Online Appendix for the paper “Saving Lives: Evidence from a
Conditional Food Supplementation Program”
Marian W. Meller and Stephan Litschig November 26, 2013
List of Figures
1 Figure 1: Impact on child mortality, maximum differential exposure cohorts, h=10 . 2 2 Figure 2: Impact on child mortality, seasonally matched non-exposed cohorts . . . 3 3 Figure 3: Impact on child mortality, between 1 and 8 months differential exposure
cohorts . . . 4 4 Figure 4: Impact on child mortality, maximum differential exposure cohorts,
be-fore PANN 2000 . . . 5 5 Figure 5: Impact on child mortality, maxium differential exposure cohorts, post
scaling up of PANN 2000 . . . 6
List of Tables
1 Table 1: Impact on fertility, kids born between April and November 2001 . . . 7 2 Table 2: Impact on fertility, kids born between September 2000 and November 2001 8 3 Table 7.1: Impact on health check-ups, quadratic speci cation . . . 9 4 Table 8.1: Impact on child mortality, seasonally matched non-exposed cohorts . . . 10 5 Table 8.2: Impact on child mortality, between 1 and 8 months differential exposure
cohorts . . . 11 6 Table 3: Impact on child mortality, maximum differential exposure cohorts, before
PANN 2000 . . . 12
7 Table 4: Impact on child mortality, maximum differential exposure cohorts, post scaling-up of PANN 2000 . . . 13
Figure 1: Impact on child mortality, maximum differential exposure cohorts, h=10
-.025 -.015 -.005 .005 .015 .025Chil
d mort
alit
y rat
e (res
idua
l, bin
mea
n)
.7905 .8105 .8305 .8505 .8705 .8905 .9105 .9305 .9505 .9705 .9905Consumption poverty index
Program provinces
-.025 -.015 -.005 .005 .015 .025Chil
d mort
alit
y rat
e (res
idua
l, bin
mea
n)
.7905 .8105 .8305 .8505 .8705 .8905 .9105 .9305 .9505 .9705 .9905Consumption poverty index
Linear fit Quadratic fit
Sierra provinces
Notes:
The unit of observation is the rural
parroquia
. The sample is restricted to
parroquias
within 10 percentage point distance to the cutoff. The residuals are computed from regressions
with the control variables listed in Table 6. Program provinces are Bolívar, Chimborazo,
Imbabura, and Manabí. Sierra provinces are Bolívar, Chimborazo, and Imbabura.
Figure 2: Impact on child mortality, seasonally matched non-exposed cohorts
-.025 -.005 .015 .035Chil
d mort
alit
y rat
e (res
idua
l, bin
mea
n)
.8305
.8505
.8705
.8905
.9105
.9305
.9505
Consumption poverty index
Program provinces
-.005 -.025 .015 .035Chil
d mort
alit
y rat
e (res
idua
l, bin
mea
n)
.8305
.8505
.8705
.8905
.9105
.9305
.9505
Consumption poverty index
Linear fit
Quadratic fit
Sierra provinces
Notes:
The unit of observation is the rural
parroquia
. The sample is restricted to
parroquias
within 6 percentage point distance to the cutoff. The residuals are computed from regressions
with the control variables listed in Table 8.1 of the Appendix. Program provinces are Bolívar,
Chimborazo, Imbabura, and Manabí. Sierra provinces are Bolívar, Chimborazo, and Imbabura.
Figure 3: Impact on child mortality, between 1 and 8 months differential exposure
cohorts
-.025 -.015 -.005 .005 .015 .025Chil
d mo
rtal
ity ra
te (
resid
ual, b
in me
an)
.8305
.8505
.8705
.8905
.9105
.9305
.9505
Consumption poverty index
Program provinces
-.025 -.015 -.005 .005 .015 .025Chil
d mo
rtal
ity ra
te (
resid
ual, b
in me
an)
.8305
.8505
.8705
.8905
.9105
.9305
.9505
Consumption poverty index
Linear fit
Quadratic fit
Sierra provinces
Notes:
The unit of observation is the rural
parroquia
. The sample is restricted to
parroquias
within 6 percentage point distance to the cutoff. The residuals are computed from regressions
with the control variables listed in Table 8.2 of the Appendix. Program provinces are Bolívar,
Chimborazo, Imbabura, and Manabí. Sierra provinces are Bolívar, Chimborazo, and Imbabura.
5Figure 4: Impact on child mortality, maximum differential exposure cohorts,
be-fore PANN 2000
-.025 -.015 -.005 .005 .015 .025Chil
d mo
rtal
ity ra
te (
resid
ual, b
in me
an)
.8305 .8505 .8705 .8905 .9105 .9305 .9505Consumption poverty index
Program provinces
-.025 -.015 -.005 .005 .015 .025Chil
d mo
rtal
ity ra
te (
resid
ual, b
in me
an)
.8305 .8505 .8705 .8905 .9105 .9305 .9505Consumption poverty index
Linear fit Quadratic fit
Sierra provinces
Notes:
The unit of observation is the rural
parroquia
. The sample is restricted to
parroquias
within 6 percentage point distance to the cutoff. The residuals are computed from regressions
with the control variables listed in Table 3 of the Appendix. Program provinces are Bolívar,
Chimborazo, Imbabura, and Manabí. Sierra provinces are Bolívar, Chimborazo, and Imbabura.
Figure 5: Impact on child mortality, maxium differential exposure cohorts, post
scaling up of PANN 2000
-.025 -.015 -.005 .005 .015 .025Chil
d mo
rtal
ity ra
te (
resid
ual, b
in me
an)
.8305 .8505 .8705 .8905 .9105 .9305 .9505Consumption poverty index
Program provinces
-.025 -.015 -.005 .005 .015 .025Chil
d mo
rtal
ity ra
te (
resid
ual, b
in me
an)
.8305 .8505 .8705 .8905 .9105 .9305 .9505Consumption poverty index
Linear fit Quadratic fit
Sierra provinces
Notes:
The unit of observation is the rural
parroquia
. The sample is restricted to
parroquias
within 6 percentage point distance to the cutoff. The residuals are computed from regressions
with the control variables listed in Table 4 of the Appendix. Program provinces are Bolívar,
Chimborazo, Imbabura, and Manabí. Sierra provinces are Bolívar, Chimborazo, and Imbabura.
Table
1:
Impact
on
fertility
,kids
born
between
April
and
No
vember
2001
Dependent var iable: num ber of liv e-births between April
and Novem ber 2001 per wo m an aged 15 -4 4 y ears in November 2001 Neighbor hood h (percentage points) 3 4 5 6 P re -program controls N Y N Y N Y N Y Compa rison m ean Panel A: B olív ar, Chim borazo, Im babura, Manabí -0. 009 0. 004 -0. 004 -0. 0 01 0. 001 -0. 0 01 0. 003 0. 0 01 Treatm ent parroquia (yes = 1) 0 .13 (0. 015 ) (0. 013 ) (0. 015 ) (0. 015 ) (0. 011 ) (0. 011 ) (0. 011 ) (0. 0 10 ) Observat ions 48 48 58 58 70 70 75 75 R 2 0. 1 10 0. 490 0. 086 0. 350 0. 062 0. 294 0. 059 0. 273 Com pari son m ean Panel B: Bolív ar, Chim borazo, I mbabura -0 .024 -0.020 -0.008 0. 002 -0. 006 -0.0 10 0. 002 0. 0 00 Treatm ent parroquia (yes = 1) 0 .15 (0. 021 ) (0. 028 ) (0. 022 ) (0. 024 ) (0. 017 ) (0. 019 ) (0. 017 ) (0. 018 ) Observat ions 30 30 39 39 46 46 49 49 R 2 0. 2 19 0. 62 0 0. 152 0. 402 0. 112 0. 335 0. 077 0. 277 Notes: OLS estimat es. The unit of observat ion is the rural parroquia . h is the per centage poi nt distance from the eligibi lity cutoff 0.8905. P re -program controls include the n u m b er o f l iv e -births between April and Novem ber 1990 per wom an aged 15 -44 y ears in Novem ber 1990 , num ber of check -ups at health posts per 0 to 4 year -old during y ear 1999 (1999 heal th census), proportion of last -born children born between April 1988 and F ebruary 1989 who had died by Novemb er 1990 (1990 census), proportion of stunted children under 5 y ears old in 1990 (199 0 census and 1988 nutr itio n survey ), an indicator for the presence of a heal th center in the parro quia in 2000 (20 00 health census ), proport io n of wom en aged 18 or older wh o com pleted prim ary schooling (2001 census), proport ion of dwelli ngs wit h access to piped water (2001 census) , proportion of households w ith toilet (2001 census), average number of persons per room (2001 census), number of last -born children bor n between A pril 1988 and F ebruary 1989 liv ing in the parroquia in 1990 (1990 census) . Comparis on m eans are the estim ated constant term s in h = 3 without contr
ols. Robust standard errors in parenthes
es. *, ** , and ** * indica te significance at 10%, 5%, a nd 1% lev el s, respecti vely.
Table
2:
Impact
on
fertility
,kids
born
between
September
2000
and
No
vember
2001
Dependent var iable: num ber of liv e -births between Septe m ber 2000 and Nov emb er 2001 per wom an aged 15 -4 4 years in Novem ber 2001 Neighbor hood h (percentage points) 3 4 5 6 P re -program controls N Y N Y N Y N Y Com par ison m ean Panel A: Bolí var, Chim borazo, Im babura, Manabí 0. 0 10 0. 0 15 0. 00 7 0. 0 01 0. 00 6 -0. 0 0 3 0. 00 5 -0. 0 0 2 Treatm ent parroquia (yes = 1) 0 .22 (0. 0 21 ) (0. 0 18 ) (0. 0 21 ) (0. 01 7 ) (0. 01 6 ) (0. 01 4 ) (0 .01 5 ) (0. 0 1 3 ) Observat ions 48 48 58 58 70 70 75 75 R 2 0. 044 0. 4 37 0. 0 41 0. 459 0. 0 45 0. 388 0. 05 8 0. 414 Com pari son m ean Panel B: Bolív ar, Chim bo razo, Imbabura 0. 009 -0.02 6 0. 009 -0. 01 5 0. 014 -0.0 1 5 0. 013 -0. 0 0 8 Treatm ent parroquia (yes = 1) 0.23 (0. 036 ) (0. 03 7 ) (0. 033 ) (0. 032 ) (0. 026 ) (0. 023 ) (0. 024 ) (0. 020 ) Observat ions 30 30 39 39 46 46 49 49 R 2 0. 06 2 0. 650 0. 060 0. 5 50 0. 041 0. 491 0. 053 0. 494 Notes: OLS estimat es. The unit of observat ion is the rural parroquia . h is the per centage poi nt distance from the eligibi lity cutoff 0.8905. P re-program controls includ
e the num ber of live -births b etween Septem ber 1989 and Novem ber 1990 per wom an aged 15 -44 y ears in November 1990 , number of check -ups at health post s per 0 to 4 year -old during y ear 1999 (1999 health census) , propor tion of last -born children bor n between April 1988 and February 1989 who had died by Novem ber 1990 (1990 censu s), proportion of stu nted childr en under 5 y ears old in 1990 (1990 cens us and 1 988 nut rit ion surv ey), an indica tor for the presence of a h ealth cent er in th e parroqui a in 200 0 (2000 he alth c ensus), pr oport ion of wom en aged 18 or older who com pleted prim ary schooling (2001 census) , propor tion of dwell ings wit h access to piped water (2001 census ), proportion of households w ith toilet (2001 census), average number of persons per room (2001 census), number of last -born childr en born between April 1988 and F ebruary 1989 liv ing in the parroquia in 1990 (1990 census) . Comparis on m eans are the estim ated constant term s in h = 3 without contr
ols. Robust standard errors in parenthes
es. *, ** , and ** * i ndicate sign ificance at 10%, 5 %, and 1% l evels , respectiv el y . 9
Table
7.1:
Impact
on
health
check-ups,
quadratic
specication
Dependent v ariable: Num ber of check -ups at healt h post s per 0 to 4 y ear -old duri ng y ear 2000 Neighborhood h (percent age points ) 3 4 5 6 P re -program controls N Y N Y N Y N Y Compar ison m ean Panel A: Bolí var , Chim borazo, Im babura, Manabí 0.00 1 0.377 0.119 0.312 0.072 0.172 0.144 0.235 Treatm ent parroquia (ye s = 1) 0.869 (0.436) (0.411) (0.373) (0.313) (0.356) (0.256) (0.333) (0.268) F -test (quadratic term s) 1.15 9 0.116 1.461 0.311 0.211 0.065 0.666 0.861 [p -value ] [0.324] [0.891] [0.241] [0.734] [0.810] [0.937] [0.517] [0.428] Observati ons 48 48 58 58 70 70 75 75 R 2 0.04 5 0.581 0.110 0.612 0.02 4 0.580 0.022 0.574 Com parison m ean Panel B: Bolí var, Chim borazo, Im babura 0.460 1.122 0.920** 1.504** 1.165** 1.340** 0.986* 1.1 80 ** Treatm ent parroquia (ye s = 1) 0.783 (0.339) (0.640) (0.439) (0.663) (0.495) (0.557) (0.499) (0.467) F -test (quadrat ic terms ) 1.949 0.519 0.569 1.252 1.069 0.995 1.284 3.275 [p -value ] [0.164] [0.606] [0.572] [0.305] [0.353] [0.382] [0.287] [0.050] Observati ons 30 30 39 39 46 46 49 49 R 2 0.196 0.770 0.220 0.741 0.188 0.747 0.159 0.728 Notes: OL S e st im at es . T he un it of ob se rv at ion i s t he ru ral parroquia . h is the perce ntage point distance from
the eligi bility cutoff 0.8905 . T h e total number of check -u p s o f 0 t o 4 y ea r-olds is norm alized by the esti m ated populat ion of 0 to 4 y ear -olds in 2000 based on F lores (2001). Pr e -program control s include num ber of check -ups at health posts per 0 to 4 y ear -old during y ear 1999 (1999 health census), proport ion of last -bo rn children born between April 1988 and F ebruary 1989 who had died by Novem ber 1990 (1990 censu s), proportion of stunted children under 5 yea rs old in 1990 (199 0 cen sus and 1988 nutri tion surv ey) , an indi cator for the p resence of a heal th center in the parroquia in 2000 (2 000 hea lth census), propor tion of wom en aged 18 or older who com pleted primary schooling (2001 census), proporti on of dwell ings wit h access to piped wate r (2001 census), propor tion of househol ds wit h to
ilet (2001 census), average number of persons per room (2001
census), num ber of last -bo rn child ren born betw een April 1999 and F ebruary 2000 living in the parroquia in 2001 (2001 census), num ber of last -born children born between April 1988 and F ebruary 19 89 livi ng in the parroquia in 1990 (19 90 censu s). Co m paris on m eans are the estim ated constant term s in h = 6 wit hout contr ols. Rob ust st andard error s in pare ntheses. *, ** , and *** indicat e significa nce at 10% , 5% , and 1% l evel s, respect iv el y .
Table
8.1:
Impact
on
child
mortality
,seasonally
matched
non-e
xposed
cohorts
Dependent variable: Proportion of last
-born chil dren born between April 1997 and F ebruary 19 9 8 who ha d died by N ovem ber 2001 Neighborhood h (percentage points) 3 4 5 6 P re -program controls N Y N Y N Y N Y Compar ison m ean Panel A: Bolív ar, Chim borazo, Im babura, Manabí 0.01 0 0.0 30* -0.00 5 0.00 0 0.00 3 0.00 7 0.0 11 0.0 1 4 Treatm ent parroquia (yes = 1) 0.0 14 (0.01 3 ) (0.01 6 ) (0.01 3 ) (0.01 6 ) (0.01 1 ) (0.01 2 ) (0.01 5 ) (0.01 6 ) Observations 48 48 58 58 70 70 75 75 R 2 0.0 28 0. 320 0.0 60 0. 195 0.0 03 0. 139 0.0 37 0. 154 Com pariso n m ean Panel B: Bol ív ar, Chim borazo, Im babura -0.0 0 4 0.0 41* -0.0 21 -0.01 1 -0.0 04 0.0 11 0.0 20 0.0 30 Treatm ent parroquia (yes = 1) 0.02 0 (0.02 3 ) (0.0 2 3 ) (0.01 9 ) (0.0 30 ) (0.01 6 ) (0.01 1 ) (0.0 23 ) (0.0 26 ) Observations 30 30 39 39 46 46 49 49 R 2 0.0 03 0. 623 0 .135 0. 345 0. 008 0. 189 0.0 80 0. 274 Notes: OLS estim ates. The unit of obser vati on is the rural parroquia . Child mort ality is the proportion of last -born children born betwee n October April 1997 and F ebruary 1998 who had died by Novem ber 2001 (2001 cen sus). P re -program controls include number of check -ups at health posts per 0 to 4 y ear -old during y ear 1999 (1999 health census), proporti on of last -born children born between October 1986 and August 1987 who had died by Novem ber 1990 (1990 census), pr opo rtion of stunted children under 5 y ears old in 1990 (1990 census and 1988 nut riti on survey ), an indicator for the presence of a healt h center in the parroquia in 2000 (2000
health census), proportio
n of wom en aged 18 or old er who compl eted prim ary schoolin g (2001 census), proporti on of dwelli ngs with access to piped water (2001 census), proportion of households with toilet (2001 census), averag e number of persons per room (2001 census), number of last -born children born betw een October 1997 and August 1998 livi ng in the parroquia in 2001 (2001 census), num ber of last -born children born between October 1986 and August 1987 liv ing in the parroquia in 1990 (1990 census) . Compari son m eans are the estim ated constant terms in h = 3 without controls. Robust stand ard errors in parentheses. *, ** , and *** indicate si gnificance at 10%, 5%, a nd 1% l evels , respectiv el y . 11
Table
8.2:
Impact
on
child
mortality
,between
1
and
8
months
dif
ferential
exposure
cohorts
Dependent var iable: Proporti on of last-born children born
between Septem ber 199 8 and August 2000 w ho had di ed by Novembe r 2001 Neighbor hood h (percentage points) 3 4 5 6 P re -program controls N Y N Y N Y N Y Com par ison m ean Panel A: Bolí var, Chim borazo, Im babura, Manabí -0.0 06 -0.0 09 -0.0 05 -0.0 08 -0.0 04 -0.0 07 -0.00 4 -0.0 07 Treatm ent parroquia (yes = 1) 0.02 2 (0.00 7 ) (0.00 7 ) (0.00 7 ) (0.00 6 ) (0.00 6 ) (0.0 0 6 ) (0.00 5 ) (0.00 6 ) Observat ions 48 48 58 58 70 70 75 75 R 2 0.0 30 0. 329 0. 008 0.4 07 0.0 07 0. 340 0.0 10 0. 338 Com pari son m ean Panel B: Bolív ar, Chim borazo, Im babura -0.0 08 -0.0 09 -0.0 07 -0.0 05 -0.0 05 -0.0 05 -0.0 04 -0.0 04 Treatm ent parroquia (yes = 1) 0.0 2 4 (0.01 0 ) (0.01 6 ) (0.01 0 ) (0.01 1 ) (0.0 09 ) (0.0 09 ) (0.0 08 ) (0.0 08 ) Observat ions 30 30 39 39 46 46 49 49 R 2 0.0 44 0. 385 0. 018 0.4 52 0.0 15 0.4 32 0.0 16 0. 438 Notes: OLS estimat es. The unit of observat ion is the rural parroquia . h is the per centage poi nt distance from the eligibi lity cutoff 0.8905. Child mort ality is the proportion of last -born chi ldren born between Sept em ber 1998 and August 2000 wh o had di ed by November 2001 (2001 census). Pre -program
controls include number of check
-ups at health posts per 0 to 4 y ear -old during y ear 1999 (1999 healt h census), proporti on of last -born children bor n bet ween Septem ber 198 7 and August 1989 who had died by November 1990 (1990 census), pro portion of stunt ed children under 5 y ears old in 1990 (1990 census and 1988 nutrit ion survey) , an indicat or for the presence of a health center in the parroquia in 200
0 (2000 health census), proporti
on of
wom
en aged
18 or older
who comple
ted primary schooling (2001
census ), proportion of dwell ings with access to piped water (2001 census), proportion of househol ds wit h toilet (2001 census) , average num ber of per sons per room (2001 census ), number of last -born children born between Septem ber 199 8 and August 2000 livi ng in the parroquia in 2001 (20 01 cens us), number of last -born children born between Septe m ber 198 7 and August 1989 living in the parroquia in 1990 (1990 cens u s). Comparison means are the estimated constant terms in h = 3
without controls. Robust standard error
s in parentheses. *, ** , and ** * indica te significance at 10%, 5%, a nd 1% lev els, respectiv el y .
Table
3:
Impact
on
child
mortality
,maximum
dif
ferential
exposure
cohorts,
before
PANN
2000
Dependent v ariable: Proportion of l ast -born childr en born between April 19 88 and F ebruary 1989who had died
by N ovem ber 1990 Neighborhood h (perce ntage points) 3 4 5 6 P re -program controls N Y N Y N Y N Y Com pa rison m ean Panel A: Bolív ar, Chim borazo, Im babura, Manabí -0.0 08 -0.0 07 -0.0 05 -0.0 0 2 -0.0 0 4 -0.0 0 3 -0.00 3 -0.0 0 2 Treatment parroquia (yes = 1) 0.0 3 1 (0.0 11 ) (0.0 1 1 ) (0.0 10 ) (0.0 10 ) (0.00 9 ) (0.00 9 ) (0.00 8 ) (0.00 8 ) Obse rvations 48 48 58 58 70 70 75 75 R 2 0.0 09 0. 0 66 0. 0 51 0. 13 7 0.0 2 3 0. 0 70 0.0 2 2 0. 06 9 Compa rison m ean Panel B: Bolív ar, C him borazo, Im babura -0.0 02 0.0 0 0 -0.0 1 1 -0.0 04 -0.0 19 -0.01 3 -0.01 4 -0.0 1 0 Treatment parroquia (yes = 1) 0.0 32 (0.0 21 ) (0.0 21 ) (0.01 4 ) (0.014) (0.01 3 ) (0.01 4 ) (0.01 1) (0.01 1 ) Obse rvations 3 0 3 0 39 39 46 46 49 49 R 2 0.0 9 9 0. 1 71 0.1 21 0. 2 37 0.0 56 0. 1 59 0.0 51 0. 1 62 Notes: OLS estimates. The unit of observation is the rural parroqui a . h is the percentage point distance from the eligibility cutoff 0.8905. Child mortality is the proportion of last -born childr en bor n between April 19 88 and F ebruary 1989 who had di ed by November 1990 ( 19 90 census). P re -program co ntrols include num ber of last -born children
born between April
19 88 an d F ebruary 1989 living in the parr oquia in 1990 (1990 census) and the avera ge age of the children in that cohort. Comparison means are the estimated constant terms in h = 6 without controls. Robust stan
dard errors in parentheses.
*, ** , and *** indicate si gnificance at 10%, 5%, and 1% l evel s, respectiv el y . 13
Table
4:
Impact
on
child
mortality
,maximum
dif
ferential
exposure
cohorts,
post
scaling-up
of
PANN
2000
Dependent var iable: Proporti on of last-born children born
between April 2008 and F ebruary 200 9 who ha d died by N ovem ber 20 10 Neighbor hood h (percentage points) 3 4 5 6 P re -program controls N Y N Y N Y N Y Com pa rison m ean Pane l A: Bolí var , Chim borazo, Im babura, Manabí -0.0 07 -0.01 3 -0.0 06 -0.0 04 -0.0 05 -0.0 06 -0.00 5 -0.0 06 Treatm ent parroquia (yes = 1) 0.0 14 (0.00 8) (0.0 11 ) (0.00 7 ) (0.00 9 ) (0.007) (0.00 8 ) (0.00 6 ) (0.00 8 ) Observat ions 48 48 58 58 70 70 75 75 R 2 0.0 15 0. 149 0. 015 0. 118 0.0 10 0. 112 0.0 12 0. 120 Com pari son m ean Panel B: Bolív ar, C him borazo, Im babura -0.0 12 -0.0 11 -0.0 03 0.0 04 -0.0 02 0.0 01 -0.0 02 0.0 01 Treatm ent parroquia (yes = 1) 0.0 14 (0.01 5 ) (0.0 29 ) (0.01 2 ) (0.01 9 ) (0.01 2 ) (0.01 6 ) (0.010) (0.01 3 ) Observat ions 30 30 39 39 46 46 49 49 R 2 0.0 3 5 0. 248 0. 021 0. 165 0.0 18 0. 153 0.0 16 0. 147 Notes: OLS estimat es. The unit of observat ion is the rural parroquia . h is the per centage poi nt distance from the eligibi lity cutoff 0.8905. Child mort ality is the proportion of last -born children bor n betwee n April 2008 and F ebruary 200 9 who had died by November 20 10 (20 10 census). Pre -program
controls include number of check
-ups at health posts per 0 to 4 y ear -old during y ear 1999 (1999 healt h census), proporti on of last
-born children born betwe
en April 1988 and February
1989
who
had died by Novem
ber 1990
(1990 census)
, proportion of stunted childr
en under 5 y ears old in 19 90 (1990 census and 1988 nutrition survey ), an indicator for the p resence of a health center in the parroquia in 2000 (2000 h ealth census), proport ion of wom en aged 18 or older w ho com pleted prim ary schooling (2001 census) , proport ion of dwell ings with access to piped water (2001 census) , proportion of household s wit h toilet (2001 census), aver age num ber of persons per room (200 1 census), number of last -born children born between A pril 1999 and F ebruary 2000 liv ing in the parro quia in 2001 (2001 census), num ber of last -born children bor n between April 1988 and F ebruary 1989 liv ing in the parroquia in 1990 (1990 census) . Comparis on m eans are the estim ated constant term s in h = 3 without contr
ols. Robust standard errors in parenthes
es. *, ** , and ** * indica te significance at 10%, 5%, a nd 1% lev els, respectiv el y .