Upload
julian-atkinson
View
214
Download
1
Embed Size (px)
Citation preview
Helping reduce poverty in the short- and long-term: The experience of Conditional
Cash Transfers
Norbert Schady
The World Bank
September, 2007
Motivation
What are conditional cash transfer (CCT) programs? CCTs transfer cash to poor households, but impose
conditions on recipients To date, these conditions have generally been related to
investments in child human capital:– Regular preventive health check-ups for children 0-5– School enrollment and attendance for children age 6-17
Motivation
Conditional cash transfer (CCT) programs have become very popular: first in Latin America and now across the world
Very attractive tool as they address twin objectives:1. Reduce current consumption poverty2. Promote accumulation of human capital, helping break a vicious
cycle whereby poverty is transmitted across generations The success of the initial programs (documented through
rigorous impact evaluations) created strong demand… But many questions remain: time for a thorough stocktaking The Bank is preparing a Policy Research Report on CCTs
Conceptual basis for CCTs
1. Why transfers? Equity as a social goal, although generally some efficiency
costs
2. Why cash? Unless markets are missing, cash transfers are more efficient
than in-kind transfers or subsidies for consumption of a particular good
Conceptual basis for CCTs
3. Why conditions? In general, if markets are perfect (and in the absence of
political economy considerations), conditional transfers will be less efficient than unconditional transfers. However, there may be:
– Imperfect private information or misguided beliefs – Intra-household principal-agent problems – Externalities across households
– Political economy considerations: “compact” between the state and recipients (rather than “entitlement”)
Structure of presentation
Evidence from existing CCTs—mostly in lower-income or lower middle-income countries in Latin America and South Asia
– Consumption poverty– Education outcomes– Health outcomesIs this evidence relevant for Eastern Europe and Central Asia? (some very brief thoughts)
– Marginal populations with high levels of poverty and under-investment in human capital?
– Using cash and conditions to change other behaviors?
Targeting and consumption poverty
The impact of CCTs on consumption poverty depends on:
1. Targeting CCTs have generally targeted their resources to the poor through some
combination of geographic targeting and individual means-testing The targeting of CCTs has been decidedly pro-poor—in particular in comparison
with other social assistance expenditures
2. Magnitude of benefits 20 % of mean household consumption in PROGRESA in Mexico 4 % of mean household consumption in PRAF in Honduras Programs that make larger transfers have larger impacts on consumption
poverty
3. Whether there are offsetting adjustments by households Reductions in remittances and in adult labor market participation are minimal In some countries, transfers have resulted in large reductions in child labor
4. Whether households productively invest transfers
CCT benefits are decidedly progressive…
0%
10%
20%
30%
40%
50%
60%
1 2 3 4 5 6 7 8 9 10
Deciles of per capita consumption minus transfer
Cambodia Girls SP 2000Mexico Oportunidades 2002/4Chile SUFChile Solidario 2003Brazil BF 2004+PreBF 2002/3Jamaica PATH 2004Ecuador BDH 2006Bangladesh FSSP 2000
Programs that make larger transfers have bigger effects on consumption and poverty…
Nicaragua (RPS) Honduras (PRAF)(transfer: 17%) (transfer: 4%)
0.2
.4.6
.8ke
rnel
den
sity
0 1 2 3 4ln pc exp
Treatment Control
Treatment and Control year 2001Log per capita daily expenditure
0.2
.4.6
.8ke
rnel
den
sity
0 2 4 6ln pc exp
Treatment Control
Treatment and Control year 2002Log per capita daily expenditure
The crowding out effects of transfers are modest, and may be positive…
1. Remittances: No crowding out of remittances in Nicaragua, Honduras, and Mexico
2. Minimal reductions in adult labor supply in Mexico and Nicaragua
3. Large reductions in child labor as a result of transfers in some countries—in part, because schooling and child labor can be substitutes Nicaragua: RPS transfers led to 2.5-4.9% point reduction in child
labor Reductions in child labor generally larger for boys than girls
In most cases, CCTs appear to have increased consumption and reduced poverty…
Table 2: Program Impact on Poverty at the National Poverty LineHead Poverty Povertycount Gap Severity
Honduras -0.0138 -0.0215 -0.0190* [0.0150] [0.0132] [0.0110]
Nicaragua 2001 -0.1017*** -0.1328*** -0.1126*** [0.0293] [0.0194] [0.0149]
Nicaragua 2002 -0.0669** -0.0982*** -0.0865*** [0.0296] [0.0200] [0.0155]
Mexico 1998 0.0007 -0.0284** -0.0393*** [0.009] [0.011] [0.012]
Mexico 1999/6 -0.0060 -0.0445*** -0.0616*** [0.011] [0.011] [0.011]
Mexico 1999/10 -0.0207** -0.0794*** -0.0938*** [0.009] [0.012] [0.013]
Data: Evaluation data from Honduras, Nicaragua, and Mexico. Each entry is the program impact on therespective poverty index at the national poverty line. Standard errors in brackets. Source: Evaluation data from PRAF, RPS, and Progresa.
Short-term comparisons may understate long-term CCT impact on consumption poverty…
Long-term CCT effects on consumption poverty
In Mexico, households appear to invest about 25 cents out of every peso transferred by Oportunidades in productive assets in micro-enterprises and agricultural production
– By investing transfers, beneficiary households increased consumption by about 24 % after six years on the program
– Short-term, “static” comparisons of consumption poverty between CCT beneficiary and comparison households may underestimate the returns to transfers
Evidence from Nicaragua that the RPS provided a stable stream of income that could be used as collateral for loans or to protect households during income fluctuations
Conclusions: Targeting and consumption poverty
Conclusion 1: CCTs have generally been well targeted Disincentive effects have been small CCTs have had an important effect on consumption poverty
Policy implication 1: CCTs can help reduce consumption poverty However, if the main objective of the program is the reduction
of (short-term) consumption poverty, UCTs are likely to be more effective unless there are political economy reasons to impose conditions
CCTs have large, significant effects on school enrollment in a number of countries…
Examples from Latin America:– Nicaragua—RPS: 17.7% points– Honduras—PRAF: 3.3% points– Mexico—Oportunidades (urban areas): 3% points
Examples from “stipends” for girls in secondary school in Asia:– Cambodia: 22-33% points– Bangladesh: 12% points– Pakistan: 9% points
The impact of CCT on enrollment is largest in countries which had low baseline enrollment rates…
Mexico
Mexico
Mexico
Nicaragua
Honduras
Ecuador
Colombia
Colombia Brazil
Cambodia
Bangladesh
010
20
30
Impact
of
CC
T o
n e
nro
llment
20 40 60 80 100Enrollment rate at baseline
CCT effects on enrollment are largest among children with low propensities to enroll in school…
1. Comparisons across countries: Larger program effects in countries with lower baseline enrollment levels
Impact of CCT on transition from primary to secondary is 2-3 times larger in Cambodia than in Mexico
2. Comparisons across grades within countries: Larger program effects in transition grades with high dropout rates
Impact of PROGRESA-Oportunidades in Mexico is ~ 4 times as large for children in 6th grade at baseline as children in other grades in primary school
3. Comparisons of households by socioeconomic status: Larger program effects among poorer households
In Nicaragua, program effect on enrollment is 26 % points among extremely poor households, 12 % points among poor, and 5 % points among non-poor
Evidence on effect of CCTs on educational attainment of adults and learning outcomes is limited…
1. In Mexico, Children who were randomly assigned to receive two more years of CCT have approximately 0.2 more years of schooling
Results hold for children who were 13-15 at baseline, and who are therefore likely to have completed their schooling
If rate of return to schooling is ~ 10 percent, this means that, on average, these children will earn approximately 2 percent higher wages as adults
2. However, in Mexico, Ecuador, and Cambodia no effect of CCT program on test scores—either in the short run, or the long run Are CCTs bringing into school children who are so disadvantaged in other
ways that the amount of learning that takes place is limited? Is the quality of the supply so low that children are learning very little—in spite
of the additional years of schooling attained?
Cross-cutting issues and conclusions: Education
Conclusion 2 CCTs significantly increase school enrollment in a variety of
settings, especially among population groups with the lowest counterfactual enrollment
Impact of conditional transfers on enrollment is larger than that of unconditional transfers (Ecuador, Mexico)
CCT program effects on final outcomes, in particular learning, is somewhat disappointing:
– Quality of supply– Other household constraints to learning
Cross-cutting issues and conclusions: Education
Policy implication 2 Clarify objectives: What is the relative importance of
increases in enrollment relative to redistribution? If enrollment is important, target intervention to low-enrollment
populations or to transition years with high school drop-out Do we have the right conditions? Is there scope for a program
that targets poor children with the highest academic potential?
Cash transfers and enrollment incentives are unlikely to be sufficient to improve learning outcomes on their own
– Identify other supply or household constraints and design policies to address them
Consistent evidence that CCTs result in increases in utilization of public health facilities…
Evidence from Mexico, Nicaragua, Honduras, Colombia, and Turkey: CCTs increase coverage of some preventive services for children (growth monitoring, immunization rates)
Colombia: 23-33 % point increase in growth monitoring Turkey: 14 % point increase in preschool vaccination rates
Mexico: Increase in use of preventive health care by adults as well as children (preventive visits; check-ups for diabetes and hypertension)
18 % point increase in the preventive health visits Increase in use of public providers may result in decrease in visits to
private providers
CCT programs have had some positive effects on child health and nutrition…
Chronic malnutrition: Decreases in stunting in Mexico and Colombia (among the poorest children in rural areas, especially younger children) but not in Brazil or Honduras
Infant mortality: A municipality with full coverage of Oportunidades has 11% lower infant mortality rate than it would have had without the program
Anemia prevalence: Children who received Oportunidades transfers had 10 percentage point lower anemia (54.9%, compared to 44.3%)
– No differences after control group had been folded into treatment for one year
Incidence of illness: After 2 years of program exposure, children in Oportunidades treatment area were 40% less likely to be reported to have been ill by parents than those in control area
CCT program had positive effects on some measures of adult health, but evidence draws only on Mexico…
Subjective measures of health status: CCT program reduced number of days adults 50+ had difficulty performing daily tasks, reduced number of days incapacitated due to illness, increased their ability to walk longer distances
Objective measures of health status: PROGRESA-Oportunidades reduced incidence of obesity (by 6.4 % points) hypertension (7.2 % points) and diabetes (4.2 % points)
Cross-cutting issues and conclusions: Health and nutritional status
Conclusion 3: CCTs have resulted in increases in utilization of preventive health
check-ups by children and (for Mexico) adults There is some evidence of CCT program effects on child nutritional
status and child cognitive development, but…– Positive program effects in some countries and population
groups and not others– Unconditional transfers seem to have positive effects of roughly
similar magnitude, if well targeted CCTs may also reduce child morbidity and mortality and improve
adult health status, but evidence draws on only one country (Mexico)
Cross-cutting issues and conclusions: Health and nutritional status
Policy implication 3: Clarify objectives: What is the relative importance of
improvements in health status relative to redistribution? If health status is important, target intervention to population
groups who are poor and have poor health status Do we have the right conditions? Provision of information,
social worker campaigns to improve parenting habits may have higher returns than visits to health centers
What role for CCTs in Eastern Europe and Central Asia?
Overall, inadequate investment in the quantity of human capital does not seem to be a problem
Safety nets are important but redistributive concerns do not in and of themselves justify imposing conditions on transfers
So… is there a role for CCTs in countries in Eastern Europe and Central Asia?
1. “Niche” programs: targeted at populations that are socially excluded or under-invest in human capital
2. Different conditions: Use the basic CCT insight—cash transfers can be paired with conditions to change behaviors—to address other problems
What role for CCTs in Eastern Europe and Central Asia?
1. “Niche” programs: not the main safety net program: The example of Chile Solidario
Targeted at the extremely poor and excluded, as identified by the Ficha CAS (a proxy means): ~ 5% of population
Social worker works with eligible families to determine assets and deprivation Families identify highest-priority needs
Beneficiaries sign a “contract” with social worker No rigid “conditions”
Social worker helps beneficiaries get appropriate documentation, access existing social programs
Small cash transfer– Monetary benefits taper off over time and, in theory, families become ineligible after
five years in the program
What role for CCTs in Eastern Europe and Central Asia?
1. “Niche” programs: The example of Chile Solidario
First round of evaluation suggests mixed results:– Increased uptake of social programs– Increased enrollment in preschool, primary, and adult education– Increase use of preventive visits to health facilities by children and adults– Greater optimism about a family’s future– However, disappointingly, no increase in labor supply or earnings—at
least in the short-term
Other examples include a program being developed for Aboriginal groups in Australia
Is there scope for a program like this in countries in ECA—for example, to address issues of social exclusion or under-investment in human capital among the Roma?
What role for CCTs in Eastern Europe and Central Asia?
2. Conditional programs—with different conditions: the 1996 US welfare reform
• Central government: Transformed entitlement program into block grant to states
1. Work requirement: States had to show that 50% of recipients were at work or in work programs to receive transfers
2. Time limits:60 months over entire lifetime
• State governments introduced a series of reforms to their transfer programs: changing the “culture of welfare”
1. Lowering earnings disregards: decreasing magnitude of overall transfers, but allowing recipients to keep a higher fraction of earnings while maintaining eligibility for transfers
2. Sanctions—for example for noncompliance with work programs or training programs
What role for CCTs in Eastern Europe and Central Asia?
2. Conditional programs—with different conditions: the 1996 US welfare reform
• Reforms are generally credited with:• Decreasing caseloads• Increasing labor market participation• Reducing poverty
• But…implemented during very healthy economic expansion—some believe that reforms have not yet really been put to the test
What role for CCTs in Eastern Europe and Central Asia?
3. Niche programs—with different conditions: The New York Opportunity program
In one sense, this is a “traditional” CCT, targeted at the poor in NYC– Family has to demonstrate income<130% of the national poverty line
Households receive payments if their children are enrolled in school, regularly attend school, and have regular preventive health care visits
However, there are a variety of NYC-specific innovations:– Includes additional payments for adequate performance, including child
performance on test scores – Includes payments to parents if they maintain sustained full-time
employment, and if they receive education and training while employed An exciting case in which a country (or city) in the developed world learned
from one in the developing world Program has just been launched by Mayor Bloomberg--no evaluation of the
impact is available, although a rigorous evaluation is planned
What role for CCTs in Eastern Europe and Central Asia?
The importance of impact evaluation The reason we know so much about CCTs, and what they
accomplish and do not accomplish, is because many programs took impact evaluation very seriously
PROGRESA in Mexico: randomly assigned communities into “treatment” and “control” groups during roll-out of national program
Smaller programs (Nicaragua, Honduras) also evaluated by random assignment
Other evaluations based on matching techniques or regression discontinuity
US Welfare reform: built on a series of state-level experiments in the 1980s and early 1990s: randomized evaluation mandated by national government