Thursday, 20 March 2003. Guangzhou.
Sun Zhigang is twenty-seven. He is a graduate of the Wuhan University of Science and Technology, from Huanggang in Hubei, and he works for a garment company, Daqi, in Guangzhou. He is in the medical clinic of a detention centre. Police had held him because he had no temporary living permit and no identity card. His household registration, his hukou户口, is with his family in Hubei.
Today he dies there. The clinic’s account is a heart attack or a stroke.
An autopsy at Sun Yat-sen University later found the marks of a savage beating about seventy-two hours before his death. The Southern Metropolis Daily published the case on 25 April 2003. The Congressional-Executive Commission on China summarised the case in one line: Sun was mistakenly detained under the custody and repatriation regulations and beaten to death in custody. Legal scholars wrote to the National People’s Congress to say the regulations were unconstitutional. On 18 June the State Council repealed them and put a set of aid measures for the destitute in their place. Forced detention was outlawed, and shelter passed from the Ministry of Public Security to the Ministry of Civil Affairs.
A death had ended a system that held people for lacking the right paper, in under three months. What it did not end was the paper. The registration system stayed. Two years earlier a State Council circular had told towns of fewer than 100,000 people to grant registration to residents with fixed jobs and homes, which admitted people already settled in small places and not the migrants building the coastal factories.
The migrants were owed more than a permit. Official sources suggested that over 70 percent of them were owed back wages at the end of 2003, about 100 billion yuan. In October Premier Wen Jiabao took up one such dispute personally, raised by a Sichuan farmer, and from 22 January 2004 Beijing required wages to be paid within 30 days. A directive of 5 January 2003 had already given migrants a legal right to work in cities. The state was beginning to treat the migrant as a constituency while the registration system still treated her as a guest.
That is the argument of this chapter, and it is about two different things a state can build. One is a floor: a minimum under people who stay where they are. The other is a ladder: the means to leave. Between 2003 and 2009 China rebuilt insurance and dropped the farm tax in its villages, and kept the 1958 registration rule. India put a legal floor under the village; no source here describes a matching mobility policy.
Sun’s story reached print during another one. On 4 April 2003 a military doctor, Jiang Yanyong, chief physician at Beijing’s 301 Hospital, emailed an 800-word letter to two television stations saying the government was drastically understating the number of SARS cases. Neither station aired it, and the information leaked to Western news organisations. On Sunday 20 April, China announced 339 Beijing cases it had not disclosed before, and by the next day the health minister and the mayor of Beijing had been removed from their Party posts. Jiang is one of the book’s warners: a person who says the number is wrong while it is still safe to say nothing. The Warner Rule has its chapters; this is not the last.
The same year China began to rebuild the village insurance that had gone with the communes; chapter 2 told the 2003 pilot.
Chain C041984 → 2003Pays off: the village insurance is rebuilt19472026This time it was paid for out of the state budget, not out of a collective that no longer existed. At the end of 2005 the state also stopped taxing farmers. On 29 December the Standing Committee of the National People’s Congress annulled the Agricultural Tax Regulations of 3 June 1958, with effect from 1 January 2006. China’s village floor was being rebuilt: insurance, then no tax.
On 12 April 2005 India launched the National Rural Health Mission, with an initial outlay of Rs 6,731 crore for 2005–06. Its parts were training health volunteers, more medicines, and stronger primary and community health centres. The volunteers were the ASHAs, on the Chhattisgarh Mitanin model that chapter 2 told.
Chain X051965 → 2005Pays off: the village health worker, in a shape that lasts19472026The Mission came with targets, and the targets were public: by 2012 it wanted infant mortality down to 30 per 1,000 live births, maternal mortality to 100 per 100,000, and the total fertility rate to 2.1. The one scheme under it whose growth can be read straight off an official table is the payment for hospital births. Janani Suraksha Yojana beneficiaries rose from 7.38 lakh in 2005–06 to 104.38 lakh in 2014–15, and about nine lakh ASHAs were by then drawing its incentives. That is the shape of the scheme: a simple task, paid per task, done at scale.
Five months later came the larger thing. The National Rural Employment Guarantee Act was notified on 7 September 2005. It obliged every state to make a scheme giving not less than 100 days of guaranteed employment a year to every rural household whose adult members volunteer for unskilled manual work. The scheme was launched on 2 February 2006 in 200 districts. It reached 330 districts in 2007–08 and every rural district in 2008–09. It did not start from nothing: two older programmes, the Sampoorna Grameen Rozgar Yojana and the National Food for Work Programme, were folded into it in the first 200 districts, and the food-for-work allocation was raised that year from Rs 4,020 crore to Rs 11,000 crore. The Survey described the Act’s purpose as livelihood security through works that attack the causes of chronic poverty, such as drought, deforestation and soil erosion.
Look at the grammar of that sentence in the Survey’s account of the Act. The household volunteers. The first move is the household’s, and the obligation is the state’s.
Its scale is easy to state. In the budget of 6 July 2009 the finance minister reported work for 4.47 crore households in 2008–09, against 3.39 crore the year before, and allocated Rs 39,100 crore for 2009–10, 144 percent above the previous year’s estimate. The same speech promised a real wage of Rs 100 a day.
So by 2009 there were two floors. China’s was built on the budget, with taxes lifted and insurance rebuilt, by decision from the centre. India’s was built on a legal right a household could claim. Neither is a ladder. What decided how each country treated a young person who wanted to leave the village was a different set of rules.
In China the question was the factory. On 29 June 2007 the Standing Committee passed the Labour Contract Law, which took effect on 1 January 2008, after eleven years of drafting and shelving. Its month of public consultation brought in close to 200,000 suggestions, and about 70 percent came from ordinary workers. A survey by the Standing Committee at the end of 2005 had found that few workers had a written contract, and that most contracts ran for under a year.
The law’s answer was blunt. An employer who went more than a month without a written contract owed double wages. After a year without one, the worker was treated as having an open-ended contract. As a general principle, a fixed-term contract could not be ended without fault, and when it ran out the employer generally owed a month’s pay for each year worked, which the article states is capped at twelve months.
The foreign business lobbies objected. On the last day of consultation the EU Chamber of Commerce wrote that the law would limit employers’ flexibility and raise production costs, and that higher costs would force foreign firms to reconsider new investment or whether to stay in China. The American Chamber of Commerce in Shanghai said it feared a negative effect on the investment climate. A multinationals’ HR association in Shanghai, representing 54 firms and 260,000 staff, said some firms might withdraw.
The All-China Federation of Trade Unions answered that the law tilted towards workers but did not over-protect them, because it had weighed the lawful interests of employers, foreign investors included. It said over 80 percent of labour disputes arose from infringed workers’ rights, and that about 90 percent of cases that went to arbitration or court were won by the worker. The percentages are the federation’s, and the firm counts the lobby’s; neither is an audit. The law also dealt with two other abuses the union federation named. Labour-dispatch agencies, which had been used to avoid direct contracts, were made employers, bound to contracts of at least two years and jointly liable with the firms they supplied. A worker owed back wages could apply to the local court for a payment order. What the episode shows is narrower: by 2007 the Chinese state was prepared to write a floor into the factory contract, over loud foreign objection, for the migrant already inside the factory.
Then, from 2004, the coast began to report that it could not find workers. The IMF says China’s labour markets first came into focus in 2004, when shortages of migrant labour and rising wages were reported in the export regions. Economists read this two ways. Garnaut and Cai and Wang saw surplus rural labour running out. Chan, and Zhang, Yang and Wang, saw markets segmented by the registration system, with surplus labour still waiting inland. The book does not settle that argument. It notes only that on the second reading the registration rule is part of the shortage.
But a contract protects the person who is inside the gate. It does nothing for the one on the wrong side of it.
The registration system that left Sun without a permit began with a 1958 rule. On 9 January 1958 the Standing Committee adopted the Regulations on Household Registration. Article 10 said that a citizen moving from the countryside to a city must hold an urban labour department’s hiring certificate, a school’s admission certificate, or a permit from the urban registry.
In 1960 the rule met a famine. In Xinyang, in Henan, according to a memoir quoted by the journalist Jasper Becker, the prefecture declared a 1959 harvest of 3.6 million tons when it was nearer one million. Grain was confiscated, militia manned roadblocks so that no one could flee, and about one person in eight died. The registration rule is only part of that account, and Becker’s is a single press retelling of a local official’s memory. It is enough to show the rule’s first purpose, which was to keep people where the state could reach them.
On 3 June 1958 the same legislature had passed the Agricultural Tax Regulations. In December 2005 it annulled those. It did not annul the registration rule that had been adopted five months earlier. Chapter 4 told what that rule did to the dagongmei, the factory worker from the villages, and her children; this chapter does not tell it again.
The question India’s floor posed was the opposite one. If a village is guaranteed a hundred days of paid work, does it still send its young to the city?
The World Bank thought it might not. Its World Development Report 2009 listed the employment guarantee among place-based programmes that can act as barriers to internal mobility.
The measurement came later, and it used the way the programme spread. Because districts joined in phases, Clément Imbert and John Papp could compare districts that had the guarantee with those that did not. They found that it reduced short-term, seasonal migration to cities by 9 percent and raised the wages for manual short-term work in cities by 6 percent. The implied elasticity of unskilled wages with respect to short-term migration was high, -0.7. Both numbers are true at once. The floor kept some people home, and by keeping them home it raised pay for the casual labourers who did go.
2017Muralidharan, Niehaus and Sukhtankar studied what happened when the guarantee was run better. In Andhra Pradesh a reform that improved how it was run was rolled out at random across sub-districts. Beneficiary households’ earnings rose by 14 percent and poverty fell by 26 percent, and 86 percent of the income gain came from earnings outside the programme. A well-run floor lifted private wages around it, too.
That is the evidence for the floor. The case against it is not in these studies, and the chain panel below says what the studies do not show.
At the end of 2023, 66.16 percent of China’s people lived in cities and towns, a count that includes those without an urban hukou; chapter 9 set the registered share beside it. Ministry of Public Security figures cited in the same report put the rural migrants who obtained an urban hukou from 2021 to 2023 at over 40 million. The category of resident without urban registration was still large enough for the state to name it.
2026On 1 July 2026 the Centre brought in a new law, VB-G RAM G, in place of the twenty-year-old guarantee. It promises up to 125 days of work a year to an eligible rural household, up from 100. Costs are shared between the Centre and the states 60:40, and 90:10 for the north-eastern and Himalayan states. States may suspend work for up to 60 days a year in peak farm seasons.
DT Next reports the pause is meant to keep farm labour available in peak seasons. The new cost split is the point of dispute: Madhya Pradesh and Bihar questioned the added burden, and Tamil Nadu estimates it at over Rs 5,000 crore a year; several other states passed resolutions opposing the Act. The book takes no side in that dispute. Whether a poorer state can afford to hold up its share of the floor is not yet measured.
Put the two countries beside each other in 2009. In China a migrant was hired under a contract the state had strengthened, with a village behind her that had insurance and no farm tax; she generally could not register in the city she worked in. In India a villager held a card that entitled her to a hundred days of work at home.
Neither country’s measures here were built as a ladder for the person at the bottom. China kept its registration rule. India’s floor measurably cut seasonal moves. Which trade is better depends on whether there are factory jobs waiting at the top.
A floor holds people on the land. The factory that might have given them a ladder needed the land itself, and in 2006 the government of West Bengal took 997 acres of it at Singur for a car plant.