According to The Economist:
Manual scavenging was banned in 1993 by a law that also forbade the unplumbed toilets that necessitate it. But implementation has been slow. So several hundred thousand scavengers are still at work.
Most "manual scavengers" work in big cities where alternative toilets are available. The solution to this social problem, involving cheap, ecologically sound latrines that are better for public health and a retraining program, is also quinessentially Indian, in the tradition of other Indian innovations like microlending:
Sulabh, a charity, has developed an ingeniously simple one, costing around $100, that empties into one pit, and then, when it is full, another. It flushes with two litres of water, compared with the ten litres required by a standard cistern toilet. It takes ten people two years to fill one pit, by which time the waste in the other has turned into composted manure, clean enough for growing vegetables.
Sulabh has built 1.2m of the latrines across India and helped 60,000 scavengers find new work. “The toilet is a tool of social change,” declares Bindeshwar Pathak, a (high-caste) brahmin who started the charity in 1970 and has developed a passion for lavatorial technology.
I've previously noted at this blog that latrines can be best from a public health perspective in the developing world, where sewage from flush toilets is often insufficiently treated before it is dumped in waterways that are used directly for drinking water and household washing.
Should Sulabh's training programs fail, India's guaranteed employment system, one of the countries core welfare programs, can provide a backstop. The system was a key Congress party coalition platform item in 2004, and this year the “National Rural Employment Guarantee Scheme” (NREGS) went from being a pilot program in 330 districts to a nationwide program key to the Congress party coalition's May 2009 general election campaign. Last year the program "provided 30m families with an average of 43 days' work." The government estimates that 260 million of its citizens are below the poverty line.
Wherever it applies, Indians may demand employment as their right. If it is not provided within 15 days, they are entitled to receive unemployment benefit. . . . To exorcise ghost workers, NREGS muster rolls are read aloud at work-sites each morning. Many of these lists are available online. To chase away dodgy contractors—a main beneficiary of some works projects—at least 60% of the funds devoted to NREGS must be spent on wages for manual labourers. . . . these poor folk are self-selecting: only a genuinely needy man would be likely to labour under an Indian sun for 60 rupees ($1.50) a day, the national minimum wage stipulated under NREGS. . . . In Rajasthan, at least, it is working well. With its history of public works, and with fairly well-established local governments to run the scheme, the state last year came closer to honouring its prescriptions than any other. In the fiscal year that ended this March, Rajasthan provided an average of 85 days' work to some 2m households, a threefold increase in public-works employment offered by the state . . . [and] NREGS employees came mostly from India's poorest groups: including 19% from dalit (formerly “untouchable”) castes, and 46% from tribal groups. Nearly 70% were women. Earning 73 rupees for a day's labour—the minimum wage set by the state—they did better than their sisters working in agriculture. In 2005 the average agricultural wage for women in Rajasthan was 48 rupees a day.
Many seasonal workers use the program to survive economically between busy seasons. Competition from NREGS has also allegedly driven up local wages in other manual work sectors like farming, although this may also be attributable to worldwide increases in food prices and a booming Indian economy.
The manual work dominanted public works projects themselves don't appear to be terribly important to the economy and suffer from a lack of expert engineering supervision, and like all government programs in India, there is inevitable corruption, although it is modest by Indian standards. "According to a recent World Bank simulation, more Indian peasants would be withdrawn from poverty if the government just handed them cash—without first making them shovel dirt." But, handing out money has it own problems: "By Congress's own rule of thumb—attributed to Rajiv Gandhi, the party's murdered leader, in the 1980s and since parroted by activist and politician alike—85% of welfare spending fails to reach its intended recipients."