Document Sample
Merrick Powered By Docstoc
					Reproductive Health and
Poverty Reduction: What Do
(can, might, don’t)We Know?

Tom Merrick
Hewlett/PRB London Research
Conference ~ November, 2006
Why study RH/poverty links:
   Financing of the "Cairo" agenda has
    fallen far short of changing needs.
   Changed funding modes: poverty-
    reduction credits, with MDG focus,
    guided by evidence about social sector
    investments and poverty reduction.
   How strong is the evidence that poor RH
    outcomes undermine poverty reduction?
Macro evidence that fertility
decline helps economies grow
   Rapid fertility declines in East Asia
    created a demographic bonus—a
    temporary bulge in working ages that
    enabled greater investment.
   Cashing in on bonus required "good"
    economic policies: open economies, job
    creation, investments in health and
    education, gender equity.
   Is there a parallel household-level story?
               Poor women get less care
                         Antenatal care       Skilled attendant at delivery


           Poorest 20%                                      Richest 20%
  Source: World Bank/DHS 1999 Summary of data for 10 countries

But does this, in turn, make them poorer?
Do poor RH outcomes keep
households poor?
   Poverty analyses by others suggests:
   Not much direct impact of RH
    outcomes (early childbearing,
    unintended pregnancy, maternal
    mortality) on poverty in households.
   Linkages are indirect—via health,
    education, consumption—see chart
Conceptual framework: early childbearing and poverty

  Adapted from work by Ruger, Jamison and Bloom 2001
Poverty measurement, concepts

 Income poverty
 Expenditure and consumption
 Capabilities (Sen):
     Education
     Health

     Social and economic inclusion
    Our review focused on
    three sets of RH outcomes

 1. Early childbearing
 2. Maternal mortality and morbidity
 3. Unintended, mistimed pregnancy &
  large family size
    Adverse effects of poor RH:
    quick summary
   Health: strong evidence on obstetric
    complications, unsafe abortion, low birth
    weight, lasting health problems affecting
    productivity, well-being.
   Schooling: evidence is good, includes
    debate on intergenerational transmission
    of poverty via early childbearing and
    school drop out.
   Well-being (consumption, productivity):
    evidence harder to find, impact affected
    by welfare and educational policies, labor
    market conditions.
Common threads:

 Context  matters (fosterage, labor
  market conditions, stage of
  demographic transition).
 Causality is very difficult to
  demonstrate (many feedbacks).
 Scarcity of information on maternal
  deaths in survey data.
   Child rearing customs: fosterage
    mitigates impact of early childbearing,
    maternal mortality in Africa
   Labor market conditions in Latin
    America affect link between women’s
    work and fertility
   Effects are more pronounced when
    conditions are changing (an echo of the
    macro story)
The causality problem

            Possible third
            causal variable

Reproductive                  Poverty
Health Outcome                Indicator
For a stronger evidence base:
   Apply analytical techniques that can
    overcome the problems of mutual
    causality ("natural experiments").
   Make more use of longitudinal data that
    enable tracking of effects over time (our
    work with Progresa/Oportunidades data).
   Get more mileage out of existing data
    sources (DHS, LSMS).
   Address knowledge gaps: for example,
    effects of morbidity associated with
    poorly managed obstetric complications.
Country-level work is needed:
   Research on P/RH consequences
    suggests that impacts affected by
    context: stage of demographic and
    epidemiological transition, political,
    economic and social contexts, including
    gender, so we need country studies
   It's not always necessary to have "gold
    standard” causal research to make the
    case in each country.
   It is important to link country evidence
    to relevant international evidence.
Using panel data from Mexico’s
Oportunidades (Progresa) to study RH
& Poverty Links
   Conditional cash transfers (CCTs) to
    poor households for education, health,
   Evaluation: baseline in 1997-98, follow-
    on surveys in 1999, 2000, 2003
   Initially controlled experiment, but
    controls lost as more localities included
    in program
   Survey covers some aspects of RH, but
    limited information in baseline; there’s
    an RH module in 2003
    Our research objectives:
   See whether a panel survey can help us study RH-
    Poverty link
   Initial focus on early childbearing:
       Do kids of early-CB mothers have worse educational
        outcomes (progression to secondary school—attendance
        by kids who’ve completed six grades)
       Existing evaluations (Schultz 2000) of CCTs showed
        improvements in secondary enrollments, especially girls
       Could CCTs have reduced enrollment gap relative to kids
        of later CBers
   Hypothesis is of interest because welfare and GED
    helped adolescent mothers (and their kids) in the
Percent of kids who progress to
secondary school (1997 baseline)
Mother’s age*    18 & under   19 & over
at first birth
All kids         41%          49%

Boys             50%          54%

Girls            33%          44%

(*mothers 25-39 in ’97)
Percent of kids who progress to
secondary school (2003 follow-up)
Mother’s age*    18 & under    19 & over
at first birth
All kids         56%           65%

Boys             58%           67%

Girls            54%           63%

(*mothers who were 25-39 in 1997)
    What we’re learning
   Enrollment gap existed in 1997-98, eight
    percentage points, large for girls
   Overall enrollments improve by 2003, probably
    because of CCTs (control problem in 2003, also
    issues of supply side)
   Early childbearing gap persists, but girls catch
    up a lot more than boys
   Difficult to show that early CB “caused” gap
    (endogeneity, trying to disentangle)
   May be able to attribute narrowing of gap to
    CCTs (of interest because of possibility of
    better targeting)
   Using existing panel surveys is very challenging

Shared By: