Collaborative Research: Policy Analysis by Revealed Preference: With an Application to Food Security

世界各国政府将其预算的很大一部分花在公共产品的提供上。因此，了解现金再分配和实物再分配之间的权衡，对政府政策至关重要。例如，实物再分配可能会减少选择，从而降低受助者的福利，但如果纳税人有家长式的偏好，更关心防止饥饿，而不是穷人的其他愿望，那么他们的福利可能会增加。私人投资机构提议进行一项严格的实验研究，以估计(1)受益人对同等价值的现金和实物转移的偏好，(2)现金和实物转移对营养的相对影响，以及(3)受益人对实物福利的估值。这一设计既可以衡量受益人的偏好(使用接受率)，也可以衡量现金与实物福利对营养的因果影响(使用随机分配的报价来估计治疗意图和治疗对营养的影响)。此外，PIs为更多随机选择的家庭提供了将其实物福利交换为实验性不同现金转移的选项，这使得能够估计不同价格下对现状的需求，并量化受益人从相对于现金转移的实物福利中获得的消费者盈余(或损失)。在方法上，绩效指标在一种设计中整合了主导现代政策评估的两种范式，一种基于影响，另一种基于显露的偏好。此外，通过衡量随机变动的价格，物价指数是第一批使用实验数据估计公共提供的私人产品的个人层面受益者估值的机构之一。揭示偏好方法的独特之处在于使人能够理解和适应受益人偏好相对于现金和实物福利的异质性，从而降低在关键领域(如穷人的粮食安全)进行政策试验的政治和道德风险。

Governments around the world spend large shares of their budgets on the public provision of private goods. Understanding the tradeoffs between redistribution in cash and in kind is thus central to government policy. For example, redistributing in-kind may potentially reduce choice and thus lower recipients' welfare, but may increase taxpayers' welfare if they have paternalistic preferences and care more about preventing hunger than about the other aspirations of the poor. The PIs propose a rigorous experimental study to estimate (i) beneficiary preferences over equally-valued cash and in-kind transfers, (ii) the relative impact on nutrition of cash and in-kind transfers, and (iii) beneficiary valuation of in-kind benefits. This design allows one to measure both beneficiary preferences (using the take up rate) as well as the causal impact of cash vs. kind benefits on nutrition (using the random assignment of the offer to estimate both intent-to-treat and treatment-on-treated effects on nutrition). Further, the PIs offer additional randomly selected households the option to exchange their in-kind benefits for an experimentally varied cash transfer, which allows to estimate the demand for the status quo at various prices and to quantify the consumer surplus (or loss) beneficiaries obtain from in-kind benefits relative to cash transfers. Methodologically, the PIs integrate in one design the two paradigms that have dominated modern policy evaluation, one based on impacts, the other based on revealed preferences. In addition, by measuring take-up at randomly varied prices, the PIs are among the first to estimate individual-level beneficiary valuations of publicly-provided private goods using experimental data. The revealed preference approach is unique in enabling one to understand and accommodate heterogeneity of beneficiary preferences over cash and in-kind benefits, thereby reducing the political and ethical risks of policy experimentation in a critical area such as food security for the poor