Table of Contents

Several commentators have commended the Economic Survey in recent years for its rigorous scrutiny of the Indian economy and the innovative use of new data for original research.1 While the chapter under consideration can be characterized along similar lines, it deserves further praise for two other underappreciated virtues. First, there is inherent value in making the internal logic of a UBI explicit by articulating the rationale for such a policy shift (reducing poverty and rationalizing welfare delivery), the instruments at India’s disposal (direct cash transfers through the JAM infrastructure), an awareness of the constraints on policy design (financing and banking infrastructure), and the tactics for bringing it to fruition (gradualism and quasi-universality). The survey’s chapter deeply enriches public debate by providing a focal point around which all discussions of an Indian UBI can revolve. This obviates the need to discuss the idea’s merits and limits by reading the tea leaves of budgetary allocations and official rhetoric.

Second, by establishing a strong normative preference for a UBI and evaluating it under the state imprimatur, the Economic Survey has brought an idea that had lingered just outside the mainstream of economic policy directly in front of agenda-setting state elites. A significant strand of the political science literature, focusing on the role of ideas and reform proposals in political economy (in contrast to institutions and vested interests), has sought to draw attention to the pliancy of the worldviews held by state actors. According to the political scientists Pratap Bhanu Mehta and Michael Walton, the institutional choices nations make and the development pathways they take are defined by “cognitive maps,” which is to say “the underlying interpretation of how the world works, and . . . the range of possibilities for action that an individual or group recognizes” along with negotiated political settlements between different social groups.2 As Mehta and Walton put it, the story of India’s political economy is best told by the policies mobilized by changes in the cognitive maps available to state elites, from the high modernism of the Nehruvian era to the inclusive growth mantras of the late 2000s. As Dani Rodrik writes, innovative ideas upend assumptions about the available menu of policy choices, institutional rules, and resource limitations, and this can “relax political constraints, enabling those in power to make themselves (and possibly the rest of society) better off without undermining their political power.”3

A UBI’s introduction within policy discourse expands the space of the possible.

A UBI’s introduction within policy discourse expands the space of the possible. This is not to say that simply talking and writing about a UBI makes it more likely to be instituted. But if the notion of a UBI and its associated narratives of large-scale poverty reduction and a collective social dividend are used by the Indian state to craft UBI-like policies, the Economic Survey’s exhaustive analysis of its implications will likely have played an outsized role.

Any meaningful analysis of the UBI proposal must, however, grapple with several assumptions and prescriptions that compel deeper scrutiny, and others that warrant active contestation. In particular, these include the semantic oddity of a “quasi-universal basic income,” the survey’s inconsistent treatment of the merits of targeting welfare benefits and India’s flagship antipoverty programs, and its limited imagination when estimating the impact of removing existing welfare schemes. Other important matters to explore are the survey’s insufficient engagement with non-cash-based poverty interventions and its inadequate acknowledgment of the current implementation deficit of the DBT program. As the debates around the idea of an Indian UBI mature, these questions should be first in line for answers.

A Trifecta of Misnomers

The Economic Survey espouses the principle of quasi-universality—targeting 75 percent of India’s eligible population—in a bid to keep costs manageable and make a potential basic income scheme politically feasible. For similar reasons, the suggested transfer would only be an income supplement to push recipients above the 2011–12 poverty line. By one estimate, guaranteeing a truly universal and basic (poverty line equivalent) income would cost close to 13 percent of GDP, more than all of the Indian central government’s expenditures in 2016–17 (12.7 percent of GDP).4 To cite a different calculation, Jawaharlal Nehru University economist Himanshu used updated versions of the C. Rangarajan Committee’s urban and rural poverty lines to estimate the cost of a UBI for all citizens at 24.2 trillion rupees in 2016–17; this amount would be greater than all the 2015–16 tax revenue for India’s central and state governments (23.4 trillion rupees).5 It was, therefore, eminently sensible for the survey to relax the assumptions of universality (the PDS in many Indian states has run under a quasi-universal system) and to provide a sufficient basic income for the purposes of modeling a viable policy alternative. But it is clear that what is left is not universal, basic, or income.6

The Economic Survey essentially proposes targeted unconditional cash transfers, examples of which abound in the developing world. Pakistan’s Benazir Income Support Program, which provides monthly unconditional cash transfers of 1,500 Pakistani rupees to more than 5 million vulnerable households, has been running since 2008.7 Similarly, China has maintained since 2007 a nationwide minimum-income guarantee through its di bao program, which provides unconditional cash supplements to households if their incomes fall below a specified level.8 To mention two other examples, Kenya delivers cash transfers to extremely poor households with orphans and vulnerable children, while Indonesia sent direct cash transfers to poor households affected by price increases in 2005 and 2008.9

In the case of India, three assumptions in the Economic Survey’s proposed annual grant prompt closer attention. First, the survey uses estimates from the 2011–12 Tendulkar poverty line to base its calculations of a monthly transfer. This is no uncontroversial premise. When the Suresh D. Tendulkar Committee released its estimates in 2009, its methodology (particularly its appraisal of the basket of goods consumed by individuals at the poverty line) was widely censured for grossly underestimating poor individuals’ calorie intake and spending on education and healthcare; these assumed values in turn generated particularly low urban and rural poverty estimates.10

Economists, civil activists, and opposition members publicly criticized the Tendulkar poverty line so vociferously that the Planning Commission appointed another expert group, the C. Rangarajan Committee, to reevaluate the Tendulkar formula. Releasing its methodological findings in 2014, the panel ended up revising the urban and rural poverty lines upward to monthly per capita expenditures of 1,407 rupees and 972 rupees respectively (in comparison to the Tendulkar Committee’s estimates of 1,000 rupees and 816 rupees).11 While the Economic Survey acknowledges that the “line is somewhat notional and one must be careful before making a value judgement on the adequacy of the line to measure well-being,” it does not put forward a rationale for not using the Rangarajan poverty lines or the higher fiscal cost for a quasi-UBI that adopting its estimates would have produced.

A second assumption the Economic Survey makes that deserves scrutiny pertains to the relationship between cash transfers and in-kind transfers. Specifically, when cash transfers are intended to substitute for in-kind transfers rather than supplement them, they throw up a host of thorny issues. The survey notes that the NSS 2011–12 data—on which its estimate for an income supplement is based—include the poor’s consumption from the PDS and MGNREGA. However, if this transfer is to be provided after these schemes are withdrawn, would the government not be expected to compensate the poor for the loss of food and wages? The survey disagrees, instead arguing that the efficiency gains from such cash transfers would outweigh the lost benefits from leaky programs, and so “not accounting for replacement would still not seriously affect the costing of UBI. After all, replacing one rupee of the fertilizer subsidy should require a compensating UBI of less than one rupee.”12

When cash transfers are intended to substitute for in-kind transfers rather than supplement them, they throw up a host of thorny issues.

But substitution is an expensive affair and would likely cost more than the Economic Survey suggests. Alternative estimates find that converting the food subsidy into cash transfers would deliver no more than 1,200 rupees annually per capita, whereas such transfers should be at least 2,200 rupees to compensate for the loss of the PDS. The survey’s claim that “not accounting for replacement would still not seriously affect” quasi-UBI estimates only holds “if the existing programmes have zero contribution to welfare.”13 Moreover, two additional variables unaccounted for in the proposed transfer amount are what Reetika Khera labels the transaction and transition costs of moving to cash. The former refers to the expenditures made for access to banking facilities and markets (a particularly acute issue in rural areas), while the latter describes the cost of learning and adapting to a new system of receiving welfare (again, a particularly acute problem among disadvantaged communities). While these costs are difficult to calculate economy-wide, survey respondents often cite them as playing a significant role in their preference for food over cash.14 When beneficiaries cannot be certain that they will receive uniform and periodic cash payments (as is currently the case with the direct cash transfers instituted in select Indian districts), such transfers cannot be termed an income in any meaningful sense of the word. Using the Rangarajan poverty line estimates, providing adequate compensation to offset the loss of consumption provided by programs like the PDS and MGNREGA, and factoring in transition costs would likely raise the needed cash-transfer amount beyond what is suggested in the Economic Survey.

Third, India’s population is so massive that paying for even these modest quasi-universal cash transfers (less than what beneficiaries can expect to receive for one hundred days of labor under the MGNREGA or the minimum wage across Indian states) would require more fiscal space than the country’s ten largest central welfare schemes, which include the Mid-Day Meal scheme, national programs for universal elementary education and rural road building, and a national sanitation drive.15 While the survey marshals evidence to demonstrate the significant misallocation of funds under these schemes, the problem is that its financing strategy takes no prisoners and identifies all such schemes as candidates for replacement.

It is doubtful that disassembling schemes—meant primarily to boost infrastructure, ensure food security for children, incentivize school attendance, and encourage toilet construction and usage—to instead distribute their budgets among citizens would help achieve any of these development goals. Unlike price subsidies for fuel, fertilizer, and electricity (that are used quite similarly to cash), such welfare programs seek to improve long-term human development outcomes and cannot simply be optimized by a cash transfer.16 While the Economic Survey’s assessment can be justified based on the scheme’s immense fiscal burden, it deserves scrutiny for not distinguishing between programs that would be reasonable or unreasonable to roll back. Further, the survey does not explore the possibility of meeting a UBI’s funding requirements through a mix of additional taxation, tax-base expansion, or the privatization of public-sector enterprises.

Given these limitations, to frame the Economic Survey–advocated policy as a UBI is misleading. The proposed “basic income” is a routine, targeted, unconditional cash transfer associated with an anomalous set of underlying assumptions. While the survey anticipates these doubts, it offers little explanation beyond acknowledgments and probabilistic assertions of a quasi-UBI’s efficiency. As Devesh Kapur wrote in 2011, when the foundations of direct cash transfers were being laid, substituting subsidies with cash is simply a change of tactics unless it is accompanied by the development of a larger strategy for food and energy security.17

Targeting Versus Universalism

In contrast to a typical UBI proposal, the Economic Survey’s proposal stops short of advocating true universality. Citing political and fiscal prudence, it recommends two approaches for implementation.

The first involves excluding the top 25 percent of the country’s income distribution—“the obviously rich”—and targeting transfers to the remaining citizens. Such targeting would be done by using asset ownership records and a voluntary opt-out akin to the Give It Up scheme for LPG subsidies, as well as by making beneficiary lists public to invite social sanctions upon the rich, or instituting regular verification procedures.

The second approach entails thinking about a UBI less in terms of a new centrally sponsored scheme and instead moving toward the goal of a basic income incrementally. The Economic Survey suggests a variety of ways of doing this: giving beneficiaries the choice of cash in place of in-kind entitlements under existing programs; only targeting women or easily identifiable and vulnerable groups like the elderly, pregnant women, widows, and the differently abled; diverting to households a portion of the fiscal resources the central government currently earmarks for certain underdeveloped “special category” states; and piloting the program in urban areas first before expanding to rural India.18 The latter set of piecemeal interventions, while intended by the Economic Survey to guide thinking about UBI pilots, can also serve as options for beneficiary selection.

But do such targeting mechanisms, even when seeking to be broadly inclusive, provide the same relief from errors of misallocation and leakage as uniform universal transfers? Are some targeting measures more accurate, cheaper, easier to implement, and more politically feasible than others? Answering questions of targeting efficiency and cost-effectiveness is important because even if the Indian exchequer were to make room for a sufficient, nondistortionary basic income, the method of disbursing said transfers and designing the interface between beneficiaries and government officials would be crucial for successful implementation.

At its core, targeting seeks to reduce poverty by concentrating transfers among a defined eligible population, subject to budget constraints and political considerations, by using an array of identification strategies.

Targeting Methods

At its core, targeting seeks to reduce poverty by concentrating transfers among a defined eligible population, subject to budget constraints and political considerations, by using an array of identification strategies. The main motivation behind targeting is to most efficiently concentrate scarce public resources so that poorer groups receive a high share of social assistance. Targeting methods can be classified into four groups based on the criteria used to determine eligibility:19

  • Individual or indicator targeting: Individual assessments can be performed by using a variety of approaches, including verified, simple, and proxy means tests. Verified means tests involve cross-checking information on household or individual income and/or wealth against independent records, whereas simple means tests involve no verification or household visits. By contrast, proxy means tests use statistical models based on observable indicators correlated with poverty to estimate income or consumption when relevant data are unavailable. The survey’s suggestion of using the Socio-Economic Caste Census as a source for gauging asset ownership is an example of this individualized targeting approach.
  • Categorical targeting: This approach involves using categories like gender, ethnicity, age, or geographical area to determine eligibility. Examples include the Economic Survey’s recommendations of universalizing social pensions and targeting women or the residents of certain states and urban areas.
  • Community-based targeting: This method relies on the local knowledge of community groups like village elders or school councils to identify the poorest households. The survey’s suggested name-and-shame scheme arguably falls under this category.
  • Self-targeting: This approach leverages the cost of program entry to make participation unappealing to the nonpoor and increase uptake by the poor, typically by instituting a work requirement or distributing less desirable types of food and other resources. The survey proposes regular verification measures for those wanting UBI benefits, such that “the rich, whose opportunity cost of time is higher, would not find it worth their while to go through this process and the poor would self-target into the scheme.”20

Each of these targeting methods imposes different burdens on the administrators and beneficiaries involved. The most significant of these burdens, to which the survey devotes significant attention, are the classic implementation errors of the unwarranted exclusion of genuine beneficiaries and the inclusion of nondeserving individuals. But even if one assumes the eligible population in theory receives the designated transfer in practice, the existing literature describes how the process of targeting, alongside its direct costs, can generate problematic negative externalities:21

  • Administrative costs: These costs are borne by the administrative bodies implementing the targeting system. They consist of the outlays on gathering, analyzing, and verifying detailed demographic and income data over time. Such costs rise along with the degree of targeting accuracy—the costs of verified means tests are particularly high given the high information thresholds involved compared to geographic- or age-based targeting.
  • Private costs: These are direct costs all beneficiaries bear while applying for or participating in welfare programs. These range from the cost of obtaining relevant applications and personal information to the opportunity cost of time and wages foregone in queuing, traveling, and even paying bribes.
  • Social costs: These community-borne costs occur when fine targeting cuts through neighborhoods and villages and divides a population into beneficiaries and nonbeneficiaries. This includes the stigma of being branded as the poor (or incorrectly recognized as the nonpoor), which can hurt participants’ self-image and self-esteem and reduce community cohesion.
  • Incentive costs: These costs arise when the design of eligibility criteria encourage individuals or households to modify their behavior (such as work effort, family size, and/or migration choices) to qualify for benefits.
  • Political costs: These costs take hold if and when finer targeting leads to less political support for, and even increased popular disapproval of, such programs, as large parts of the population—especially middle-class beneficiaries—stand to be excluded from receiving benefits.

Targeted social transfers became prominent within development policy circles in the 1980s, following the economic crises of the 1970s and concomitant ideological shifts in favor of neoliberal policies and structural-adjustment programs and away from universalistic policies.22 By 1990, the World Bank was stating that a “comprehensive approach to poverty reduction, therefore, calls for a program of well-targeted transfers and safety nets as an essential complement to the basic strategy.”23

Targeted social transfers became prominent within development policy circles in the 1980s, following the economic crises of the 1970s and concomitant ideological shifts in favor of neoliberal policies.

As targeting programs and their evaluations have proliferated in the decades since, this deep and growing literature can help Indian policymakers mulling targeted unconditional transfers assess the suitability of such targeting instruments. This body of research finds that methods of beneficiary identification like proxy means testing or community-based targeting vary quite widely in their impact on reducing exclusion or broadening coverage. While they may result in budgetary savings, these savings can be offset by high administrative, private, and social costs. However, a more inclusive approach relying on relatively cruder methods like untargeted, uniform transfers to all beneficiaries or simple categories has delivered results comparable to, and it seems in some cases even better than, more sophisticated methodologies. From the options outlined in the Economic Survey, those that come closest to this latter set of polices present a promising avenue for further investigation.

A series of multinational literature reviews reveal that the performance of targeted antipoverty programs is highly sensitive to policy design and information constraints. Successful examples, while not infrequent, are difficult to generalize.

A seminal 2004 study by economists David Coady, Margaret Grosh, and John Hoddinott examined 122 such interventions (40 percent of them cash-transfer programs) implemented across forty-eight middle- and low-income countries between 1985 and 2000.24 The median targeted program funneled approximately 25 percent more resources to the poor than untargeted programs did, but there was a striking degree of variation across developmental contexts. Fully one-quarter of all programs left the poor worse off; in each of these cases, a random distribution of benefits would have yielded better outcomes. No single method was predominant: 80 percent of the disparity in targeting outcomes was due to variations in program design within a given targeting mechanism rather than based on which targeting method was used. Implementation played a decisive role. Targeting results improved in settings with higher income levels (correlated with a capacity to design and implement better targeting interventions), with greater inequality (which made it easier to distinguish between differing income groups), and with higher government accountability, along with the use of more than one targeting method at a time.

Other reviews have sought to define general rules for the use of certain targeting methods. In 2009, Rachel Slater and John Farrington of the Overseas Development Institute reviewed forty-nine social transfer programs in low-income countries and found that, while no targeting mechanism stands out across the board, income-based methods like means testing are demanding both in terms of the administrative capacity required and the associated costs of data collection and verification. Targeting social categories, then, offers a useful alternative that enjoys high political and community-level buy-in. The authors warn, however, that unintended results are pervasive; structural forces like a “weak information base, and poor targeting decisions, may mean that the value for money that targeting generates is below its potential.”25

Looking to pick up from where Coady, Grosh, and Hoddinott left off, Stephen Devereaux and his coauthors at the Institute of Development Studies undertook a 2015 study on targeting accuracy and cost-effectiveness.26 To avoid overlap, the assessment only included studies published or using data collected after the year 2000—a total of eighty-five studies from forty-one developing countries. While the authors, unsurprisingly, found that no “best” method exists and that the effectiveness of any particular mechanism is tied wholly to context, a set of guiding principles did emerge.

They found that means testing is expensive and yields high errors of inclusion and exclusion, while proxy means tests vary depending on how well the indicators used correlate with income or consumption. Meanwhile, categorical targeting does well at identifying and reaching the eligible population, but does worse at identifying the poor if they do not fall into defined categories. Much the same goes for geographical targeting—it is efficient if poverty is spatially concentrated, but otherwise errors of inclusion and especially of exclusion persist. Community-based targeting requires more administrative legwork to guard against elite capture and calls for high social cohesion, but once these conditions are satisfied the mobilization of local knowledge can minimize errors, limit costs, and improve program acceptability among stakeholders. Finally, self-targeting, when instituted in high-poverty settings, may be rendered inadequate by high demand (such as in the case of public works programs) and fall prey to high exclusion errors.

A range of method-specific reviews reiterate how targeting performance is ultimately mediated by program objectives, design, and implementation.

A range of method-specific reviews reiterate how targeting performance is ultimately mediated by program objectives, design, and implementation. A thorough examination of the empirical and theoretical literature on community-based testing finds that rent-seeking tendencies among local elites may override gains from local knowledge and social capital; moreover, communities’ preferences vary between being pro-poor and being expressly inegalitarian, and intended targeting outcomes may be undermined by communities gaming the system in response to funding and evaluation criteria.27 A 2012 paper on the effectiveness of three social cash-transfer programs in Kenya, Malawi, and Mozambique—each of which used some level of community-based testing—found that all the programs performed better than the mean score for programs in the Coady review. That said, qualitative surveys found some evidence that perceptions of fairness varied widely and generated tensions around unclear eligibility criteria, program design, and exclusion errors.28

As for the use of proxy means tests in developing countries like Bangladesh, Indonesia, Rwanda, and Sri Lanka, the evidence indicates faults in theory and practice: built-in design errors where the statistical methodology used to predict household incomes is flawed, and implementation issues arising from the use of static household surveys that grow rapidly out of date. Such targeting errors frequently impose social costs by exacerbating tensions between beneficiaries and nonbeneficiaries, and these inaccuracies limit national program budgets when they exclude large sections of the populations from the benefits, resulting in smaller transfers to deserving households.29

As a corrective to these targeting pitfalls, recent papers by economists Caitlin Brown, Martin Ravallion, and Dominique van de Walle demonstrate the benefits of simplifying beneficiary identification using simple categorical targeting or even a universal basic income.30 Using survey data from nine countries in sub-Saharan Africa, the authors assessed the impact of proxy means tests on poverty reduction, for a given budget set at the country’s aggregate poverty gap, against two counterfactuals: 1) uniform, untargeted transfers and 2) targeted transfers to households with different categories of people such as the elderly, the differently abled, and children.

In countries where undernutrition and food insecurity is highly prevalent, approaches that rely on universal or near-universal nutritional interventions should be adopted over household-targeted interventions.

Two key results emerge from this analysis. First, none of these approaches is a panacea: no method does better than reducing the baseline headcount index of poverty from 20 percent to 15 percent. Second, all three targeting methods are locked in a dead heat: while the best-performing variant of a proxy means test (a poverty quantile regression) brought the mean headcount index to 15.4 percent, both a UBI and targeted transfers to the elderly, widows, the differently abled, and children reduced it to 17.1 percent respectively. To be sure, that is not a negligible gap in performance. In the words of World Bank economist Berk Özler, “in a country of 25 million people, such as Cameroon, reducing the Headcount Index from 17.1 percent to 15.4 percent allows close to half a million people [to] escape poverty.”31 But the appeal of uniform or loosely targeted transfers grows substantially considering the implementation lags of a proxy means test; its exclusion errors (sizable, as the study finds); and its high administrative, social, and political costs.

The advantages of broad coverage are found in comparison to targeting particular geographic areas or even targeting poor households. In a 2007 study, Özler and his coauthors estimated the impact of transferring a predetermined budget to geographically defined subpopulations in Cambodia, Ecuador, and Madagascar using “poverty maps.” They found that while geographically based transfers can substantially lower the poverty rate (and yield large savings), uniform transfers or simpler geographic targeting also performed well when the available budget and poverty line were both comparatively low.32 A 2017 paper by Brown, Ravallion, and van de Walle questions the assumption that poor individuals (identified by their nutritional status) are found in poor households. Using data for thirty sub-Saharan countries, the paper finds that the poorest 20 percent of households are where approximately only 25 percent of underweight women and undernourished children reside. This suggests that in countries where undernutrition and food insecurity is highly prevalent, approaches that rely on universal or near-universal nutritional interventions should be adopted over household-targeted interventions.33

Returning to India, there are empirical echoes of these findings in the literature on targeting within the country. In a 2016 study, World Bank economist Rinku Murgai along with Ravallion and van de Walle studied the cost-effectiveness of the MGNREGA in Bihar using two rounds of survey data from 150 villages.34 They found that after factoring in the scheme’s costs (40 percent of the total budget was devoted to administrative costs), the program reduced the state’s poverty rate less than a basic income scheme would have by simply distributing its budget among every (rich or poor) household. This result held even when the simulation accounted for a leakage of 10 percent, and estimated the impact of transferring the budget only to households verified to be below the poverty line. That said, the margin of poverty reduction was the same as in the previous case, a fact that underscores the conventional wisdom that below-poverty-line targeting has limited efficiency.

In India, it is next to impossible to verify incomes given the country’s pervasive informal economic sector and large poor population. The practice of targeting households below the poverty line for welfare benefits using a proxy means test has a long history in India—four censuses were held respectively in 1992, 1997, 2002, and 2011. The methodologies designed for households below the poverty line have been criticized in terms of survey design, methodological inaccuracies, data quality, and policy relevance and implementation.35

In 2010, Khera and Drèze examined the impacts of employing simple inclusion and exclusion criteria to construct a list of households eligible for social assistance. One of the main methods they considered was an “exclusion approach,” whereby all households that met simple exclusion criteria (such as owning assets like cars or televisions, amenities like piped water and electricity, and durable housing) would be removed from the eligibility list. In contrast, an “inclusion approach” would involve selecting all households that met any inclusion criteria: Scheduled Caste or Scheduled Tribe households, landless households, households with no adult educated beyond grade five, households headed by single women, and households where at least one member worked as an agricultural laborer.36 According to the authors, the exclusion approach “can be described as a quasi-universal system, that is, universal except for a slab of privileged households.” This is quite similar to the Economic Survey’s intent of “approaching targeting from an exclusion of the non-deserving perspective.”37 Using data from the National Family Health Survey 2005–06 for rural India, Khera and Drèze’s analysis showed that, while all these alternative approaches (four in total, depending on the strictness of the inclusion or exclusion criteria) do well at including the poorest fifth of households on the eligibility list, the proposed exclusion approach performed best at minimizing the proportion of the richest fifth of households included on the list while maximizing the share of the former group.38

If India must rule out a truly universal basic income due to fiscal or political constraints, the evidence suggests that targeting based on simpler and fewer criteria, not complicated scoring techniques, deserves further exploration.

If India must rule out a truly universal basic income due to fiscal or political constraints, the aforementioned evidence suggests that targeting based on simpler and fewer criteria, not complicated scoring techniques, deserves further exploration. A 2013 study buttresses findings in favor of modest targeting rules.39 Examining the use of proxy means tests for allocating cards designating households below the poverty line in rural Karnataka, the authors found that expanding the eligibility criteria increases the likelihood of manipulation by corruptible officials, if enforcement is weak and the government officials tasked with determining eligibility do not have strictly pro-poor preferences. Through a survey of over 14,000 households, the authors found that 70 percent of ineligible households managed to secure a card indicating they were below the poverty line, while 13 percent of those eligible did not. Small bribes were frequent. They also estimated an economic model to compare a targeted PDS to a universal PDS; given evidence of weak enforcement, they infer that pro-poor administrators would likely prefer universal eligibility to targeting.

There is no easy way to assess the competing targeting methods of a large-scale unconditional cash transfer. But the vast and heterogeneous literature on the subject does allow for a weak ranking of the Economic Survey’s recommendations on the basis of cost and accuracy.

As a starting point, there is significant evidence from Indian and international contexts that a proxy means test based on household surveys like the Socio-Economic Caste Census allows room for corruption and produces high inclusion and exclusion errors by using outdated data and arbitrary methodologies. Though policymakers and administrators are familiar with this approach, the deficiencies associated with its real world implementation and high administrative costs make it an unappealing option.

Meanwhile, relying on community sanctions to deter the well-off from accepting transfers—while a cheaper alternative—could backfire in settings with weak accountability mechanisms and may even aggravate social divisions. According to official statements, more than 12 million Indian LPG consumers (out of a total of 200 million) have voluntarily foregone their subsidy following the Give It Up campaign.40 But it is not clear if enough of the country’s well-off would opt out of a similar basic income transfer. As a corollary from another geographic setting, current efforts in Iran to restrict beneficiaries reinforce the difficulty of withdrawing subsidies. The Iranian parliament instructed the government to halt cash transfers to one-third of the population (comprising government officials, recipients of alternative welfare benefits, and armed service members) in April 2016 in response to a growing fiscal burden, but only about 9 percent of subsidy recipients were struck off the rolls as of January 2017.41

The likelihood of social and/or incentive costs is also high for self-targeting methods, where cash transfers would either be inordinately small or require passing a series of bureaucratic hurdles and likely subvert program objectives in the process. Giving the beneficiaries of existing programs the choice to switch between in-kind benefits and cash may in principle protect them from a cumbersome transition to a new system, but doing so also, as the survey acknowledges, runs the risk of reinforcing existing inefficiencies.42 Similarly, targeting on the basis of geographical units like urban areas does little by itself to improve program design and functioning if the transfers are to be routed through existing delivery mechanisms. Beyond determining the size of a given eligible population based on transparent and verifiable indicators—a valuable service—much the same is true for categorical targeting approaches.43

Despite these various limitations, the survey’s suggestion to introduce a UBI by starting with specific social groups—like women, the elderly, widows, and the differently abled—is highly likely to improve targeting accuracy and result in relatively low incentive, social, and administrative costs.44 As Reetika Khera has written, this approach combines three key benefits. First, these are easily identifiable populations that do not require extensive means testing to determine their eligibility. Second, taken together, maternity benefits and universal social pensions cost an affordable 1.5 percent of GDP. Third, these interventions have been the subject of extensive evaluations that show that they reach the intended populations and have a positive impact on poverty reduction.45

The counterpoint is that selecting demographic categories only solves the problem of who to target, not how to target. Several design features require deeper exploration. The survey does not consider using multiple targeting interventions at a time, like demographic targeting only for rural households, a common practice that has produced good results, or universalizing coverage within a specific geographic unit like poorer districts or states.46 It also does not take into account a bare-bones method of indicator targeting like a poverty or demographic scorecard.47

While there is no single optimal targeting mechanism, all future efforts at identifying the right approach should seek to rigorously examine the trade-offs between targeting accuracy and its myriad associated costs.

While there is no single optimal targeting mechanism, all future efforts at identifying the right approach should seek to rigorously examine the trade-offs between targeting accuracy and its myriad associated costs, navigate the tension between minimizing leakages and avoiding beneficiary exclusion, and balance the use of sophisticated methodologies against the capacity of service-delivery agencies.

The Politics of UBI: Abroad and in India

Beyond ensuring a rigorous and clear policy design and sufficient organizational capacity for implementation, policymakers must also consider the political feasibility of different transfer mechanisms. While empirical research on the political economy of redistribution is limited compared to that on targeting efficiency, economic models of targeted transfers that build in the impact of politics nevertheless yield three broad, deeply relevant insights.

First, when policymakers use targeting for efficient cash transfers, they should not assume that budgets will remain fixed through the life of a program. After all, externally imposed budgetary constraints are ultimately expressions of political priorities, and program budgets almost invariably tend to shrink after targeting is initiated.48 As economists Jonah Gelbach and Lant Pritchett showed in an economic model where transfer budgets are determined by majority voting, forging ahead with targeting while assuming budgets will remain fixed produces the worst possible outcome for the poor. This is because the rich and middle class eventually tend to choose to limit taxation rates, which in turn shrinks the budget available for redistribution.49 To maximize utility for the poor, it is best to do away with targeting altogether and provide a uniform transfer. Other scholars have inferred a “paradox of redistribution,” observing: “The more we target benefits at the poor only and the more concerned we are with creating equality via equal public transfers to all, the less likely we are to reduce poverty and inequality.”50

Second, political support for targeted programs depends largely on the priorities of powerful constituencies, since the poor, by themselves, may be unable to generate widespread political consensus.51 Yet universal schemes can build broad coalitions across income classes. This, in turn, increases political rewards for politicians as well as the quality of program implementation and the size of transfers. Some amount of leakage, in such frameworks, may be preferable to compensate middle-income constituencies for their political support.52

Finally, the degree of political support for particular targeted programs and the extent of redistributive policies pursued can be influenced through a variety of alternative mechanisms. These include concerns about the fairness or effectiveness of a particular program (arising from corruption or preferential treatment), differing attitudes about the cause of poverty (if seen as a product of individual failure, transfers will likely be perceived as unfair), the rate of upward mobility in society (if perceived to be high, today’s poor are unlikely to support future redistribution), the extent of ethnic or religious divisions (the more they are deeply entrenched, the less likely a large coalition will call for universal transfers), and the propensity of a large middle class to capture the benefits of redistribution.53

For example, a 2009 attitudinal survey asked more than 1,300 Zambian respondents to choose between offering universal benefits for all children, all elderly, and all differently abled; targeting the extremely poor; and targeting no one. In an apparent contrast with the prescriptions of political-economy models of targeting, the survey found that most respondents preferred targeting benefits to the extremely poor than more universal solutions. It attributes these views to many aforementioned factors, such as “voters’ attitudes towards the poor, their understanding of social justice, the level of cohesion in society, the degree to which a program is perceived as procedurally fair and effective, as well as . . . whether a program is designed from scratch or has already been in existence.”54

In India, the political appeal of a UBI is difficult to gauge because there is little consensus on its final form. Much depends on which welfare programs it would replace (a move likely opposed by the poor) or which nonmerit subsidies and corporate tax exemptions would be axed (a move likely opposed by the middle class and business groups). Expenditures on implicit subsidies have declined in recent years, as have corporate tax exemptions; these declines indicate that fiscal room for funding a UBI from these sources is rapidly diminishing. Barring the imposition of new taxes, the budget for such a program would likely have to come at the cost of existing schemes.55

Statements by Finance Minister Arun Jaitley following the release of the 2017–18 national budget make two political realities certain.56 First, a UBI could only be introduced when existing subsidy programs are swapped out, and when legislators do not demand that existing subsidies continue alongside such a UBI. In response to questions about a universal basic income in parliament, the Indian Ministry of Finance has stated that it has no plans to institute such a scheme.57 Arvind Subramanian echoed this rationale for a UBI’s political infeasibility, noting the difficulty of rolling back existing schemes and stating that the government’s finances would “go bust” if it were to be added on.58

Second, Finance Minister Jaitley contended that a basic income program must be targeted only to the poorest households. Such statements demonstrate that a UBI in India would be plagued by patterns of politics and development that militate against a redistributive reform of this scale and suppress its radical potential. A UBI would have to contend with a lack of redistributive pressures for universal benefits, as increasingly intense electoral competition makes channeling benefits to narrow constituencies more attractive for politicians. It would also need to grapple with a lack of redistributive capacities, given that the combination of India’s weak subnational bureaucratic capabilities and greater fiscal capacity to fund targeted transfer schemes makes universal safety net programs less feasible.59

Further, the passing of justiciable rights to food, employment, information, and education under the erstwhile Congress-led alliance created a resilient legislative framework that has survived the subsequent change in government more or less intact.60 The Modi government’s drive to institute DBT across several welfare programs under these pieces of legislation, however, has blurred the distinction between the use of targeted cash transfers as a replacement for in-kind benefits and a UBI’s welfare-enhancing potential for plugging gaps in social protection. This, consequently, has weakened the case for the latter.

Yet there is cause for cautious optimism. In a recent discussion at the Center for Global Development, Chief Economic Adviser Subramanian argued that “in principle, nothing prevents a state government from doing [a UBI] on its own.”61 One path to making this happen, he suggested, would be for a state with a reasonably efficient infrastructure for identifying and reaching the poor to ask the central government for what is in essence an unconditional cash transfer, that is to say federal funds not tied to the implementation of any particular scheme. It is likely that Jammu and Kashmir will be the first Indian state to answer Subramanian’s call. According to news reports, it has asked for the freedom to spend its share of funds devoted to central schemes, and the state’s finance minister, Haseeb Drabu, has made presentations on implementing statewide direct cash transfers to state residents below the poverty line to Finance Minister Jaitley, and to the prime minister’s office with Jaitley’s support.62 The government of Telangana is reportedly interested in a similar program.63 Several Indian states have moved faster than the central government to implement social policy reforms, expand welfare coverage, increase public expenditure, and attempt policy innovations. Scholars differ on what is driving these changes. In any case, the space for a UBI is far more likely to be found in one of India’s many diverse regional political economies.64

But any hope of bringing such initiatives to fruition must guard against two dangers. First, there is the risk of reifying existing inequities by implementing such programs in states that already have a high administrative capacity rather than in states with less progressive governance machineries that nonetheless have a greater need for a universal social safety net. Second, the rush toward implementation may obscure the fact that policy proposals in the guise of a UBI, including that of the government of Jammu and Kashmir, may introduce little more than targeted cash transfers in lieu of in-kind benefits under existing welfare schemes.

None of the aforementioned design principles or political dynamics should be construed as promoting a sense of policy fatalism about an Indian UBI. Indeed, Pranab Bardhan, a key UBI proponent, has recommended that political mobilization start with a coalition of informal workers’ unions and organized-sector unions in support of the idea. But if UBI promoters want to move one step closer to policy formulation, and if UBI detractors wish to authoritatively dismiss the validity of the idea, they need to mobilize public opinion based on actionable evidence on program performance.

This is the Achilles’ heel of India’s UBI debate, and it evokes the inevitable question: Where is India’s UBI pilot?65 around the implementation design and political economy of a UBI. Such an effort is unlikely to shed light on many fundamental questions that a UBI gives rise to—the reorganization of economic and political power throughout Indian society, and its effect on India’s long-term growth prospects, for example. But it can offer new answers to the growing challenges of poverty and inequality. Variations of the experiment could also test the impact of more effective targeting approaches (including tests of eligibility), so that the best version of the existing welfare system serves as a benchmark for a pure UBI.

General Principles for Piloting an Indian Basic Income

The Big Picture

  • A new basic income pilot in India must seek to significantly advance knowledge about how to implement such a program in a large developing country. An ambitious experiment could generate new evidence on the impact of large-scale, unconditional transfers on state capacity and the ability of such transfers to alleviate poverty relative to that of India’s existing welfare schemes.
  • Unlike past experiments in India, this pilot should be a long-term study implemented by state authorities. Monitoring and evaluation, however, must be the responsibility of an external organization with sufficient competence and experience in running large-scale experimental evaluations.
  • Experiment designers must ensure robust ethical standards for participants—such as ensuring voluntary participation, shielding beneficiaries from potential economic harm, and defining strict privacy protections. Data should be made anonymous on an individual level to communicate trends in aggregate and must be released to the public at regular intervals to maintain transparency.

What to Test

For beneficiaries, a pilot must determine the impact of unconditional transfers on:

  • Household and individual finances: changes to personal finances and consumption patterns, ranging from food and medical expenses to investments in productive assets
  • Access costs: gauge financial and time costs that participants incur as they apply for such a program, have their enrollment information verified, travel, and collect cash transfers
  • Labor market: effects on wages for both informal- or formal-sector workers, labor supply, and entrepreneurship
  • Healthcare: the incidence of sickness or injury, mental health outcomes, and visits to private and public healthcare providers
  • Food security: changes to nutritional profiles and dietary diversity as evidenced by spending on different food groups (such as cereals, fish, and eggs) and so-called temptation goods like alcohol and tobacco
  • Individual preferences, aspirations, and anxieties: beneficiary views on program performance and transfer sufficiency relative to existing welfare services, as well as their opinions on their current economic circumstances and future economic and social mobility
  • Community outcomes: changes to social cohesion in a given community; changes to aggregate economic indicators like inflation, market wages, and productivity; and disaggregating the above effects by gender, caste, and income level

For the implementing agency, a pilot must evaluate:

  • Implementation quality: the proportion of beneficiaries that verifiably receive transfers of the intended size and at the expected time, and the performance of a mechanism for redressing grievances
  • Implementation costs: the expenses associated with financing transfers, as well as the administrative costs of targeting, identifying, and authenticating beneficiaries
  • Targeting performance: for variants of the experiment ranging from truly universal transfers to those conditioned on strict eligibility criteria, how well a particular methodology can minimize inclusion and exclusion errors and maximize coverage of deserving beneficiaries
  • Transfer size: the degree to which different payout amounts successfully meet beneficiary expectations, or at least avoid causing a net reduction in well-being
  • Duration and frequency: the differential impact of larger, lump-sum payments paid annually or biannually relative to that of smaller, monthly payments
  • Payment channel: the reliability of (non–Aadhaar linked) electronic bank transfers, mobile money, and other digital payment mechanisms relative to Aadhaar-based payments

Next Steps

  1. Naming a state government agency—or several, with the central government playing a coordinating role—to initiate pilots across administrative districts or blocks with varying socioeconomic conditions and degrees of financial inclusion.
  2. Identifying scholars, research organizations, and evaluation agencies to collaborate on pilot design with government officials.
  3. Releasing a public consultation paper seeking comments on pilot design from potential beneficiaries and citizens, policy experts and scholars, government authorities, international and nongovernmental organizations, and members of civil society.
  4. Conducting a preliminary study in the target geographical area to establish baseline indicators for all measures of interest.

The State of India’s Existing Welfare Programs

If a UBI is to compete with India’s existing social welfare programs, it is necessary to understand their targeting effectiveness and trends in institutional performance. While the MGNREGA and PDS are key protagonists in the Economic Survey’s story about fiscal misallocation and the exclusion of deserving beneficiaries, whether these programs are heroes or villains remains unclear. By the survey’s own admission, both schemes have improved significantly of late in expanding their coverage of deserving beneficiaries. The Economic Survey cites two studies—a 2016 survey of 3,600 households across six Indian states that found beneficiaries received an average of 92 percent of their entitled PDS food grains, and another study that estimated PDS leakages shrank from 54 percent in 2004 to 34.6 percent in 2011—and goes on to state that extrapolating from the latter study’s results indicates that the overall leakage for the PDS throughout India may have fallen further to 20.8 percent.66

These trends are echoed in the broader literature on PDS performance. A 2015 literature review described many studies that found an improvement in PDS functioning across several Indian states since 2004.67 Estimates from two rounds of the India Human Development Survey (IHDS) found a decline in the proportion of grain that did not reach beneficiaries from 49 percent to 32 percent between 2004–05 and 2011–12, a national trend buoyed by extensive PDS reforms in Bihar, Chattisgarh, and Odisha.68 In 2016, a NITI Aayog study on the PDS using IHDS data found that targeting efficiency has progressed over time, with a decline in exclusion errors from 54.9 percent to 41.4 percent between 2004–05 and 2011–12 through expanded coverage.69 Another 2015 evaluation of the PDS by the National Council for Applied Economic Research also noted performance improvements, and the lowest likelihood of leakage was observed among the poorest of households. Bihar, Chattisgarh, and Karnataka emerged as high performers, although a significant share of food grain allocated to households above the poverty line in states like Assam, West Bengal, and Uttar Pradesh was diverted.70

To be sure, a common thread in this literature is that problems—such as the shortfall in beneficiaries’ entitlements, patchy grievance-redressal mechanisms, cross-state discrepancies in the price of key food grains, and still-substantial leakages—remain significant. But as the PDS has grown increasingly progressive and expansive in its coverage over the years, and in its established positive impact on household welfare, several states have found that implementing reforms in the areas of administration, distribution, information, and identification bears fruit.71

The story of the National Rural Employment Guarantee Scheme’s targeting efficiency is similar. The Economic Survey notes that the scheme has made several improvements in terms of technology and program design since 2014, including the geotagging of public assets and the digitization of job cards. Even after accounting for what a 2016 literature review termed the “‘third law of [MGNREGA]’: [namely that] every result has an equal and opposite result,” research on the scheme’s self-targeting mechanism has found that it improved pro-poor access significantly.72

This efficiency in targeting, however, was tempered by a relative inability to meet the high demand for work, a shortcoming attributed largely to funding constraints and the limits of local administrative capacity.73 IHDS survey data from 2011–12 revealed that 30 percent of poor rural households participate in the MGNREGA relative to 21 percent of the nonpoor, while 30 percent of households with no literate adult take part in the scheme compared to 13 percent of households with at least one adult college graduate.74 The data also demonstrated that poverty among the scheme’s beneficiaries in 2011–12 fell by 6.7 percentage points because of consumption facilitated by MGNREGA.75 A 2015 paper on MGNREGA’s targeting accuracy found that nonpoor households were more likely to receive work, although there was evidence of a fall in the rationing rate and a meaningful increase in the probability of Scheduled Tribe households and marginal farmers getting work, which seemed to indicate that the scheme’s administration had improved over time.76 Meanwhile, a 2014 World Bank study, using NSS 2009–10 data, found that demand for work under the scheme is higher in India’s poorer states and among poorer families, including the Scheduled Castes, Scheduled Tribes, and Other Backward Classes, as are participation rates in the scheme for the same groups.77 Further, while targeting efficiency differed across states, it improved with an increase in the overall participation rate. The program did not fare as well in the provision of guaranteed work—the research indicated that unmet demand was the single biggest factor constraining the scheme’s impact on poverty reduction.

A few other studies also contain findings of relevance to the MGNREGA’s targeting performance. A 2013 study of the program’s targeting in the states of Madhya Pradesh and Tamil Nadu stated that “several correlates of poverty (for example, illiteracy, affiliation to disadvantaged groups such as Scheduled Castes and Scheduled Tribes, and landlessness) are associated with higher probabilities of participation,” although high MGNREGA wages relative to agricultural wages meant that large numbers of the nonpoor also self-selected themselves into the scheme.78 Another 2013 paper, using 2009–10 NSS data, discovered that self-targeting made it possible for poorer, Scheduled Caste and Scheduled Tribe households to participate in the MGNREGA at higher rates nationally. When cross-state data were examined, however, approximately half of twenty-seven states demonstrated pro-poor targeting, while in the other half the negative impact of administrative rationing (the denial or restriction of work to beneficiaries seeking assistance) dominated.79

The evidence assembled above on these two flagship programs can be interpreted in many ways. One inference is that pockets of targeting efficiency do exist, and that the observed fall in corruption levels is likely to continue with renewed political will, increased awareness, and administrative reforms. But this literature can also be read as sufficient confirmation that the way forward is to dismantle both the schemes and the perverse political economies surrounding them.

Discussions of targeting efficiency also obscure the larger question about cash transfers in comparison to food aid or public-works employment, given that households will likely have to bear the brunt of volatile commodity prices and wages.

Given the significant variation in these schemes’ performances, it may not be a bad thing in some parts of India to deeply restructure faulty systems of basic public service delivery and even scrap those broken beyond repair. But winding down key components of India’s social protection machinery while granular data are still being assembled on what works in in-kind aid programs, and as reformist states undertake active policy experimentation (including the substitution of cash for certain in-kind benefits), goes against the grain of evidence-based policymaking. Doing away with these social welfare programs would be entirely premature when no district administration, let alone a state, has tested the impact of making cash grants the exclusive component of welfare.80

Discussions of targeting efficiency also obscure the larger question about the appropriate role of cash transfers in comparison to food aid or public-works employment, especially given the high likelihood that households will have to bear the brunt of volatile commodity prices and wages.81 And if focused poverty reduction is the ultimate goal of instituting such programs, then the set of policy options goes beyond traditional welfare programs comprising cash or in-kind transfers or a universal basic income. For example, the all-of-the-above “graduation approach” pioneered by the Bangladesh-based nonprofit BRAC used a version of community-based targeting to identify the poorest households in a village. Over a two-year period, it provided participants with an income-generating asset like livestock and training to boost revenue, weekly coaching visits, consumption support in the form of cash or food, a savings account, and basic information on healthcare. A randomized controlled trial conducted across 11,000 households in six countries found that the program substantially improved household consumption levels up to a year after its conclusion. And it was cost-effective: every dollar spent on the program in India generated long-term benefits worth $4.33 for ultra-poor households.82 The siren song of a UBI should thus be heard as a call for subnationally administered pilots that generate comparable data and provide a more substantive rationale to assess its suitability for India writ large.

UBI: A Logical Extension of Direct Benefit Transfer?

The Economic Survey makes it clear that “the success of the UBI hinges on the success of JAM”—the delivery of government benefits using Aadhaar-linked bank accounts and authentication systems.83 Since one of the survey’s recommendations for phasing in a UBI involves introducing it in urban areas first, it is instructive to examine the government’s experience with pilot projects, starting in 2015, to replace PDS-provided food grain with a DBT system of cash transfers in the union territories of Chandigarh, Puducherry, and urban parts of Dadra and Nagar Haveli.84 Noting how the DBT system was initiated in Puducherry, then temporarily discontinued after two months due to difficulties in implementation and public opposition before being restarted, the survey acknowledges the magnitude of the task. It calls such undertakings a “cautionary tale” and states that “independent evaluations emphasize the need for an improved digital financial infrastructure, even in these relatively urban settings.”85

But complications with administering DBT may run deeper than the Economic Survey suggests. Two new studies examine the results of these ambitious pilots and the DBT’s long-term potential. The first was an evaluation conducted by the South Asia office of the Abdul Latif Jameel Poverty Action Lab (J-PAL), with the Development Monitoring and Evaluation Office of the NITI Aayog and the Department of Food and Public Distribution; it comprised three rounds of household surveys in all three union territories between January 2016 and March 2017.86 This study examined the quality of policy implementation, the sufficiency of the cash-transfer amount, and shifts in beneficiary attitudes toward the scheme. It did not analyze the nutritional impact of these transfers nor the exclusion of genuine beneficiaries from the scheme. The second, an Indian Council for Research on International Economic Relations (ICRIER) working paper, made the case for a phased nationwide rollout of DBT for food over a five-year period. Based on the international experience with cash transfers and the Chandigarh and Puducherry pilots, the paper analyzed the ability of Indian states to shift away from food-grain distribution to cash and, perhaps most notably, stated that the road to implementing a universal basic income in India runs through the DBT.87

Both of the aforementioned papers found significant room for improvement in last-mile delivery, the size of the subsidy and associated private costs borne by beneficiaries, and grievance redressal for beneficiaries. The J-PAL South Asia evaluation found that the average share of beneficiaries verifiably receiving DBT as intended improved over the course of the three survey rounds, though it topped out at 78 percent as of March 2017. While the average proportion of transfer recipients who reported not receiving DBT declined over time, their relatively high share among the total population was at odds with the government’s records, which noted a failure rate of less than 1 percent. The reason for this is unclear. While the authors appear to rule out leakages, they suggest that these discrepancies likely arose from insufficient and irregular updates from implementing authorities through text messages (mostly in English) or database errors. Any of these factors may have prevented beneficiaries from receiving transfers.

Meanwhile, the ICRIER study found that as of May 2017 approximately 7 percent of eligible beneficiaries in Puducherry were not receiving the subsidy—largely due to delays in the Aadhaar seeding of their bank accounts.88 Further, it corroborated that several beneficiaries in Puducherry did not receive SMS updates, while, in Chandigarh, banks sent nonstandardized messages to recipients. This latter study also found signs of confusion among recipients with multiple Aadhaar-linked bank accounts or new phone numbers, and when different banks recorded DBT credits differently in customer passbooks (beneficiaries surveyed by the J-PAL researchers encountered these hurdles as well).

The J-PAL study determined that relative to the PDS, the DBT demanded a larger investment of both time and money to first access transfers from the bank and then go to the market to purchase food items. While ATM use reduced time spent, only 37 percent of all beneficiaries possessed the requisite ATM card. Beneficiaries chose to purchase better quality and more expensive grain relative to what they received under the PDS. Taken together, the researchers estimated that with these additional transaction costs, the subsidy size was short by up to 20 percent in Chandigarh and Puducherry, while beneficiaries expected to receive close to 900 rupees per household more than what the transfer provided. ICRIER’s researchers ascertained that while the majority of Chandigarh’s and Puducherry’s beneficiaries felt the amount was too small, they were unaware that the subsidy was less than the prices they paid to procure food grain under the previous system.

The quality of implementation mattered deeply for beneficiary preferences. Studies seemed to demonstrate that beneficiary preferences often were swayed in favor of the DBT when it functioned reliably, added convenience, and made better quality food consumption possible. Communication breakdowns, high associated access costs, and the irregular size or delivery of transfers, however, tended to diminish support for the scheme. In total, 38 percent of all recipients reported concern with DBT in the latest survey, up from 25 percent in the second round.89 An inadequate subsidy, fluctuating transfer amounts, missing transfers, and poorly functioning grievance-redressal mechanisms—such as a toll-free number only used by one beneficiary in Chandigarh, two in Puducherry, and none in Dadra and Nagar Haveli in the latest survey round—caused growing consternation among beneficiaries. Yet preference for the DBT over the PDS improved by 26 percentage points between the earliest and most recent round of surveys—approximately two-thirds of all beneficiaries preferred cash transfers in the latest iteration. What explained this apparent contradiction? A regression analysis revealed that beneficiaries content with PDS performance tended not to prefer the DBT, while those who successfully received multiple transfers and were satisfied with the subsidy amount were more likely to favor the DBT.

In a slightly older study, the financial inclusion consulting firm MicroSave assessed the Chandigarh and Puducherry pilots from August 2015 to January 2016 and found similar results.90 A total of 36 percent of survey respondents in Chandigarh and 59 percent of those in Puducherry spent more time accessing cash and purchasing rations than in the earlier system, while 42 percent in Puducherry and 57 percent in Chandigarh asserted that the switch to DBT resulted in the loss of wages. The study found that the transfer received by an Antyodaya Anna Yojana family (households identified as the poorest of the poor) would require spending twice as much compared to what the family had previously spent to access the same amount of rice.

All three studies underline the need for an urgent course correction in DBT’s operation and technology. This illuminates the long path its infrastructure has yet to traverse. Many Indian states still fall well short of linking the Aadhaar numbers of all beneficiaries to PDS ration cards and MGNREGA bank accounts, so much work remains to be done before this architecture could be used to support a more demanding UBI scheme. (See figures 3 and 4 for representative official data on cross-state variation in which percentage of participants’ PDS ration cards and MGNREGA wage payments have been linked to their Aadhaar numbers.) Indeed, the desirability of greater Aadhaar linkage for these schemes is itself called into question by reports of exclusion and denial of benefits. At one level, reforms to improve DBT functioning involve relatively low-cost tweaks, such as improving the frequency and clarity of beneficiary communication, maintaining standardized administrative records on transfers and the relevant failure rates, and updating the formula for calculating subsidies to more accurately reflect market costs.91

Yet a well-functioning cash transfer architecture also demands far heavier lifts like ensuring a sufficient supply of food grain in the open market, as well as substantially expanding electricity and internet coverage, financial inclusion, and the supply of basic public services. Without institutionalizing substantive grievance redressal mechanisms and transition procedures for beneficiaries and administrators to shift from one system of entitlements to another, a blinkered approach to welfare reform will continue to force India’s poor to bear the burdens of policy experimentation.

Finally, beyond the policy minutiae of cash transfers, there are several unresolved issues with the Aadhaar infrastructure undergirding this massive reconfiguration of service delivery. Technological lapses, the exclusion of genuine beneficiaries, and high rates of authentication failure (which the Economic Survey also notes) have been documented in states like Delhi, Gujarat, Rajasthan, Jharkhand, Andhra Pradesh, and Telangana where the Aadhaar program is involved in both the distribution of benefits like food entitlements and social security pensions as well as beneficiary authentication.92 The Indian government’s assertions of enhanced efficiency in welfare schemes with the use of Aadhaar have been difficult to verify independently, with insufficient public data on nationwide and scheme-wide program performance.93

More broadly, the weak privacy and security protections and the absence of substantive grievance redressal mechanisms in the 2016 Aadhaar Act raise significant concerns about data security.94 While the act mandates that “the Aadhaar number of an individual shall not be published, displayed or posted publicly by any person or entity or agency,” more than 200 central and state government agencies have publicly displayed the personal information and Aadhaar numbers of more than 100 million beneficiaries on their websites in recent months.95

The program faces a number of legal challenges stemming from its security vulnerabilities and concerns about its implications for privacy and mass surveillance.96 The Indian Supreme Court is yet to rule on the petitions challenging the validity of the Aadhaar program, though it unanimously declared that privacy is a fundamental right under the Indian Constitution in a recent landmark ruling. Next, a five-judge court will decide if the Aadhaar platform violates this right.97 Given that questions of first principle are still unsettled, coupled with issues such as unreliable beneficiary coverage and performance in cash-transfer applications as well as high associated private costs, it would be short-sighted for the Indian government to rely exclusively on the Aadhaar-seeded bank accounts for large-scale welfare disbursal. A balanced assessment of UBI transfer modalities calls for cost-benefit analyses that consider existing alternatives like electronic transfers through the National Electronic Funds Transfer system (used in the Madhya Pradesh pilots) or digital payments through mobile wallets.98

Getting cash transfers right, as Yamini Aiyar of the Center for Policy Research has written, requires that India’s unwieldy state apparatus work at “getting targeting right, adapting to market fluctuations, dealing with supply constraints and building a functioning banking system.”99 This demands a steep learning curve for both beneficiaries and administrators, and one made steeper yet with the addition of Aadhaar-based DBT to the mix. A UBI, whatever its final configuration, would likely reproduce the pathologies of the welfare system it is meant to replace unless the Indian state resolves existing flaws in Aadhaar and cash-transfer design and implementation.


1 This is in addition to the survey’s literary aspirations, and one would be hard-pressed to find another Ministry of Finance document described as “an avant-garde act.” Indrajit Hazra, “Economic Survey 2017 Reminds Us of Vikram Seth’s ‘A Suitable Boy’ but Without Any Characters,” Economic Times, February 1, 2017, The Economic Survey 2015–16 provided new evidence on the deeply segmented nature of Indian agricultural markets: See Ministry of Finance, “Agriculture: More From Less,” in Economic Survey 2015–16,, 68–83. The Economic Survey 2016–17 surpasses its predecessors with the use of big data to generate state-level estimates of trade, measure inter-state labor migration, and calculate Bengaluru and Jaipur’s property tax potential. Roshan Kishore, “Data and Innovation in the Economic Survey,” LiveMint, January 31, 2017,

2 Pratap Bhanu Mehta and Michael Walton, “Ideas, Interests and the Politics of Development Change in India: Capitalism, Inclusion and the State,” Effective States and Inclusive Development Research Center, Working Paper no. 36, July 2014.

3 Dani Rodrik, “When Ideas Trump Interests: Preferences, Worldviews, and Policy Innovations,” Journal of Economic Perspectives 28, no.1 (Winter 2014): 189–208.

4 Based on author’s calculations. Taking 893 rupees as the base amount and adjusting for inflation, the annual transfer comes out to approximately 15,000 rupees per year. According to the survey, an annual transfer of 15,226 rupees costs 12.9 percent of GDP (if targeted to 75 percent of the population, the fiscal cost falls to 9.7 percent). See Ministry of Finance, “Universal Basic Income: A Conversation With and Within the Mahatma,” table 3, page 205. See also Manas Chakravarty, “Government Expenditure as a Share of GDP Shrinks,” LiveMint, February 7, 2017,

5 Himanshu, “A Proposal for Universal Basic Services,” Ideas for India, May 21, 2017,

6 Reetika Khera, “Trends in Diversion of Grain From the Public Distribution System,” Economic & Political Weekly 46, no. 21 (May 21, 2011),; Himanshu, “Is India Ready for Universal Basic Income,” LiveMint, November 8, 2016,; Paolo Verme, “Subsidy Reforms in the Middle East and North Africa Region: A Review,” Policy Research Working Paper no. WPS7754, World Bank,

7 “BISP’s Impact on Vulnerable People’s Livelihood: Evidence From Impact Evaluation Report,” BISP Policy Brief,, accessed July 5, 2017; “Reaching the Poorest Through Strengthening the Social Safety Net System in Pakistan,” World Bank, April 22, 2015,

8 “China’s New Approach to Beating Poverty,” Economist, April 29, 2017,; “OECD Economic Surveys—China,” Organization for Economic Cooperation and Development (OECD), March 2017,

9 “The Cash Transfer for Orphans and Vulnerable Children (CT-OVC),” Ministry of East African Community (EAC), Labor and Social Protection,, accessed on July 5, 2017; “Kenya Cash Transfer for Orphans and Vulnerable Children,” World Bank, December 31, 2016,; “BLT Temporary Unconditional Cash Transfer: Social Assistance Program and Public Expenditure Review 2,” working paper, World Bank, 2012,

10 Planning Commission, “Report of the Expert Group to Review the Methodology for Estimation of Poverty,” Government of India, November 2009,; Sreelatha Menon and Indivjal Dhasmana, “New Methods Needed to Answer Old Controversy in Poverty Measurement,” Business Standard, April 9, 2012,; Madhura Swaminathan, “A Methodology Deeply Flawed,” Hindu, February 5, 2010,; S. Subramanian, “The Poverty Line: Getting It Wrong Again,” Economic & Political Weekly 46, no. 48 (November 26, 2011).

11 Planning Commission, “Rangarajan Report on Poverty,” Press Information Bureau, National Informatics Center, August 7, 2014, The Rangarajan Committee’s recommendations drew flak as well—a testimony to how poverty estimates are a lightning rod for controversy in India. See Angus Deaton and Jean Drèze, “Squaring the Poverty Circle,” Hindu, July 25, 2014, In 2016, the Task Force on Elimination of Poverty constituted by the NITI Aayog found that there was no consensus among Indian states over continuing with the Tendulkar line or shifting to the Rangarajan estimates or other higher poverty lines, and recommended instituting a new expert committee to identify the population below the poverty line. “Key Highlights of NITI Aayog’s Occassional Paper on Eliminating Poverty: Creating Jobs and Strengthening Social Programs,” NITI Aayog, Government of India,, accessed July 5, 2017; PTI, “Task Force on Poverty Files Report, Proposes New Panel on BPL,” India Today, September 11, 2016,

12 Ministry of Finance, Economic Survey 2016–17, 189.

13 Sourindra Mohan Ghosh and Imrana Qadeer, “Undermining Welfare,” Indian Express, February 8, 2017,

14 Reetika Khera, “Cash vs In-Kind Transfers: Indian Data Meets Theory,” IEG Working Paper no. 325, 2013,

15 Ministry of Rural Development, “Notification of the Revised Wage Rates Under Section 6(1) of the Mahatma Gandhi NREG Act, 2005,” Government of India, March 2017,; Ministry of Labor and Employment, “Minimum Wages,” Press Information Bureau, May 4, 2016,

16 For more extensive critiques of the survey’s method of financing a UBI, see Himanshu, “A Proposal for Universal Basic Services”; Madhura Swaminathan, “Getting the Basics Wrong,” Hindu, March 1, 2017,

17 Devesh Kapur, “The Shift to Cash Transfers: Running Better But on the Wrong Road?,” Economic & Political Weekly 46, no. 21 (May 21, 2011).

18 Ministry of Finance, Economic Survey 2016–17, 191–3. The top ten recipients of redistributive resource transfers (defined by the survey as gross devolution to the state adjusted for the respective state’s share in aggregate GDP) are Sikkim, Arunachal Pradesh, Mizoram, Nagaland, Manipur, Meghalaya, Tripura, Jammu and Kashmir, Himachal Pradesh, and Assam. Ministry of Finance, “Redistributive Resource Transfers (RRT) Should Be Significantly Linked to Fiscal and Governance Efforts on the Part of the States: Economic Survey 2016-17,” Press Information Bureau, January 31, 2017,

19 David Coady, Margaret Grosh, and John Hoddinott, “Targeting of Transfers in Developing Countries: Review of Lessons and Experience,” World Bank, 2004,; Rachel Slater and John Farrington, “Cash Transfers: Targeting,” Department for International Development, Overseas Development Institute, November 2009,

20 Ministry of Finance, “Universal Basic Income: A Conversation With and Within the Mahatma,” 191.

21 Coady, Grosh, and Hoddinott, “Targeting of Transfers in Developing Countries”; Stephen Devereux, et al., “Evaluating the Targeting Effectiveness of Social Transfers: A Literature Review,” IDS Working Paper vol. 2015, no. 460, July 2015,;jsessionid=B8CCC48BFA4C6BD4FBE0C39815B7C1C6?sequence=1;
Amartya Sen, “The Political Economy of Targeting,” World Bank, 1992,

22 For historical overviews of this period, see Walter Korpi and Joakim Palme, “The Paradox of Redistribution and Strategies of Equality: Welfare State Institutions, Inequality, and Poverty in the Western Countries,” American Sociological Review 63, no. 5, (1998): 661–87,; Thandika Mkandawire, “Targeting and Universalism in Poverty Reduction,” United Nations Research Institute for Social Development, Program Paper no. 23, December 2005,$FILE/mkandatarget.pdf. Examples of the early literature on targeting can be found in Irwin Garfinkel, ed., Income-Tested Transfer Programs: The Case For and Against (New York: Academic Press, 1982),; George A. Akerlof, “The Economics of ‘Tagging’ as Applied to the Optimal Income Tax, Welfare Programs, and Manpower Planning,” American Economic Review 68, no. 1 (March 1978): 8–19,; Ravi Kanbur and Nick Stern, “Transfers, Targeting and Poverty,” Economic Policy 2, no. 4 (April 1987) 111–47,

23 World Bank, World Development Report, 1990: Poverty (New York: Oxford University Press, 1990), 3,

24 Coady, Grosh, and Hoddinott, “Targeting of Transfers in Developing Countries.”

25 Rachel Slater and John Farrington, “Targeting of Social Transfers: A Review for DFID,” Overseas Development Institute, September 2009,

26 Devereux, et al., “Evaluating the Targeting Effectiveness of Social Transfers.”

27 Jonathan Conning and Michael Kevane, “Community-Based Targeting Mechanisms for Social Safety Nets: A Critical Review,” World Development 30, no. 3 (2002): 375–94,

28 Sudhanshu Handa, et al., “Targeting Effectiveness of Social Cash Transfer Programmes in Three African Countries,” Journal of Development Effectiveness 4, no. 1 (April 2012): 78–108,

29 Stephen Kidd, Bjorn Gelders, and Diloá Bailey-Athias, “Exclusion by Design: An Assessment of the Effectiveness of the Proxy Means Test Poverty Targeting Mechanism,” International Labor Office, Social Protection Department, 2017,; Stephen Kidd and Emily Wylde, “Targeting the Poorest: An Assessment of the Proxy Means Test Methodology,”  Australian Agency for International Development, 2011,

30 Caitlin S. Brown, Martin Ravallion, and Dominique van de Walle, “Are Poor Individuals Mainly Found in Poor Households? Evidence Using Nutrition Data for Africa,” National Bureau of Economic Research, working paper no. 24047, 2017,

31 Berk Özler, “Fact Checking Universal Basic Income: Can We Transfer Our Way Out of Poverty?,” World Bank, February 27, 2017,, 205–8.

32 Chris Elbers, et al., “Poverty Alleviation Through Geographic Targeting: How Much Does Disaggregation Help?,” Journal of Development Economics 83, no. 1 (2007): 198–213.

33 Brown, Ravallion, and van de Walle, “Are Poor Individuals Mainly Found in Poor Households?”

34 Rinku Murgai, Martin Ravallion, and Dominique van de Walle, “Is Workfare Cost-Effective Against Poverty in a Poor Labor-Surplus Economy?,” World Bank Economic Review 30, no. 3 (2016): 413–45.

35 For a review of issues, see Sachin Kumar Jain, “Identification of the Poor: Flaws in Government Surveys,” Economic & Political Weekly 39, no. 46/47 (November 20–26, 2004), 4981–4,; Indrajit Roy, “‘New’ Lists for ‘Old’: (Re-)constructing the Poor in the BPL Census,” Economic & Political Weekly 46, no. 22 (May 28, 2011),; Santosh Mehrotra and Harsh Mander, “How to Identify the Poor? A Proposal,” Economic & Political Weekly 44, no. 19 (May 9, 2009),

36 Jean Drèze and Reetika Khera, “The BPL Census and a Possible Alternative,” Economic & Political Weekly 45, no. 9 (February 27–March 5, 2010), 54–63,

37 Ibid.; also Ministry of Finance, Economic Survey 2016–17, 191. Such exclusion criteria have since become part of official Indian government policy, and have found mention in the National Food Security Act (2013), with some states moving forward with implementation after adopting this approach. See Drèze and Khera, “Recent Social Security Initiatives in India,” World Development 98, (October 2017): 555–72.

38 For a critical review of this approach, see M. R. Sharan, “Identifying BPL Households:

A Comparison of Competing Approaches,” Economic & Political Weekly 46, nos. 26 and 27 (June 25, 2011),

39 Paul Niehaus, Antonia Atanassova, Marianne Bertrand, and Sendhil Mullainathan, “Targeting With Agents,” American Economic Journal: Economic Policy 5, no. 1 (2013): 206–38,

40 Ministry of Petroleum and Natural Gas, “Text of Speech of Minister of State (I/C) for Petroleum and Natural Gas, Shri Dharmendra Pradhan During Curtain Raiser of WLPGA 2017 Asia LPG Summit Organized by World LPG Association in New Delhi,” Press Information Bureau, February 6, 2017,; Ministry of Petroleum and Natural Gas, “‘Mahilao ko mila Samman, Yahi hai Ujjwala ki Pehchan’ Says Shri Dharmendra Pradhan,” Press Information Bureau, May 5, 2017,

41 Thomas Erdbrink, “Iranian Parliament Cancels Cash Subsidies to 24 Million People,” New York Times, April 13, 2016,; “Cash Subsidies Hit $60 Billion in 7 Yrs.,” Eghtesad Online, January 22, 2017,

42 Ministry of Finance, Economic Survey 2016–17, 192. An ongoing experimental study seeks to estimate the preferences of beneficiaries when they are offered cash and food transfers of equal value. See

43 Further, news reports that the government is considering universal coverage of social pensions using the Socio-Economic and Caste Census make it less likely that this route will be explored for a UBI. Subodh Ghildiyal, “National Social Assistance Programme: Fiscal Burden May Force Modi Govt to Prune Scope of Welfare Schemes,” Economic Times, June 12, 2017,

44 For a review of the performance of India’s pensions schemes, see Drèze and Khera, “Recent Social Security Initiatives in India,” 11. On the economic vulnerability of India’s elderly, see Debasis Barik, Sonalde Desai, Tushar Agrawal, “After the Dividend: Caring for a Greying India,” Economic & Political Weekly 50, no. 24, (June 13, 2015),

45 Reetika Khera, “A Phased Approach Will Make a ‘Basic Income’ Affordable for India,” Wire, December 20, 2016,; Jessica Pudussery and Saloni Chopra, “Social Security Pensions in India: An Assessment,” Economic & Political Weekly 49, no. 19 (May 2014); Christopher Garroway, “How Much Do Small Old Age Pensions and Widows’ Pensions Help the Poor In India?,” United Nations Economic and Social Council, September 2013,

46 Rachel Sabates-Wheeler, Alex Hurrell, and Stephen Devereux, “Targeting Social Transfer Programmes: Comparing Design and Implementation Errors Across Alternative Mechanisms,” Journal of International Development, November 2015, 1523

47 Mark Schreiner, “Simple Poverty Scorecard® India,” 2011,; Caitlin Brown, Martin Ravallion, and Dominique van de Walle, “A Poor Means Test? Econometric Targeting in Africa,” no. w22919, National Bureau of Economic Research, 2016,

48 Thandika Mkandawire, “Targeting and Universalism in Poverty Reduction,” United Nations Research Institute for Social Development, Program Paper no. 23, December 2005,$FILE/mkandatarget.pdf.

49 Lant Pritchett, “The Political Economy of Targeted Safety Nets,” World Bank, Social Protection Paper Discussion Series, January 2005,

50 Walter Korpi and Joakim Palme, “The Paradox of Redistribution and Strategies of Equality: Welfare State Institutions, Inequality, and Poverty in the Western Countries,” American Sociological Review (1998): 661–87.

51 Sen, “The Political Economy of Targeting”; Timothy Besley and Ravi Kanbur, “The Principles of Targeting,” World Bank, working paper, March 1990,

52 See Pritchett, “The Political Economy of Targeted Safety Nets”; Stephen Kidd, “Pathways’ Perspectives on Social Policy in International Development: The Political Economy of ‘Targeting’ of Social Security Schemes,” Development Pathways, no. 19 (October 2015),

53 For a review of this literature: Daron Acemoglu, Suresh Naidu, Pascual Restrepo, and James A. Robinson, “Democracy, Redistribution, and Inequality,” in Handbook of Income Distribution, vol. 2B (Amsterdam: Elsevier, 2015),; Carol Graham, “Public Attitudes Matter: A Conceptual Frame for Accounting for Political Economy in Safety Nets and Social Assistance Policies,”  World Bank, SPDP Series 0233, 2002,

54 Esther Schüring and Franziska Gassmann, “The Political Economy of Targeting—A Critical Review,” Development Policy Review 34, no. 6 (2016): 809–29,

55 Remya Nair, “Union Budget 2016-17: Arun Jaitley Begins Phasing Out Corporate Tax Exemptions,” LiveMint, February 29, 2017,; Arun Jaitley, “Budget 2017-2018,” speech, Ministry of Finance, February 1, 2017,; Gireesh Chandra Prasad, “Eliminating LPG Subsidy to Almost Free Union Budget From Oil Price Volitality,” LiveMint, August 1, 2017,

56 “Union Budget 2017: Will Go for Universal Basic Income When Indian Politics Matures: Arun Jaitley,” Economic Times, February 1, 2017,; “Universal Basic Income May Not Be Politically Feasible: Jaitley,” Hindustan Times, June 12, 2017,

57 Shri Arjun Ram Meghwal, “Finance, Question No. 4184,” Lok Sabha, Parliament of India, August 11, 2017,; Shri Arjun Ram Meghwal, “Finance, Question No. 1773,” Lok Sabha, Parliament of India, March 10, 2017,

58 PTI and Maulik Pathak, “Universal Basic Income Only After Withdrawal of Existing Schemes: Arvind Subramanian,” LiveMint, February 25, 2017,

59 Atul Kohli, “State and Redistributive Development in India,” United Nations Research Institute for Social Development, accessed January 22, 2018,; Devesh Kapur and Prakirti Nangia, “A Targeted Approach: India’s Expanding Social Safety Net,” World Politics Review, September 24, 2013,; Devesh Kapur, “India’s Economic Development,” in International Development: Ideas, Experience, and Prospects, eds. Bruce Currie-Alder, Ravi Kanbur, David M. Malone, and Rohinton Medhora (Oxford: Oxford University Press, 2014),

60 Yamini Aiyar and Michael Walton, “Rights, Accountability and Citizenship: Examining India’s Emerging Welfare State,” Accountability Initiative, Center for Policy Research, October 2014,

61 “Demonetization, Digital Identity and Universal Basic Income,” YouTube video, 2:32, Center for Global Development, April 18, 2017,

62 “Full Text of FM Haseeb Drabu’s J&K Budget Speech,” Kashmir Monitor, January 11, 2017,; Prasanta Sahu, “Jammu and Kashmir May Become India’s First State to Roll Out Universal Basic Income Plan; Haseeb Drabu Makes Presentation to FM Arun Jaitley,” Financial Express, March 25, 2017,; Anil Padmanabhan, “‘By Hosting GST Council Meeting, J&K Will Be a Part of Indian Economic History,’” LiveMint, May 17, 2017,; Arup Roychoudhary, “PM Modi to Take Call on Universal Basic Income in J&K,” Business Standard, May 10, 2017,

63 Venkat Ram Reddy, “TS Looks to Jaitley for Universal Basic Income,” Deccan Chronicle, January 31, 2017,

64 Explanations range from party contestation, the strength of subnational identity, and the interplay of political leadership, coalitions, and policy legacies. Pradeep Chhibber and Irfan Nooruddin, “Do Party Systems Count? The Number of Parties and Government Performance in the Indian States,” Comparative Political Studies 37, no. 2 (2004): 152–87; Prerna Singh, How Solidarity Works for Welfare: Subnationalism and Social Development in India (New York, NY: Cambridge University Press, 2015); Rajeshwari Deshpande, K. K. Kailash, and Louise Tillin, “States as Laboratories: The Politics of Social Welfare Policies in India,” India Review 16, no. 1 (2017): 85–105; Sajjid Z. Chinoy and Toshi Jain, “How State Finances Could Delay Investment Cycle,” Business Standard, July 20, 2017,

65 For reasons described previously, the encouraging results from the Madhya Pradesh basic income pilot cannot solely determine the evidence base for an Indian UBI, especially given the small sample size, relatively short duration, and external funding and program administration.

66 Ministry of Finance, Economic Survey 2016–17, 181; Himanshu and Abhijit Sen, “In-Kind Food Transfers—I: Impact on Poverty,” Economic & Political Weekly 48, nos. 45 and 46 (2013): 29; Himanshu and Abhijit Sen, “In-Kind Food Transfers—II: Impact on Nutrition and Implications for Food Security and Costs,” Economic & Political Weekly 48, no. 47 (2013); Jean Drèze, Prankur Gupta, Reetika Khera, and Isabel Pimenta, “Food Security Act: How Are India’s Poorest States Faring?,” Ideas for India, June 29, 2016,

67 Sudha Narayanan and Nicolas Gerber, “Social Safety Nets for Food and Nutritional Security in India,” Indira Gandhi Institute of Development Research, Working Paper no. 031, 2015,, 14. The studies cited include Reetika Khera, “Trends in Diversion of Grain From the Public Distribution System,” Economic & Political Weekly 46, no. 21 (May 21, 2011); Andaleeb Rahman, “Expansion and Outreach: Revival of Rural Public Distribution System,” Economic & Political Weekly 49, no. 20 (May 17, 2014); Jean Drèze and Reetika Khera, “Rural Poverty and the Public Distribution System,” Economic & Political Weekly 48, no. 45–46 (November 16, 2013).

68 Jean Drèze and Reetika Khera, “Understanding Leakages in the Public Distribution System,” Economic & Political Weekly 50, no. 7 (February 14, 2015).

69 The 40 percent exclusion rate and an increase in inclusion errors (the incidence of non-poor receiving PDS benefits) from 28.8 percent to 36.9 percent in this period, however, indicates high potential for further improvement. Development Monitoring and Evaluation Office, “Evaluation Study on Role of Public Distribution System in Shaping Household and Nutritional Security India,” NITI Aayog, DMEO Report no. 233, December, 2016,, 40.

70 “Evaluation Study of Targeted Public Distribution System in Selected States,” National Council of Applied Economic Research, September 2015,, 79–81.

71 Sonalde Desai, “Public Distribution System: More People Rely on the PDS Than Ever Before,” India Human Development Survey Research Brief,, accessed July 5, 2017.

72 See Sandip Sukhtankar, “India’s National Rural Employment Guarantee Scheme: What Do We Really Know About the World’s Largest Workfare Program?,” India Policy Forum, July 13, 2016, The review describes the difficulty with estimating annual national leakage or corruption in the scheme, although it puts rough estimates of fiscal losses at 190 billion rupees in 2009–10, and 75 billion rupees in 2011–12.

73 Ibid.

74 Sonalde Desai, Prem S. Vashishtha, and Omkar Sharad Joshi, “Mahatma Gandhi National Rural Employment Guarantee Act: A Catalyst for Rural Transformation,” National Council of Applied Economic Research, August 2015,, 34.

75 Ibid, 59.

76 Upasak Das, “Rationing and Accuracy of Targeting in India: The Case of the Rural Employment Guarantee Act,” Oxford Development Studies 43, no. 3 (2015): 361–78,

77 Puja Dutta, Rinku Murgai, Martin Ravallion, and Dominique van de Walle, “Right to Work? Assessing India’s Employment Guarantee Scheme in Bihar,” Equity and Development series, World Bank, February 26, 2014,

78 Raghbendra Jha, Raghav Gaiha, Shylashri Shankar, and Manoj K. Pandey, “Targeting Accuracy of the NREG: Evidence From Madhya Pradesh and Tamil Nadu,” European Journal of Development Research 25, no. 5 (2013): 758–77,

79 Yanyan Liu and Christopher B. Barrett, “Heterogeneous Pro-Poor Targeting in the National Rural Employment Guarantee Scheme,” Economic & Political Weekly 48, no. 10 (2013): 46–53.

80 There is a persuasive argument for “benchmarking” the performance of traditional antipoverty programs and international aid against cash transfers. See Christopher Blattman and Paul Niehaus, “Show Them the Money,” Foreign Affairs, June 2014,

81 On how the public distribution system shields households from price risk, see Lucie Gadenne, Sam Norris, Monica Singhal, and Sandip Sukhtankar, “Price Risk and Poverty,” May 15, 2017, For a review of the literature see Melissa Hidrobo, et al., “Cash, Food, or Vouchers? Evidence From a Randomized Experiment in Northern Ecuador,” Journal of Development Economics 107 (March 2014): 144–56,; Reetika Khera, “Cash vs In-Kind Transfers: Indian Data Meets Theory.”

82Abhijit Banerjee, et al., “A Multifaceted Program Causes Lasting Progress for the Very Poor: Evidence From Six Countries,” Science, May 15, 2015,

83 Ministry of Finance, Economic Survey 2016–17, 194.

84 Ministry of Consumer Affairs, “States Asked to Expedite the Reforms So as to Bring Transparency in the Functioning of PDS,” Press Information Bureau, September 16, 2016,

85 Ministry of Finance, Economic Survey 2016–17, 193.

86 K. Muralidharan, Paul Niehaus, and Sandip Sukhtankar, “Direct Benefits Transfer in Food: Results From One Year of Process Monitoring in Union Territories,” UC San Diego, 2017,, 15.

87 Shweta Saini, et al., “Indian Food and Welfare Schemes: Scope for Digitization Towards Cash Transfers,” Indian Council for Research on International Economic Relations, Working Paper no. 343, August 2017,

88 Ibid., 20.

89 Muralidharan, Niehaus, and Sukhtankar, “Direct Benefits Transfer in Food.”

90 Isvary Sivalingam and Lokesh Kumar Singh, “Feeding India’s Poor: Plugging Leakages, Without Doing Any Harm,” MicroSave, May 2016, The survey cites this report while describing external studies of these pilots, in addition to a series of qualitative interviews of beneficiaries, government officials, and independent evaluators conducted by the news magazine Governance Now and the news website See Pratap Vikram Singh, “DBT Chandigarh: An Idea Whose Time Hasn’t Come,” Governance Now, March 25, 2016,; Anumeha Yadav, “Modi Government’s Cash for Food Experiment Gets a Quiet Burial in Puducherry,”, May 18, 2015,

91 Saini, et al., “Indian Food and Welfare Schemes,” 39–44; Muralidharan, Niehaus, and Sukhtankar, “Direct Benefits Transfer in Food,” 19; Sivalingam and Singh, “Feeding India’s Poor,” 5.

92 A range of government documents, scholarly work, and news reporting has documented such issues in a variety of Aadhaar applications. For a representative selection, see Ministry of Finance, Economic Survey 2016–17, 194.; Committee on Digital Payments, “Medium Term Recommendations to Strengthen Digital Payments Ecosystem,” Ministry of Finance, Government of India, December 2016,, 128; Saini, et al., “Indian Food and Welfare Schemes,” 9–10; Anmol Somanchi, Srujana Bej, and Mrityunjay Pandey, “Well Done ABBA?,” Economic & Political Weekly 52, no. 7 (February 18, 2017); Nandini Nayak and Shikha Nehra, “Accessing the Right to Food in Delhi,” Economic & Political Weekly 52, no. 23 (June 10, 2017),; Ronald Abraham, Elizabeth S. Bennett, Noopur Sen, and Neil Buddy Shah, “State of Aadhaar Report 2016–17,” IDinsight, May 2017,, 59–61; Prerna Kapoor, Remya Nair, and Elizabeth Roche, “Aadhaar Fails MGNREGS Test in Telangana,” LiveMint, April 7, 2017,; Jaideep Deogharia, “Technology Glitches Deprives Many From PDS Ration in Jharkhand,” Times of India, September 7, 2016,; Vimukt Dave, “Bumps in Gujarat’s PDS Highway,” Business Standard, February 26, 2017,; Anumeha Yadav, “Rajasthan Presses on With Aadhaar After Fingerprint Readers Fail: We’ll Buy Iris Scanners,”, April 10, 2016,; “FP Shops Left Over Beneficiaries Report,” Society for Social Audit, Accountability and Transparency, May 29, 2015,; Ankita Aggarwal, “Ten Ways MGNREGA Workers Do Not Get Paid,” Economic & Political Weekly 52, no. 6 (February 11, 2017),; Niha Masih, “Lost in Transition: Has Linking Aadhaar to Government Welfare Schemes Made It Difficult for Beneficiaries to Avail of Aid?,” Hindustan Times, October 8, 2017,

93 Anumeha Yadav, “How Efficient Is Aadhaar? There’s No Way to Know Since the Government Won’t Tell,”, July 6, 2017, See also Rahul Lahoti, “Questioning the ‘Phenomenal Success’ of Aadhaar-linked Direct Benefit Transfers for LPG,” Economic & Political Weekly 51, no. 52 (December 24, 2016),

94 See Shweta Agrawal, Subhashis Banerjee, and Subodh Sharma, “Privacy and Security of Aadhaar: A Computer Science Perspective,” Economic & Political Weekly 52, no. 37 (September 16, 2017),; Vrinda Bhandari and Renuka Sane, “Towards a Privacy Framework for India in the Age of the Internet,” NIPFP Working Paper Series, Working Paper no. 179, November 2016,

95 “The Gazzette of India,” Government of India, September 14, 2016,, 76; Komal Gupta, “Govt Departments Breach Aadhaar Act, Leak Details of Beneficiaries,” LiveMint, April 25, 2017,; Amber Sinha and Srinivas Kodali, “(Updated) Information Security Practices of Aadhaar (or Lack Thereof): A Documentation of Public Availability of Aadhaar Numbers With Sensitive Personal Financial Information,” Center for Internet and Society, May 1, 2017,; Aman Sethi, Samarth Bansal, and Saurav Roy, “Details of Over a Million Aadhaar Numbers Published on Jharkhand Govt Website,” Hindustan Times, July 19, 2017,; “Over 200 Government Sites Reveal Aadhaar Details; No Leakage From UIDAI: Minister,” Economic Times, July 20, 2017,

96 Ananth Padmanabhan, “The Three Sins of Aadhaar,” Open Magazine, August 4, 2017,; Aman Sethi, “Right to Privacy: Data Shows States Using Aadhaar to Build Profiles of Citizens,” Hindustan Times, August 25, 2017,

97 “Supreme Court to Hear Aadhaar Pleas in November,” Hindu, August 30, 2017,

98 Reserve Bank of India, “Frequently Asked Questions,” updated July 24, 2015,

99 Yamini Aiyar, “Three Years On, the Modi Government Still Has Gaping Holes in Its Social Policy,” Wire, February 1, 2017,