Few debates in development economics carry as much ideological baggage as the question of whether to give poor households cash or deliver support in kind. Behind the technocratic framing lies a deeper tension: do we trust the poor to make good decisions with resources, or do we believe program designers know better? For decades, the default assumption in most aid architectures favored in-kind transfers—food baskets, subsidized inputs, voucher programs—partly on efficiency grounds, partly on paternalistic ones.

The experimental revolution in development economics has given us something rare in this debate: actual evidence. Over the past two decades, a substantial body of randomized controlled trials has accumulated across Sub-Saharan Africa, South Asia, and Latin America, comparing cash and in-kind modalities head-to-head. The results are more nuanced than either camp typically acknowledges, but they do point in a clear direction.

This article synthesizes what the experimental evidence actually shows across three critical dimensions: how cash recipients spend their transfers, how the two modalities compare on cost-effectiveness, and how conditionality interacts with the choice of transfer type. The goal isn't to declare a universal winner—context specificity matters enormously—but to identify the empirical regularities that should discipline program design. If we're serious about evidence-based development, we need to follow the data even when it challenges institutional preferences.

The most persistent objection to cash transfers is that recipients will waste the money on alcohol, tobacco, or other temptation goods. This concern is not trivial—it reflects genuine worry among policymakers and taxpayers about the efficacy of public spending. But the experimental record is remarkably consistent on this point. A landmark 2014 meta-analysis by Evans and Popova, covering 19 studies across multiple continents, found no systematic evidence that cash transfers increase spending on temptation goods. In many cases, expenditure on alcohol and tobacco actually decreased.

Consider the GiveDirectly experiments in Kenya, among the most rigorously evaluated unconditional cash transfer programs in existence. Haushofer and Shapiro's 2016 RCT found that large lump-sum transfers led to significant increases in spending on durable assets—metal roofing, livestock, furniture—and food consumption. There was no detectable increase in spending on alcohol or tobacco. Recipients behaved, in aggregate, as rational economic agents investing in high-return assets and addressing binding consumption constraints.

This pattern holds across diverse contexts. Blattman and colleagues found similar results among high-risk youth in Liberia—a population that many would consider maximally likely to misuse cash. Even among ex-combatants and substance users, cash grants were overwhelmingly invested in business activities and basic needs. The Progresa/Oportunidades evaluations in Mexico showed that cash recipients increased caloric intake and dietary diversity at rates comparable to or exceeding those achieved by food-based programs.

The mechanism behind these results is straightforward but important. Poor households face severe liquidity constraints and have extensive knowledge of their own needs. When given fungible resources, they allocate them to their highest-value uses—which are almost always basic consumption, productive assets, or human capital investments. The informational advantage that households hold over program designers regarding their own circumstances is substantial and systematically underestimated in paternalistic transfer frameworks.

Where temptation good concerns do have some empirical support is in specific subpopulations and contexts—particularly where substance dependency is clinically significant rather than recreational. But even here, the evidence suggests that labeling or light-touch behavioral nudges may address the issue more efficiently than replacing the entire transfer modality. The blanket assumption that cash will be misused is not supported by the weight of experimental evidence, and program design premised on that assumption imposes real costs.

Takeaway

The informational advantage poor households hold over program designers regarding their own needs is large and consistent. Designing transfers around the assumption of misuse imposes costs that the evidence rarely justifies.

Even if cash and in-kind transfers produced identical outcomes for recipients, the cost-effectiveness comparison would still favor cash in most contexts—because delivering physical goods is dramatically more expensive than delivering money. The logistics of procurement, storage, transportation, and distribution for in-kind programs generate administrative costs that typically consume 20-40% of total program budgets. Cash transfers delivered through mobile money or banking systems can reduce administrative costs to single-digit percentages of total outlay.

The World Food Programme's own evaluations provide striking evidence. A series of randomized comparisons across Ecuador, Uganda, Niger, and Yemen found that cash and voucher modalities achieved equivalent or superior nutritional outcomes at 18-35% lower cost per beneficiary than equivalent food transfers. In Ecuador, Hidrobo and colleagues' 2014 RCT showed that cash, vouchers, and food transfers all improved dietary diversity and caloric intake—but cash and vouchers did so at substantially lower cost. The food arm required warehousing, spoilage management, and dedicated distribution infrastructure that simply evaporated in the cash modality.

Leakage and diversion compound the cost disadvantage of in-kind transfers. Physical commodities are vulnerable to theft, spoilage, and diversion at every node in the supply chain—from central warehouses to local distribution points. India's Public Distribution System, one of the world's largest in-kind transfer programs, has historically experienced leakage rates estimated between 40-60% depending on the state. Cash transfers routed through biometric-authenticated bank accounts or mobile money reduce these vulnerabilities dramatically, though they introduce different risks around digital exclusion and financial literacy.

Market effects represent another dimension of the cost comparison. Large-scale in-kind food distribution can depress local agricultural prices, harming the producers who are often themselves among the poor. Cash transfers, by contrast, tend to stimulate local market activity—the GiveDirectly general equilibrium study by Egger and colleagues found significant positive spillovers to non-recipient households in treatment areas, driven by increased local demand. The transfer modality doesn't just affect recipients; it shapes the market environment for entire communities.

The cost-effectiveness case for cash is strongest where markets function reasonably well—where recipients can actually purchase goods and services with the cash they receive. In contexts of acute supply disruption, conflict, or extreme market thinness, in-kind transfers retain a comparative advantage because the binding constraint is availability, not purchasing power. The evaluation evidence thus supports a conditional conclusion: default to cash where markets are functional, reserve in-kind for genuine supply-side failures.

Takeaway

The relevant comparison is not just what recipients receive but what it costs to get resources to them. Administrative overhead and supply chain leakage in in-kind programs often mean that a dollar of cash delivers more value than a dollar budgeted for physical goods.

The cash versus in-kind debate often collapses a second, distinct policy dimension: whether transfers should be conditional or unconditional. Conditional cash transfers (CCTs) like Progresa/Oportunidades, Bolsa Família, and their many successors require recipients to meet behavioral conditions—typically school enrollment, health check-ups, or vaccination schedules—in exchange for the transfer. The interaction between transfer type and conditionality is analytically important and empirically underexplored relative to its policy significance.

The strongest evidence for conditionality comes from contexts where the targeted behavior has significant positive externalities or where households systematically underinvest in specific human capital dimensions. Baird, McIntosh, and Özler's 2011 experiment in Malawi compared conditional and unconditional cash transfers for adolescent girls' schooling. Conditional transfers produced significantly larger effects on school enrollment than unconditional ones—but unconditional transfers outperformed on broader welfare measures including pregnancy rates and psychological well-being. The choice of conditionality optimizes for different outcomes, and the right choice depends on what you're trying to achieve.

In-kind transfers embed a form of implicit conditionality—you can only consume what's delivered. School feeding programs condition nutrition on attendance. Subsidized fertilizer programs condition input access on agricultural activity. This implicit conditionality is both a feature and a limitation. It achieves behavioral targeting without monitoring costs but constrains recipient choice in ways that may reduce overall welfare. The experimental comparisons suggest that when explicit conditionality is feasible and affordable to monitor, pairing it with cash transfers typically dominates pairing it with in-kind delivery—because you achieve the behavioral targeting without sacrificing the flexibility and cost advantages of cash.

Monitoring costs are the critical constraint. Conditional programs require verification systems—confirming school attendance, tracking health visits, auditing compliance. In institutional environments with weak administrative capacity, these monitoring costs can be prohibitive and prone to corruption. Filmer and Schady's 2014 work in Cambodia found that a labeled cash transfer—cash explicitly designated for education but without formal monitoring or enforcement—produced schooling gains nearly as large as formally conditional transfers. The labeling effect suggests that framing and social expectations can substitute for expensive verification infrastructure.

The practical implication for program design is a hierarchy. Where monitoring infrastructure exists and the targeted behavior has clear positive externalities, conditional cash transfers represent the evidence-based default. Where monitoring is costly or infeasible, labeled or soft-conditional cash transfers capture much of the behavioral benefit at lower cost. In-kind transfers occupy a narrower optimal niche: situations where supply constraints bind, where markets are severely disrupted, or where the targeted good has specific properties—like therapeutic nutrition for acute malnutrition—that cash cannot reliably replicate through markets.

Takeaway

Conditionality and transfer type are separate design choices that interact in important ways. The most cost-effective combination is usually cash paired with the lightest conditionality structure that achieves the behavioral objective—not heavier in-kind modalities used as implicit conditions.

The experimental evidence on cash versus in-kind transfers points toward a clear but context-dependent conclusion. Cash transfers are the efficient default in most development settings—they respect recipient agency, cost less to deliver, stimulate local markets, and do not systematically lead to misuse. The burden of proof should fall on those advocating more expensive and restrictive in-kind modalities to demonstrate specific conditions under which they outperform.

But defaults are not universals. Acute supply failures, specific nutritional emergencies, and thin markets create genuine niches for in-kind delivery. The evidence demands design precision, not ideological commitment to either modality.

What the experimental record ultimately reveals is that the most consequential design choices are often the ones we treat as settled. Revisiting transfer modality assumptions with rigorous evaluation—rather than institutional inertia or donor preference—remains one of the highest-return investments in development program design.