Gibson, JohnBeegle, KathleenDe Weerdt, JoachimFriedman, Jed2013-04-112013-04-112013-02https://hdl.handle.net/10986/13153This paper uses data from eight different consumption questionnaires randomly assigned to 4,000 households in Tanzania to obtain evidence on the nature of measurement errors in estimates of household consumption. While there are no validation data, the design of one questionnaire and the resources put into its implementation make it likely to be substantially more accurate than the others. Comparing regressions using data from this benchmark design with results from the other questionnaires shows that errors have a negative correlation with the true value of consumption, creating a non-classical measurement error problem for which conventional statistical corrections may be ineffective.en-USCC BY 3.0 IGOAGRICULTURAL ECONOMICSBENCHMARKBIASESCANDLESCIGARETTESCOMMODITYCONSUMPTION MEASUREDEMAND FOR FOODDEPENDENT VARIABLEDEVELOPING COUNTRIESDEVELOPMENT ECONOMICSDEVELOPMENT POLICYEARNINGS LEVELSECONOMETRICSECONOMIC ACTIVITYECONOMIC REVIEWECONOMIC STUDIESECONOMIES OF SCALEEMPIRICAL RESEARCHEMPIRICAL RESULTSENGEL CURVEENGEL CURVESEQUATIONSEXPENDITURE DATAEXPLANATORY VARIABLEFOOD ^ CONSUMPTIONFOOD CONSUMPTIONFOOD DEMANDFOOD RECALLFOODSHIGH CORRELATIONHOUSEHOLD BUDGETHOUSEHOLD BUDGET SURVEYSHOUSEHOLD CHARACTERISTICSHOUSEHOLD CONSUMPTIONHOUSEHOLD SIZEHOUSEHOLD SURVEYHOUSEHOLD SURVEYSHOUSEHOLDSHYPOTHESESINCOME GROWTHINCOME INCREASEINEQUALITYINSTRUMENTAL VARIABLESLABOR ECONOMICSLABOR MARKETLIVING STANDARDSMEASURED CONSUMPTIONMEASUREMENT ERRORMEASUREMENT ERRORSNEGATIVE CORRELATIONNEGATIVE EFFECTOBSERVED VALUEPER CAPITA CONSUMPTIONPERMANENT INCOMEPOLICY DISCUSSIONSPOLICY RESEARCHPOLITICAL ECONOMYPOOR COUNTRIESPOVERTY REDUCTIONPUBLIC GOODSRANDOM ERRORSRELIABILITYSIGNIFICANT DIFFERENCESSIGNIFICANT IMPACTSIGNIFICANT NEGATIVESTANDARD ERRORSSTATISTICAL SIGNIFICANCESUBSTITUTION EFFECTTOTAL CONSUMPTIONTOTAL EXPENDITUREWorld Bank10.1596/1813-9450-6372