Cohort analysis using LIS data
The Lexis table US - us86 us91 us94 us00 us04 us10 FR - fr84 fr89 fr94 fr00 fr05 fr10 Age 25-65 5X5 years Lexis table Variables - nhhmem Educlev year5 pweight iso2 female partner immigr page ter leqdhi File name: $mydata/prosoc/us_fr_APC.dta
Variables Iso2 – Country code Year5 – period (5 years interval based on the actual time of the wave) Page – age (5 years interval) Pweight – weights Female – dummy variable for sex (male omitted) Ter – tertiary education dummy (based on Educlev) Leqdhi – (log) ppp adjusted DHI divided by the square root of HH members
APCD Cohort bumps in leqdhi (deviation from the overall trend) Age 25 to 55 For each country separately
Lissy syntax (stata) use $mydata/prosoc/us_fr_APC.dta , clear levelsof iso2, local(cnty) foreach l of local cnty { capt noi di "`l'" apcd leqdhi [w=pw] if page>=25 & page<=55 & iso2=="`l'" , age(page) period (year5) }
The output
Easier to plot output use $mydata/prosoc/us_fr_APC.dta , clear levelsof iso2, local(cnty) foreach l of local cnty { capt noi di "`l'" capture apcd leqdhi [w=pw] if page>=25 & page<=55 & iso2=="`l'" , age(page) period (year5) capture estout , cells(b ci_u ci_l) matlist r(coefs) }
The output
Plotting the output Copy the entire output (including everything) and paste it into a .txt file Open “output to graph.do” In line 1, change the folder name to the folder in which your .txt file is In line 2, change the file name to your .txt file name (don’t forget the .txt suffix!)
Inter-cohort income inequalities – the US and France
APCtlag Cohort trend in leqdhi (deviation from the overall trend) Age 25 to 55 For each country separately First, with no controls, next with controls for partner, female and tertiary education
Syntax (no control varaibles) use $mydata/prosoc/us_fr_APC.dta , clear levelsof iso2, local(cnty) foreach l of local cnty { capt noi di "`l'" apctlag leqdhi [w=pw] if page>=25 & page<=55 & iso2=="`l'" , age(page) period (year5) rep(2) }
Results (using the same .do file as before)
Syntax (with control variables) apctlag leqdhi partner female ter [w=pw] if page>=25 & page<=55 & iso2=="`l'" , age(page) period (year5) rep(2)
Results
APCGO apcgo leqdhi partner female [w=pw] if page>=25 & page<=55 & iso2=="`l'" , age(page) period (year5) gap(ter) rep(2)
Output
Output to graph We use a different .do file: output to graph (apcgo).do Again, we need change the folder and file names
Results
APCTlag of the Gini coeffcient We calculate the Gini for each 5X5 cell We give each case the value of Gini of its cell APCTlag of the new varaible
Syntax gen gini=. foreach l of local cnty { capt noi di "`l'" qui { forvalues ii=1985(5)2010 { forvalues jj=25(5)55{ fastgini leqdhi [w =pw] if year5==`ii' & page==`jj' & iso2=="`l’” replace gini=r(gini) if year5==`ii' & page==`jj' & iso2=="`l'"
Syntax (.cont) } apctlag gini [w=pw] if page>=25 & page<=55 & iso2=="`l'" , age(page) period (year5) rep(2)
Results
Exercise $mydata/prosoc/APC_WS.dta is similar dataset (Lexis table) for more countries: ca de es fi fr it no dk lu nl us il au Using only two countries, estimate: APCD and APCTlag models of leqdhi APCGO model of gender gap in leqdhi APCTlag model of Gini inequality