压力测试调查到调查推断：了解贫困预测何时可能失败（英）

Stress Testing Survey to Survey Imputation Understanding When Poverty Predictions Can Fail Paul CorralAndres HamPeter LanjouwLeonardo LucchettiHenry Stemmler Poverty and Equity Global DepartmentAugust 2025 Policy Research Working Paper11192 Abstract Accurate and timely poverty measurement is central todevelopment policy, yet the availability of up-to-datehigh-quality household survey data remains limited—par-ticularly in countries where poverty is most concentrated.Survey-to-survey imputation has emerged as a practicalresponse to this challenge, allowing practitioners to updatepoverty estimates using recent surveys that lack directwelfare measures by borrowing information from othercomprehensive surveys. A critical review of the method isprovided, revisiting its statistical underpinnings and testingits limitations through extensive model-based simulations.Through these simulations, the analysis demonstrates howviolations of parameter stability, omitted variable bias, and shifts in survey design can introduce substantial errors—particularly when imputing across time or under economicand structural change. Results show that standard correc-tions such as re-weighting or covariate standardization mayfail to eliminate these biases, especially when imputingacross time or under structural change. The performance ofalternative model specifications is also evaluated under vari-ous methods, including performance under heteroskedasticerrors, non-normality. The findings offer practical guidancefor practitioners on when survey-to-survey imputation islikely to succeed, when it should be reconsidered, and howto communicate its limitations transparently in the contextof poverty monitoring and policy design. This paper is a product of the Poverty and Equity Global Department. It is part of a larger effort by the World Bank toprovide open access to its research and make a contribution to development policy discussions around the world. PolicyResearch Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contactedat pcorralrodas@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about developmentissues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry thenames of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely thoseof the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank andits affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Stress-Testing Survey-to-Survey Imputation: UnderstandingWhen Poverty Predictions Can Fail∗ Paul Corral, Andres Ham, Peter Lanjouw, Leonardo Lucchetti, and Henry Stemmler† Key words:Poverty, Inequality, Poverty imputation, missing data JEL classification:I32, C53 1Introduction The measurement and monitoring of poverty are central to assessing global development progress. Forthe World Bank, whose mission is to eradicate extreme poverty on a livable planet, the ability to trackpoverty reduction is fundamental for measuring institutional effectiveness and guiding policy decisions.Yet, a persistent challenge hampers this crucial task:the limited availability of recent, high-qualityhousehold survey data that includes information that allows for comparable measures of expenditure orincome across time. This challenge is particularly acute in countries where poverty is most concentrated(Dang et al., 2017; World Bank, 2024). The issue affects many countries, and hinders poverty monitoring due to the lack of data in some ofthe world’s most populous and poverty afflicted economies. For example, India and Nigeria account fora substantial share of the world’s poor, meaning that they heavily influence global poverty trends.InIndia, official consumption survey data was unavailable between 2011 and 2022.1In Nigeria comparablesurvey data for poverty monitoring was absent between 2009 and 2018, and as of 2025 the country hasno new publicly available consumption survey data. These data gaps not only affect our understandingof poverty at the country level but compromise our ability to produce reliable global poverty estimatesand assess progress toward poverty reduction goals. Traditional poverty measurement relies on household surveys that collect detailed consumption or incomedata.These surveys represent substantial investments in both financial and human resources.Theyrequire extensive preparation, careful implementation, and can place considerable burden on respondinghouseholds, who must either maintain detailed consumption diaries or participate in comprehensive recallinterviews. The method of data collection itself can introduce significant biases into poverty estimateswhich