Post on 07-Mar-2018
transcript
Population Density, Fertility, and DemographicConvergence in Developing Countries - Appendix
David de la Croix∗ Paula E. Gobbi†
February 21, 2017
∗IRES & CORE, Universite catholique de Louvain, Place Montesquieu 3, B-1348 Louvain-la-Neuve, Belgium.Email: david.delacroix@uclouvain.be†National Fund for Scientific Research (Belgium) and IRES, Universite catholique de Louvain. Email:
paula.gobbi@uclouvain.be
1
A Theory
Proof of Proposition 1.
Using the mean value theorem for derivatives, one has:
∃δ ∈ (0, 1) such thatΦ(Pt)− Φ(0)
Pt
= Φ′(δPt),
It follows that:Pt+1
Pt
=Φ(0)
Pt
+ Φ′(δPt).
As Φ′′() < 0, population growth Pt+1/Pt is negatively correlated with density Pt.
Reminder: convergence rate and half-life
Consider a sequence {xt} converging to a long-run value x. Its rate of convergence is:
limt→∞
|xt+1 − x||xt − x|
< 1.
A low rate of convergence implies that xt is converging quickly. Assume that the dynamicbehavior of xt is governed by the difference equation:
xt+1 = f(xt).
If f(·) is differentiable, we can take a first order Taylor expansion around x,
xt+1 − xxt − x
= f ′(x).
When dynamics are monotonic, xt+1 − x and xt − x have the same sign, and we can relate thespeed of convergence to the first order derivative of f(·) evaluated at steady state. We can alsodefine the half-life of xt, T , as the time it takes to fill half the gap with the steady state. It isgiven by:
xt+T − x =1
2(xt − x) ,
and can be computed from:f ′(x)T = 1/2.
2
B Sample
Country Year Phase Sharescluster individual
Sub-Saharan Africa
Benin BJ 2001 IV 0.010 0.013Burkina Faso BF 1998-99 III 0.008 0.013Burundi BU 2010 VI 0.015 0.019Cameroon CM 2004 IV 0.019 0.022Central African Republic CF 1994-1995 III 0.009 0.012Comoros KM 2012 VI 0.010 0.01Congo Democratic Republic CD 2007 V 0.012 0.02Cote d’Ivoire CI 1998-99 III 0.006 0.006Ethiopia ET 2000 IV 0.022 0.031Gabon GA 2012 VI 0.013 0.017Ghana GH 1998 IV 0.016 0.01Guinea GN 1999 IV 0.012 0.014Kenya KE 2003 IV 0.016 0.017Lesotho LS 2004 IV 0.015 0.014Liberia LB 2007 V 0.012 0.014Madagascar MD 1997 III 0.011 0.014Malawi MW 2000 IV 0.023 0.027Mali ML 2001 IV 0.016 0.026Mozambique MZ 2011 VI 0.025 0.028Namibia NM 2000 IV 0.010 0.014Niger NI 1998 III 0.011 0.015Nigeria NG 2003 IV 0.015 0.015Rwanda RW 2005 V 0.018 0.023Senegal SN 2005 IV 0.015 0.029Sierra Leone SL 2008 V 0.014 0.015Swaziland SZ 2006-2007 V 0.011 0.01Tanzania TZ 1999 IV 0.007 0.008Togo TG 1998 III 0.012 0.017Uganda UG 2000-2001 IV 0.011 0.013Zambia ZM 2007 V 0.013 0.015Zimbabwe ZW 1999 IV 0.009 0.012
3
Country Year Phase Sharescluster individual
Middle East and North Africa (MENA)
Egypt EG 2000 IV 0.040 0.032Jordan JO 2002 IV 0.020 0.012Morocco MA 2003-2004 IV 0.019 0.034
Latin America
Bolivia BO 2008 V 0.040 0.034Colombia CO 2010 VI 0.196 0.099Honduras HN 2011-2012 VI 0.046 0.045Peru PE 2000 IV 0.057 0.057
South and South East Asia
Bangladesh BD 1999-2000 IV 0.014 0.021Cambodia KH 2000 IV 0.019 0.031Indonesia ID 2002-2003 IV 0.053 0.057Nepal NP 2001 IV 0.010 0.018Pakistan PK 2006-2007 V 0.039 0.02Philippines PH 2003 IV 0.033 0.028
Total number of Observations 24,769 490,669
Table B.1: Countries with Corresponding Year and DHS Phase, and the Shares of Clusters andIndividuals of Each Country.
4
−50 0 50 100
−40
−20
020
40
0
2
4
6
8
10
Figure B.1: Map of ln(1+population density) in 1990
−50 0 50 100
−40
−20
020
40
0.0
0.5
1.0
1.5
Figure B.2: Map of Land Productivity (Maximum Potential Caloric Yield)
−50 0 50 100
−40
−20
020
40
−20
−15
−10
−5
0
Figure B.3: Map of GDP per Capita
5
Histogram of cl$n
cl$n
Fre
quen
cy
0 20 40 60 80 100
010
0020
0030
0040
0050
00
Histogram of log(1 + cl$dens90)
log(1 + cl$dens90)
Fre
quen
cy
0 2 4 6 8 10
050
010
0015
0020
00
Histogram of cl$rate
cl$rate
Fre
quen
cy
0 2 4 6 8
010
0020
0030
0040
00
Figure B.4: Distribution of Clusters’ Characteristics: Number of Women (Top), log(1+density)(Bottom-Left) and Birth Rate (Bottom-Right)
6
Histogram of cl$age
cl$age
Fre
quen
cy
15 20 25 30 35 40 45 50
010
0030
0050
00
Histogram of cl$educ
cl$educ
Fre
quen
cy
0 5 10 15
050
010
0015
0020
0025
00
Histogram of cl$mortal
cl$mortal
Fre
quen
cy
0.0 0.2 0.4 0.6 0.8 1.0
020
0060
0010
000
Histogram of cl$marriage
cl$marriage
Fre
quen
cy
0.0 0.2 0.4 0.6 0.8 1.0
010
0030
0050
00
Figure B.5: Distribution of Clusters’ Characteristics: Age (Top-Left), Education (Top-Right),Infant Mortality (Bottom-Left), Marriage Rates (Bottom-Right)
7
15 19 23 27 31 35 39 43 47
050
0010
000
1500
020
000
0 2 4 6 8 11 14 17 20 23 27
020
000
6000
010
0000
1400
00
0 0.1875 0.375 0.5 0.6 0.7 0.875
010
0000
2000
0030
0000
0 2 4 6 8 10 12 14 16 18
020
000
6000
010
0000
Figure B.6: Distribution from the Individual Recode of DHS: Age (Top-Left), Education (Top-Right), Infant Mortality (Bottom-Left), Number of Births (Bottom-Right)
8
C The Instrument
We encoded the location of all buildings and cities belonging to UNESCO World Heritage Sitesbuilt between the neolithic revolution and 1900. The list below retains those who ended upbeing relevant for at least one cluster. The number of clusters that has a given site as its closestsite is indicated in the last column. Some sites are located in countries outside of our sample.
Country UN World Heritage Site long. lat. # clusters
Afghanistan Minaret and Archaeological Remains of Jam 64.52 34.40 1Argentina Quebrada de Humahuaca -65.35 -23.20 102Bangladesh Historic Mosque City of Bagerhat 89.80 22.67 207
Ruins of the Buddhist Vihara at Paharpur 88.98 25.03 150Benin Royal Palaces of Abomey 1.98 7.18 292Bolivia Fuerte de Samaipata -63.82 -18.17 185
Potosi -65.75 -19.58 126Sucre -65.26 -19.04 237Jesuit Missions of the Chiquitos -60.50 -16.00 20Tiwanaku: Spiritual and Political Centre
of the Tiwanaku Culture -68.68 -16.56 328Botswana Tsodilo 21.73 -18.75 358Burkina Faso Ruins of Loropeni -3.58 10.25 512Cambodia Angkor 103.83 13.43 418
Temple of Preah Vihear 104.68 14.39 40Colombia Cartagena -75.53 10.42 383
Historic Centre of Santa Cruz de Mompox -74.43 9.23 1362National Archeological Park of Tierradentro -76.03 2.58 2340San Agustın Archaeological Park -76.23 1.92 438
Costa Rica Precolumbian Chiefdom Settlementswith Stone Spheres of the Diquıs -83.48 8.91 120
Cote d’Ivoire Historic Town of Grand-Bassam 3.74 5.20 519Ecuador City of Quito -78.50 0.00 56
Historic Center of Santa Ana de los Rıos de Cuenca -78.98 -2.88 149Egypt Abu Mena 29.67 30.85 137
Ancient Thebes with its Necropolis 32.60 25.73 182Historic Cairo 31.26 30.05 476Memphis and its Necropolis 31.13 29.98 171Saint Catherine Area 33.98 28.56 16
El Salvador Joya de Ceren Archaeological Site -89.37 13.83 37Ethiopia Rock-Hewn Churches, Lalibela 39.04 12.03 68
Fasil Ghebbi, Gondar Region 37.47 12.61 64Aksum 38.72 14.13 38Tiya 38.61 8.43 247Harar Jugol, the Fortified Historic Town 42.14 9.31 126
continued on next page
9
Country UN World Heritage Site long. lat. # clusters
Gambia Kunta Kinteh Island and Related Sites -16.36 13.32 50Ghana Forts and Castles, Volta, Greater Accra,
Central and Western Regions 0.49 5.39 189Asante Traditional Buildings -1.63 6.40 346
Guatemala Archaeological Park and Ruins of Quirigua -89.04 15.27 432Honduras Maya Site of Copan -89.13 14.85 290India Rani-ka-Vav at Patan, Gujarat 72.10 23.86 5
Red Fort Complex 77.24 28.66 13Rani-ki-Vav (the Queens Stepwell)
at Patan, Gujarat 72.10 23.86 5Iran Shahr-i Sokhta 61.33 30.59 4
Bam and its Cultural Landscape 58.37 29.12 2Indonesia Borobudur Temple Compounds 110.20 -7.61 478
Prambanan Temple Compounds 110.49 -7.75 531Israel Incense Route - Desert Cities in the Negev 35.16 30.54 15
Masada 35.35 31.31 25Caves of Maresha and Bet-Guvrin
in the Judean Lowlands 34.90 31.60 14Jordan Petra 35.44 30.33 66
Um er-Rasas (Kastrom Mefa’a) 35.92 31.50 274Kenya Lamu Old Town 40.85 -2.28 87
Fort Jesus, Mombasa 39.68 -4.06 152Libya Archaeological Site of Cyrene 21.86 32.83 2Madagascar The Royal Hill of AmbohimangaMalaysia Melaka and George Town, historic cities
of the Straits of Malacca 100.35 5.42 44Archaeological Heritage of the Lenggong Valley 100.97 5.07 172
Mali Timbuktu -3.00 16.77 26Old Towns of Djenne -4.55 13.91 349Tomb of Askia 0.04 16.29 129
Morocco Medina of Fez -4.98 34.06 82Medina of Marrakesh -7.99 31.63 52Ksar of Ait-Ben-Haddou -7.13 31.05 38Historic City of Meknes -5.56 33.88 35Archaeological Site of Volubilis -5.56 34.07 23Medina of Tetouan (formerly known as Titawin) -5.37 35.57 45Medina of Essaouira (formerly Mogador) -9.77 31.52 55Portuguese City of Mazagan (El Jadida) -8.50 33.26 43Rabat, Modern Capital and Historic City:
a Shared Heritage -6.82 34.02 106Mozambique Island of Mozambique 40.74 -15.03 819Nepal Kathmandu Valley 85.31 27.70 133
Lumbini, the Birthplace of the Lord Buddha 83.28 27.47 89
continued on next page
10
Country UN World Heritage Site long. lat. # clusters
Nicaragua Ruins of Leon Viejo -86.61 12.40 91Leon Cathedral -86.88 12.44 278
Niger Historic Centre of Agadez 7.99 16.97 172Nigeria Sukur Cultural Landscape 13.57 10.74 654
Osun-Osogbo Sacred Grove 4.55 7.76 234Pakistan Archaeological Ruins at Moenjodaro 68.14 27.33 207
Buddhist Ruins of Takht-i-Bahi andNeighbouring City Remains at Sahr-i-Bahlol 71.95 34.32 143
Taxila 72.89 33.78 76Fort and Shalamar Gardens in Lahore 74.31 31.59 283Historical Monuments at Makli, Thatta 67.90 24.77 158Rohtas Fort 73.59 32.96 78
Palestine Birthplace of Jesus: Church of the Nativityand the Pilgrimage Route, Bethlehem 35.21 31.70 115
Peru City of Cuzco -71.98 -13.52 140Historic Sanctuary of Machu Picchu -72.58 -13.12 143Chavin (Archaeological Site) -77.18 -9.59 176Chan Chan Archaeological Zone -79.08 -8.10 239Historic Center of Lima -77.04 -12.05 333Historical Center of the City of Arequipa -71.53 -16.40 176Sacred City of Caral-Supe -77.52 -10.89 48
Philippines Baroque Churches of the Philippines 120.97 14.59 821Historic Town of Vigan 120.39 17.58 89
Senegal Island of Goree -17.40 14.67 101Island of Saint-Louis -16.50 16.03 72Stone Circles of Senegambia -15.52 13.69 727
Spain Alhambra, Generalifeand Albayzın, Granada -3.59 37.18 1
Tanzania Ruins of Kilwa Kisiwaniand Ruins of Songo Mnara 39.52 -8.96 221
Stone Town of Zanzibar 39.19 -6.16 71Togo Koutammakou, the Land
of the Batammariba 1.13 10.07 328Uganda Tombs of Buganda Kings at Kasubi 32.55 0.35 1496Venezuela Coro and its Port -69.68 11.40 173Vietnam My Son Sanctuary 108.57 15.52 12Zimbabwe Great Zimbabwe National Monument 30.93 -20.28 904
Khami Ruins National Monument 28.38 -20.16 426South Africa Robben Island 18.37 -33.80 220
List of the relevant UNESCO World Heritage Sites
11
Figure C.1: Clusters’ Shortest Distance to UNESCO Site
Histogram of cl$dist_site
cl$dist_site
Fre
quen
cy
0 5 10 15
010
0030
0050
00
Figure C.2: Distribution of Clusters’ Shortest Distance to UNESCO Sites
12
D Examples of Different Densities
Figure D.1: Densities of 0.1, 1, 10, 100, 1000, and 10000 Inhabitants/km2
13
E Analysis at the Cluster Level, by Continent
Dependent variable:children ever born, per woman (average in cluster)
Sub-Saharian Africa (1) (2) (3) (4) (5)
ln(1+density) −0.177∗∗∗ −0.121∗∗∗ −0.107∗∗∗ −0.096∗∗∗ −0.044∗∗∗
(0.010) (0.009) (0.008) (0.008) (0.007)marriage 2.121∗∗∗ 1.688∗∗∗ 1.649∗∗∗ 0.961∗∗∗
(0.091) (0.080) (0.079) (0.073)infant mortality 3.036∗∗∗ 2.926∗∗∗ 2.076∗∗∗
(0.194) (0.189) (0.210)ln(GDP per capita) −0.104∗∗∗ −0.053∗∗∗
(0.017) (0.014)woman’s education −0.059∗∗∗
(0.015)(woman’s education)2 −0.006∗∗∗
(0.001)
Observations 10,262 10,262 10,262 10,262 10,262Adjusted R2 0.569 0.640 0.669 0.674 0.720
Middle East & N. Africa (1) (2) (3) (4) (5)
ln(1+density) −0.134∗∗∗ −0.123∗∗∗ −0.106∗∗∗ −0.107∗∗∗ −0.040∗∗∗
(0.032) (0.032) (0.020) (0.019) (0.013)marriage 2.451∗∗ 1.889∗∗∗ 1.873∗∗∗ 1.171∗∗∗
(0.431) (0.316) (0.315) (0.249)infant mortality 9.194∗∗∗ 9.175∗∗∗ 4.945∗∗∗
(0.876) (0.858) (0.680)ln(GDP per capita) −0.015 0.017
(0.029) (0.025)woman’s education −0.137∗∗∗
(0.019)(woman’s education)2 −0.001
(0.002)
Observations 1,973 1,973 1,973 1,973 1,973Adjusted R2 0.541 0.553 0.638 0.638 0.737
Notes: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01. Robust standard errors, clustered at the regionallevel (247 clusters & 24 clusters respectively), in parentheses. All specifications includecountry fixed effects, geographical controls (the Caloric Suitability Index and distance toa large body of water) and a polynomial of order 2 in mean age.
Table E.1: OLS Estimates at the Cluster Level – Sub-Saharan Africa & Middle East and NorthAfrica
14
Dependent variable:children ever born, per woman (average in cluster)
South and S.-E. Asia (1) (2) (3) (4) (5)
ln(1+density) −0.141∗∗∗ −0.118∗∗∗ −0.099∗∗∗ −0.084∗∗∗ −0.025∗
(0.015) (0.014) (0.011) (0.011) (0.014)marriage 2.444∗∗∗ 2.094∗∗∗ 2.067∗∗∗ 1.564∗∗∗
(0.225) (0.239) (0.238) (0.219)infant mortality 4.053∗∗∗ 3.977∗∗∗ 2.606∗∗∗
(0.828) (0.817) (0.808)ln(GDP per capita) −0.102∗∗∗ −0.091∗∗∗
(0.029) (0.029)woman’s education −0.099∗∗∗
(0.023)(woman’s education)2 −0.002
(0.001)
Observations 4,151 4,151 4,151 4,151 4,151Adjusted R2 0.480 0.513 0.555 0.560 0.616
Latin America (1) (2) (3) (4) (5)
ln(1+density) −0.211∗∗∗ −0.159∗∗∗ −0.136∗∗∗ −0.130∗∗∗ −0.052∗∗∗
(0.017) (0.015) (0.012) (0.013) (0.011)marriage 2.223∗∗∗ 2.016∗∗∗ 2.002∗∗∗ 1.269∗∗∗
(0.142) (0.107) (0.107) (0.094)infant mortality 6.800∗∗∗ 6.645∗∗∗ 3.836∗∗∗
(0.506) (0.475) (0.297)ln(GDP per capita) −0.084 −0.066∗
(0.058) (0.035)woman’s education −0.234∗∗∗
(0.022)(woman’s education)2 0.005∗∗∗
(0.001)
Observations 8,383 8,383 8,383 8,383 8,383Adjusted R2 0.412 0.534 0.593 0.599 0.727
Notes: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01. Robust standard errors, clustered at the regionallevel (81 clusters & 57 clusters respectively), in parentheses. All specifications includecountry fixed effects, geographical controls (the Caloric Suitability Index and distance toa large body of water) and a polynomial of order 2 in mean age.
Table E.2: OLS Estimates at the Cluster Level – South and South-East Asia & Latin America
15
F Analysis at the Cluster Level for Countries Grouped
by Income
Dependent variable:children ever born, per woman (average in cluster)
least developed otherseconomies
(1) (2) (3) (4)
ln(1+density) −0.179∗∗∗ −0.030∗∗ −0.190∗∗∗ −0.056∗∗∗
(0.016) (0.012) (0.012) (0.005)marriage 1.131∗∗∗ 1.137∗∗∗
(0.125) (0.107)infant mortality 2.255∗∗∗ 2.911∗∗∗
(0.387) (0.506)ln(GDP per capita) −0.057∗∗∗ −0.050∗∗∗
(0.018) (0.012)woman’s education −0.089∗∗∗ −0.102∗∗∗
(0.030) (0.033)(woman’s education)2 −0.003 −0.003
(0.002) (0.002)
Observations 8,479 8,479 16,290 16,290Adjusted R2 0.562 0.705 0.552 0.741
Notes: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01. Robust standard errors, clustered at thecountry level, in parentheses. All specifications include country fixed effects,geographical controls (the Caloric Suitability Index and distance to a largebody of water) and a polynomial of order 2 in mean age.
Table F.3: OLS Estimates at the Cluster Level, by Income Groups
16
G Robustness of the Analysis at the Individual Level
G.1 Weighting
Table G.1 shows the same estimates than those in Table 4 but when including weights to theregressions.
Dependent variable: Children ever born
(1) (2) (3) (4) (5) (5-IV)
ln(1+density) −0.071∗∗∗ −0.052∗∗∗ −0.047∗∗∗ −0.045∗∗∗ −0.019∗∗∗ −0.082∗∗∗
(0.001) (0.001) (0.001) (0.001) (0.001) (0.008)married 1.527∗∗∗ 1.517∗∗∗ 1.517∗∗∗ 1.453∗∗∗ 1.453∗∗∗
(0.018) (0.017) (0.017) (0.017) (0.017)mean marriage 0.317∗∗∗ 0.172∗∗∗ 0.164∗∗∗ −0.048∗∗∗ −0.150∗∗∗
(0.015) (0.016) (0.015) (0.014) (0.020)mortality 0.472∗∗∗ 0.472∗∗∗ 0.432∗∗∗ 0.433∗∗∗
(0.007) (0.007) (0.007) (0.007)mean mortality 0.539∗∗∗ 0.510∗∗∗ 0.157∗∗∗ 0.175∗∗∗
(0.035) (0.035) (0.033) (0.036)ln(GDP per capita) −0.028∗∗∗ −0.014∗∗∗ −0.016∗∗∗
(0.003) (0.002) (0.004)woman’s educ 0.004∗∗∗ 0.003∗∗∗
(0.001) (0.001)(woman’s educ)2 −0.003∗∗∗ −0.003∗∗∗
(0.000) (0.000)educ in cluster −0.008∗∗∗ 0.011∗∗∗
(0.003) (0.003)educ2 in cluster −0.001∗∗∗ −0.001∗∗∗
(0.000) (0.000)
Observations 490,669 490,669 490,669 490,669 490,669 490,669
Notes: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01. Robust standard errors, clustered at the cluster level, inparentheses. All specifications include country fixed effects, geographical controls (the CaloricSuitability Index and distance to a large body of water) and age dummies.
Table G.1: Poisson and IV Poisson Estimates at the Individual Level, with Individuals Weights
G.2 Restricted sample and another dependent variable
Table G.2 shows the estimated coefficients of an IV Poisson regression when restricting thesample to a subsample of women aged 40+ and to countries with good quality of data (seebelow for details on this issue). The last column of the table also shows the estimate wheninstead of using “children ever born” as the dependent variable, we use “children born in thelast 5 years”.
17
Data Quality: Misreporting date of birth or underreporting number of births are commonsources of error in surveys that look at birth history (Schoumaker (2014)). These errors arevery much linked to low education levels of respondents (Pullum (2006)), and can affect ageat first birth in three ways. The first is the so-called the “Potter effect” when the womanreports that an earlier birth occurred later than it actually did (Potter (1977)). This will likelyincrease the age at first birth for older women. The second source of error is adjustment ofbirth date by interviewers or respondents in order to avoid completing the health section ofthe DHS questionnaire (for children younger than 5 or 3). This will cause a reduction in theaverage age at first birth for younger women. The last problem is omission of earlier births,which most likely occurs with older respondents and is likely to increase the average age at firstbirth in a population.
Schoumaker (2014) explores the quality of the data using three approaches. The first consistsof reconstructing trends in the total fertility rate (TFR) using a Poisson regression, and relyingon one survey per country (see Schoumaker (2013b) for details on this method). The secondapproach consists of pooling all the surveys conducted in the same country and then recon-structing fertility trends from the pooled dataset (Schoumaker (2013a)). The third approachaims to correct birth histories by adjusting or adding births.
Table 5 in Schoumaker (2014) distinguishes between good, moderate, and poor quality data.As a robustness check of our results in Section 4.1, we run the Poisson regression only for thosecountries with good quality data. Those countries are Colombia, Egypt, Gabon, Honduras,Indonesia, Morocco, Lesotho, Namibia, Nepal, Peru, Philippines, and Zimbabwe. Results areshown in Table G.2.
18
Depen
dentvariable:
childre
nev
erb
orn
childre
nb
orn
inth
ela
st5
year
s
all
wom
enw
omen
aged
40+
good
qual
ity
dat
aon
lyal
lw
omen
ln(1
+den
sity
)−
0.05
8∗∗∗
−0.
058∗∗∗
−0.
047∗∗∗
−0.
070∗∗∗
(0.0
05)
(0.0
07)
(0.0
05)
(0.0
08)
mar
ried
1.42
2∗∗∗
1.18
6∗∗∗
1.36
9∗∗∗
1.61
4∗∗∗
(0.0
14)
(0.0
31)
(0.0
22)
(0.0
14)
mea
nm
arri
age
−0.
149∗∗∗
−0.
289∗∗∗
−0.
122∗∗∗
−0.
109∗∗∗
(0.0
17)
(0.0
24)
(0.0
23)
(0.0
25)
mor
tality
0.42
8∗∗∗
0.38
0∗∗∗
0.57
1∗∗∗
0.32
9∗∗∗
(0.0
06)
(0.0
11)
(0.0
12)
(0.0
08)
mea
nm
orta
lity
0.17
0∗∗∗
0.13
0∗∗∗
0.49
3∗∗∗
0.23
1∗∗∗
(0.0
29)
(0.0
44)
(0.0
60)
(0.0
41)
ln(G
DP
per
capit
a)−
0.00
7∗∗∗
0.00
0−
0.01
4∗∗∗
−0.
023∗∗∗
(0.0
02)
(0.0
03)
(0.0
03)
(0.0
04)
educ
0.00
3∗∗∗
0.00
3∗∗
0.00
9∗∗∗
−0.
008∗∗∗
(0.0
01)
(0.0
01)
(0.0
01)
(0.0
01)
educ2
−0.
003∗∗∗
−0.
002∗∗∗
−0.
002∗∗∗
−0.
000∗∗∗
(0.0
00)
(0.0
00)
(0.0
00)
(0.0
00)
mea
ned
uc
0.00
4∗∗∗
0.00
6∗−
0.01
6∗∗∗
−0.
015∗∗∗
(0.0
02)
(0.0
03)
(0.0
03)
(0.0
04)
(mea
ned
uc)
2−
0.00
1∗∗∗
−0.
002∗∗∗
−0.
000
−0.
000∗∗
(0.0
00)
(0.0
00)
(0.0
00)
(0.0
00)
Obse
rvat
ions
490,
669
95,0
5320
8,51
049
0,66
9
Notes:∗ p<
0.1;∗∗
p<
0.05
;∗∗∗ p<
0.01
.R
obust
stan
dar
der
rors
,cl
ust
ered
atth
ecl
ust
erle
vel,
inpar
enth
eses
.A
llsp
ecifi
cati
ons
incl
ude
countr
yfixed
effec
ts,
geog
raphic
alco
ntr
ols
(the
Cal
oric
Suit
abilit
yIn
dex
and
dis
tance
toa
larg
eb
ody
ofw
ater
)an
dag
edum
mie
s.
Table G.2: IV Poisson Estimates, Restricting the Sample to Women Aged 40+, Restricting theSample to Countries with Good Quality of Data, and Using “Children born in the last 5 years”as Dependent Variable
19
References
Potter, Joseph. 1977. “Problems in using birth-history analysis to estimate trends in fertility.”Population Studies: A Journal of Demography 31 (2): 335–364.
Pullum, Thomas W. 2006. “An Assessment of Age and Date Reporting in the DHS Surveys,1985-2003.” Technical Report, Methodological Reports No. 5. Calverton, Maryland: MacroInternational Inc.
Schoumaker, Bruno. 2013a. “Reconstructing Long Term Fertility Trends with Pooled BirthHistories.” Paper presented at the XXVII International Population Conference, Busan(South Korea).
. 2013b. “A Stata Module to Compute Fertility Rates and TFRs from Birth Histories:tfr2.” Demographic Research 28 (38): 1093–144.
. 2014. “Quality and Consistency of DHS Fertility Estimates, 1990 to 2012.” TechnicalReport, DHS Methodological Reports No. 12. Rockville, Maryland, USA: ICF Interna-tional.
20