--------------------------------------------------------------------------------------- name: log: /Users/ivan/Dropbox/ivanf/ExtrapoLATE/Data/tables1&2.log log type: text opened on: 20 Dec 2014, 15:18:42 . use m_d_806.dta, clear . . . ********************************************* . ********1. Cleaning the raw data******************* . ********************************************* . . gen multi2nd = (ageq2nd == ageq3rd) . . gen educm = gradem - 3 . replace educm = gradem - 2 if fingradm == 2 | fingradm == 1 (803347 real changes made) . replace educm = max(0,educm) (2458 real changes made) . . . gen blackm= ( racem==2) . gen hispm= ( racem==12) . gen whitem= ( racem==1) . gen othracem = 1 - blackm - hispm - whitem . . . gen boy1st = (sexk==0) . gen boy2nd = (sex2nd==0) . gen boys2 = (sexk==0 & sex2nd==0) . gen girls2 =(sexk==1 & sex2nd==1) . gen samesex =(boys2==1 | girls2==1) . . gen morekids = 1 if kidcount>2 (676104 missing values generated) . *not sure if the sas code makes morekids a dummy variable or just equal to ki > dcount. . replace morekids = 0 if kidcount<=2 (676104 real changes made) . . . gen illegit=0 . gen yom = . (927267 missing values generated) . replace qtrmar = qtrmar - 1 (927267 real changes made) . replace yom = yobm + agemar if (qtrbthm <= qtrmar) (340508 real changes made) . replace yom = yobm + agemar + 1 if (qtrbthm>qtrmar) (586759 real changes made) . gen dom_q = yom + (qtrmar/4) . gen dolb_q = yobk + ((qtrbkid)/4) . replace illegit = 1 if ((dom_q - dolb_q)>0 ) (108294 real changes made) . . gen yobd=79 - aged (164424 missing values generated) . replace yobd = 80 - aged if qtrbthd==0 (187601 real changes made) . . gen agem1 = agem*1 . gen aged1 = aged*1 (164424 missing values generated) . gen ageqm = 4*(80 - yobm)-qtrbthm-1 . gen ageqd = 4*(80 - yobd) - qtrbthd (164424 missing values generated) . gen agefstd = int((ageqd - ageqk)/4) (164424 missing values generated) . gen agefstm = int((ageqm - ageqk)/4) . gen msample = 0 . replace msample = 1 if ((aged!=.) & (timesmar==1) & (marital==0) & (illegit==0) & (ag > efstd >=15) & (agefstm >= 15) & !mi(agefstd)) (596315 real changes made) . . . gen weeksm1 = weeksm*1 . gen weeksd1 = weeksd*1 (164424 missing values generated) . gen workedm = 0 . replace workedm = 1 if weeksm>0 (565443 real changes made) . gen workedd = 0 . replace workedd = 1 if weeksd>0 (905599 real changes made) . gen hourswd = hoursd*1 (164424 missing values generated) . gen hourswm = hoursm*1 . . *All women sample: . keep if ((agem1>=21 & agem1<=35) & (kidcount>=2) & (ageq2nd>4) & (agefstm>=15) /*& (a > gefstd>=15 | agefstd==.)*/ & (asex==0) & (aage==0) & (aqtrbrth==0) & (asex2nd==0) & ( > aage2nd==0)) (532427 observations deleted) . . . . . ******************************************* . ******** Table 1 ************************ . ******************************************* . . . *Means: . sum weeksm1 Variable | Obs Mean Std. Dev. Min Max -------------+-------------------------------------------------------- weeksm1 | 394840 20.83419 22.28601 0 52 . sum workedm Variable | Obs Mean Std. Dev. Min Max -------------+-------------------------------------------------------- workedm | 394840 .5654873 .4956935 0 1 . . . *OLS: . reg weeksm1 morekids agem1 agefstm boy1st boy2nd blackm hispm othracem Source | SS df MS Number of obs = 394840 -------------+------------------------------ F( 8,394831) = 4161.77 Model | 15250419.8 8 1906302.48 Prob > F = 0.0000 Residual | 180852844394831 458.051277 R-squared = 0.0778 -------------+------------------------------ Adj R-squared = 0.0777 Total | 196103263394839 496.666397 Root MSE = 21.402 ------------------------------------------------------------------------------ weeksm1 | Coef. Std. Err. t P>|t| [95% Conf. Interval] -------------+---------------------------------------------------------------- morekids | -8.978191 .0724282 -123.96 0.000 -9.120148 -8.836234 agem1 | 1.466036 .0108691 134.88 0.000 1.444732 1.487339 agefstm | -1.423913 .0132728 -107.28 0.000 -1.449928 -1.397899 boy1st | -.1153498 .0681431 -1.69 0.091 -.2489083 .0182087 boy2nd | -.1773649 .0681446 -2.60 0.009 -.3109263 -.0438036 blackm | 6.451669 .1079086 59.79 0.000 6.240171 6.663166 hispm | -.7810209 .2009055 -3.89 0.000 -1.17479 -.3872521 othracem | 2.860371 .2048039 13.97 0.000 2.458961 3.26178 _cons | 8.280615 .3316669 24.97 0.000 7.630558 8.930672 ------------------------------------------------------------------------------ . reg workedm morekids agem1 agefstm boy1st boy2nd blackm hispm othracem Source | SS df MS Number of obs = 394840 -------------+------------------------------ F( 8,394831) = 2798.11 Model | 5205.2351 8 650.654388 Prob > F = 0.0000 Residual | 91811.4601394831 .232533565 R-squared = 0.0537 -------------+------------------------------ Adj R-squared = 0.0536 Total | 97016.6952394839 .245712038 Root MSE = .48222 ------------------------------------------------------------------------------ workedm | Coef. Std. Err. t P>|t| [95% Conf. Interval] -------------+---------------------------------------------------------------- morekids | -.1764489 .0016319 -108.12 0.000 -.1796473 -.1732504 agem1 | .0241995 .0002449 98.82 0.000 .0237195 .0246794 agefstm | -.0291002 .0002991 -97.31 0.000 -.0296863 -.0285141 boy1st | -.0005312 .0015354 -0.35 0.729 -.0035404 .0024781 boy2nd | -.0040863 .0015354 -2.66 0.008 -.0070956 -.001077 blackm | .1060263 .0024313 43.61 0.000 .101261 .1107916 hispm | -.0309759 .0045267 -6.84 0.000 -.039848 -.0221037 othracem | .0420805 .0046145 9.12 0.000 .0330363 .0511248 _cons | .4829654 .0074729 64.63 0.000 .4683188 .497612 ------------------------------------------------------------------------------ . . *First stages: . reg morekids multi2nd Source | SS df MS Number of obs = 394840 -------------+------------------------------ F( 1,394838) = 5135.94 Model | 1218.87641 1 1218.87641 Prob > F = 0.0000 Residual | 93704.0413394838 .237322753 R-squared = 0.0128 -------------+------------------------------ Adj R-squared = 0.0128 Total | 94922.9177394839 .240409174 Root MSE = .48716 ------------------------------------------------------------------------------ morekids | Coef. Std. Err. t P>|t| [95% Conf. Interval] -------------+---------------------------------------------------------------- multi2nd | .6030987 .0084155 71.67 0.000 .5866046 .6195927 _cons | .3969013 .0007786 509.75 0.000 .3953753 .3984274 ------------------------------------------------------------------------------ . reg morekids samesex Source | SS df MS Number of obs = 394840 -------------+------------------------------ F( 1,394838) = 1460.96 Model | 349.934714 1 349.934714 Prob > F = 0.0000 Residual | 94572.983394838 .239523508 R-squared = 0.0037 -------------+------------------------------ Adj R-squared = 0.0037 Total | 94922.9177394839 .240409174 Root MSE = .48941 ------------------------------------------------------------------------------ morekids | Coef. Std. Err. t P>|t| [95% Conf. Interval] -------------+---------------------------------------------------------------- samesex | .059544 .0015578 38.22 0.000 .0564907 .0625973 _cons | .3719712 .0011075 335.87 0.000 .3698006 .3741418 ------------------------------------------------------------------------------ . . *Wald estimates (twins) . ivregress 2sls weeksm1 (morekids = multi2nd) Instrumental variables (2SLS) regression Number of obs = 394840 Wald chi2(1) = 26.71 Prob > chi2 = 0.0000 R-squared = 0.0138 Root MSE = 22.132 ------------------------------------------------------------------------------ weeksm1 | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- morekids | -3.276339 .6339241 -5.17 0.000 -4.518807 -2.033871 _cons | 22.15149 .2573002 86.09 0.000 21.64719 22.65579 ------------------------------------------------------------------------------ Instrumented: morekids Instruments: multi2nd . ivregress 2sls workedm (morekids = multi2nd) Instrumental variables (2SLS) regression Number of obs = 394840 Wald chi2(1) = 28.92 Prob > chi2 = 0.0000 R-squared = 0.0123 Root MSE = .49263 ------------------------------------------------------------------------------ workedm | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- morekids | -.0758759 .0141105 -5.38 0.000 -.103532 -.0482199 _cons | .5959943 .0057272 104.06 0.000 .5847691 .6072194 ------------------------------------------------------------------------------ Instrumented: morekids Instruments: multi2nd . . *Wald estimates (samesex) . ivregress 2sls weeksm1 (morekids = samesex) Instrumental variables (2SLS) regression Number of obs = 394840 Wald chi2(1) = 29.00 Prob > chi2 = 0.0000 R-squared = 0.0173 Root MSE = 22.093 ------------------------------------------------------------------------------ weeksm1 | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- morekids | -6.359578 1.181014 -5.38 0.000 -8.674324 -4.044833 _cons | 23.39115 .4761434 49.13 0.000 22.45792 24.32437 ------------------------------------------------------------------------------ Instrumented: morekids Instruments: samesex . ivregress 2sls workedm (morekids = samesex) Instrumental variables (2SLS) regression Number of obs = 394840 Wald chi2(1) = 25.11 Prob > chi2 = 0.0000 R-squared = 0.0142 Root MSE = .49217 ------------------------------------------------------------------------------ workedm | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- morekids | -.1318477 .0263098 -5.01 0.000 -.183414 -.0802815 _cons | .6184985 .0106072 58.31 0.000 .5977088 .6392882 ------------------------------------------------------------------------------ Instrumented: morekids Instruments: samesex . . *2sls (twins & samesex) . ivregress 2sls weeksm1 (morekids = multi2nd samesex) Instrumental variables (2SLS) regression Number of obs = 394840 Wald chi2(1) = 50.59 Prob > chi2 = 0.0000 R-squared = 0.0154 Root MSE = 22.114 ------------------------------------------------------------------------------ weeksm1 | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- morekids | -3.966153 .5576396 -7.11 0.000 -5.059107 -2.8732 _cons | 22.42884 .2269522 98.83 0.000 21.98402 22.87365 ------------------------------------------------------------------------------ Instrumented: morekids Instruments: multi2nd samesex . estat overid Tests of overidentifying restrictions: Sargan (score) chi2(1) = 5.27224 (p = 0.0217) Basmann chi2(1) = 5.27227 (p = 0.0217) . ivregress 2sls workedm (morekids = multi2nd samesex) Instrumental variables (2SLS) regression Number of obs = 394840 Wald chi2(1) = 50.69 Prob > chi2 = 0.0000 R-squared = 0.0133 Root MSE = .49239 ------------------------------------------------------------------------------ workedm | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- morekids | -.0883985 .0124164 -7.12 0.000 -.1127343 -.0640628 _cons | .6010292 .0050533 118.94 0.000 .5911248 .6109335 ------------------------------------------------------------------------------ Instrumented: morekids Instruments: multi2nd samesex . estat overid Tests of overidentifying restrictions: Sargan (score) chi2(1) = 3.50455 (p = 0.0612) Basmann chi2(1) = 3.50456 (p = 0.0612) . . . . ******************************************* . ******** Table 2 ************************ . ******************************************* . . gen age_c=1 if ageq2nd<=16 (271011 missing values generated) . replace age_c=0 if age_c!=1 (271011 real changes made) . gen hsgrad_c = 1 if educm==12 (205022 missing values generated) . replace hsgrad_c = 0 if hsgrad_c!=1 (205022 real changes made) . gen somecol_c = 1 if (educm>12 & educm<=15) (318887 missing values generated) . replace somecol_c = 0 if somecol_c !=1 (318887 real changes made) . gen colgrad_c = 1 if educm>15 (354535 missing values generated) . replace colgrad_c = 0 if colgrad_c!=1 (354535 real changes made) . . . * PANEL A for married sample (as reported in the paper) . . *Column 1 . foreach var in age_c hsgrad_c somecol_c colgrad_c { 2. quietly sum `var' if msample == 1 3. local `var'_1 = r(mean) 4. di ``var'_1' 5. } .34309175 .48770086 .20235851 .13167381 . . . *Column 3 (must be estimated first to estimate column2 ) . foreach var in age_c hsgrad_c somecol_c colgrad_c { 2. quietly reg morekids multi2nd if (`var'==1 & msample == 1) 3. mat beta1 = e(b) 4. quietly reg morekids multi2nd if msample == 1 5. mat beta2 = e(b) 6. local num = beta1[1,1] 7. local denom = beta2[1,1] 8. local `var'_3 = `num'/`denom' 9. di ``var'_3' 10. . } 1.3077702 1.0210578 1.0488898 1.1442391 . . *Column 2 . foreach var in age_c hsgrad_c somecol_c colgrad_c { 2. local `var'_2 = ``var'_3'*``var'_1' 3. di ``var'_2' 4. } .44868516 .49797079 .21225179 .15066633 . . . *Column 5 (must be estimated first to estimate column2 ) . foreach var in age_c hsgrad_c somecol_c colgrad_c { 2. quietly reg morekids samesex if (`var'==1 & msample == 1) 3. mat beta1 = e(b) 4. quietly reg morekids samesex if msample == 1 5. mat beta2 = e(b) 6. local num = beta1[1,1] 7. local denom = beta2[1,1] 8. local `var'_5 = `num'/`denom' 9. di ``var'_5' 10. . } .56451028 1.0566473 1.0465806 .70191626 . . *Column 4 . foreach var in age_c hsgrad_c somecol_c colgrad_c { 2. local `var'_4 = ``var'_5'*``var'_1' 3. di ``var'_4' 4. } .19367882 .51532778 .21178449 .09242399 . . . * PANEL A for all women sample (not reported in the paper) . . *Column 1 . foreach var in age_c hsgrad_c somecol_c colgrad_c { 2. quietly sum `var' 3. local `var'_1 = r(mean) 4. di ``var'_1' 5. } .31361817 .48074663 .192364 .10207932 . * means of ageq2nd and educm (age in years): . quietly sum ageq2nd . di r(mean)/4 6.5912224 . quietly sum educm . di r(mean) 12.125707 . . *Column 3 (must be estimated first to estimate column2 ) . foreach var in age_c hsgrad_c somecol_c colgrad_c { 2. quietly reg morekids multi2nd if `var'==1 3. mat beta1 = e(b) 4. quietly reg morekids multi2nd 5. mat beta2 = e(b) 6. local num = beta1[1,1] 7. local denom = beta2[1,1] 8. local `var'_3 = `num'/`denom' 9. di ``var'_3' 10. . } 1.3437758 1.0339078 1.0826552 1.1900255 . . *Column 2 . foreach var in age_c hsgrad_c somecol_c colgrad_c { 2. local `var'_2 = ``var'_3'*``var'_1' 3. di ``var'_2' 4. } .42143252 .4970477 .20826388 .12147699 . . . *Column 5 (must be estimated first to estimate column2 ) . foreach var in age_c hsgrad_c somecol_c colgrad_c { 2. quietly reg morekids samesex if `var'==1 3. mat beta1 = e(b) 4. quietly reg morekids samesex 5. mat beta2 = e(b) 6. local num = beta1[1,1] 7. local denom = beta2[1,1] 8. local `var'_5 = `num'/`denom' 9. di ``var'_5' 10. . } .6279348 1.0933298 1.0024985 .77931072 . . *Column 4 . foreach var in age_c hsgrad_c somecol_c colgrad_c { 2. local `var'_4 = ``var'_5'*``var'_1' 3. di ``var'_4' 4. } .19693176 .52561463 .19284463 .07955151 . . . * PANEL B for all women sample (as reported in the paper) . . * means of ageq2nd and educm (age in years): . quietly sum ageq2nd . di r(mean)/4 6.5912224 . quietly sum educm . di r(mean) 12.125707 . . . * kappa-weighted means: . /* > Here's what I usually do: > > reg Z X1 X2 X3 X4 X5 > predict p_Z, xb > gen kappa = 1-D*(1-Z)/(1-p_Z)-(1-D)*Z/p_Z > summ X1 [iweight=kappa] > */ . . . gen second_ageq2nd = ageq2nd^2 . gen third_ageq2nd = ageq2nd^3 . gen fourth_ageq2nd = ageq2nd^4 . . gen second_educm = educm^2 . gen third_educm = educm^3 . gen fourth_educm = educm^4 . . * age of second child (age in years) . . quietly logit multi2nd ageq2nd second_ageq2nd third_ageq2nd fourth_ageq2nd . quietly predict p_Z_multi2nd . gen kappa_multi2nd = 1-morekids*(1-multi2nd)/(1-p_Z_multi2nd)-(1-morekids)*multi2nd/p > _Z_multi2nd . quietly sum ageq2nd [iweight=kappa_multi2nd] . di r(mean)/4 5.5106271 . . quietly logit samesex ageq2nd second_ageq2nd third_ageq2nd fourth_ageq2nd . quietly predict p_Z_samesex . gen kappa_samesex = 1-morekids*(1-samesex)/(1-p_Z_samesex)-(1-morekids)*samesex/p_Z_s > amesex . quietly sum ageq2nd [iweight=kappa_samesex] . di r(mean)/4 7.137047 . . * mother's schooling: . * reset the weights: . drop p_Z_multi2nd p_Z_samesex kappa* . . quietly logit multi2nd educm second_educm third_educm fourth_educm . quietly predict p_Z_multi2nd . gen kappa_multi2nd = 1-morekids*(1-multi2nd)/(1-p_Z_multi2nd)-(1-morekids)*multi2nd/p > _Z_multi2nd . quietly sum educm [iweight=kappa_multi2nd] . di r(mean) 12.429212 . . quietly logit samesex educm second_educm third_educm fourth_educm . quietly predict p_Z_samesex . gen kappa_samesex = 1-morekids*(1-samesex)/(1-p_Z_samesex)-(1-morekids)*samesex/p_Z_s > amesex . quietly sum educm [iweight=kappa_samesex] . di r(mean) 12.073305 . . log close name: log: /Users/ivan/Dropbox/ivanf/ExtrapoLATE/Data/tables1&2.log log type: text closed on: 20 Dec 2014, 15:19:06 ---------------------------------------------------------------------------------------