Fertility, the Demographic Dividend, and Economic Growth David E. Bloom, David Canning Günther Fink Jocelyn E. Finlay Harvard School of Public Health Fourth Annual Research Conference on Population, Reproductive Health, and Economic Development Cape Town, January 2010
Effects of Fertility on Income per Capita Investigate the effects of fertility on income Control for endogeneity by using changes in abortion law as an instrument. About 26% of pregnancies end in abortion. Look at the mechanisms through which fertility operates Income per worker Labor force participation Working age share of population
Not Population Growth! Population growth is fertility rate mortality rate + net migration. Fertility rate and mortality rate have very different economic effects. The effect of population growth is not well defined unless we know the source of the growth.
Caveats Income per capita is not a welfare measure Focus on average income not distribution and poverty Macro income per capita still interesting. Macro can capture effects micro misses Social norms in behavior Thresholds and critical value effects
Social Spillovers Micro model y x y Difficult to estimate since y endogenous and common to everyone in the community Macro model y x y y x 1
6 7 8 9 10 11 Figure 4: Income per Capita and Fertility in 2000 Luxembourg Norway United States United Qatar Arab Emirates Austria Singapore Canada Switzerland Australia Denmark Germany Belgium France Ireland Iceland Brunei Italy Japan Sweden United Netherlands Finland Kingdom Kuwait Cyprus Israel SpainMalta New Bahamas, Zealand Slovenia BahrainThe Portugal Korea, Barbados Czech Greece Antigua Mauritius Rep. Oman Saudi Arabia Republic St. and Kitts Barbuda and Nevis Hungary Estonia ChileArgentina Belarus Uruguay Malaysia Gabon Russian Slovak Croatia Lithuania Republic Poland Federation Dominica Costa Mexico Panama South Rica Africa Swaziland Bulgaria St. Brazil Vincent and Botswana the Grenadines Kazakhstan Thailand St. Tunisia Lucia Venezuela, Turkmenistan RB Cuba Iran, Lebanon Algeria Colombia Islamic Dominican Rep. Belize Republic Equatorial Guinea Macedonia, Turkey Ukraine Romania FYR El Salvador Cape Namibia Verde Jamaica Suriname Djibouti China Ecuador Fiji Egypt, Arab Rep. Maldives Georgia Sri Lanka Peru Papua New Guinea Albania Armenia Azerbaijan Indonesia Morocco JordanMicronesia, Guatemala Fed. Sts. Kyrgyz Uzbekistan Philippines Republic Nicaragua Tonga Bosnia and Herzegovina Zimbabwe Samoa Vanuatu India Pakistan Cameroon Guinea Moldova Vietnam Honduras Haiti Cote d'ivoire Bangladesh Syrian Arab Solomon Republic Islands Lesotho Tajikistan Korea, Mongolia Senegal Dem. Rep. Kiribati Nepal Mauritania Sao Comoros Ghana Tome Kenya Lao and PDR Congo, Principe Benin Rep. Sudan Mozambique Nigeria Mali Gambia, Central The African Rwanda Yemen, Uganda Rep. Burkina Republic Faso Togo Tanzania Madagascar Zambia Malawi Chad Ethiopia Burundi Guinea-Bissau Niger Sierra Somalia Leone Cambodia Eritrea Liberia 0 2 4 6 8 TFR, 2000
Mechanisms Most economic models focus on income per worker effects Malthus, Solow We also have effects on workers per capita Age structure working age share Female labor force participation Workers per capita is bounded cannot explain long run growth but can vary a lot in the medium term
Components of Income per Capita Identity Y P Y L W t t t t L W P t t t t Y Y L W log log log log P L W P t t t t t t t t
Income per worker Land per worker Malthusian effect, number of workers Capital per worker Solow effect, growth rate of workforce Savings rather than children for old age security Human capital per worker Investment in children, quality quantity tradeoff
Income per worker: Timing Most effects only occur when fertility affects growth in labor force. Time lag of 15-20 years before children enter the labor force. Large effects after long run adjustment to steady state income to investment - several generations.
Working Age Share Lower fertility always reduces the youth dependency rate. Lower fertility lowers the number of worker age people in 20-60 years, increasing old age dependency Turning point in overall effect is close to replacement fertility. Youth dependency effect is immediate. Old age dependency effect is longer run.
Working age share Figure 1: Relationship between fertility and steady state working age share 70% 60% 50% 40% 30% 20% Life expectancy 40 years (Zambia, 2005) Life expectancy 60 years Yemen, 2005) Life expectancy 80 years (France, 2005) 10% 0% 0 1 2 3 4 5 6 7 8 9 Total fertility rate
40 50 60 70 80 Figure 2: Total fertility rates and working age shares in 2000 United Arab Emirates Korea, Rep. Kuwait Qatar Singapore Czech Slovenia Bosnia and Herzegovina Russian Cuba Republic Slovak BarbadosBahrain Belarus Greece Bulgaria Hungary Poland Austria Canada China Republic Federation Ukraine Spain Italy Moldova Romania Japan Germany Croatia Malta Netherlands Estonia Finland Luxembourg Macedonia, Thailand Switzerland Portugal Mauritius Korea, Dem. FYRRep. Cyprus Australia Denmark Ireland Sri Lanka Lithuania Georgia Belgium Brunei France Kazakhstan United States United Chile Bahamas, Brazil The Armenia Norway New Iceland Kingdom Guyana Indonesia Zealand Sweden Tunisia Turkey Azerbaijan Albania Costa Argentina Colombia Rica Fiji Libya Uruguay Lebanon Mexico Morocco Panama South Algeria Dominican Israel Malaysia Africa Suriname Vietnam St. Lucia Mongolia Venezuela, Republic RB Iran, Islamic Ecuador India Oman St. Vincent Peru and Rep. the Grenadines Kyrgyz Jamaica Turkmenistan El Republic Salvador Bangladesh Egypt, Arab Rep. Uzbekistan Philippines Botswana JordanSaudi Arabia Belize Syrian Tonga Paraguay Arab Nepal Bolivia Haiti Cambodia Papua Micronesia, Sudan Republic Honduras Ghana New Guinea Fed. Sts. Nicaragua Gambia, The Lesotho Samoa Solomon Comoros PakistanIslands Zimbabwe Tajikistan Djibouti Cape Gabon Sao Vanuatu Tome Lao Verde Kenya Cameroon and Cote PDR Principe Namibia d'ivoire Maldives Central Eritrea African Equatorial Republic GuatemalaMadagascar Guinea Mauritania Sierra Somalia Leone Swaziland Tanzania Senegal Togo Mozambique Ethiopia Nigeria Benin Guinea Congo, Zambia Rwanda Rep. MalawiLiberia Chad Angola Guinea-Bissau Yemen, Burkina Burundi Faso Mali Rep. Niger Uganda 0 2 4 6 8 Total fertility rate
Utility Time Constraint Consumption Household Model e U ( c, d, f, e) log c c 0 log d f ( ) k( N f ) f 1 l d bf f c w l w e f f m
Household Decisions Female labor supply 1 ( e c0 w ) m l f 1 bf (1 ) w f Fertility 1 l f f b k Investment in children 1 e wf l f wm c0
Effects of Fertility Female labor supply adjusted for investment in children l f w f 1 bf Investment per child adjusted for labor supply e f t t w m wf (1 ) / f t wb f
Female Labor Force Participation High labor force participation in poor countries possible to work and care for children at the same time. Fertility to female labor supply effect may appear when women have formal sector work where child care and work time are separated. Migration to urban areas may split extended family links that provide childcare.
Female Labor Force Participation Rate 2000 Income per Capita and Female Labor Force Participation, 2000 100 90 Tanzania Mozambique 80 70 60 Thailand Iceland US 50 40 30 20 Sudan Egypt 10 0 6 6.5 7 7.5 8 8.5 9 9.5 10 10.5 11 Log (Real GDP per Capita 2000)
Quality Quantity Tradeoff Increased investment in children as fertility declines May be household effect or effect via government spending Influences income per worker in the long run Short run effect on school enrollment Enrollment may not reflect all investment educational quality, health
-6-4 -2 0 2 Figure 3: Change in the Total Fertility Rate 1960-2000 Guinea-Bissau Equatorial ChadGuinea Ethiopia Uganda Niger Mali Congo, Burkina Faso Guinea Rep. Japan Sweden Luxembourg Denmark Uruguay Switzerland Cameroon Greece Belgium France Israel Mozambique Malawi Romania Benin Italy United Kingdom Nigeria Madagascar Tanzania Portugal Netherlands China Gambia, Togo Spain The Australia United States Ireland Guatemala Senegal New Zealand Lesotho Ghana Canada Pakistan Rwanda Barbados Jamaica Indonesia Cape VerdeKenya Panama Sri Chile Lanka IndiaPhilippines Zimbabwe Egypt, Honduras Arab Rep. Mauritius Brazil Turkey South Venezuela, Malaysia Africa El Peru Salvador Nicaragua Syrian RB Singapore Jordan Arab Republic Korea, Rep. Thailand Mexico Morocco Algeria Dominican Republic Iran, Islamic Rep. 1 2 3 4 5 6 7 8 9 10 Total fertility rate, 1960
Abortion Laws
Index (sample average) Figure 5: Abortion Index: Average 1960-2005 5 4.5 4 3.5 3 2.5 2 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 Year
Are Abortion Laws Exogenous? Timing of changes may be exogenous. Contingent factors in many examples. Extreme views and majority rule give sharp discontinuities in legal changes. French and UK liberalization had spillover effects to former colonies. Laws often used as templates for changes to laws inherited at independence. We use laws not enforcement.
Table 3 The effect of Fertility on Income per Capita Dependent variable: log GDP per capita (4) (5) (6) Total fertility rate -0.369*** -0.551*** -0.196** (0.026) (0.160) (0.086) Year dummies Yes Yes Yes Country fixed effects No Yes Yes Regional time trends No No Yes Estimation method IV IV IV Observations 1169 1169 1169 R-squared 0.541 0.897 0.957 Cragg-Donald F-stat 373.0 16.15 35.05
Table 4 First Stage: The effect of Abortion Laws on Fertility (1) (2) (3) Dependent variable: Total fertility rate Abortion index -0.410*** -0.072*** -0.096*** (0.020) (0.020) (0.018) Year dummies Yes Yes Yes Country fixed effects No Yes Yes Regional time trends No No Yes Observations 1169 1169 1169 R-squared 0.354 0.928 0.950 Note: Robust standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1
Table 5: The Effect of Fertility on the Components of Income per Capita (1) (2) (3) (4) Dependent variable: ln(gdp/p) ln(gdp/l) ln(l/w) ln(w/p) Total fertility rate -0.196** -0.061-0.071*** -0.068*** (0.089) (0.085) (0.019) (0.010) Year dummies Yes Yes Yes Yes Country fixed effects Yes Yes Yes Yes Regional time trends Yes Yes Yes Yes Estimation method IV IV IV IV Observations 1169 1105 1129 1145 R-squared 0.957 0.965 0.897 0.941 Cragg-Donald F-stat 35.05 32.38 32.75 34.49 Note: Robust standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1
Table 6: The Effect of Fertility on Income per Capita: Mechanisms (1) (2) (3) (4) Dependent variable: Log capital per worker Population growth rate Female labor force participation rate Male labor force participation rate Total fertility rate 0.021 0.645*** -9.947*** 0.495 (0.099) (0.292) (2.210) (0.695) Year dummies Yes Yes Yes Yes Country fixed effects Yes Yes Yes Yes Regional time trends Yes Yes Yes Yes Estimation method IV IV IV IV Observations 1105 999 1129 1129 R-squared 0.976 0.690 0.906 0.897 Cragg-Donald F-stat 32.38 14.87 32.75 32.75
Table 6: The Effect of Fertility on Income per Capita: Mechanisms Dependent variable: (5) (6) (7) Working age share Youth dependency rate Old-age dependency rate Total fertility rate -4.076*** 12.38*** -0.964** (0.646) (1.780) (0.416) Year dummies Yes Yes Yes Country fixed effects Yes Yes Yes Regional time trends Yes Yes Yes Estimation method IV IV IV Observations 1145 1145 1145 R-squared 0.935 0.957 0.955 Cragg-Donald F-stat 34.49 34.49 34.49
Table 7: The Effect of Fertility on Education
Why is the Macro Labor Force Effect Social Spillovers. Work is contagious. so Large? Life course decisions different with the possibility of fertility control. Abortion laws affect women who are at he margin of working (local average treatment effect). Effect mainly in middle and high income countries?
Future Directions Add laws on access to contraceptives as well as abortion. Use micro data (DHS) at different levels of aggregation. Interaction with demand side Unemployment, underemployment