Document(s) 7 de 16

Introduction

This chapter examines two approaches to promoting employment and social security for the labor force in India: (i) public works programs; and (ii) initiatives to provide social security for unorganized workers. A common objective of both types of interventions is consumption smoothing for unorganized-sector workers. While the modes of worker protection differ, the common theme is exploring the potential for public interventions to address market failures which may contribute directly and indirectly to poor welfare and labor-market outcomes in India.

The structure of the chapter is: the next section looks at public-works schemes in India. It first outlines the objectives of public works, before exploring spending trends and employment and other outcomes. A discussion of factors driving outcomes follows, before a review of the National Rural Employment Guarantee Act (NREG) and conclusions. The following section focuses on social security for unorganized workers. It first outlines international experience in expansion of social protection. A review of recent initiatives in India follows. This is followed by review of evidence on membership of intermediary organizations of unorganized workers. Conclusions and recommendations follow.

The motivation for this chapter includes the following:

• One of the factors driving chronic and transient poverty in India is uninsured risks faced by households. There is thus an equity rationale for understanding the role for public policy in helping households deal with risks and shocks related to unemployment, health and death, maternity and sickness, and unprotected old age.
  
  
• Uninsured risks and associated credit and insurance market failures may contribute to low productivity in the unorganized sector, particularly in agriculture. If this is so, there is a dynamic efficiency motive for understanding the role of policy in risk management.²
  
  
• GoI (Government of India) has placed an increased policy emphasis on both public works and social security for unorganized-sector workers in its CMP (Common Minimum Program). This makes it important to understand what are the realistic expectations for public programs, in the light of experience in India and internationally.
  
  
• One objection to greater flexibility in labor regulations in India is the absence of a functional safety net. Understanding more about public interventions to provide social-security and employment support to workers may therefore be relevant to debates on labor regulation.

The objectives of the chapter are:

• To assess the role of both public-works and social-security initiatives in promoting consumption smoothing by unorganized-sector households in the face of predictable fluctuations in income-related to seasonal labor demand, and unpredictable fluctuations due to covariate and idiosyncratic shocks.
  
  
• To draw lessons for initiatives which may help in improving the impact of public interventions on the social risk-management strategies of unorganized-sector workers.

Public works schemes³

Objectives of public works

It is important to examine the objectives of public-works programs in order to assess workfare in India. They include:

• The transfer impact, i.e., direct transfer benefits to participating households. This impact is often equated with the employment impact, though the latter is properly considered as an intermediate activity.
  
  
• Impacts on overall wage levels, which may be felt by participating and non-participating households whose incomes are below the post-program area wage. The positive impact may be offset by negative impacts on labor demand from higher wages.
  
  
• Indirect impacts from the economic benefits of assets created under works schemes. These would typically be spread among a wider population than participants, so that the economic returns on assets to the poor will affect the distributional impact of workfare.
  
  
• The consumption smoothing or insurance function that schemes perform by acting as an effective wage floor. Whether or not households participate in the program, insurance benefits may promote less risk-averse behavior in production and portfolio choices.
  
  
• Skills development impacts for workfare participants, through work or training provided.⁴
  
  
• Impacts on gender and other socio-economic relations due to features such as equal pay for equal work, increased female labor-force participation, etc.
  
  
• Community-empowerment impacts which may arise through reliance on communities at different stages of sub-project planning and identification, beneficiary selection, works execution and monitoring.

Evolution of policy and expenditures on public works in India

Since the 1970s, public works have been an important component of the Indian safety net, with a succession of programs, both centrally sponsored and state-specific. This assessment focuses on the major central-works programs. While changes in programs have been frequent, much of the development has been rebranding rather than fundamental reform. The most significant policy shift in the 1990s has been the increased role for PRIs (Panchayat Raj Institutions). However, even this shift, which is clear in the guidelines of workfare programs (and very thorough in NREG) has been more mixed than policies might indicate. The more recent policy shift has been from scheme-based provision of works to a legislatively backed 100-day rural-employment guarantee under NREG.

Figure 12.1 shows spending for recent years on works schemes as a share of total central government expenditure and GDP, and in real terms. There was a clear fall in total spending as a share of GDP and government spending between 1995 and 2003-2004, to around a third of spending shares in the early 1990s. Spending in real terms tracked the decline in government spending share. While this trend has reversed in most recent years, spending shares remain below their high point. While the decline and recent reversal is the most notable feature, it is interesting to note the upticks in 1993–1994 and 2003–2004, i.e., years preceding national elections.⁵

Figure 12.1 Spending on main public works programs, various indicators (sources: Rural Development Ministry to 2002/2003; MOF revised budget estimates for 2003-2005 and GoI total spending).

Note
Real 1993 Rs. adjusted by agricultural laborer consumer prince indices (CPI) (RBI series).

Impacts of public-works schemes

What has been the performance of public-works programs with respect to their objectives? Unfortunately, both administrative and household data allow only partial answers. This is due to two reasons:

• Administrative data on employment generated are 'based only on arithmetic calculation' (CAG 1997), i.e., work days generated are calculated backwards from total spending on an assumed 60:40 split between wages and materials. Survey-based assessments throw doubt on these estimates.
  
  
• The NSS has to date asked households only about participation in public works of more than 60 days.⁶ NSS coverage rates for public works can therefore be considered the lower bound.

Direct-employment effects

Figure 12.2 shows the administrative estimates of work days from the major works programs from 1993 to 2004, as well as aggregate rainfall data. As with spending share, employment-generation performance declined between the mid-1990s and early 2000s.

The level of public-works employment generated has been far lower than program objectives would suggest. Average annual employment per BPL household under SGRY in 2001–2004 came to only around 6.7 days, assuming that all workdays were for BPL.⁷ The same work days spread across the agricultural workforce would mean around two days work per worker.⁸ Such employment-generation figures are lower than JRY and EAS, which had an average of 16 days employment per BPL household under JRY and 15 days per registered EAS worker in 1992–1999. The fall in employment seems broadly consistent with real spending since the mid-1990s.

Figure 12.2 Work days of public employment and rainfall, various years (source: RD Ministry; Indiastat.com for rainfall data).

Note
Rainfall is measured by variation from annual mean rainfall.

Statewise employment-generation figures for SGRY suggest that the aggregate employment effect of workfare in all states is very low, with even outliers such as Assam and Orissa generating less than ten days work per agricultural worker, and none but Assam and Karnataka generating more than 15 days SGRY employment per BPL household.

At the household level, NSS allows for examination of works coverage. As noted, NSS is an imperfect source on the issue. Nevertheless, comparisons across states and time are possible, and include any public works undertaken. A few observations emerge:

• At the all-India level, the coverage rate is lower in each subsequent period, from 4.8 percent in 1987/1988, to 4.3 percent in 1993/2004 to 2.9 percent in 1999/2000. Given that works reliance is likely to vary by year depending on exogenous factors, not too much should be read in by way of a trend. This can be seen in state-specific coverage rates, which in a number of cases exhibit more pronounced inter-temporal variance, with a different pattern to the all-India.
  
  
• For 1987–1988, public works were able to respond well to a severe drought year. Drought data on a disaggregated basis needs to be explored to examine the responsiveness of works programs to more localized covariate shocks.
  
  
• Average coverage rates in the North-East have remained notably higher, reaching as high as 22 percent in Tripura and 30 percent in Arunachal Pradesh in some years.

Targeting of public works

Targeting of public works is more progressive than all other major anti-poverty programs with the relatively high coverage rate in the poorest quintile and among ST households.⁹ Results from a 2004/2005 national survey for SGRY/FFW indicate progressive coverage across wealth groups, though low overall coverage even for the poorest. Nonetheless, the high relative coverage rates in the bottom quintile and among ST households places public works as the best program performer in distributional terms (see Table 12.1).

The positive targeting performance of public works from national data is supported by recent evidence from a three-state study in Orissa, Karnataka and MP in 2006. Particularly for the Food-for-Work program, coverage in the lowest quartile was relatively high (around 57 percent higher than the population average), while coverage among ST households was well above the population average, by almost 80 percent in the case of SGRY (Dev et al. 2001).

The above analysis assesses targeting in terms of average incidence. This may be misleading where there are marginal adjustments in budgets. Analysis from 1993–1994 indicates that average benefit incidence underestimates the gains to poor households from increased spending on works (Lanjouw and Ravallion 1999). While the marginal odds of participation are lower for nearly all groups than averages, the relative reduction in odds is significantly less for the poor than for higher income quintiles (Table 12.2).

What has been driving employment outcomes in India's public works?

What are some of the explanations for such a wide gap between the employment commitments of public-works programs and outcomes? Several factors appear to be at work.

In employment terms, falling real spending since the mid-1990s has been reinforced by rising real wages on works in the 1990s (Table 12.3). Between 1993/1994 and 1999/2000, average annual growth in real earnings on works was 3.8 percent, which was higher than growth rates for casual rural laborers in and out of agriculture (Sundaram 2001). By the late 1990s, public-works wages were 21 percent (for men) and 38 percent (for women) higher than average wages in casual agricultural labor (Sundaram 2001b using NSS 1999–2000 data). While public-works schemes generally do not aim at paying less than minimum wages, around 75 percent of all work days on rural casual labor were below agricultural minimum wages, with the share above 90 percent in AP, Bihar and Orissa (Murgai and Ravallion 2005).

Although above-market wages should induce high demand for participation, budgetary resources are in practice constrained. Indian and international experience suggests that program wages above market rates result in rationing of employment on works (Ravallion et al.(1993); re Maharashtra EGS; and Sub-barao 2003). It is interesting to compare information on wages in public works internationally, which shows mixed experience on the relationship. Generally, programs considered to be more successful (e.g., Trabajar; Korea; MEGS pre-1988) were more likely to have program wages below market wages. However, the position in India is subject to legal precedent that program wages below minimum wages are illegal (Papola 2005).

Public-works programs do not appear to have met the 60:40 labor to capital ratio, although the official method of calculating work days makes this hard to confirm. However, a survey-based EAS evaluation for 1993–1997 found that no category of works met the 60 percent wage component criterion, with the national average share of wages to total expenditure at 47.5 percent. Furthermore, only three of 14 major states (West Bengal, Orissa and Maharashtra) met the 60 percent wage target across activities. In contrast, several states had much lower wage shares, with Bihar at 27 percent, Haryana at 30 percent and Rajasthan at 33 percent (Program Evaluation Organization of Planning Commission 2002).

Another factor which may impact the labor intensity of public works is the extent of reliance on contractors. Though little evidence exists of contractors systematically using more labor-intensive methods, a GoI review of EAS indicates that even a 10–15 percent profit margin for contractors would reduce the labor budget per unit of spending.¹⁰ Evidence in India on the extent of reliance on contractors in public works is mixed. A 2005 MRD evaluation of SGRY found use of contractors was substantially lower than previous evidence had indicated. At national level, contractors were reported to be involved in only 14 percent of assets, with significant state variations (e.g., Rajasthan, in only 2.4 percent of assets created, while Kerala reported for two-thirds of assets and Karnataka just over a third).

The findings should be viewed in light of other evidence, which seems difficult to reconcile. A 2005 study for RD [Rural Development] Department in Orissa found contractors in over 92 percent of SGRY assets (against a figure of 25 percent for Orissa in the national study).¹¹ A village study in AP of the state's food-for-work program in 2001–2002 found that all works were executed by contractors, and that, in the large majority of cases, sarpanches were the contractors.¹² Finally, a recent study of SGRY in Kerala, Karnataka, Rajasthan and West Bengal found regular use of contractors, despite the prohibition in guidelines (World Bank 2006).

Figure 12.3 Share of villages and village population covered by public-employment programs in previous year, 2002 (source: NSS 58th round, 2001/2002. Bank staff estimates).

Despite commitments to full rural coverage, the share of villages covered by works schemes has been far less than complete. Assessment of EAS for 1993–1997 found that only 53 percent of villages had any works, with some states much lower. More strikingly, the proportion of villages covered in all four years was only 5.4 percent.¹³ Analysis from 2002 NSS village data confirms that partial spatial coverage has continued, with only 48.5 percent of villages (56.4 percent of population) reporting any public-employment program in the past year (Figure 12.3). Overall, there is substantial inter-state variation in coverage, with village (population) coverage as low as 20 (25) percent in Jharkhand and 11 (18) percent in Punjab. In addition, there is for all states (except Kerala) a higher likelihood of works having been available in the previous year in larger villages. At the all-India level, rural population coverage is around 16 percent higher than village coverage unweighted for population, with the large village effect particularly pronounced in states such as Bihar, Gujarat and Punjab.

Timing of public works has often not been well matched to seasonal employment needs, with difficulties in matching peak periods of public-works provision with periods of lowest market demand. This can be seen in Figure 12.4 for Maharashtra, which shows public-works employment dropping sharply in the monsoon when market-based work is least available. Recent analysis of MEGS indicates that this pattern has continued.¹⁴ It is consistent with findings in Rajasthan that works have typically been carried out during January–March, when opportunity costs of labor are high.¹⁵ Part of the issue is that certain works cannot be executed during the lean season due to monsoon. This is reinforced by budgeting practices which concentrate disbursements for works in the final quarter of fiscal.

Figure 12.4 The seasonality of MEGS employment (source: Subbarao, 1993 and 1997).

The persistence of implementation problems in public works has detracted from employment (and other) impacts. The problems are in general related to lack of accountability in workfare design, financing and management. Issues in GoI's assessments of public works and external assessments include (CAG 1997 and 2000; PEO 2000; Nayak et al. 2002; World Bank, forthcoming):

• Leakage and parking of funds. CAG reports point both to leakage of funds and parking of significant funds in deposit and other accounts. Parked funds show up in public accounts as executed but in fact remain undisbursed.
  
  
• Prioritization of works by line departments rather than PRIs.
  
  
• Lack of reliable records, in particular muster rolls. Survey evidence on the NFFW program from six states indicates that unavailable and fudged muster rolls continue to be a serious issue. At the same time, the study found that the practice had reduced sharply in Rajasthan in the face of better public information and community-level activism, suggesting that simple improvements in access to information can improve outcomes (Roy and Dreze, The Hindu, August 15, 2005).
  
  
• Lack of both strong administrative monitoring and evaluation and absence of "bottom up" monitoring through mechanisms such as social audits.

Other impacts from public-employment schemes

Measuring other impacts of public-employment schemes is more challenging. For several impacts, there are simply no data available. This includes impacts on overall wage levels, economic benefits of assets created, community empowerment and skill development impacts. On other indicators, research provides suggestions on the direction and scale of impacts, but is typically very much localized. The evidence on other impacts of works includes the following.

Female labor-force participation in SGRY appears to be lower than general rural female participation rates. Data from a 2005 evaluation of SGRY¹⁶ show that the all-India share of SGRY female participants was only 12 percent, which is similar to CAG findings on JRY and EAS of only around 16 percent female beneficiaries (the target share was 30 percent). In states such as UP, Bihar and Punjab, females' participation in public works is very low, at less than 2 percent. Interestingly, the survey-based results above appear to diverge from administrative data for the same year, which report around 26 percent of person days generated for women.¹⁷ The figures compare unfavorably with estimates of females by usual status in the rural workforce of around 35 percent in 2000.¹⁸

It is difficult to estimate the economic impact from assets in Indian works schemes. Given the importance of rates of return on assets in assessing program impact, this is unfortunate, though a common problem in workfare worldwide. Qualitative evidence available from GoI and other evaluations largely relates to quality of assets rather than rates of return.¹⁹ A common criticism of public works is that they are 'washed away the next monsoon'. In light of this view, the findings of a beneficiary survey are intriguing.²⁰ Both at all-India level and in all states, beneficiaries finding the quality of works very good or good dominated. This needs to be interpreted with caution, both because of possibly low expectations of beneficiaries, and because the survey found that only 31 percent of respondents were aware of quality specifications for works. Evidence from EAS also indicates that public-works-assets maintenance is poor, so that even decent-quality assets may deteriorate quickly (CAG 2000). A challenge for the NREG will be developing cost-effective methods for estimating quality and rates of return on assets.

The insurance and associated productivity impacts of public works are also difficult to measure absent rigorous evaluations. However, what evidence there is for India suggests a positive insurance function of public works, with attendant impacts on production decisions. A study from the 1980s in Maharashtra found that income variability among landless agricultural households in villages where MEGS was available was half the level of villages where it was not (Walker et al. 1986). Farmer studies in Maharashtra also find greater adoption of higher risk/return agricultural practices (Devereux 2005), though how much this can be attributed to MEGS is unclear.

The national rural employment guarantee (NREG)

GoI has in recent years committed to major expansion in public works, initially with the National Food-for-Work program from 2004, and since 2006 through NREG. Overall, it is too early to make conclusions on the effects of NREG on parameters such as poverty, labor markets and the local economy. Nonetheless, relative to previous public-works schemes, there are a number of design features of NREG which are very sensible, and in many states there has been greater political and institutional commitment to trying to 'make the scheme work'. That said, the summary of initial implementation experience suggests that implementation is highly variable across (and even within) states, and that there remain major challenges in implementation.

Ex ante estimates of NREG impacts

Given the early stage of implementation, simulations of scheme impacts carried out prior to initiation remain worthy of consideration to look at the potential impacts of NREG. The main findings are the following (Murgai and Ravallion 2005).

ESTIMATED LABOR-SUPPLY EFFECTS

In terms of labor supply, an NREG wage rate close to Rs.60 at current prices²¹ could induce a 5 percentage point increase in lean-season labor supply among casual rural workers. Interestingly, the incremental labor supply impact of a wage rate above Rs.60 is negligible for all groups. The overall wage elasticity of labor supply is estimated to be 0.17.

The estimated labor-supply impact can also be presented in terms of expected number of rural people presenting for work on a typical day in the lean season, and this can be broken down by gender. This is presented in Table 12.4, together with estimated fiscal costs at three wage rates. Some interesting observations emerge:

• If the estimated supply response is realized, it would imply a total annual provision of work days of around 3.9 billion once NREG is rolled out nationwide. This compares to an annual average in 1999–2004 of closer to 600 million days annually on workfare programs – a more than six times expansion once NREG reaches scale. Such expansion would not happen immediately, due to initial implementation in backward districts, and other factors. NREG thus implies huge additional demands on public-works delivery systems in India.
  
  
• Fiscal costs are potentially substantial for a nationally implemented scheme. The wage rates for NREG suggest fiscal costs of a nationwide program of around 1.7–1.9 percent of GDP. The costs estimates are gross, and there would be netting out of SGRY and NFFWP spending (which nationally ran to around 0.2 percent of GDP in 2004–2005).
  
  
• The benefits of inducing higher female labor supply are fully realized at a wage rate lower than state minimum agricultural wages. Nonetheless, in absolute terms, more men than women would be expected to take up NREG lean-season employment (though predicted female share is higher than under SGRY) – see Table 12.5.

Poverty and distributional impacts of NREG

The estimated lean-season poverty-reduction impacts of NREG are significant, and the distribution of gains progressive. At a wage equivalent to Rs.63 in 2006 prices, lean-season rural poverty could be reduced from 37 percent pre-NREG to 23 per cent or from 34 percent to 30 percent on an annualized basis. The distribution of gains would also be progressive, with around 54 percent of gainers in the bottom two expenditure quintiles, and less than 10 percent in the top quintile.

More notably, direct income gains from NREG would be around 51 percent of pre-NREG lean season income for the bottom expenditure quintile (Murgai and Ravallion 2005).

How has NREG done so far in practice?

This section presents administrative data on performance, before a summary of several studies on initial implementation experience with NREG. Looking at administrative data for the end of 2006 (as given in the organization's website), a few points are worth noting:

• NREG generated significant demand from rural households in its first nine months of implementation, with 33.7 million rural households issued with a job card, or approaching one quarter of all rural households. Even the lower figure of 13.7 million households being provided employment approaches 10 percent of rural households. This is a major achievement in program coverage relative to prior public-works schemes.
  
  
• There is a wide gap between those registered with NREG job cards and those actually demanding work, with the latter around 42 percent of the former. At the same time, the gap between those demanding and provided work is negligible. This points to two features of demand manifestation. The first is that households may register with no current intent to undertake public works, exercising an implicit insurance function of the program. The second – and more worrying – is the process through which work is "demanded" under NREG. While the scheme is framed as on-demand, in practice worksites have to be opened in a locality before people are likely to come forward for work. This in turn requires a quorum of people expressing a demand for work. This 'chicken and egg' issue suggests the importance of the supply side even in such a demand-driven program.
  
  
• Program outcomes have been quite variable across states. Coverage of NREG registration is high in absolute terms largely where one would expect based on district concentration across states, though AP, MP and West Bengal stand out on the upside, and North-Eastern states as laggards. Greater variability can be seen on the supply of actual employment to those registered, which may reflect both demand - and supply-side effects. Of particular note is Rajasthan, with easily the highest employment ratio of major NREG states, reflecting a massive public awareness effort on the part of CBOs.
  
  
• Administrative data also indicate impressive inclusion of ST (scheduled Tribe) households, and of women relative to previous public-works schemes. Nationally, ST households account for around 39 percent of workdays generated in 2006 (and as high as two-thirds in Rajasthan), a share that far outstrips their general population share. In addition, the share of women in total work days generated in 2006 was around 38.5 percent, and as high as two-thirds in Rajasthan. This is a major improvement in female participation relative to SGRY.

Beyond the administrative data, survey results from initial NREG implementation experience in a number of states provide useful insights, and identify several key challenges in ensuring program effectiveness. The studies were carried out in the first six months of implementation, so should be seen as identifying very initial performance.

In the states surveyed, the composition of households who had received job-card work under NREG indicates that the program has managed to sustain good penetration among ST households that is seen in previous schemes. The coverage rate among ST households is high in absolute and relative terms (though not in Bihar, where ST are a much smaller share of the population). Overall, it is too early to comment reliably on the targeting outcomes of NREG, but the results are promising.

Apart from the above, the surveys identify several implementation challenges that are consistent with anecdotal reports from other NREG districts. They include:

• Program wage rates. There has been tension between the program commitment to payment of agricultural minimum wages and the reliance on piece rates for wage payment. This has resulted in most states in actual wage payments which fall short of official minimum agricultural wages (Table 12.6). This is due to rural schedules of rates for work which assume a level of capital intensity not appropriate to a labor-intensive scheme such as NREG. A number of states – e.g., Gujarat, AP, and Rajasthan – are addressing this issue through general revisions of rural schedules of rates or setting of NREG-specific rates (based on new time and motion studies), including in AP a specific time and motion study for disabled workers.
  
  
• Role of communities in selection of works. In principle, gram sabhas have a key role to play in identification of works under NREG. As of mid-2006, such consultation was happening rarely in MP, Chhattisgarh and Jharkhand, and in AP the GP role was being undertaken by the Village Organization which operates a CDD initiative (CBGA 2006). In Bihar, most GP-level
  
  

• sub-projects were selected by the mukhiya without gram sabha consultation, and in Gujarat, households reported a gram sabha input to works identification in only about a third of cases.
  
  

• The generally low capacity of PRIs to carry out a range of NREG functions expected of them is a difficult challenge. Apart from works selection, GPs are expected to mobilize field assistants and beneficiaries, keep asset, beneficiary and financial records, and play an important role in social audits.
  
  

• There is no evidence of any state having paid an unemployment allowance (UA) in cases where work has not been provided within the specified period. This is important for two reasons. The first is that the 'guarantee' of the Act is ineffective unless UA is paid in a timely manner. The second is that states – which have to finance UA from their own budgets – will have stronger incentives to improve implementation if UA is implemented.

Conclusions and recommendations on public works

Many of the appropriate reforms of public-works policy are reflected in NREG, which represents the most serious effort to date to address the institutional and implementation problems encountered in works schemes. Nonetheless, areas that warrant particular attention as the program matures include:

• Continue and deepen awareness raising efforts on NREG entitlements, in close collaboration with civil society. Initial results indicate that basic awareness is still lacking in several states, and even in well-performing states that knowledge of specific entitlements can be improved.
  
  
• Complete time and motion studies in major NREG states to assess the appropriateness of piece-rate-payment schedules. The experiences of AP, Gujarat and others are useful.
  
  
• Explore options for a wider range of authorized work activities which may be done under NREG, in particular for those not capable of hard manual labor. The NREG Act allows for this, but the option of expanding categories of works has yet to be exercised. As experience grows, this would be worth looking at from several perspectives. The first is whether 'softer' work options – e.g., provision of community services for vulnerable populations – could be introduced which have high positive externalities. The second is looking at the needs of sub-groups of job-card holders for whom special efforts may be needed to provide appropriate work, e.g. disabled people.
  
  
• Deepen the direct involvement of communities in identification of works undertaken under NREG. While it is understandable that the initial implementation would rely on shelves of works already identified, it will be important to ensure that the role of gram sabhas anticipated under the Act is made a reality, and that groups such as SHG and CBOs (Community-based organizations) are mobilized to have their priorities reflected through the gram sabha process.
  
  
• Assess the ways in which demand for work is manifesting itself, and the constraints – formal and informal – on matching that with supply of worksites and employment opportunities. This is probably the most difficult area of NREG to operationalize effectively, and goes to the heart of what a demand-driven approach means in practice. This will involve a principled but practical look at such questions as the minimum number of workers required at a worksite, how the bureaucratic demands of authorizing worksite opening can be matched to demand, etc.
  
  
• Make states and communities aware of the unemployment allowance provision in the Act. This is essential if the guarantee function is to be credible. Continue to develop the capacity-building function allowed for under NREG. Under the Act, a portion of funds are available for capacity building of PRIs and others involved in NREG implementation. To date, this is still to be developed, but will be a critical element of program success.
  
  
• Get 'baselines' in place for impact assessment. Unfortunately, there is no solid pre-program baseline on NREG. However, efforts are being made to provide for credible impact evaluations of poverty, labor-market and other outcomes.
  
  
• Operationalize the central and state NREG councils to oversee implementation, and use them to promote exchange of good practice between states.

An additional issue for consideration is whether any element of direct human-capital formation can be factored into NREG as it matures. Presently there is no provision under NREG for skill formation among workers. This could be considered, along the lines of the South African public-works scheme. While such an approach obviously requires a supply side – probably on a contracted basis – which can provide useful training, it seems an interesting option to consider.

A final issue is that public works for the poor remain restricted to rural areas. An issue for the consideration of policymakers is whether an expanded self-targeted public-works program could be designed for implementation among the urban poor. Such programs exist in the works schemes of a number of developing countries. While a small scheme (SJSRY) exists, its impacts to date appear negligible and funding remains marginal (World Bank forthcoming).

Social insurance for the unorganized sector

As in many developing countries, unorganized-sector workers in India face an array of uninsured risks, making them highly vulnerable to both idiosyncratic and covariate shocks. While family, jati and other informal support networks play an important role in smoothing consumption in the face of shocks, the risk-sharing function they perform is far from perfect (and more effective for idiosyncratic than covariate shocks, and for small as opposed to catastrophic shocks).²² Recent work also demonstrates a link between reliance on jati (caste) networks of mutual informal insurance and low rates of occupational and geographical mobility found in rural India, with migration and marriage outside the jati triggering loss of access to finance in the face of shocks (Munsi and Rosenzweig 2005). Credit- and insurance-market failures may thus be limiting the ability of rural workers to seek more productive employment opportunities.

Survey evidence confirms the impact of uninsured shocks on household welfare. This is particularly the case for health shocks, and for the poor.²³ For example, at least 24 percent of Indians who are hospitalized fall into poverty as a result (Peters et al. 2002). There are also concerns that credit-market failures drive coping mechanisms which may turn transient into long-term and even inter-generational poverty, e.g. withdrawal of children from school, debt-bondage, etc. Such effects have been analyzed for countries such as Indonesia, which stress that the benefits of social insurance in poor countries may come less from the direct contribution to consumption smoothing than the reduction of coping strategies to smooth consumption (Chetty and Looney 2005).

International context

Despite efforts, many developing countries have failed to extend insurance coverage to the majority of workers typically found in the unorganized sector. At the same time, international evidence since the nineteenth century indicates that social security spending tends to increase sharply with rising income levels, or at least to increase sharply once a threshold level of GDP is attained (Lindert 2004). Important facts on this transition include:

• As countries become richer, the share of social protection spending in GDP and total public spending tends to increase (Palacios and Paralles 2002; Lindert 2004).
  
  
• Within social protection spending, social insurance typically accounts for an increasingly higher share of GDP as countries become richer, while the share of social assistance stabilizes at 2–3 percent of GDP (a level India already approaches). The transition also involves an increased role for cash-based interventions relative to food-based.
  
  
• Voluntary social-insurance systems have not resulted in major coverage expansion, indicating that more formalized labor markets – where mandated participation can better be enforced – are closely associated with higher coverage rates. The association between GDP per capita and mandatory pension coverage is shown in Figure 12.5.²⁴ Experience with health-insurance expansion in East Asia conforms to this pattern, with universal coverage being achieved at income levels between US$5,000 and $10,000 and where formal-sector employment dominates.²⁵

Figure 12.5 Social insurance and assistance spending shares by region, and pension coverage by GDP (sources: WDR (2006)).

Current status of unorganized-sector social security in India²⁶

Given the high rate of informality in Indian labor markets and level of income, coverage of social insurance is predictably low and concentrated heavily in the organized sector. Figure 12.6 provides coverage estimates of different social insurance types for 2004/2005 across the distribution, showing not only the failure of formal health and pension insurance systems to expand coverage, but also the growing penetration of life insurance driven by the commercial-insurance sector.

Recent approaches to social insurance (SI) for the unorganized sector

In the face of very low SI coverage of unorganized-sector workers, the public (central and state), non-government and private sectors are all involved in efforts to expand coverage. Various approaches are being tried with differing degrees of success, in terms of reaching scale, achieving financial viability, and providing financial protection to households. Each approach has advantages and drawbacks, some inherent (e.g. viability in the face of covariate or catastrophic idiosyncratic shocks for community-based initiatives), and others a product of specific design and implementation features. The main initiatives described are:

• Welfare and provident funds for unorganized-sector workers at central and state levels, which include both occupation-specific and generalized unorganized sector funds.
  
  
• GoI national schemes for unorganized-sector workers outside the welfare-fund model.
  
  
• NGO and community-based initiatives, which range from coverage of specific SI types (usually health) to more bundled social-insurance packages.

Figure 12.6 Coverage rates of health, life and pension insurance by quintile, 2004/2005 (source: World Bank (forthcoming)).

Welfare and provident funds

Welfare funds (ILO 2004) have traditionally been occupation-specific schemes providing a range of benefits for members. The earliest fund dates to 1946, though they have proliferated since the late 1970s. It is difficult to obtain a comprehensive picture of funds due to their decentralized nature, both geographically and occupationally. Despite the lack of comprehensive coverage, patterns emerge in terms of key features and challenges of welfare funds.

Most attention in discussion of welfare funds has been paid to the Keralan funds, of which there were more than 20 by 2000. (Dev 2000; Kannan 2002), and central government funds, of which there are five with central financing and another with a central act but to be implemented and financed at state level.²⁷ However, in terms of total coverage, funds in other states – initially in southern and western states, but spreading in recent years to the north and east – have accounted for an increasing share of participating workers since the 1990s.

The financing of welfare funds follows two basic models: (i) cess-financing, which is used for all the central government funds, and may be a cess-proper or excise duty; and (ii) contributory financing, which is either tripartite between government, employers and workers, or from employer and employee only.²⁸ In cases where government contributes to the second category of funds, the dominant method is direct contribution. Management of the funds is either tripartite or government only (in roughly equal proportion for documented funds). A few things are apparent from looking at contribution rates and benefits of funds:

• Contribution rates exhibit dramatic variations across funds, from very low amounts (e.g. AP and Karnataka Labor Welfare Funds), through more significant absolute amounts (e.g. Kerala Cashew Workers), to quite substantial shares of worker compensation more akin to formal sector contribution rates (e.g. Maharashtra Mathadi Boards). More detailed analysis is needed of the correlation between contribution rates and benefits offered. However, initial examination suggests no systematic relationship.
  
  
• It appears that no financial/actuarial analysis has preceded the setting of contribution rates and packages of entitlements. This is likely to show up more starkly as the membership of funds ages. The long-term finances of the Beedi Workers' Fund demonstrate the impacts of membership ageing, with expenditures in most of the 1980s well below income, but moving into significant deficits after 1992 (Rajan 2001).

Although welfare funds tend to be characterized primarily as instruments of social insurance, the picture with respect to benefits and services offered is more complex. Education is the most commonly provided entitlement across funds, followed by medical cover and then pensions. The benefits provided spread beyond typical social insurance, raising concerns about the ability to provide financial protection in the face of shocks. For the pensions portion of schemes, the large majority are defined benefit programs, typically with benefits expressed in nominal rupees and hence subject to deterioration in real value.

There is also major heterogeneity in the generosity of entitlements provided by different funds. Given this, establishing typical expenditure patterns is not straightforward, as detailed spending information is available only for a limited number of funds. For the central funds, the available data suggest the following (Rajan 2001):

• Health overwhelmingly dominates spending for all central funds except Mica, where education is significant. This has been the case since at least the early 1990s also.²⁹
  
  
• Administrative costs are extremely high in the smaller funds. This poor performance is remarkable given that financing is cess or excise based, and should not demand the intensity of administration of funds based on contributions.
  
  
• None of the central funds provide for old age, disability, sickness or life insurance. The exception is the Building and Construction Workers' Fund, which is delegated to states. Below gives a recent example of a state-level initiative to promote old-age pensions: a defined contribution provident fund for unorganized-sector workers in West Bengal.

Since 2001, West Bengal has operated a Provident Fund for workers in the unorganized sector, not including agricultural workers. In 2006, enrolment was around 700,000. Membership is open to wage and self-employed workers in designated industries/activities (which are expanding over time) whose family income is no more than Rs.3,500 per month (though the means test appears to be rather informal). The key feature of the Fund is that it is a defined contribution scheme. Members contribute Rs. 20 per month, with a matching contribution from the state government. Interest accrues on the account at a rate designated by government (the EPFO [Employee Provident Fund Organization] rate to date). Accrued principal and interest is paid to the worker at age 55 or at death or when there is no account activity for six months.

Rapid expansion has been driven by a highly decentralized distribution network and governance structure, with support from trade unions and political parties in mobilizing awareness and interest. Collection agents are typically public workers who are incentivized by a one rupee payment for each enrollee. The state government covers the administration costs of the Fund. While the West Bengal experience has great promise, particularly given its DC basis, it remains to be seen whether the promotion and distribution success achieved is replicable in states with less stable political institutions (LO 2004).

Welfare funds have potential as vehicles for expanding social insurance to some segments of the unorganized sector, i.e. those amenable to a cess and/or with strong presence of intermediary organizations which can share the transaction costs of wide membership/low contribution schemes. They provide interesting parallels to occupational social-insurance schemes in OECD countries at earlier stages of development. Despite their potential, however, the evidence suggests that they face a number of design and implementation challenges which may constrain their ability to provide adequate and sustainable financial protection for members, including:

• The small and occupationally concentrated risk pools of a number of funds risk driving high administrative costs and weak capacity to absorb significant covariate shocks. These challenges can be surmounted through group insurance with commercial insurers, but it remains unclear to what extent this has been exploited.
  
  
• The scope of benefits in many funds is likely to contribute to inadequate financial protection for any specific type of benefit. This in turn risks benefits becoming supplements to smaller and/or more predictable expenditures rather than being focused on larger shocks.
  
  
• For the insurance-like benefits in welfare-fund-benefit packages, there appears to be no actuarial basis for contribution determination. In light of this, the extent to which the schemes can be considered insurance rather than partial pre-payment schemes is an issue.
  
  
• There appears to be no example of a rigorous impact study of welfare-scheme participation.

Other government initiatives on social security for the unorganized sector

The government has in recent years increased efforts to expand social security to unorganized-sector workers through non-welfare fund mechanisms (Planning Commission 2002). This section provides a brief review of initiatives in recent years.³⁰ By way of context, it is important to note a couple of points on the insurance industry in India. First, public and private insurance companies have obligations under the IRDA (Insurance Regulatory and Development Authority) for a minimal level of both social and rural coverage (Breman and Ahuja 2005). Second, to date life and non-life insurance provision remains separated.

In 2004, GoI introduced a social-insurance scheme for unorganized-sector workers (excluding agriculture), piloted in 50 districts nationwide and targeting around 2.5 million workers.³¹ The scheme was managed by EPFO in collaboration with ESIS for health services. It was voluntary and contributory for those with monthly incomes under Rs.6,500 and provided for old-age, medical and accident insurance. GoI contributed around Rs.250 annually per worker. Premia for workers were Rs.50 per month for workers to 35 years and Rs.100 above that. Employers were meant to contribute Rs.100 monthly. Workers covered the employer contribution themselves in addition to their basic contribution when no employer is identified. The scheme appears to have had negligible penetration, with an estimated enrollment in mid-2005 of less than 10,000 and the scheme is now dormant.

Another major initiative was the Universal Health Insurance Scheme (UHIS), launched in 2003 (Ahuja and Narang 2005). This is a voluntary contributory scheme for BPL households, covering medical costs of hospitalization, loss of income during short-term-illness and death. There is a contribution subsidy from GoI ranging from Rs.200–400, with net contribution from the worker ranging from Rs.165 to Rs.300, depending on household size. In parallel, a scheme for unorganized non-BPL households was introduced. Outcomes on UHIS appear also to have been limited in the initial phases, with only around 400,000 households covered in the first year of operation (less than 5 percent of them BPL) and a further 31,000 households up to January 2005.

Another important scheme targeted to the unorganized sector is offered by the government-owned Life Insurance Corporation (LIC) called the Janashree Bima Yojana (JBY) scheme. It covers 44 occupational groups, chosen to target those living near the poverty line. The scheme pays Rs.20,000 for natural death, 50,000 for accidental death or permanent disability and 25,000 for partial permanent disability. There is also a scholarship of Rs.300 per quarter per child paid to workers who send their children (up to two) to grades 9–12 for a maximum of four years. The package is financed by a premium of 200 rupees collected through 'nodal' agencies, i.e., groups that must include at least 25 workers. A number of groups, ranging from SHGs to small occupational groups, have signed up as 'nodal agencies', reflecting a growing tendency to rely on the partner–agent model.

As in West Bengal, there is a matching premium payment, in this case subsidized by the central government. Financing comes from a social-security fund at LIC that was set up in 1989 and has received two grants from the central government. In 2005, there were approximately 3.5 million lives covered. There was a massive expansion of the program in mid-2006 when the state government of Rajasthan purchased the policy on behalf of workers in all 2.2 million Below Poverty Line (BPL) households.

JBY is undoubtedly the single largest insurance scheme for the informal sector in India. The design of the scheme is interesting from a policy perspective in several ways. First, it is an attempt incrementally to increase coverage, in this case on the basis of occupations or groups considered to have lower incomes, but with a capacity to contribute. Second, there is a transparent subsidy intended to incentivize voluntary take up. On the other hand, the target group-size – 25 – seems small, and financing of the subsidy over the long run does not seem to have been considered carefully. Moreover, the government subsidy has only been available to one (state-owned) provider and there is no record keeping mechanism to allow the government to monitor whether its money is reaching the intended target.

The above schemes are by no means a comprehensive picture of central and state government initiatives to expand social insurance. However, even the selective presentation suggests a few preliminary conclusions:

• The rollover of insurance schemes at central level has been frequent. When trust in the notion of insurance may be low and information dissemination is weak, such policy uncertainty is problematic. The frequent introduction of schemes appears to be driven more by political imperatives than a strategic approach to expansion.
  
  
• While the need is high, taking on health insurance confronts the most challenging area of social insurance first, where moral hazard risks and administrative demands are high.
  
  
• Most central schemes have to date failed to address the fundamental issue of high transactions costs and the need for decentralized distribution networks.
  
  
• As with welfare funds, it is not apparent that the contributions for various schemes are based on an informed assessment of claims information from existing insurance products.

NGO and community-based social-insurance initiatives for the unorganized sector

Estimates on the scale of social-insurance provision by NGOs and other community-based actors (e.g. microfinance institutions, health facilities) vary from around three to around five million (Breman and Ahuja 2005). Given the very low coverage rates achieved by government initiatives, community-based micro-insurance (CBMI) initiatives assume significance as a potential channel for expansion of social security. At the same time, the experience to date is mixed, with questions on the capacity to go to scale, given the reliance on subsidies from founding institutions and donors. In addition, CBMI to date have focused primarily on health insurance, and not yet addressed other types of insurance, particularly old age. Finally, there are legal issues with respect to CBMI, as the IRDA Act does not provide for such schemes as part of the broader insurance market (Devadasan et al. 2004). There are a few basic models of community-based social insurance:³²

• The insurer–agent model, where the NGO/MFI or other founder acts as intermediary between members and the insurer, in India typically one of the public insurance companies (though SEWA has also involved a private commercial insurer). This model has been relied on by large organizations such as SEWA and Buldhana, and smaller ones such as Navsarjan in Gujarat and BAIF in Maharashtra (Acharya and Ranson 2005).
  
  
• The founding organization acts as the direct insurer, but is not the provider of the insured services. This applies both to some NGO schemes (Yeshasvini in Karnataka; Dhan in TN) and occupationally based programs (e.g., Tribhuvandas Foundation).³³
  
  
• The founding organization is the direct insurer and the main provider of the insured services. Examples include ACCORD in TN, Kasturba Hospital scheme in Maharashtra; Students' Health Home in West Bengal.

The different approaches exhibit both common features and differences. They include:

• In terms of types of risks covered, the primary focus has been health insurance. Some schemes cover associated costs such as loss of income, and life and accident insurance.³⁴ However, old-age insurance is not a CBMI product.
  
  
• Nearly all CBMI work on voluntary participation. This approach is consistent with the experience from Asia on coverage of health-insurance schemes for the informal sector (Bernett et al. 1998).
  
  
• In terms of financing, all CBMI schemes require a contribution. However, there is a range in the level of contribution required and the extent to which it is subsidized. In the best cases, contributions cover the bulk of medical claims and administration is covered by donor support (e.g., SEWA Vimo). At the other end of the spectrum, contributions are covered for many of the members from the NGO/funder's own resources. An additional innovation in some schemes is provision for deposit of a lump sum by the participant, from which interest generated covers the annual premium.
  
  
• Although community/NGO involvement controls administrative costs, most schemes rely on an external subsidy. The subsidy may come from Government, the NGO or donors, cooperatives or other sources. This is very typical of rural health-insurance initiatives in developing countries, virtually none of which cover services from contributions only (O'Keefe 2003).
  
  
• The experience on whether benefits are provided on reimbursement or up-front deduction/exemption is mixed. The majority of schemes are based on reimbursement, but those that both insure and provide services are cashless, as is the Yeshasvini scheme in Karnataka, which uses a third-party administrator. Experience internationally suggests that reimbursement-based schemes dominate in community health insurance (Bennett et al. 1998).
  
  
• While most of the schemes are relatively small, the largest have achieved significant membership (e.g., Yeshasvini had around two million farmers registered in 2004–2005, having started in 2003; the West Bengal Students' Health Home, which covers around 1.6 million students).
  
  
• Many of the schemes are relatively new and yet to be evaluated, both in terms of short-term impacts and with respect to financial trajectory.

Of the prevailing models, the insurer–agent approach appears to have the most potential for broadening coverage of social insurance. It appears to combine the benefits of large-scale pooling of risks (both within the membership group when group insurance is purchased, and beyond through the risk pool of the end-insurer), and the cost-reducing benefits of an intermediary organization close to the client. At the same time, the Yeshasvini experience cautions against being very prescriptive on a preferred model.

A further important recent development is the 'micro-pension' product of the Unit Trust of India (UTI). The first client in this partner–agent arrangement was SEWA. In 2006, around 30,000 women joined their DC pension scheme where contributions of around 200 rupees per month are invested in a balanced fund invested in bonds and equities. Individuals must maintain a savings account with SEWA Bank. The SEWA experience may not be easy to replicate on a large scale, however. First, preliminary analysis shows that administrative costs are still relatively high given the flows involved. Second, the arrangement relies on one provider, UTI, which is the only asset manager licensed to offer this kind of product. Nevertheless, the scheme was recently taken up by a dairy cooperative in Bihar and other groups of unorganized sector workers are considering it.³⁵

Looking ahead: new GoI initiatives on unorganized-sector social security

The GoI in its Common Minimum Program has committed itself to expansion of social security. To this end, the National Commission for Enterprises in the Unorganized Sector (NCEUS) produced a report in 2006 on behalf of MoLE, which recommended the introduction of a national social-insurance scheme offering old-age pensions, health insurance and maternity benefits, as well as life and disability insurance.

The NCEUS proposal is ambitious in that it seeks to offer insurance for several major risks to 300 million informal sector workers in a span of five years. The ultimate objective of universal coverage is shared by many developing-country governments as well as donors. However, such an expansion of coverage through voluntary participation is unprecedented in terms of international and Indian experience. The administrative and recordkeeping challenges alone suggest that the proposed time frame is not feasible (O'Keefe and Palacios 2006).

In addition to the scope of the proposal, elements of the design require closer consideration. In the area of old-age pensions, for example, the proposal would pay an indexed monthly pension of Rs.200 to all individuals aged 65 and over that were BPL card holders. Conceptually, the idea of providing a transfer to those that cannot afford to participate in a contributory scheme is sensible and many countries have chosen to implement this type of program.³⁶

On the positive side, for the first time, there is an effort to grapple with the issue of high transactions costs in unorganized-sector social security, through allowing a role for intermediary organizations such as NGO/MFIs, PRIs, unions and worker associations between the state/insurer and workers in raising awareness of the scheme, identifying and registering workers, collecting contributions and payment of benefits. The scheme recognizes the need for a significant contribution subsidy to incentivize participation.

Building on the NCEUS report, the GOI is currently considering various options both in terms of program design as well as sequencing. In some areas, there appears to be growing consensus. For example, there appears to be support for scaling up the NOAPS social-pension program so that it would reach a larger proportion of the elderly in India. There is also an emerging vision that involves combining government subsidies with provision by nongovernmental entities on a competitive basis, perhaps through the partner–agent model. For example, MoLE is studying programs like the JBY scheme for life insurance.

Membership of potential intermediary organizations

One of the key facilitating preconditions for participation in contributory social-insurance schemes is membership of some organization which could play an intermediary role between workers and the ultimate insurer. As a result, membership in potential intermediary organizations is of interest in assessing the institutional scope for SI expansion.³⁷ Figure 12.7 presents survey-based findings on membership at the all-India level. Group membership remains low, and distributional analysis confirms that coverage is concentrated in the upper end of the distribution for all but SHGs.

Most states also have very low membership of organizations, though with standouts. The most significant is Kerala, with high union and coop/SHG membership rates, and higher welfare fund membership. AP also stands out for the high share of workers in SHGs. There are also a number of states with higher than average membership rates of specific types of organization, including West Bengal for unions/associations, and Orissa and Chattisgarh on coops/SHGs.

Figure 12.7 Membership rates in organizations, all workers and by organized/unorganized, 2004 (source: ADB/MOF survey, 2004. Author estimates).

Conclusions and recommendations

While the market failures that suggest a role for public intervention in social insurance for unorganized-sector workers are clear, it is not axiomatic that 'more is better'. Badly designed schemes can have negative impacts for the poor (Gertler 1998). A common problem is health-service-cost escalation if moral-hazard problems among users and providers are not controlled. The losses from moral hazard are higher where the price elasticity of demand for services is significant, which international evidence suggests is the case in developing countries.³⁸

With respect to types of social insurance, there is a tension between household needs, and operational feasibility of rapid expansion coverage for different types of insurance. The profile of household shocks reveals a high demand for income smoothing due to health shocks. However, due to supply-side constraints, moral hazard and adverse selection, as well as the complexities of administering health insurance, this type of scheme is the most difficult to implement. In India, experience with the Universal Health Insurance Scheme illustrates these difficulties. In comparison, life insurance and defined contribution pensions are easier (and with life insurance, cheaper) to design and implement. The contingencies involved, age and death, are easier to monitor and less subject to moral hazard and adverse selection. Unlike health care, the benefits are simple and standardized cash payments. Disability insurance lies somewhere in between, given the need to verify the condition and potential for moral hazard. GoI has implicitly recognized the relative difficulties in its 2007/2008 budget, which seeks expansion of social security starting with life and permanent disability insurance for rural landless workers under the proposed Aam Admi Bima Yogana (though the scheme does not unfortunately have a clear strategy on aggregation of workers through groups to control transactions costs and raise awareness).

In addition, large-scale coverage expansion will require government subsidies to address the question of affordability, and oversight to mitigate fraud and/or mismanagement. However, this does not necessarily suggest a 'top-down' approach whereby new layers of bureaucracy are created to implement schemes. On the other hand, given likely economies of scale, the need for portability of benefits and the exigencies of supervision, the government's role may include creating robust and harmonized record-keeping and payments/collection infrastructure.

The second dimension of sequencing to be considered relates to the target-covered population. In this regard, it is necessary to recognize that some individuals are past the point where insurance or savings instruments are useful. For households with disabled, elderly, widows and already suffering from catastrophic health shocks or disease, the appropriate intervention is to alleviate poverty through direct transfers and other types of assistance. This distinction between those too poor to contribute and others is less obvious in the case of certain approaches to health insurance.

For most households with unorganized-sector workers, insurance is possible, at least with a degree of subsidy. However, voluntary schemes that require individuals to make contributions or pay premia in order to be eligible for coverage may not be easily expanded to the lowest income workers unless they are heavily subsidized. At the same time, experience of NGO and other schemes in India and internationally suggests that some small contribution even for the poorest is an important tool in improving the accountability of schemes to their beneficiaries.

The main alternative to direct public provision and administration is to use existing non-governmental entities and restrict the role of government to: (i) providing targeted subsidies; and (ii) regulating these entities and setting basic standards. This model already exists in India in several forms. In addition, many CBMI schemes could be incorporated under an umbrella program that provided matching contributions or premia but set standards in terms of minimum-benefit targets, eligibility conditions, investment policy and record-keeping, among others.

This coordinated partner–agent approach has recently become more attractive due to financial-sector reforms that have resulted in dynamic asset-management and insurance sectors as well as supervisory agencies that are becoming more experienced. There are several potential advantages to this approach. First, by harnessing existing groups including SHGs, cooperatives and MFIs, transaction costs could be kept low, especially where individual recordkeeping is already taking place. A second advantage would be the promotion of competition on the basis of cost and quality of services. Third, many unorganized-sector workers have no experience with formal financial-sector institutions so that groups can serve as an effective intermediary.

For health, pension, life and disability insurance, a viable plan to extend coverage based on unorganized-sector workers paying premiums or making contribution would need to meet the following criteria:

• long-term financial sustainability;
  
• transparent costing based on actuarial principles;
  
• reasonable incentives to encourage widespread take up on a voluntary basis;
  
• ability to cope with irregular income streams;
  
• harmonization, in terms of incentives and portability, with formal-sector programs;
  
• low transaction costs;
  
• accurate and timely record-keeping;
  
• effective risk pooling (in the case of insurance);
  
• effective asset management (in the case of long-term savings/pensions);
  
• accessibility on a national scale.

Perhaps the greatest challenge for any approach will be that of tracking participants and the financial flows associated with each of them. While universal national registration in India in the near future appears unlikely, alternative solutions could be explored in the meantime. Computerized record management makes any program more transparent and flexible. Programs with systematic personalized record-keeping provide much greater capacity for adding new benefits and improving existing schemes. The access gap can further be reduced by means of various technological innovations.

Cash-collection and disbursement systems in the unorganized sector pose serious challenges. Conventional financial intermediaries are often ill prepared to deal with mass transactions of very small amounts in remote, disbursed and poorly educated communities. Penetration of commercial banks in the rural areas in India remains low. While India's extensive postal system operates in almost every corner of the country, the adequacy of its accounting mechanisms and capacity to assume new tasks remains to be evaluated. Both in India and abroad, a number of interesting and relevant innovations provide lessons for adapting to the needs of both public and private sector in extending the coverage of financial services.

Conclusion

This chapter has reviewed evidence on two tools for promoting consumption smoothing among unorganized-sector households in India. While public works and social security for unorganized workers have considerable potential, this has yet to be realized for the large majority. Moving from the big picture of policy initiatives to the 'nuts and bolts' will remain the biggest challenge in realizing this potential and the benefits it may hold for the poor. At the same time, even well-designed and executed programs will remain only some tools in a much wider array of policy initiatives needed to improve the lot of unorganized-sector workers.

Appendix 1

Major central rural-employment programs

• Jawahar Rozgar Yogana (JRY – from 1989–1999), a CSS targeted to BPL households in rural areas, with preferences for SC/ST and women, and 80:20 financing split between centre and states. The wage/material ratio was 60:40, and DRDA/ZP the main implementing agencies through line departments and GPs.
  
  
• Employment Assurance Scheme (EAS – from 1993–2001), a CSS initially focused on 261 districts, and gradually extended to nationwide rural coverage by 1997. EAS committed to 100 days of lean-season employment for up to two adults per rural family, and was also financed on an 80:20 central and state split. As with JRY, the wage/material ratio was 60:40 and DRDA/ZP the main implementing agencies through line departments and GP.
  
  
• In 1999, JRY was restructured into the Jawahar Gram Samidhi Yogana (JGSY), with the main difference being that creation of demand-driven community infrastructure moved from being a secondary objective to the primary objective, with rural-employment generation conversely becoming the secondary objective.
  
  
• From 2001, the EAS and JGSY were merged into Sampoorna Grameen Rozgar Yogana (SGRY), which had a target of 100 days employment (per rural household). Financing is split 75:25 between the centre and the states, and PRIs have in principle been given an increased implementation role.
  
  
• In late 2004, the National Food for Work Program (NFFWP) was introduced in 150 backward districts, with a massive increase in funding to Rs.5,400 crore in the 2005–2006 budget.
  
  
• The National Rural Employment Guarantee Act (NREG) was passed in August 2005 and implemented from 2006. It initially covers 200 backward districts, with commitment to national rural coverage within five years. This is the first legislated national rural-employment program, and commits to 100 days guaranteed employment per rural household annually. The funding split is 90:10 between centre and states, and PRIs are considered the primary implementing agencies.

Appendix 2

Abbreviations used in this chapter

Document(s) 7 de 16