Balcazar Salazar, Carlos2015-08-172015-08-172015-07https://hdl.handle.net/10986/22463When measuring inequality of opportunity, researchers usually opt to eliminate within-type variation. Provided that in practice it is impossible to observe all circumstances, this implies that the researcher estimates a lower bound of the true level of inequality of opportunity. By using data drawn from 27 Demographic Household Surveys (circa 2008), it is found that lower bound estimates can have substantial measurement error, and that measurement error can vary considerably across countries. As a consequence, lower bound estimates of inequality of opportunity can demand too little redistribution to equalize inequalities due to circumstances and can make the “traditional” cross-country comparisons misleading.en-USCC BY 3.0 IGOMETHODSEMPIRICAL APPLICATIONINEQUALITY OF OPPORTUNITYDECOMPOSABLE INEQUALITY MEASURESEMPIRICAL APPROACHESTIMERESEARCH WORKING PAPERSINCOMENORMAL DISTRIBUTIONAGEGENDERDISTRIBUTION FUNCTIONBRAIN RESEARCHINEQUALITIESPOLITICAL ECONOMYRESEARCHERSEFFECTSMEAN LOG DEVIATIONPOLICY DISCUSSIONSSAMPLESSIZEDOWNWARD BIASHIGH CORRELATIONSURVEYSRESEARCH PAPERSMEASURESECONOMIC INEQUALITYSEXWEIGHTECONOMIC DEVELOPMENTLABOR MARKETTHEORYCHILDRENCROSS-COUNTRY COMPARISONSDISTRIBUTION OF OUTCOMESESTIMATINGCHILDMETHODOLOGIESEQUALITYEDUCATIONAL ATTAINMENTCLEAN WATERINEQUALITY MEASURESEQUALITY OF OPPORTUNITYESTIMATESMEASUREMENT ERRORPOLICY RESEARCHHOUSEHOLD SURVEYSMEASURING INEQUALITYRESEARCHSOCIAL ISSUESCOMPENSATIONPRODUCTUNEQUAL OPPORTUNITIESTOTAL INEQUALITYFEMALELINEAR REGRESSIONINEQUALITYWorking PaperWorld Bank10.1596/1813-9450-7379