Ambitious welfare programs and promises will likely be a central issue in India’s 2024 general elections. Over the past decade, there has been a sea change in how governments deliver welfare benefits to citizens. Technological advancements and state capacity improvements have allowed both central and state governments to transfer benefits directly to voters, weakening meddlesome middlemen. Unlike the discretionary sops of the past, these rule-based direct transfers are supposed to reduce favoritism and corruption. In so doing, they are widely perceived to be popular vote winners. Yet, in actuality, there is limited hard evidence of “schemes translating into votes.” This uncertainty presents a question: how, if at all, have transformative changes in India’s welfare state affected voting behavior?
Transformative Changes and Their Implications
India has experienced at least three interrelated transformations in the functioning of the welfare state: less discretion in the selection of beneficiaries, the direct transfer of benefits, and a proliferation of benefits.
India’s impressive digital public infrastructure allows the government to reduce discretion in beneficiary selection and transfer benefits directly to citizens. The socioeconomic census data and big data triangulated from multiple sources provide the government with granular, household-level information to frame objective rules and criteria. Near-universal biometric identification via the Aadhaar program allows the government to confirm that the intended recipient of a program is the actual beneficiary. An expansion of mobile banking services has allowed the government to transfer money directly into the accounts of beneficiaries, who can access their funds with the push of a button. The ubiquitous use of mobile phones has unlocked a host of features such as the geotagging of assets and the ability to verify that beneficiaries are using transfers for their intended purposes. All told, the scale of welfare provision is staggering. There are more than 300 programs delivering benefits to citizens across the country, programs that range from a $10 cooking gas cylinder to a $2,000 house. These programs reach nearly 950 million people and have accounted for $270 billion in government spending since 2017.
The three transformative changes mentioned above—reduced discretion in the selection of beneficiaries, the direct transfer of benefits, and a proliferation of benefits—have important implications for voting behavior.
Less discretion implies less favoritism, which potentially undercuts the logic of caste voting. Conventional wisdom holds that voters prefer a politician from their caste group because they believe the politician is more likely to favor them in the distribution of state largesse. The expectation of preferential access to resources and opportunities—often phrased as humara hai, humare liye karega (the politician is one of us and will do something for us)—is at the heart of caste appeals in elections. When middlemen lose their discretion, the logic of caste voting is weakened, at least in theory. In practice, however, do India’s voters adjust their beliefs about who benefits them?
The direct transfer of benefits also reduces opportunities for corruption. But corruption is a multifaceted phenomenon; in addition to high-level political corruption, it also includes common citizen malpractices, such as people claiming benefits more often than they are supposed to, ineligible or nonexistent people receiving benefits, and people using welfare money for other purposes. Although high-level corruption is a politically salient issue, it is far from clear whether voters reward transparency measures that hinder their own ability to claim from the state. This leads to a second question: do India’s voters reward efficient welfare implementation along with the receipt of benefits? In other words, do they value process (how welfare is delivered) independent of outcomes (whether a benefit is received)? The answer to this question shows whether politicians, in the long term, have an incentive to change how they deliver welfare.
Finally, the proliferation of benefits brings into focus trade-offs in allocation. For instance, should politicians distribute an expensive benefit to fewer people or a cheaper benefit to more people? Do expensive benefits have a greater impact on voters’ preferences than cheap benefits? The answer to this question determines, in part, the politically optimal allocation of resources, which in turn has implications for the kind of redistribution occurring through welfare programs.
Figure 1 provides an overview of the transformation of India’s welfare state and the electorally minded questions they raise.
Does Welfare Trump Caste?
Historically, India’s welfare state has been characterized by ethnic clientelism. Politicians used middlemen to deliver benefits to citizens, and those middlemen favored members of their own ethnic group when distributing benefits. As a result, citizens preferred in-group politicians to out-group politicians. Ethnic or caste voting in this case operated on an instrumental logic: vote your caste and you will get more material benefits.
Rule-based direct transfers reduce middleman discretion and ethnic favoritism. When governments rely on socioeconomic indicators to identify beneficiaries, they end up distributing welfare benefits across caste or ethnic boundaries. As an example, nearly six out of ten beneficiaries in the Pradhan Mantri Awas Yojana-Gramin, a housing program run by the central government, belong to Scheduled Castes, Scheduled Tribes, or religious minorities like the Muslim community. In contrast, several earlier housing programs implemented through village pradhans or party workers tended to favor kin (co-ethnics) and co-partisans.
Voters’ response to this transformation can be gauged along two lines: in-group backlash and out-group reward. Consider a case in which Party 1 champions the interests of Caste A and Party 2 of Caste B. Party 1 comes to power and implements a welfare program that uses objective rules to select beneficiaries. As a result, the welfare program significantly benefits voters from Caste B. This raises two questions. First, do voters from Caste A (in-group voters) engage in backlash against Party 1 because they expected preferential treatment? Second, do voters from Caste B (out-group voters) who benefit from the welfare program reward Party 1? These two dynamics—in-group backlash and out-group reward—capture what is at stake politically.
To assess voters’ responses, in April–May 2022 the author interviewed more than 6,000 voters using telephone and online surveys. In addition, in February 2020 the author interviewed 530 Scheduled Caste/Dalit beneficiaries of Pradhan Mantri Awas Yojana-Gramin (the prime minister’s affordable housing scheme) using face-to-face surveys in fifty-seven villages in the state of Bihar.
The findings are striking: upper and dominant caste voters, who for the purposes of this analysis may be regarded as core supporters of the ruling Bharatiya Janata Party (BJP), did not punish the BJP for running welfare programs that substantially benefited Dalits, Adivasis, and Muslims. Figure 2 shows the results from a survey experiment that randomly assigned voters to receive information either about how many citizens benefited from a welfare program (“positive” condition) or about the number of beneficiaries from the Dalit, Adivasi, and Muslim communities (“positive + out-group” condition).
Figure 2 presents the responses from core supporters, who similarly evaluated the BJP government’s performance and made hypothetical donations to the party even when they learned that many welfare program beneficiaries came from Dalit, Adivasi, and Muslim communities. In other words, there was no backlash from in-group voters.
Dalits who benefited from the housing program recognized that the BJP had done something for them, expressed gratitude to the party for that benefit, and seemed aware that they received the benefit based on objective rather than discretionary criteria. However, Dalits continued to believe that local politicians from their own community were more likely to benefit them. The instrumental logic of caste voting persisted, particularly at the local level.
Two substantial implications follow: first, the BJP is not constrained by its traditional support base when it implements welfare and redistributive policies that substantially benefit voters outside of that core; and second, the reputational gains from welfare programs are contained and have not yet weakened the edifice of caste mobilization in politics.
Do Voters Prioritize Process or Outcomes?
A second transformation in welfare delivery is the use of the so-called JAM trinity: Jan Dhan Yojana bank accounts (intended to universalize consumer banking access), Aadhaar-based biometric authentication (used to verify users’ identities), and mobile phone–enabled features (like the geotagging of assets and digital banking services). These technologies reduce the possibility of many of the typical malpractices associated with claim-making. Do voters, who otherwise punish high-level corruption, reward parties for efficient implementation?
To assess this question, the author studied how voters reacted to performance information in survey experiments. These telephone-based and online survey experiments were conducted in twelve different languages using the random digit dialing method, as well as online in Hindi and English. The experiments randomly varied the type of information shared with participants. In the “negative” condition, voters were told about high unemployment, fuel inflation, and income inequality. In the “positive” condition, voters were told about a welfare program and the number of people who benefited from it. A third, “efficiency” condition provided additional information about the use of direct benefit transfer, biometric authentication, and geotagging of assets. After listening to (or reading) the information, participants evaluated the current government’s performance on a scale from 0 to 10. They were also asked: if they hypothetically had to donate 10 rupees to political parties, how much would they give the major parties in their state? (The list always included the BJP.) Figure 3 presents the results from this experiment.
Comparing voters’ evaluation of the government in the positive and negative conditions revealed how they rewarded good performance and punished bad performance. The results indicated the government’s rating improved by 11.4 percentage points, and hypothetical donations to the ruling party increased by 2.8 percentage points, when voters simply learned about welfare programs.
Similarly, comparing voters’ evaluations in the efficiency and positive conditions revealed how much they rewarded implementation. Specifically, the results indicated how much they valued process (how welfare benefits are delivered), once outcomes were fixed (how many people received that benefit). Voters’ evaluation of the government improved modestly by 1.9 percentage points, and donations to the ruling party increased by 2.9 percentage points, when voters learned about transparency measures.
The main takeaway from this analysis is that, to the extent that welfare programs shape voters’ preferences, the actual delivery of benefits (outcomes) matters more than efficient implementation (process). For there to be durable change in how welfare benefits are distributed, voters would need to reward efficient implementation more than they currently do.
Do Cheap Benefits Deliver Politically?
Finally, the expansion of welfare programs has meant that governments deliver an increasingly diverse set of benefits to citizens, ranging from homes to home appliances. With the proliferation of benefits comes the question of impact and future allocation: which welfare programs are most politically impactful, and how should future resources be apportioned to different programs?
In this context, an interesting question is whether expensive benefits have greater impact on political preferences than cheaper benefits. There is an implicit trade-off in this question: because governments have finite resources, a cheap benefit like a $10 cooking gas cylinder can be distributed to many more voters than an expensive benefit like a $2,000 house. But how does the political impact of the two welfare programs compare?
Data from the National Election Studies 2019, conducted by the Lokniti Program of the Centre for the Study of Developing Societies, provide an interesting answer. In the last parliamentary elections, beneficiaries of the Ujjwala (clean cooking gas) scheme were 4.6 percentage points more likely to report voting for the BJP, 2.7 percentage points more likely to say that the BJP works for the poor, and 4.5 percentage points more satisfied with the central government. In addition, program beneficiaries reported that they were less likely to vote for narrow, caste-based parties and more likely to say the incumbent should be reelected, holding constant other predictors of political preferences.
These results are consistent with anecdotal evidence that Ujjwala was a popular vote winner in those elections. What is striking is that the Pradhan Mantri Awas Yojana-Gramin housing program, which involves a considerably more expensive benefit, has comparable political impact. In other words, a relatively inexpensive benefit seems to have as much political impact as an expensive benefit.
Similar evidence emerges in survey experiments that randomly vary information about government performance. Participants were shown either positive or negative information about the government’s performance. In the “positive” condition, participants were told about the government’s welfare programs. A further randomized element varied whether that information was about a cheap or an expensive benefit. In the “negative” condition, participants were told about high unemployment, fuel inflation, and income inequality. The results can be seen in Figure 4.
First, positive information of either kind affected political attitudes and preferences. Compared to participants who received negative information, those that received information about the distribution of a cheap benefit evaluated the government’s performance 11.3 percentage points higher, and hypothetical donations to the ruling party increased by 3.6 percentage points. The effect of information about the expensive benefit was broadly comparable. Compared to participants that received negative information, those that learned of an expensive benefit being distributed evaluated the government’s performance 13.3 percentage points higher and donated 4.6 percentage points more to the ruling party.
Crucially, positive information about the expensive benefit did not have any greater impact. When comparing participants that learned about the distribution of cheap and of expensive benefits, there were only small differences in performance evaluation (2.1 percentage points) and donations to the ruling party (1 percentage point).
Put together, the evidence suggests that relatively inexpensive benefits can have as great a political impact as expensive benefits. This finding has implications for how politicians design redistributive welfare programs. In particular, recent discussions on “freebies” (revdis, in Hindi parlance) have focused on normative aspects, potentially overlooking the electoral logic for distributing certain types of benefits.
The Future of Welfare Politics
The transformative changes in India’s welfare state are impacting voting behavior in nuanced ways, with signs of both change and continuity. Rule-based selection of beneficiaries has produced cross-ethnic distribution at scale, disturbing the prevailing ethnic-clientelistic equilibrium. Despite this trend, the instrumentalist logic of identity voting persists at many levels of electoral competition. Even though direct benefit transfer has potentially improved last-mile delivery, voters appear to privilege outcome over process. Welfare expansion has led to innovative benefits but also to allocation challenges, particularly when voters comparably reward the delivery of inexpensive and expensive benefits. As parties race to the hustings to campaign in the general elections, these findings can shape popular understanding of the electoral payoffs of the “new welfarism.”
In the months ahead, Carnegie scholars and contributors will be analyzing various dimensions of India’s upcoming election battle—including the role foreign policy plays, the strength of partisan ties, and how technology has reshaped campaigning. Keep up to date with the project at https://carnegieendowment.org/specialprojects/indiaelects2024.
Acknowledgments
The author thanks Sanjay Kumar of the Centre for the Study of Developing Societies (CSDS) for providing access to the National Election Studies data, Yashwant Deshmukh and Gaura Shukla of the Centre for Voting Opinion and Trends in Election Research (CVoter) for fielding the telephone survey experiments, and Atulesh Shukla for fielding the face-to-face survey.
