220 likes | 320 Views
A first estimate of LCD by gender (Uruguay). Marisa Bucheli Cecilia González dECON , FCS, Udelar. In Uruguay we are doing NTA by SES We have estimations of labor income, consumption, LCD and public transfers We have preliminary estimations of RA and private transfers
E N D
A first estimate of LCD by gender (Uruguay) Marisa Bucheli Cecilia González dECON, FCS, Udelar
In Uruguay we are doing NTA by SES • We have estimations of labor income, consumption, LCD and public transfers • We have preliminary estimations of RA and private transfers • We recently began to think of doing estimations by gender • Wehavenotworkedonunpaidactivities
NTA by SES groups • To my best knowledge, in the Latin American team we used (at least at the beginning) different procedures for estimation • In our case, we have estimations basead on two differente procedures (the one we used at the beginning and the proposed late by CELADE) • But we have not compared the sensitivity of the results to the procedures
First procedure • We estimate the profiles (mean and smooth mean) as usual but for each SES group separately • Note that all members of the hh belong to the same group so the only challenge is define groups with a “good” size in all ages • We estimate the aggregated value (AV) of each group (g) and age (a), where P is the population and XS is the smooth microdata value: • The Total AV is the sum of the Total AV of groups
First procedure • In ordertocalculatethe formula, weneedtoknowthepopulation of eachage-group • Weestimateitusingitsproportion in thesurvey • Note: in thedefinition of theclassification, wetookintoaccount (¿?) thesize of theage-grouppopulation in thesurvey
Second procedure • We estimate the total AV of each group using the weight of the group in the microdata • We estimate the total AV of the age-group using the weight of the age-group in the microdata (X is the mean value in the microdata): • We estimate the mean value as AV of the age-group / Population in the age-group • In the analysis of the data we work with five-year-age group
Up tonow… • Wehave a complete NTA estimation (though a preliminaryversionparticularly of privatetransfers and RA) followingthefirstprocedure (usingthe “educationallevel of thehhadults” as the proxy of SES) • Manychallenges: 1) ¿inter-hhtransfers?; 2) public RA; …. • We have estimations of labor income, consumption and public transfers following the second procedure (using the “educational level of the hh head 2” as the proxy of SES)
NTA by gender • Our first idea was to follow the first procedure to estimate NTA by gender • Two differences: • In the gender classification we know the population of each age-group. We used it • In the SES classification, all the members of a hh belong to the same group. In the classification by sex, it is possible that members of the same hh belong to a different group • This issue is not important in the estimation of accounts for which we have individual information in the surveys: labor income, some components of private consumption, public inflows and some public outflows
Butiftheinformationisgiven at household-level … • Privateeducation: wefollowexactlythesameprocedurethan in NTA: • In thesurvey, weidentifythestudents (and theirlevel of education) thatattendprivateschool. Weassigntoeachonetheamount of thetuitionspaidbythehh. • In the case otherspending (books, courses of language, computation, etc.) we use themethodproposedby NTA • Weclassifythepersonsbyage and sex in ordertocalculate mean and smooth mean values
Thedifferenceisduetospendingnotrelatedtoattendance (apparently, to “enseñanza no curricular” -language, computation, specialteachers, etc.-) ¿Isitthemethod? ¿Shouldwetakeintoaccountthe sex-composition of thehhwhenwehavetoassignspendinginformed at hhlevel?
Butiftheinformationisgiven at household-level … • Thisisthe case of most of theprivateconsumption and indirecttaxes
Butiftheinformationisgiven at household-level … • Privatehealth: wefollowexactlythesameprocedurethan in NTA: • In thesurvey, weidentifythepersonswhowereill. Weassigntoeachonetheamount of thespendingrelatedtobeill. • In the case otherspendingwe use themethodproposedby NTA • Weclassifythepersonsbyage and sex in ordertocalculate mean and smooth mean values
We do notknowwhichcomponentsexplaintheincreasing gap Weshould explore ifitisdueto a componentassignedtoan individual throughanindirectmethod (notanill-relatedcomponent)
Butiftheinformationisgiven at household-level … • Rest of privateconsumption: wefollowexactlythesameprocedurethan in NTA: • Weusedanequivalencescaletocalculatetherest of privateconsumption per hhmember • Weassignedtoeach individual of thehhthesameamount • Weclassifythepersonsbyage and sex in ordertocalculate mean and smooth mean values
Somequestions • Wewouldliketoknow more aboutthegenderdifferencein theprivatehealthand privateeducation. • Are theysensitivetothemethod of allocation of spendinginformed at householdlevel? • Ifthereis a genderdifference in private / healtheducation, shouldwe use thetraditionalmethod of imputation of therest of privateconsumption? • Anotherchallenge: privatetransfers
Unpaidwork • Wehavenotworkedin thisissue in thelastyear • In thepast, weperformedsomeestimations of thevalue of unpaidactivities in whichweimputed a wagetounpaidwork: • Results quite sensitiveto use theopportunitycostcriteriaorreplacementcriteria • Alsosensitivetoconsiderspecialist / non-specialistwage in thereplacementcriteria • Thereis a new survey (2009) butwehadnotworkedwithityet