict shadow wages and market wage rates for all households. Table 4.3 shows the OLS estimates of the shadow wage equation for household agricultural production. Agricultural equipment, education level of household head, and town and village enterprises present in local area all help to improve agricultural labour productivity. Productivity is higher in Liaoning and Zhejiang provinces but lower in Sichuan and Hubei provinces. Table 4.4 reports the OLS estimates of the market wage equation. Individual level data are used in the regression. Female labour has lower wage rates. As expected, age and education are both significant and have a positive effect for wage rates. Hebei and Shanxi are negative and significant. After the regression, we project market wage rates and for all workers and then aggregate all the individuals to get household-level wage data. We also predict the shadow wages for all households.
21
Table 4.3: Shadow wage equations in household agricultural production (OLS estimation) Dependent variable: log shadow wage Land Equipment Number of adults Number of dependents Family non-labour income Male labour ratio Head age Head age squared Head education Phone present in village Enterprise present in village Village size Shanxi Liaoning Zhejiang Sichuan Hubei Constant Observations R-squared Absolute value of t statistics in parentheses * significant at 5%; ** significant at 1%
Coefficient 0.004 0.000 -0.085 0.156 0.000 -0.63 0.018 0.000 0.049 -0.129 0.702 0.057 -0.13 0.527 0.486 -0.386 -0.368 -1.96 989 0.38
t-statistic (1.47) (3.86)** (3.36)** (3.48)** (0.95) (4.39)** (0.87) (0.87) (5.41)** (1.42) (9.42)** (1.00) (1.47) (5.96)** (4.89)** (4.30)** (4.12)** (4.56)**
Table 4.4: Employment wage equation (OLS estimation) Dependent variable: log market wage rate Female Number of dependents Marriage Urban hukou Age Age squared Education Village size Enterprise present in village Phone Distance to town Hebei Shanxi Liaoning Sichuan Hubei Constant Observations R-squared Absolute value of t statistics in parentheses * significant at 5%; ** significant at 1%
Coefficient -0.146 0.033 0.011 -0.03 0.057 -0.001 0.046 -0.09 0.098 0.069 -0.05 -0.26 -0.335 -0.062 -0.15 0.03 -0.478 986 0.12
22
T-statistic (2.92)** (0.83) (0.42) (0.34) (3.22)** (3.01)** (5.95)** (1.64) (1.42) (0.8) (1.47) (3.05)** (3.72)** (0.71) (1.84) (0.37) (1.26)
The table below shows the comparison between observed market wage rates per hour and the estimated shadow wages for agricultural production and all the predicted wages. There is a big gap between the returns to labour in the agricultural and non-agricultural sectors, which reflect labour market imperfections in China. Due to the extreme values in observed market wage rates and estimated endogenous agricultural wages, on average the mean of the predicted values are lower than those observed. Table 4.5: Comparison of the shadow wage and market wage rate Unit: Yuan/hour in 2000 prices Variable Shadow wage for agricultural production Shadow wage for agricultural production, predicted Employed wage rate Employed wage rate, predicted
Obs Mean 989 0.883 1128 0.336 986 3.032 3442 2.323
Std. Dev. 6.987 0.198 2.534 0.629
Min Max 0.029 159.036 0.059 1.456 0.002 25.833 0.874 4.931
4.6 Estimation of the labour supply equations The labour supply functions of agricultural production and market wage employment are estimated separately as in Sichular and Zhao (2004). The dependent variable is the hours worked per working age adult. For explanatory variables we focus on log of agricultural shadow wage, market wage rate, non-labour income, a set of household characteristics, village level variables and regional dummies. The supply equations are specified as below:
log H A = α 0 + α A ln W A + α M ln WM + α I I + α X X + ε A log H M = β 0 + β A ln W A + β M ln WM + β I I + β X X + ε M Where WA is the estimated agricultural shadow wage rates, WM is the market wage rates. HA and HM are hours worked on farm and off farm respectively. I is non-labour income such as land rent and transfers. X is a vector of individual or household-specific characteristics. The labour supply error terms, which represent unobserved heterogeneity in preference for leisure, are correlated with shadow wages. Instruments used for the marginal products here are the same as in Section 4.3, including household composition variables and community
23
level variables. The IV estimation results are presented in the above table. The own price elasticity of agricultural work is positive and significant, which means a 1% increase in agricultural shadow wage will increase the agricultural labour supply by 2.176%. Market employment wage is not significant. Other income has a small negative but significant effect on farm hours. In the other equation, hours worked off farm are not sensitive to the market wage rate either. It is negatively related to the agricultural shadow wage, i.e., a 1% increase in shadow wage will decrease market supply by 1.364%. Not surprisingly, enterprises present in village are positive for market labour supply and negative for agricultural labour supply. Table 4.6 Estimation of labour supply equations
Log agricultural shadow wage Log employment wage Other income Number of dependents Head age Head age squared Head education Enterprise present in village Male labour ratio Shanxi Liaoning Zhejiang Sichuan Hubei Constant Observations R-squared Robust t statistics in parentheses * significant at 5%; ** significant at 1%
Log agricultural hours IV Standard error 2.176 (4.58)** 0.05 (0.15) -0 (2.14)* -0.562 (4.71)** -0.042 (1.67) 0.001 (1.9) -0.126 (4.31)** -1.855 (5.32)** 1.496 (3.59)** 0.573 (3.83)** -1.355 (4.73)** -1.641 (6.27)** 1.246 (5.27)** 1.181 (4.84)** 11.593 (12.48)** 1091 0.18
Log market employment hours IV Standard error -1.364 (4.12)** 0.152 (0.58) -0 (0.53) 0.191 (2.86)** 0.05 (2.57)* 0 (1.87) 0.062 (3.07)** 0.993 (4.08)** -1.018 (4.21)** -0.482 (3.96)** 0.725 (3.55)** 0.548 (3.11)** -0.578 (3.15)** -0.665 (3.44)** 3.264 (4.80)** 845 0.10
4.7 Estimation of the labour supply elasticity Following the procedure in Sichular and Zhao (2004), we calculate the labour mobility elasticity in response to wage ratios in agricultural and wage employment sectors. The following equations are derived from the above two labour supply equations: Log(HA /HM) = (α0 -β0) +(αA-βA) log(WA /WM )+ (αM-βM+αA-βA) log (WM )+… Log(HM /HA ) = (β0-α0 ) +( βM- αM) log(WM /WA)+ (βA-αA+βM-αM) log (WA )+…
24
(αA-βA) is the responsiveness of Log(HA /HM) relative to log(WA /WM ), which is equal to (2.716+1.364)=4.08. It implies that a one percent increase in the ratio of agricultural wage to employment wage rate raises the ratio of on farm hours by about 4.08%. Similarly, ( βM- αM) =0.1 is the responsiveness of Log(HM /HA ) to log(WM /WA), which means that a one percent increase in the ratio of market wage rate to agricultural shadow wage increase the market employment hours by 0.1%. Due to the presence of log (WM) in the first equation and log (WA) in the second equation, the two elasticities are quite different each other. We use the
elasticity of 4.08 for later analysis because it is significant.
V. Modified GTAP analysis with constrained labour mobility in China 5. l Modelling labour market mobility in GTAP Like all the other primary factors, the mobility of unskilled labour in GTAP is specified with a constant elasticity of transformation (CET).
L A / LW = (W A / WW ) − ETRAE where LA is the number of non-farm unskilled workers, LW is the number of farm unskilled workers, ETRAE is the elasticity of transformation and it is non-positive. In the standard GTAP model, unskilled labour is perfectly mobile between sectors. To reflect Chinese labour market distortions, here we set the value of ETRAE to be 4.08 as estimated above. 5.2 Simulation results 5.2.1 Welfare Changes Effects Table 5.1 reports the welfare changes from Doha trade liberalisation after modifying the model to incorporate labour market imperfections in China. It also allows us to compare the results with the initial experiment. Overall the welfare gain to the world economy is 10% smaller in the new scenario. For countries like EU and Japan, this is mostly due to a smaller allocative efficiency effect, which can be explained by unskilled labour not being able to move to non-agricultural sector in China. Similar to before, Japan, EU, ASEAN, Korea, Brazil are among the biggest winners from a successful Doha Round. Surprisingly, China’s welfare stays pretty much the same. 25
Table 5.1 Total Welfare Effects (Measured as Equivalent Variation in 2001 US$ Millions) Welfare with perfect Welfare with imperfect WELFARE labour mobility labour mobility EU 4,003 3,864 USA -1,501 -1,494 Japan 16,331 13,888 China 3,031 3,057 Hongkong 654 648 ASEAN 2,436 2,470 XASIA 401 388 Taiwan 727 732 Brazil 2,028 2039 Korea 2,886 2,883 AUandNZ 943 943 NewEU -252 -254 India 868 862 SSA 282 -910 Canada -103 -113 ROW 3,384 3,283 Total 37,369 33,510 Source: GTAP simulation results
26
5.2.2 Sectoral Effects- Output, exports and imports The sectoral level story is similar to the initial experiment, except that the magnitude of the effect is slightly smaller than in the previous case. Expansion in production and exports are found for some manufacturing sectors such textile, apparel and leather products, while other more capital intensive manufacturing industries see a big increase in imports and decrease in output. The changes in agricultural sectors are mostly negative and very small. Table 5.3 Resource Allocation Effects for China with imperfect labour mobility (2001 US$ Millions) Sector Rice Wheat Oilseeds Sugar Plant based fibre Live animals and meat Other Agricultural Products Other Primary Products Processed Food Beverage and Tobacco Textile Apparel Leather Products Light Manufacturing Chemical & Petroleum Products Automobiles and parts Electronic machinery Metals Other Manufacturing Services Source: GTAP Model Simulation Results.
Output 825 -21 -45 -2 170 -110 -246 -819 -118 35 3877 4986 3453 -969 -3958 -1459 -5494 -2679 -465 781
Exports 609 -4 -11 2 3 -728 -259 -89 220 40 2743 5451 3004 -252 751 -33 751 601 1301 -654
Imports -1 6 37 2 9 126 49 -52 323 37 2895 548 406 374 2708 788 4549 1202 1354 381
Table 5.4 Food self-sufficiency in China after simulated trade liberalisation with imperfect labour mobility Product Rice Wheat Oilseeds Sugar Plant based fibre Live animals and meat Other Agricultural Products
Sufficiency rate 102.7% 98.2% 56.0% 81.4% 97.9% 99.4% 99.8%
Source: GTAP model Simulation results
27
Compared to the pre-simulation numbers in table 3.1, there is not much change in food selfsufficiency rate after trade liberalisation. China stays self-sufficient in rice, wheat, cotton, live animals and meat and other agricultural products. Imports of oilseeds and sugar account for a great share of domestic consumption, but the self-sufficiency rate is not affected by further liberalisation. 5.2.3 Poverty effects - wages and prices In this section we look at the implications of labour market imperfection on the distribution of welfare changes following further trade liberalisation in China. We find further trade liberalisation would have a positive impact on poverty reduction. The wage rate of unskilled and skilled labour both increase by more than 1%. Table 5.5 Percentage changes in wages and prices Agricultural Sector
% change Manufacturing Sector
% Production Change Factor
Rice Wheat Oilseeds Sugar
1.19 0.91 1.10 1.11
Processed Food Beverage and Tobacco Textile Apparel
0.98 0.87 0.34 0.38
Plant based fibre Live animals and meat Other Agricultural Products Other Primary Products
1.02 1.28
Leather Products Light Manufacturing Chemical & Petroleum Products
0.60 0.56 0.25
Automobiles and parts Electronic machinery Metals Other Manufacturing
-0.08 0.18 0.41 0.54
1.24 0.04
Unskilled Labour Skilled Labour Land Capital Natural resources
% Change 1.40 1.27 1.82 1.21 -3.45
Source: GTAP model Simulation results Table 5.6 below provides the changes in value added after trade liberalisation. Value added is lower for agriculture while higher for non-agriculture than the previous experiment. This is due to imperfect labour mobility prevents unskilled labour to move to manufacturing sector so that value added for agriculture has decreased significantly.
28
Table 5.6: Changes in sectoral value added from Doha Development Agenda (%) Sector Agriculture Non-agriculture Source: GTAP model Simulation results
Initial experiment 0.53 0.63
Experiment after labour market modification 0.47 0.74
VI. Conclusion This paper examines the welfare effects of further trade liberalisation in China when account is taken of the institutional features of the labour market restricting full labour mobility. The elasticity of labour market mobility in China is estimated using a household survey sample. By comparing the results of an experiment where perfect labour mobility is assumed, we analyse how labour market distortions in China would impact the implications of further trade liberalisation. Special attention has been paid to the implications in a broad way for poverty and rural urban inequality. Our empirical analysis provides new estimates of agricultural shadow wages. There is a big gap between shadow wages and market employment wages. We find that labour allocation between agricultural and wage employment is not sensitive to changes in wages, but rather sensitive to changes in agricultural marginal productivity. This is in line with the findings in Sicular and Zhao (2004) in that labour is not pulled out of agriculture by higher market wages, but pushed out of agriculture by lower agricultural returns. This means that labour movement is constrained in China. The results also show that labour mobility is influenced by education, household characteristics, province dummies and the development of town and village enterprises. The CGE analysis shows that results from the two experiments- with perfect and imperfect labour mobility- are quite similar except that the magnitude of the changes in the second experiment is smaller. The analysis suggests that overall trade liberalization brings welfare gain to China and the world. Japan is the biggest winner. EU, ASEAN, Brazil and Korea are among the bigger winners as well. But, there are also potential losers from a partial Doha Round liberalization, like the USA and Eastern European countries.
29
Poverty and inequality are reduced in China according to the simulations. In both cases there are increases in agricultural prices which would generate more income for rural households. The return to unskilled labour increases more than skilled labour, which means the income inequality gap could be narrowed.
Although reform of China’s labour market allowing
greater rural-urban migration would enhance the gains to farmers from further trade liberalisation, we nevertheless find that farmers benefit even with the current institutional restrictions in place.
30
References: Anderson, K 2004, “Agricultural trade reform and poverty reduction in developing countries” The World Bank Policy Research Working Paper No. 3396 Anderson, K, Huang, J & Ianchovichina, E 2004, “The Impact of WTO Accession on Chinese Agriculture and Rural Poverty”, in China and the WTO, The World Bank and Oxford University Press Anderson, K & Velenzuela, K 2006, “Do global trade distortions still harm developing countries”, The World Bank Policy Research Working Paper No. 3901 Bouet, A 2006, “What can the poor expect from trade liberalization? –Opening the black box of trade modeling”, MTIC Discussion Paper No. 93, March 2006 Brockmeier, M 2001, “A Graphical Exposition of the GTAP Model”. GTAP Technical Paper No. 8, https://www.gtap.agecon.purdue.edu/ Chen, S & Ravallion, M 2004, “Welfare impacts of China’s accession to the World Trade Organization”. The World Bank Economic Review, Vol. 18, No. 1 Diao, X, Diaz-Bonilla, E, Robinson, S & Orden, D 2005, “Tell me where it hurts, and I will tell you who to call: industrialized countries agricultural policies and developing countries”, MTIC Discussion Paper No. 84, April 2005 Eisenburger and Patel (2003): “Agricultural Liberalization in China: Curbing the State and Creating Cheap Labor”. Food First Policy Brief No. 9 “GTAP 13th Annual Short Course”. 2004 Hertel, T 1997, Global Trade Analysis- Modelling and Applications. Cambridge University Press Hertel, T& Keeney, R 2005, “What is at stake? The relative importance of import barriers, export subsidies and domestic support”, in “Agricultural Trade Reform & the Doha Development Agenda Book Chapters”, Anderson K and Martin W, the World Bank and Palgrave Macmillan Hertel, T & Reimer, J 2004, “Predicting the Poverty Impacts of Trade Reform”. The World Bank Policy Research Working Paper 3444 Hertel, T& Winters, A 2005, “Putting Development Back into the Doha Agenda: Poverty Impacts of a WTO Agreement”, New York: Palgrave MacMillan Hertel, T & Zhai, F 2004, “Labor Market Distortions, Rural-Urban Inequality and the Opening of China’s Economy” The World Bank Policy Research Working Paper No. 3455 Huang, J, Li, N & Rozelle, S 2003, “Trade reform, household effects, and poverty in rural China”, American Journal of Agriculture, Volume 85, Number 5, December 2003, pp. 1292-1298(7) 31
Huang, J, Rozelle, S & Chang, M 2004, “Tracking distortions in agriculture: China and its accession to the World Trade Organization”. The World Bank Economic Review, Vol. 18, No. 1 Ianchovichina, E & Martin, W 2004, “Impacts of China’s Accession to the World Trade Organization”. The World Bank Economic Review, Vol. 18, No. 1 Kuiper, M & Tongeren, F 2004, “Growing together or growing apart?-A village level study of the impact of the Doha round on rural China”. The 7th Annual Conference on Global Economic Analysis, Trade, Poverty and the Environment Li, S & Zhai, F 2000, “The impacts of accession to WTO on China’s Economy”, The 3rd Annual Conference on Global Economics Lejour (2001). “China and the WTO: The Impact on China and the World Economy”. Greater China and WTO International Conference Matthews, A 2001. “The possible impact of China’s WTO membership on the WTO agricultural negotiations”. Trinity Economic Papers, No. 15, 2001 OECD. 2002. “Agricultural policies in China after WTO accession.” Polaski, S 2006, “Winners and Losers: Impact of the Doha Round on Developing Countries”, Carnegie Endowment Report, March 2006 Roberts, I & Andrews, N 2005, “Development in Chinese Agriculture” ABARE eReport Sicular, T & Zhao, Y 2004, “Earnings and Labor Mobility in Rural China: Implications for China’s Accession to the WTO”, in “China and the WTO”, World bank and Oxford University Press Whalley and Zhang (2003). “Inequality Change in China and (Hukou) Labor Mobility Restrictions”. NEBR Working Paper Series No. 10683 WTO documentation on the Doha Development Agenda, www.wto.org Yu, W & Frandsen, S 2002, “China’s WTO commitments in agriculture: Does the impact depend on OECD agricultural policies?” The 5th GTAP Conference. Zhang, X 2006, “Armington elasticities and terms of trade effects in global CGE models”, Staff working paper, Australian Government Productivity Commission, January 2006
32
