The Effect of Interventions to Reduce Fertility on Economic Growth Quamrul Ashraf Ashley Lester David N. Weil Brown University December 2007
Goal: analyze quantitatively the economic effects of interventions that reduce fertility in developing countries. What we are not doing Cost benefit analysis (because we have no measure of costs) Welfare analysis (because we have no welfare criteria)
Why this is hard 1) Macroeconomic Approaches We are interested in the effects of an intervention in other words, and exogenous change in fertility This is very different observing the natural co-evolution of fertility and economic development. In other words: cross country regressions fatally flawed due to omitted variables and reverse causation
2) Microeconomic approaches microeconomic studies examine the link between fertility at the household level and various outcomes for individuals in that household (for example, wages, labor force participation, education, etc.). Proper identification much easier to achieve at the household level than at the national level. Can t address channels external to the household such as externalities and change in market prices (general equilibrium effects)
What we do Use microeconomic estimates on link from fertility household behavior/outcomes Account for general equilibrium effects using standard economic theory. Explicit dynamic model for population size, age structure, human capital, physical capital, etc. Tailor parameters of model to a specific country (Nigeria) All working parts visible.
The Demographic Setup We use year 2005 age-specific mortality and fertility from Nigerian females to build a stable population. (TFR = 5.32, e 0 =48.2, population growth rate = 2.28%) We consider an intervention that lowers age-specific fertility to the UN low variant. TFR falls by 0.25 We compare paths of all variables in case of intervention to case where TFR remains constant.
Nigeria 2005 Life Table Survivorship Function 1.20 1.00 0.80 Survivorship Rate 0.60 0.40 0.20 0.00 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 Beginning Age of 5-yr Age Interval
Nigeria 2005 Age-Specific Fertility Schedule 0.30 0.25 0.20 Fertility Rate 0.15 0.10 0.05 0.00 0-15 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50+ Age Group Pre-Shock (UN Medium Variant) Post-Shock Baseline (UN Low Variant) Post-Shock Alternative
1.100 1.000 0.900 0.800 0.700 0.600 0.500 0.400 Population -15-5 5 15 25 35 45 55 65 75 85 95 105 115 125 135 145 155 165 Years Since Shock Reduce TFR by 0.25 (Baseline) Reduce TFR by 0.50 Relative to No-Shock
How we can/will do this better: Use actual population age structure in initial year rather than stable population Use a forecast path of TFR (rather than a constant level) as a no-intervention path Allow intervention to produce a temporary rather than permanent deviation of TFR below forecast path
Channels from Demographic to Economic Dependency ratio (Bloom-Williamson effect) Labor force growth rate (Solow effect) Age structure effect on saving rate (life cycle effect) Age structure of the labor force (experience effect) Reduced fertility effect on labor supply (Bloom, Canning, Findlay, and Fink) Schooling (quality-quantity effect) Congestion of fixed factors (Malthus effect)
Dependency Ratio Reduction in TFR by 0.25 leads to rise in steady state Working age fraction of population from.552 to.562 Corresponds to a 1.8% rise in consumption per capita. Phases in over 15 years
1.020 1.000 0.980 0.960 0.940 0.920 0.900 0.880 Dependency Ratio -15-5 5 15 25 35 45 55 65 75 85 95 105 115 125 135 145 155 165 Years Since Shock Reduce TFR by 0.25 (Baseline) Reduce TFR by 0.50 Relative to No-Shock
Basic Output Model α β 1 α β Y = AK H X t t t t Y = output K = physical capital H = human capital aggregate X = land Base case parameters: α =.3 β =.6
Human capital aggregate: H = h N t i, t i 15 i 65 s e Where h, = h, h, it it it h = human capital from schooling s it, h = human capital from experience e it,
Solow Effect Reduced fertility implies slower labor force growth (with a lag) With fixed saving rate (and economy closed to capital flows) this raises physical capital per worker. Steady state effect: Decline in TFR by 0.25 reduces annual growth rate of labor force from 2.28% to 2.10%.
Effect on steady state output is given by equation intervention no intervention α 1 α no intervention + δ + = intervention + δ + y n g y n g n = population growth, δ =deprecation g = technological progress y y intervention no intervention.3 1.3.0228 +.08 +.01 = = 1.0069.0210 +.08 +.01
Life Cycle Effect on Saving Life Cycle theory implies hump shaped pattern of saving over the course of the life cycle Decline in TFR leads to concentration of population in higher saving years However, empirical evidence on age-specific saving rates is mixed We don t implement this
Age structure of the labor force Slower population growth implies an increase in average age of workers Experience is an important form of human capital, so that older work forces (unless they get too old) will be more productive.
Age structure of the labor force (continued) We use estimates of the return to experience from Bils and Klenow (2000), based on the average return to experience from a sample of 48 countries. h= x+ x 2 exp( φ ϕ ) Where x is experience, φ =.0495, and ϕ =.0007.
Age Profile of Earnings 3 2.5 2 1.5 1 0.5 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 0 Age Ashraf, Lester & Weil (Brown) Does Health Raise GDP? 08/07 19 / 66
Age Structure of the Stable Working-Age Population Pre and Post Baseline Fertility Intervention 0.21 0.18 Percent of Working-Age Population 0.15 0.12 0.09 0.06 0.03 0.00 15-20 20-25 25-30 30-35 35-40 40-45 45-50 50-55 55-60 60-65 Age Group Pre-Intervention Post-Intervention
Steady state experience effect raises labor input per worker by 0.55%
Reduced Fertility Effect on Labor Supply Less child care allows for more market work Simple example: o one child requires 4 years of adult labor market time o pair of adults has 80 years of potential labor market time o reducing TFR by 1 raises labor market input per adult by 5% We don t implement this
1.250 1.200 1.150 1.100 1.050 1.000 0.950 Income Per-Capita -15-5 5 15 25 35 45 55 65 75 85 95 105 115 125 135 145 155 165 Years Since Shock Reduce TFR by 0.25 (Baseline) Reduce TFR by 0.50 Relative to No-Shock
Effect of TFR Reduction on Human Capital Joshi and Schultz (2007) examine a randomized family planning intervention in Matlab, Bangladesh. Treated villages saw an extra 15% decline in TFR and 0.52 extra years of schooling for males In our experiment, TFR falls by only 4.7%, so we scale accordingly:.047 Rise in Schooling =.52 =.16 years.15
Translate this into units of effective labor using Mincerian return of 10.1% from Hall and Jones (1999)..16 years x 10.1% =.016% increase in human capital per worker.
Human Capital Per-Capita 1.060 1.050 1.040 Relative to No-Shock 1.030 1.020 1.010 1.000 0.990-15 -5 5 15 25 35 45 55 65 75 85 95 105 115 125 135 145 155 165 Years Since Shock No Schooling Change (Baseline) With Schooling Change
1.120 1.100 1.080 1.060 1.040 1.020 1.000 0.980 Income Per-Capita -15-5 5 15 25 35 45 55 65 75 85 95 105 115 125 135 145 155 165 Years Since Shock No Schooling Change (Baseline) With Schooling Change Relative to No-Shock
Malthusian Effects: Fixed Resources Could be renewable (soil fertility, groundwater, etc.) or non-renewable (oil) Key to population interaction will be substitution in production: o Land: capital and labor can substitute for it o Oil: not much substitution
Output is a CES aggregate of other inputs and fixed factor θ α 1 α θ 1 θ 1 θ 1 θ θ Y = AK H + ax where θ is the elasticity of substitution. We examine a range of values for θ θ =1 (Cobb Douglas our baseline) θ =.75 less substitution than base θ =2 more substitution higher θ implies rents/gdp fall as population rises
For a given value of θ we calibrate the production function using information on resource rents as a share of national income World Bank (2006) resource rents as a fraction of GDP, 1970-2000: Nigeria 32% Mexico 8% India 3.5%. (doesn t include implicit rent on owned land, use of unpriced resources like groundwater, etc.)
Income Per-Capita Land Share of Income = 10% 1.120 1.100 1.080 Relative to No-Shock 1.060 1.040 1.020 1.000 0.980-15 -5 5 15 25 35 45 55 65 75 85 95 105 115 125 135 145 155 165 Years Since Shock EOS = 1.00 (Baseline) EOS = 2.00 EOS = 0.75
Income Per-Capita Land Share of Income = 30% 1.210 1.180 1.150 Relative to No-Shock 1.120 1.090 1.060 1.030 1.000 0.970-15 -5 5 15 25 35 45 55 65 75 85 95 105 115 125 135 145 155 165 Years Since Shock EOS = 1.00 EOS = 2.00 EOS = 0.75
Conclusions In a very high fertility country, reducing TFR by 0.25 ( permanently ) raises income per capita by about 6% after 50 years. Biggest piece (about 1/3) is via dependency ratio. About 1/4 is higher schooling (quality-quantity) Rest is roughly split among capital, land, and worker experience effects Results sensitive to land share and elasticity