世界银行-压力测试调查到调查推断：了解贫困预测何时可能失败（英）-2025.8.pdfVIP

d

e

z

i

r

o

h

t

u

A

e

rPolicyResearchWorkingPaper11192

u

s

o

l

c

s

i

D

c

i

l

b

u

P

d

e

z

i

r

o

h

t

u

A

e

r

uStressTestingSurveytoSurveyImputation

s

o

l

c

s

i

D

c

i

lUnderstandingWhenPovertyPredictionsCanFail

b

u

P

PaulCorral

dAndresHam

e

z

i

rPeterLanjouw

o

h

t

uLeonardoLucchetti

A

e

rHenryStemmler

u

s

o

l

c

s

i

D

c

i

l

b

u

P

d

e

z

i

r

o

h

t

u

A

e

r

u

s

o

l

c

s

i

D

c

i

l

b

u

P

PovertyandEquityGlobalDepartment

August2025

PolicyResearchWorkingPaper11192

Abstract

Accurateandtimelypovertymeasurementiscentraltoshiftsinsurveydesigncanintroducesubstantialerrors—

developmentpolicy,yettheavailabilityofup-to-dateparticularlywhenimputingacrosstimeorundereconomic

high-qualityhouseholdsurveydataremainslimited—par-andstructuralchange.Resultsshowthatstandardcorrec-

ticularlyincountrieswherepovertyismostconcentrated.tionssuchasre-weightingorcovariatestandardizationmay

Survey-to-surveyimputationhasemergedasapracticalfailtoeliminatethesebiases,especiallywhenimputing

responsetothischallenge,allowingpractitionerstoupdateacrosstimeorunderstructuralchange.Theperformanceof

povertyestimatesusingrecentsurveysthatlackdirectalternativemodelspecificationsisalsoevaluatedundervari-

wel