Targeted Human Development Programs_ Investing in the Next Generation

7/21/2019 Targeted Human Development Programs_ Investing in the Next Generation

1/30

Targeted HumanDevelopment Programs:

Investing in the Next Generation

Arianna LegoviniFerdinando Regalia

Inter-American Development Bank

Washington, D. C.

Sustainable Development DepartmentBest Practices Series


2/30

Cataloging-in-Publication provided by

Inter-American Development BankFelipe Herrera Library

Legovini, Arianna.. Targeted human development programs : investing in the next generation /Arianna Legovini, Ferdinando Regalia..

. p. ; cm. (Sustainable Development Dept. Best practices series ; POV-110)

September 2001Verso t.p.

1. Human capital--Latin America. 2. PoorLatin AmericaSocial conditions. 3.

Latin AmericaSocial policy 4. Health services accessibilityLatin America.5. Education of the PoorEconomic aspects. I. Regalia, Ferdinando. II. Inter-

American Development Bank. Sustainable Development Dept. Poverty and Ine-quality Unit. III. Title. IV. Series.

331.11 L8468--dc21

Arianna Legovini is an Economist in the Poverty and Inequality Unit. Ferdinando Regaliais a Junior Professional in the Social Programs Division of the Regional Operations De-

partment 2 of the Inter American Development Bank.The authors thank Jere Behrman,Carola Alvarez, Ricardo Moran, Amanda Glassman and Nora Lustig for their thoughtfuland comprehensive comments on the initial draft of this document. Presentations byCristovam Buarque and Santiago Levy at the Seminar on Targeted Human DevelopmentPrograms (June 1, 2000) provided valuable insights for this report.

The opinions and conclusions expressed herein are those of the authors and do not neces-

sarily reflect the official position of the Inter-American Development Bank.

September 2001

This publication (Reference No. POV-110) can be obtained from:

Poverty and Inequality UnitMail Stop E-0421Inter-American Development Bank1300 New York Avenue, N.W.Washington, D.C. 20577

E-mail: [email protected]: 202-623-3299Web Site: http://www.iadb.org/sds/pov


3/30

Foreword

Intergenerational persistence of poverty is generally linked to thereduced capacity of poor families to foster human capitalaccumulation of their children and pull them out of poverty.

Supply-side interventions, which increase the availability andquality of schooling and health services, might be insufficient toimprove this capacity when resource-constrained low-income

families cannot bear the direct and indirect private costs ofacquiring these services.

During the past decade a new generation of integrated povertyreduction programs, Targeted Human Development Programs

(THDPs), have been implemented in Latin America to tackle thisproblem by addressing the demand side in the use of social

services. The Inter-American Development Bank has played andcontinues to play an active role in establishing these programs andensuring that they incorporate sound impact evaluation

mechanisms.

This report sets out the rationale for implementing THDPs andoutlines instructions for their design and implementation. Inaddition a list of appropriate country conditions for the

implementation of THDPs is presented. The report also describesways of incorporating impact evaluation mechanisms into project

design and presents some of the results based on existing empiricalevidence. Finally, a logical framework model for a THDPoperation is included in the Annex.

Nora C. Lustig

Senior Advisor and ChiefPoverty and Inequality Unit


4/30

Contents

What Is a Targeted Human Development Program?

1

Motivation for an Integrated Approach

3

. Designing and Implementing a Targeted Human Development Program

5

Impact Evaluation and Expected Results

15

Country Conditions

19

Bibliography

20

Annex:

Logical Framework for a Targeted Human Development Program

24


5/30

1

What Is aTargeted Human Development Program?

Targeted Human Development Programs(THDPs) are integrated poverty reduction pro-grams designed to increase the capacity of the

poor to accumulate human capital. The programsare directed primarily to poor and vulnerablefamilies with pre-school and school-age chil-dren. Their main long-term objective is to eradi-cate the structural causes of poverty by fosteringinvestment in the next generations humancapital. A secondary objective is to alleviate

poverty in the short term, mainly throughmonetary transfers.

This report sets out the motivation for imple-menting THDPs and step-by-step instructionsfor design and implementation. It describes theimplementation of impact evaluation mecha-nisms and results that can be expected and offersa list of appropriate country conditions for aTHDP. A model logical framework for a THDPoperation is presented in the Annex.

WHY FOCUS ON HUMAN

CAPITAL ACCUMULATION?

There is a strong correlation between incomeand human capital levels. Poor families are gen-erally unable to accumulate sufficient humancapital and earn enough to pull themselves outof poverty. In addition, supply-side interventionsaimed at increasing the availability and qualityof schooling and health services may not be suf-ficient to level the playing field for poor fami-lies. Resource constraints may prevent low-income families from providing adequate nutri-tion, health care, and schooling for their chil-

dren. For this reason, THDPs address the de-mand side in the use of social services.

The indivisible combination of short- and long-term objectives distinguishes THDPs from thetraditional concept of transfer programs focused

primarily on short-term income support. Strictenforcement of requirements ensures that the

long-term objectives are met. Current transfersare conditioned on behaviors that increase hu-man capital accumulation, including prenatalcare, early childhood development, and chil-drens health care, nutrition (including nutri-tional courses to parents), and school attendance.

An enhanced human capital asset base increasesearnings potential and the capacity to move outof poverty. The intention is to permanentlychange the conditions of poor families by edu-cating a whole generation of children and en-

gendering a virtuous cycle of greater humancapital and higher earnings. Educated childrenmake educated parents who look after their ownchildrens education; educated girls have lowerfertility rates and better child feeding practices;and widespread primary and secondary educa-tion reduces inequalities.

While THDPs may help poor families withstandincome shocks by increasing transitory and per-manent household income; they should not berelied on as the main safety net mechanism for

responding to macroeconomic shocks. This isbecause THDPs are designed to target thestructurally poor and may not be easily ex-

panded during downturns to incorporate newpoor, partly because targeting relies on indica-tors of permanent poverty that are not affected

by temporary income downswings.

EXPERIENCE WITH

TARGETED HUMAN DEVELOPMENT

PROGRAMS

Several THDPs have been implemented in LatinAmerica during the past decade. The first ex-periment with the concept was the 1992 Pro-gresapilot program in Mexico. Its success led tothe establishment of a countrywide program in1997 that integrates education, health, and nutri-tion activities for pre-school and school-agechildren. Progresa is also the first social pro-


6/30

2

gram to have developed a statistically sound im-pact evaluation system for assessing the impactof the program on various dimensions of povertyreduction.

The Brazilian Bolsa Escola programs focus onschooling and provide monetary incentives for

attendance and performance.

PRAF in Honduras combines supply-side anddemand-side interventions in an experiment to

measure the relative effectiveness and synergyof supply and demand actions.

Red de Proteccin Social in Nicaragua was de-signed along the lines of the Progresa modeland modified to reflect the late enrollment andearly dropout characteristics of the Nicaraguan

countryside.

The Apoyo Familiar program in Colombia alsocombines different social sector interventions.1

1The Inter-American Development Bank has played

and continues to play an active role in the establish-ment of these programs and in ensuring that they

incorporate sound impact evaluation systems. IDBloans have been extended to support the programs inColombia, Honduras, and Nicaragua. Another is in

preparation for Brazil.


7/30

3

Motivation for an Integrated Approach

APPLYING AN INTEGRATED HUMAN

CAPITAL DEVELOPMENT APPROACH

Targeted Human Development Programs adoptan integrated approach to developing the humancapital of the poor by addressing the nutritional,educational, and health care needs of poor fami-lies. Nutrition, education, and health services arecomplementary in the generation of humancapital.

Health status affects nutritional interventions.Efforts to increase food intake in poor families

may fail to improve the nutritional status offamily members whose health is weak. If theincidence of diarrhea and intestinal diseases ishigh, for example, interventions to increase foodintake must be accompanied by improvements insanitary and health conditions and good feeding

practices.

Female education and training affect the healthand nutritional status of family members. Edu-cated women postpone childbearing, have fewer

pregnancies, and provide better nutrition and

health care to their children.

Nutrition and health status affect cognitiveachievement.Several experimental studies find asignificant positive association between nutritionand health (in particular iron status and lack of

parasitic infections) and learning ability andcognitive achievement (see Behrman 1996 for areview).Pollitt and others (1993) find significanteffects of early childhood supplementary feedingon various measures of cognitive skills in ado-lescents in Guatemalan villages. Using a longi-

tudinal data set for the Philippines, Glewwe,Jacoby, and King (2000) find that malnourishedchildren are less likely to be successful inschool.2

2They find that heterogeneity in learning endow-ments, home environments, or parentalpreferences

cannot fully account for the fact that malnourished

Nutrition and health affect enrollment and

earnings. Glewwe and Jacoby (1995) find thatnutrition and health affect the age at which chil-dren enroll in school and that early childhoodmalnutrition delays enrollment. They also dem-onstrate that delayed enrollment has a sizableimpact on lifetime individual earnings.

RELAXING DEMAND

AND SUPPLY CONSTRAINTS

The rationale for implementing THDPs is theexistence of demand constraints on the ability of

poor families to invest in human capital. De-mand-side interventions would complementnot substitute forsupply-side interventions.

Demand Constraints

Poor households cannot afford to send theirchildren to school, pay for health care, or pro-vide them with a balanced diet, and they mustoften send children to work to help support thehousehold.

Income fluctuations induce discontinuity inschool attendance. Jacoby and Skoufias (1997)use panel data from rural India to analyze howhuman capital investment in children respondsto fluctuations in household income. They findthat child labor and school attendance fluctuatesignificantly in response to variations in house-hold income. Further, variability in school atten-dance might have a long-term negative impacton the human capital accumulation of poor chil-dren.

children perform poorly in school. While there is nostrong negative impact of malnutrition on learning

efforts (homework time and school attendance), ma l-nutrition delays primary school enrollment. Theirresult supports a causal link between nutrition and

academic success.


8/30

4

Children are sent to work to bridge the povertygap.Bhalotra (2000) estimates the wage elastic-ity of childrens labor supply using householddata from rural Pakistan. The wage elasticity for

boys is significantly negative: as their wage risestheir labor supply declines. This implies thatchildren work to bridge a poverty gap. The re-

sults support the claim that the primary reasonfor child labor is insufficient household income(below subsistence levels), rather than, for ex-ample, low returns to education.

Poor households cannot afford to send theirchildren to school. Legovini and Lustig (2000)show evidence from household surveys in LatinAmerica indicating that the most frequent reasoncited for not sending children to school is thehigh cost of attendance (transport, clothes, mate-rial, and the opportunity costs of remunerated or

unremunerated child labor). This signals the ex-istence of resource constraints.

Supply Constraints

The accessibility and quality of school facilities,or the returns of child labor over the returns toeducation, play an important role as well.

Ray (2000) finds that poverty increases childlabor hours and decreases school attendance inPakistan but fails to find significant results forPeru. He speculates that the higher quality ofschools in Peru may explain this difference.

Using household survey data from Mozambique,

Handa (1999) finds that both demand-side inter-ventions (increasing per capita consumption,improving adult literacy) and supply-side inter-ventions (reducing the number of children perteacher) have large positive impacts on chil-drens enrollment.

These findings suggest a need to address bothdemand and supply constraints. In addition, de-mand interventions will increase the pressure onexisting structures. Experiments, such as the onedeveloped by PRAF II in Honduras, can be de-

signed to measure the relative efficiency andsynergy of supply and demand interventions.


9/30

5

Designing and Implementinga Targeted Human Development Program

PROGRAM PRECONDITIONS

Four important conditions should be met beforeestablishing a targeted human development pro-gram. First, the governments political and fiscalcommitment should be ensured. Second, existingsocial programs should be thoroughly analyzed.Third, intersectoral coordination mechanismsshould be established. Finally, a THDP requiresan adequate institutional capacity.

Ensuring Political and Fiscal Commitment

Poverty targeted programs require politicalcommitment to reach the very poor. This re-quirement is intensified for THDPs because theysubstantially affect the allocation of socialspending, moving funds from general to targetedsubsidies and from urban to rural populations.

Because THDPs are intended to increase thehuman capital of a whole generation, so thatchildren can become educated parents and in

turn educate their children, governments shouldmake a medium-term financing commitment(10-15 years) before implementing a THDP.

THDPs should replace programs with similarobjectives that failed to deliver positive results,to avoid duplication of effort and conserve re-sources.

Analyzing Existing Social Programs

The efficacy of existing social programs (food

programs, vouchers) should be analyzed beforeimplementing a THDP. Though few social sec-tor programs have reliable evaluation results oradequate data to undertake such an evaluation,some analysis is essential to obtain a standardagainst which THDP performance can be meas-ured and to reallocate resources efficiently.

THDPs are one element of a wider arrayof gen-eral education, health, social security, labortraining, and housing policies to increase theasset base of the poor, increase their earning ca-

pacity through employment and productive ac-tivities, improve their living conditions through

basic infrastructure and housing, and reduce dis-parities. It is thus important to recognize the roleand limitations of a THDP and ensure its con-sistency with a countrys poverty reductionstrategy and overall sector policies. For exam-

ple, if a government charges user fees for schooland health services, adjusting fees in poor areasmight be less costly than channeling transfersthrough the demand side.

Establishing Intersectoral Coordination

Mechanisms

Adequate supply must be ensured before inter-ventions are undertaken to stimulate demand foreducation, heath care services, and food. MostTHDPs are first implemented as pilot programs

in areas of easy access and few supply con-straintsthis was the case with Progresa inMexico. The Red de Proteccin Social pilot

program in Nicaragua followed the same ap-proach, at least in terms of availability ofschools. Bolsa Escola programs in Brazil havenot required supply interventions because theyoperate in urban areas where school coverageand quality are considered adequate. However,when THDPs are expanded to more remote ar-eas, extensive supply-side interventionsquan-tity and quality of health and education provi-

sionsmay be indispensable.

Combining demand- and supply-side activitiesrequires substantial coordination between lineministries and the executing agency. For exam-

ple, agreement is needed on the action plan de-fining the schedule of investment in social infra-


10/30

6

structure, the targeted population, the menu ofinterventions, and the mechanisms for serviceand benefit delivery. Interministerial coordina-tion should also include finance ministers to en-sure compatibility between program objectivesand resource availability.

Building Adequate Institutional Capacity

Building the institutional capacity to manage aTHDP is essential. Data collection and analysisand development of a targeting system (if one isnot already available) require substantial finan-cial and technical resources. Delivery andmonitoring mechanisms need to be tested. Staffrequirements are intensive at start-up (informa-tion system specialists, statisticians, and sectorspecialists, and external consultants can meetmuch of that need). Synergies with national sta-

tistical institutes should be fully exploited.

Once systems are in place, however, the pro-gram requires a relatively small staff to service alarge number of beneficiaries. For example,

Progresas administrative costs are less than 5percent of the total program costs. In Red deProteccin Social, a much smaller program at itsbeginnings, administrative costs will represent12 percent to 13 percent of total costs after thefirst year of implementation.

PROGRAM DESIGN

Several issues should be taken into considerationin the design of a targeted human development

program. Program objectives should be clearlystated and include measurable outcomes. In ad-dition, the targeting and selection of beneficiar-ies are vital components of these programs. Tar-geted Human Development Programs shouldinclude education as well as health and nutritioncomponents. Other issues important to considerin the design of these programs are exit rules,

transparency, supply-side interventions, privatesector providers, the effects on local markets,expansion to urban areas and the programs fis-cal sustainability.

Setting Clear Program Objectives and Meas-

urable Outcomes

Long-Term Objective. The long-term objectiveof THDPs is poverty reduction. Indicators (suchas poverty headcount and measures of the depthand severity of poverty) need to be selected for

assessing achievement of the objective. Wherethe objective is to bridge social disparities (forexample, in education levels) across differentgroups (gender, geographic, ethnic), measures ofeducation and wage inequality can be used asindicators.

Medium-Term Outcomes. The medium-termoutcomes of THDPs vary according to the de-sign of the program. Defining program outcomesthat can be easily monitored and measured iskey to evaluating program success. Medium-

term outcomes could include, for example, anincrease in the populations average years ofschooling (education); a decline in maternal andinfant mortality rates (child and maternal care),and a decline in the incidence of child malnutri-tion or illness (health and nutrition). Also, theactive involvement of beneficiary families in

program execution (participation) may promotepositive behavioral changes in the medium termsuch as better hygiene and feeding practices.

Short-Term Outcomes. Short-term outcomes

should be monitored to insure achievement oflonger-term objectives. For example, schoolgrants are expected to increase school enroll-ment and attendance and reduce dropout rates.

Repetition rates should fall if scholarships in-crease with grades or if deposits are made inindividual savings accounts after children havecompleted a given educational cycle (as in Bra-zilsBolsa Escola). The program should reducechild labor supply, at least the number of hourschildren work. Sharper reductions can be ex-

pected if after-school programs and attendance

monitoring are included, as in the BrazilianPETI.3Prenatal and post-natal carecomponents

3The PETI is the Brazilian program for the eradica-

tion of child labor. In addition to providing scholar-ships to poor children, it implements an integratedstrategy in communities with a high incidence of

child labor. The strategy includes changes in produc-


11/30

7

should lead to an increase in the number ofhealth care visits. Nutritional components, suchas food supplements, food security, and nutri-tional training, are expected to improve the nu-tritional intake of parents and children in theshort run, and their nutritional status in the me-dium run as well as compensate for imbalances

in household food allocation. The income sup-port component is expected to alleviate thedepth and extent of poverty in the short term.Jointly with nutritional components, incomesupport is expected to increase the quantity andvariety of consumption goods acquired. Whenthe target population is extremely poor, an in-crease in income should produce a more than

proportional increase in food consumption. Theproportion of total expenditures on child-care,childrens clothing, and school supplies can also

be expected to increase.

The program might generate other outcomesdirectly or indirectly. For example, it couldchange the labor supply of adult members of

beneficiary families, affect the intra-householdbalance of power (since women receive thegrants, their power to take decisions may in-crease), increase business start-ups and crowdout inter- and intra-householdprivate transfers.

Establishing a Targeting System

and Selecting Beneficiaries

Targeting and selection of beneficiariesare cru-cial components of THDPs. Some of the steps tofollow in this regard are: (a) defining the target

population, (b) selecting the beneficiaries and (c)developing an experimental design.

Defining the Target Population

Since the objective of THDPs is to reduce pov-erty through human capital development, thetarget population is poor families with young

children. The aged poor and poor families with-out children are immediately excluded as poten-tial beneficiaries; they can be assisted by other

programs. The number of families covered de-pends on the financial resources available and

tion technology and campaigns to promote cultural

change.

the size of the transfer per family. Considerationmust be given to the trade-off between size ofthe transfer and coverage as well as to the possi-

bility of phasing in the program. Both institu-tional limitations and budget constraints mayrequire gradual implementation.

Selecting Beneficiaries

THDPs often use a combination of geographicand household targeting. These methods rely onclear selection rules based on data from nationalhousehold surveys and censuses. Geographictargeting is less expensive to administer butdoes not differentiate between poor and non-

poor households in a community, so poorhouseholds in non-poor areas would be excludedwhile non-poor households in poor areas would

be included.Household targeting requires more

information but can discriminate, albeit imper-fectly, between households. Geographic andhousehold targeting should be combined in dif-ferent geographic areas or stages of a program.When the program focuses on homogeneously

poor communities, geographic targeting is mostcost-effective as long as the cost of covering thenon-poor is smaller than the administrative costsof surveying each household, adjusted by theexpected leakage rate. As the proportion of non-

poor increases in targeted communities, house-hold means testing should be used.

Geographic areas can be ranked according to apoverty or marginality index, with the index de-pending on available data and government pri-orities. Communities can be ranked by a malnu-trition index if reducing malnutrition is a priority(PRAF II in Honduras) or by the incidence ofchild labor in certain types of agricultural activ i-ties if reducing child labor is the priority (PETIin Brazil). A common poverty or marginalityindex uses census data to calculate a weightedlinear combination of demographic and housing

characteristics. These may include the percent-age of illiterate population ages 15 and older, thepercentage of dwellings without running wateror sewerage systems, and the average number ofoccupants per room. Census data contain littleinformation on household consumption and in-come, however.


12/30

8

Combining census information with householdincome and expenditure surveys delivers betterresults. Household survey data can be used toestimate an econometric (probit) model to pre-dict the poverty status of a household, as meas-ured by per capita family consumption (prefera-

bly) or per capita family income (when con-

sumption data are not available). The explana-tory variables must be available in the censusdata so that the model can be extrapolated to thewhole population. Estimated coefficients fromthe poverty probit model are applied to the cen-sus data at the region or locality level to deter-mine the predicted incidence of poverty (draw-ing a poverty map). Localities can then beranked by degree of marginalization. The rank-ing provides a schedule of implementation forthe program.

Some restrictions apply to the choice of locali-ties for a first phase. If the program contains nosupply interventions (and none have been agreedto with line ministries), a THDP will be re-stricted to areas where the provision of educa-tion and health services is adequate. Otherwise,the conditions of the program cannot be en-forced. In many cases THDPs are started as pilot

programs in areas where road access is alsorelatively easy.

Warning: Results Cannot Be Automatically

Extended

The lessons from the evaluation of pilot pro-grams must be interpreted with caution. Resultscannot be automatically extended to areas thatrequire supply interventions. For this reason, a

pilot program should try to cover areas thathave different supply and demand needs. Ex-

periments testing the relative effectiveness ofsupply and demand interventions (PRAF II) mayalso be desirable.

Rules can be instituted to reduce the incentive tomigratefrom a nonbeneficiary community to thebeneficiary communitya minimum residencyrequirement, for example. This is a concern oflocal governments, which do not want to attract

poor people from other districts (Bolsa Escola).In national programs, migration of poor families

from control to beneficiary communities maybias evaluation results.

Household Targeting. Socioeconomic informa-tion used to test income or consumption levels isobtained through a questionnaire designed tocollect data on household characteristics and

the well-being of its members. Interviewees sign(or fingerprint) the questionnaire to certify itsauthenticity. The information should be saved inelectronic files for processing. Each householdand its members are assigned identificationnumbers, to protect the confidentiality of theinformation.

Data collection and processing should be super-vised through control of sample mechanisms ofcoverage and quality assurance. Municipalauthorities and local representatives should

provide support, helping to identify the geo-graphic borders of the locality and authorizingthe activities of those conducting the question-naires.

Once household information has been obtained,beneficiary families are selected on the basis ofa standardized evaluation of socioeconomiccharacteristics.

Warning: Program Admission RulesShould Not Be Disclosed

Program admission rules should not be dis-closed to avoid distorting family behavior. Forexample, if the proxy for testing income is the

floor or roof material of the dwelling, a familycould change them to qualify for the program.

Nor should the rule selected distort behavior.For example, if the rule for program admissionis the working status of children, reported childlabor may suddenly increase to win access to the

program. For this reason, THDPs target poorfamilies with school-age children, regardless of

whether the children attend school or work ordo both.

Community Involvement. To minimize errors ofinclusion (of non-poor families) and exclusion(of poor families), the list of selected familiesmay be submitted to the community for scrutiny,in some cases by public posting of selections.


13/30

9

Surveyors revisit any households identified bythe community to ascertain the veracity of theclaim and update the list accordingly. In Co-lombia, community committees (veeduras ciu-dadanas), which include the local priest, teacher,and doctor, review the lists to assess errors.

Experimental Design

If the program is gradually expanded and an ex-perimental evaluation design is used to assess itsimpact (see impact evaluation section), somegeographic areas or households might be ex-cluded from the programs initial phase. Therandom selection of villages or households forcontrol and treatment groups should be transpar-ent and open to all stakeholders. PRAF IIused alottery supervised by community leaders to se-

lect beneficiary and nonbeneficiary localities fortreatment and control groups among localitiesthat could qualify to receive the program. Chil-dren conducted the lottery drawing in the pres-ence of stakeholders and the state institutions incharge of administrative probity.

Education Component

This component provides monetary incentives tobeneficiary families conditional on childrensschool enrollment and attendance.

Scholarship Design

A grant that covers the opportunity and direct(transport and material) costs of a child attend-ing school will fully compensate families forsending children to school. For most parents,who positively value their childrens educationmore than their childrens earnings, a smallergrant will induce them to send their children toschool.

An analysis of female-male and rural-urban en-rollment, delayed enrollment, and repetition anddropout behavior should be undertaken to designa structure of scholarships that maximizes theimpact of the program. Common problems aredelayed enrollment and dropout after first orsecond grade, high dropout rates between pri-mary and secondary education, and high repeti-

tion rates. Gender differences are also important:in rural areas girls are often withdrawn fromschool earlier than boys to help in the house.

Strategies to deal with country- or area-specificproblems include differential grants by gradelevel and gender and rewards for successful

grade or school-cycle completion.

In Mexico, where analysis showed that enroll-ment declined and dropout increased signifi-cantly between primary and secondary school-ing,Progresaawards scholarships to children inthe last years of primary and first years of sec-ondary school. The grants increase with gradelevel and between primary and secondary schoolto avoid dropout at this critical stage. To closethe gender gap,Progresaawards larger grants togirls, which has resulted in greater female than

male school attendance. The gender gap indexneeds to be selected carefully. In Mexico, thegap in attendance favored boys, but the gap inattainment favored girls. The effect of Progresawas to reverse the gender gap in attendance andstrengthen the gender gap in attainment, effec-tively hurting boys.

In Brazil, a schedule of awards for successfulgrade completion and school cycle completion

provides incentives for performance and com-pletion of full school cycles (primary and secon-

dary).

InNicaragua, where the problem is delayed en-rollment and high dropout in primary school, the

Red de Proteccin Social awards grants to poorrural children in grades 1 though 4 to support theachievement of functional literacy. This choicewas determined in part by the absence of grade 5and above in many rural areas, a restriction thatmust be addressed through supply-side interven-tions.

Grants can be provided to individual children(as in Mexico) or families (as in Brazil). Familygrants require every school-age child in thefamily to attend school. They are easier to ad-minister but harder to monitor than individualgrants. They motivate families to send all oftheir children to school rather than select amongthem, but enforcement penalizes all children


14/30

10

when one child fails to attend school. Both sys-tems have some incentives for increased fertil-ity: in the family grant, to extend the period of

benefits, and in the individual grant, to increasethe number of grants and extend the period of

benefits. Imposing a cap on the number of chil-dren who can receive benefits and a time limit

per benefit per family can help curb fertility in-centives. Family grants and caps on number ofchildren may, however, provide incentives forfamilies to separate into different households.

Some basic principles can be used to calculategrant amounts. Household surveys such as theLiving Standards Measurement Surveys (LSMS)can be used to estimate direct and opportunitycosts of attending school. Direct costs can bederived from household expenditure modules;opportunity costs, from wage equations based on

information from the employment and educationmodules.

The opportunity cost for young children may behard to estimate because wage work is rarelyreported for young children. When no informa-tion is available, the project team should investi-gate the socioeconomic conditions and time al-location of the target population. A second butinferior alternative is to use rules of thumb thatset the opportunity cost equal to some arbitraryfraction of average earnings of unskilled work-

ers.

Because the opportunitycostof attending schoolincreases with age and school achievement, thegrant amount of the grant should increase ac-cordingly. The increase should be tied to gradelevels, to provide positive incentives for per-formance. A cap on the number of allowedrepetitions might be set as an additional per-formance incentive.

Indexing the grant to inflation can preserve the

real value of the cash grant or the income-support component.

Delivery

School grants are delivered to beneficiary fami-lies conditional on childrens school attendance.Grants are usually delivered to the mother, on

the assumption that women will spend more ofthe grant on the children. The practice changesthe balance of power in the household towardthe woman. By-products, such as changes inhousehold violence, family composition, and

break-up patterns should be monitored.

Payments are delivered through the bankingsystem or the network of post offices, whereavailable, or to an alternative delivery agency.Brazils Bolsa Escola, an urban program, usesthe banking system. Rural areas, however, lackadequate coverage by financial institutions. The

pilot phase of Nicaraguas Red de ProteccinSocialalso relies on the banking system or secu-rity agencies, but the expanded program mayneed an alternative delivery mechanism. InMexicos Progresa, school grants and incomesupport components are paid to beneficiary

families by authorized agencies that specialize inmanaging cash transfers or by temporarily in-stalled localProgresaunits,Mdulo de Atencin

Progresa. The expanded Bolsa Escola program,now in the planning stage, would deliver thegrants through automated teller machines(ATM), offering poor families access to the

banking system and greater flexibility in cashingtheir grants.

Monitoring School Attendance

School grants are conditional on childrens en-rollment and minimum attendance (usually 85-90 percent of the time). Temporary noncompli-ance results in a suspension of the benefits; re-

peated noncompliance results in automatic ex-pulsion from the program.

Teachers Keep Attendance Records. Under onemodel (Bolsa Escola), teachers deliver the in-formation to principals, who inform the agencythat issues the attendance certificates to the

banking system. External supervisors run ran-

dom checks on student attendance and teachersrecording practices. In Misso Criana, a pri-vately funded variation of Bolsa Escola, chil-drens attendance is monitored by contractednongovernmental organizations (NGOs) and thelocal church. Since the scholarship constitutes asizable proportion of poor families income,


15/30

11

mothers have a strong incentive to monitor theirchildrens school attendance.

Warning. To ensure tru th in r eporting, noincenti ves should be given to teachers toincrease attendance.

The Health and Nutrition Component

This component provides monetary and limitedin-kind transfers to beneficiary families to in-crease their food consumption, diversify theirdiets, and improve their nutritional intake.Monetary transfers are conditional on an agreedschedule of medical visits and attendance at nu-tritional courses.

Design of the Health and Nutrition Package

In Progresa, PRAF II, and Red de ProteccinSocial, the health component takes a preventivecare approach. The Progresapackage includes

basic sanitation; family planning; pre-natal, as-sisted delivery, and post-natal care; vaccina-tions; childrens development care; antiparasitetreatment for the whole family; prevention andcare of chronic tuberculosis; blood pressure anddiabetes care; treatment of diarrhea and respira-tory infection; control and detection of uterinecancer; and community training in nutrition,heath care, and hygiene. Nicaraguas Red de

Proteccin Social provides a basic health carepackage of pre-natal, post-natal, and maternalcare, childrens vaccinations, and early child-hood development.

Programs establish aschedule of visitsfor moth-ers and children. Mexicos Progresa protocol

prescribes five pre-natal medical visits, two vis-its for breast feeding women, three visits in thefirst four months of life, eight visits frommonths 4 through 24, three annual visits forchildren 2-4 years old, two for children 4-16,

and one for adults 17 and older. Usually thenumber and type of visits are based on Ministryof Health protocols. Sometimes they are toocomprehensive and not cost-effective. Only in-terventions of proven cost-effectiveness should

be considered. Experiments on the effectivenessof alternative health packages can be developedat the trial stage of project implementation,

though that may not provide adequate time toobtain definitive answers.

Food supplements may be provided to addressspecific nutritional deficiencies. Progresa pro-vides food supplements to breast-feeding and

pregnant women, to children 2 years old andyounger, and to malnourished children up to 5

years old. The supplements meet 20 percent ofcaloric requirements and 100 percent of micro-nutrient requirements. Anthropometric parame-ters are monitored.

A curriculum of basic health care, hygiene, andnutritional trainingfor mothers is agreed withthe administrators of the program and the healthministry.

Different options are available to calculate theamount of the transfer for this component. The

trade-off between consumption needs and num-ber of beneficiaries covered should be assessed.Most programs use the gap between averagefood consumption of beneficiary families andthe value of a food basket that ensures minimumcaloric intake for good health. By construction,these transfers will not fully cover the consump-tion needs of the extremely indigent and soshould be adjusted upward when this group is a

priority concern. The average value of food con-sumption of beneficiary families should be com-

puted on a per capita basis. Consumption needs

can be adjusted to account for consumption re-quirements of different family members and foreconomies of scale of larger families. But thereis little empirical evidence on the numeric valueof adult equivalencies and economies of scale.Because per capita calculations will result inmore generous transfers than adult equivalentcalculations, per capita calculations will simplycover the full consumption needs of more bene-ficiary families. The total value of the transfercan then be calculated by multiplying the indi-vidual average consumption gap by the average

number of individuals of the benchmark family.

A national program may set different transferamounts by geographic or other area, such asurban and rural, with different poverty indices.The transfers may also be adjusted annually toaccount for inflation and for regional price indexchanges. These adjustments may come at the


16/30

12

cost of administrative simplicity, however, andare not necessarily recommended.Service Delivery

The health care package is delivered by publichealth posts or contracted out. In addition to ba-sic health services, providers distribute food

supplements and training in basic health care,hygiene, and nutrition. In the Red de ProteccinSocial, contracted providers are paid a fixedamount to cover the first agreed number of visitsand a variable amount for each additional visit.

Monitoring

Health providers record visits and inform theprograms executing agency. Families receive areceipt or a health care history for each child.The executing agency maintains an electronic

database of the health care history of benefic i-aries, which is accessible to health care provid-ers, and compiles a list of families that complywith program conditions and retain eligibility forcontinued benefits. A local committee is estab-lished to supervise service delivery and control-lers dispatched to random inspection of localexecuting units and providers in the field bycontrollers.

Exit Rules

Families lose eligibility if they fail to complywith program conditions, if their income risesabove the established poverty threshold, or, forgeographic targeting, if the area no longer meetsmarginality criteria. Families poverty status isreassessed every two to three years.

Warning. To avoid underreporting of in-come by famil ies trying to retain eligibil ity,the rules (proxi es and weights) used to de-termine poverty status should not be di -vulged and may be changed periodicall y.

Some poverty indicators, such as nutr i-tional status, are immune to misreporting,so moni toring the nutr iti on status of chil -dren may be a useful means of assessingfamily poverty level.

Transparency

Transparency is crucial for maintaining programcredibility and preventing corruption, politiciza-tion, and clientelism.Selection of Beneficiaries

Targeting procedures and criteria should bepublished in the official bulletin of the govern-

ment and on the government Web page. Al-though proxy-means testing rules (proxies andweights) should not be divulged to avoid misre-

porting, procedures should be subject to externalauditing. The executing agency should also in-form community leaders of the criteria used inthe selection process.

The risk of corruption and political interferencein the selection process is no greater in THDPsthan in other government programs. Survey-

based targeting systems help avoid discretionary

selection of beneficiaries. Controlling selectionat the national rather than state or local level isanother mitigating mechanism.

Accuracy and fairness require a systematic andreliable mechanismasocial comptrollershipfor community participation in reviewing andrevising selection results. While the involvementof local key informants who know the commu-nity well, such as doctors, teachers, and schooldirectors, can be helpful, care must be taken toavoid the discretionality that survey-based tar-

geting systems are designed to prevent. It is alsohelpful to inform the community at large aboutthe program and to give beneficiaries the right tochallenge decisions (Adato and others 2000).

Information for Beneficiaries

Information booklets and presentations at com-munity meetings should provide beneficiariesand nonbeneficiaries alike with complete detailsabout status, rights, and obligations related to

program benefits.

Beneficiaries and nonbeneficiaries should beinformed that eligibility is not conditional onvoting behavioror political affiliation. To ensure

privacy, they should know what informationthey are required to share and what they are not.InProgresa, beneficiaries in each locality elect acommunity promoter to help disseminate infor-


17/30

13

mation and monitor the program. The promotermeets periodically with executing agency stafffor training and exchange of information on howthe program is operating. The promoter providesa link between beneficiaries and administrators

by channeling families suggestions and ideas toimprove service provision.

Delivery of Grants

All transactions are tracked and monitored, re-ducing the risk of diversion of funds and otherresources.

School grants and income support transfers aredistributed through banks, postal offices, andcontracted security agencies.

Impact evaluation mechanisms greatly improve

transparency and accountability. Progresa,PRAF II,Red de Proteccin Social, and ApoyoFamiliarhave well-designed evaluation mecha-nisms and external evaluators . The independ-ence of the agency responsible for evaluation isessential for minimizing bias. Independenceneeds to be established from the start, with thedesign and implementation of the baseline sur-vey.

Social Monitoring

During program execution, the social controlexerted by beneficiaries, particularly mothers, isimportant. Information is crucial for motivating

participants and reducing the risk of discretion-ary use of funds.

Supply-Side Interventions

Because THDPs are designed to increase thedemand for health and education services, gov-ernments should plan for increased investmentsin these sectors. Even where schools and health

posts can cope physically with increased de-mand, the quality of service could suffer.THDPs must either include a supply-side com-

ponent or coordinate with other programs thatdo.

In 1997-2000, Mexicos National Council forEducational Development (CONAFE) organized

more than 70,000 courses for parents associa-tions and more than 210,000 for teachers to im-

prove educational activities. One-quarter of theteachers received a 20 percent bonus for theiroutstanding school assistance record and extra-curricular activities. Parents associations re-ceived financial support for classroom mainte-

nance, and some 7,232 new classrooms werebuilt. Education supply-side interventions duringthis period cost 0.3 percent of GDP, about halfthe cost of the demand-side component. On thehealth supply side, the number of doctors andnurses rose and their average pay increased by26.4 percent and 15.8 percent, respectively.Some 8,783 health units were re-equipped.

Private Sector Providers

Programs should avoid discriminating against

private sector provision. Beneficiary childrenshould be allowed to attend private schools, forexample. Even though few private providersmay be available, scholarships may encouragenew entry, thus easing supply constraints andimproving options for the poor.

Effects on Local Markets

THDPs are expected to increase the demand forfood and other goods. If the supply is inelastic,inflationary pressures in local markets might

reduce the net welfare gains of beneficiaries andhurt nonbeneficiaries as well.

Expansion to Urban Areas

Except for Bolsa Escola programs in Brazil,THDPs in Latin America have been imple-mented only in rural areas. From an administra-tive perspective, program implementation should

be easier in urban areas, because of the greateravailability of financial intermediaries. Supply-side constrains are also less severe. Targeting,

however, poses greater challenges. Becausehousehold incomes are more heterogeneous thanin rural areas and migration to participatingneighborhoods is more likely, household ratherthan geographic (community or neighborhood)targeting should be done.

Fiscal Sustainability


18/30

14

From an economic standpoint, policymakersneed to insure that the THDP investment in hu-man capital, like any other public investment,has a positive net expected value. In otherwords, the costs of the investment must at leastequal the increase in the discounted future

stream of earnings of beneficiaries relative to thecontrol group.

While net expected value is difficult to evaluatein advance, results from pilot impact evaluations

provide useful inputs for making informed deci-sions about the reasonableness of continuing orexpanding the program.

THDP targeting mechanisms should help gov-ernments improve the incidence and cost-effectiveness of social spending, enabling the

same objectives to be achieved with fewer re-sources.

To maintain macroeconomic stability, mostcountries will need to reallocate funds in orderto finance a THDP. Options range from drop-

ping ineffective programs to consolidating themenu of basic preventive health, education,early childhood, and maternal care and nutrition

programs or eliminating general subsidies.

Several factors influence costs. Where poverty is

widespread and average GDP per capita is low,implementing a THDP on a large scale is proba-bly not feasible. Costs rise with the number ofbeneficiaries and decline with the level of pov-ertythe poorer the country, the lower the op-

portunity cost and the smaller the transferneeded to induce investment in human capital.On the supply side, however, needs are greaterin poorer countries.

In low- and middle-income countries alike, thescale of the program will be determined by fi-

nancial resources. In most countries, coveragewill be smaller than the number of potentialbeneficiaries. The resources devoted to humancapital investment depend on net expected re-turns of this investment relative to that of other

public investments, and not on whether the pro-gram can achieve full coverage of the targeted

population.

The financing costs of full-scale THDPs as apercent of GDP are small in middle-incomecountries and significant in low-income coun-

tries.Progresa is financed entirely by the Mexi-can government. Its cost in 2000 was 0.2 percentof GDP, or 1.9 percent of total social expendi-tures. Only 4.4 percent were spent on admini-stration. In-kind transfersincluding medicalvisitsrepresented 18.5 percent of the total cost,and monetary transfers, 77.1 percent. The pro-gram covered 2.56 million families or about 40

percent of rural households. In addition, supply-side interventions have ranged between 0.15

percent and 0.18 percent of Mexican GDPyearly since 1997. Bolsa Escola started in Bra-

silia and has been extended to seven states andmore than 200 municipalities, benefiting800,000 children. The program is financedlargely by the federal government and executed

by local governments. Extending Bolsa Escolato 10 million children would cost about 0.36

percent of GDP, or 1.2 percent of the publicsector budget in Brazil. The federal governmenthas recently allocated US$800 million to financethe first stage of this expansion of the program.Full coverage by theRed de Proteccin Social in

Nicaragua would cost between 2 percent and 2.2

percent of GDP, or 6 percent of public expendi-ture (excluding supply-side interventions). Re-allocating resources going to food programs (2

percent of GDP) would be sufficient to coverthese costs.

Public and Private Donors

THDPs are easily adapted to incorporate externaldonor financing. The transparency and accountability

in the design and implementation of these programsand recent positive impact evaluation results shouldboost donor confidence in the programs. Providing

scholarships to poor student is not new, but doing sowithin a comprehensive framework of investing inchildren and measuring results is.


19/30

15

Impact Evaluation and Expected Results

Impact evaluation is an indispensable tool for

assessing whether a program is achieving itsobjectives. Impact evaluations at intermediatestages of a program can uncover important le s-sons for improving the effectiveness of programdesign and execution. While impact evaluationscan be time and resource intensive, the costs aresmall relative to the scale of most programs, andthe returns in increased effectiveness of socialspending and greater accountability are high.

For most THDPs, evaluation of targetingmechanisms and program impact has been care-fully designed. That enables policies to be

judged not on the basis of some speculative as-sessment but through a quantitative analysis oftheir impacts, costs, and benefits. While evalua-tion is costly, it also produces tremendous effi-ciency gains by guiding governments and do-nors decisions on future resource allocations.

IMPACT EVALUATION DESIGN

Selecting the Design

Impact evaluations compare a treatment group(program beneficiaries) with a control group(nonbeneficiaries), using either an experimentalor a quasi-experimental design. Experimentaldesign is the most robust. Program beneficiariesand nonbeneficiaries are selected randomlywithin the target population. Randomization en-sures that there are no systematic differences inthe observed characteristics between program

participants and the control group. Data on bothgroups must be collected before (baseline sur-vey) and after (follow-up surveys) the program.The impact of the intervention is assessed by

subtracting the mean outcomes of the treatmentgroup from the mean outcomes of the controlgroup before and after the intervention. Accurateimpact evaluation also uses multivariate regres-sion techniques to control for individual observ-able and unobservable characteristics.

Randomizing beneficiariesis feasible and offers

an ethically sound basis for proceeding (all tar-geted individuals have the same probability ofbeing selected) when budget constraints requirerationing of program benefits. Even when the

programs are national in scale and aim for 100percent coverage, institutional capacity andtesting often require that coverage be expandedgradually. Phased coverage provides an oppor-tunity for random assignment of targeted indi-viduals to control and treatment groups. Indi-viduals assigned to control groups at an earlystage of program implementation become bene-ficiaries at a later stage. Optimally (if often po-litically unfeasible), control groups should not

be aware of future benefits, to avoid affectingtheir expectations and behavior.

Progresa in Mexico used an experimental de-sign impact evaluation system. Communitiesrather than individuals were randomized toavoid social tensions within communities. As aresult, all groups of individuals of interest maynot have been randomly assigned because of

possibly uneven distribution of groups acrosscommunities.

Experimental methods can also be used toevaluate program components and synergiesamong components. Evaluating components re-quires the development of multiple beneficiarygroups, each with access to different benefit

packages. This greatly increases the insights thatmay be gained from evaluation. PRAF in Hon-duras uses an impact evaluation system that as-sesses supply and demand interventions sepa-rately and jointly.

Quasi-experimental methodsare used when pro-grams were developed before an evaluationsystem was put in place. Methods range fromcomparing beneficiaries before and after the

program (reflexive comparison), to comparingbeneficiaries and nonbeneficiaries by either arti-ficially creating a control group that resembles


20/30

16

the group of beneficiaries (matching each bene-ficiary with one or more observationallyequivalent nonbeneficiaries), or using regressionmethods based on instrumental variables. Thesemethods suffer from selection bias and difficultycontrolling for external factors that have affectedthe changes in monitored outcomes. Care should

thus be taken in interpreting results.

Setting Up an Impact Evaluation System

Some of the steps to follow in setting up an im-pact evaluation systeminclude the establishmentof an evaluation team, assessments of targetingefficiency and outcomes, quantitative indicators,qualitative analysis, general equilibrium analysisand tests of different service packages.

An evaluation teamshould be involved from the

early stage of the project to support identifica-tion and design by

helping to develop a consistent and moni-torable logical framework, including ob-

jectives, inputs, outputs, timing, perform-ance indicators, risks, and underlying as-sumptions (see logical framework annex);

supporting the development and imple-mentation of the targeting methodology;

working with the statistical office to en-sure that the baseline and follow-up sur-veys appropriately sample the populationand include all relevant questions for pro-gram evaluation;

supervising data collection, training andcontracting of the interviewers, and qual-ity control; and

supervising lottery selection of familiesinto control and treatment groups to en-

sure transparency.

The evaluation system should be able to assesstargeting efficiency and short- to long-term out-comes. Additionally, it might assess the opera-tional efficiency of the program. Most of theinstruments and data used for evaluation becomeimportant operational tools for monitoring pro-

gram implementation. The final outcome of animpact evaluation should be a clear analysis ofthe cost-effectiveness of the program, which isindispensable for assessing its sustainability.

Program administrators and evaluation teammembers should work together to develop a list

of quantitative indicators divided into short-term(1-2 years), medium-term (2-3 years), and long-term (3 years or more) measures. The indicatorscan be defined separately for any group of inter-est (gender, age, and ethnic background). Thelist should be tailored to the specific program(see logical framework annex).

Qualitative analysis can complement the quan-titative analysis to assess difficult to measureoutcomes, such as THDP impact from the me-dium to the long term on the empowerment of

some member of the family relative to others,intra-household violence, and community cohe-sion.

A full-scale program with national coveragemight include analysis of the general equilib-rium effects of the program (for example,change in returns to education). This type ofanalysis is being carried out for Progresa, withits extended national coverage.

During a pilot phase, administrators may want to

test different service packages, for example, tomeasure the relative effectiveness of educationand health interventions, the transfer amountrequired to achieve a certain increase in enroll-ment, or the relative effectiveness of supply anddemand interventions. The costs of impactevaluation, however, will increase with eachexperiment.

EXPECTED RESULTS

Evaluating Targeting Efficiency

Evaluation results of the targeting mechanismforProgresa in Mexico and preliminary resultsforPRAF IIin Hondurasand Red de ProteccinSocialin Nicaragua are available. A preliminarycomparison of their targeting efficiency suggeststhat all three succeed in targeting extremely poorhouseholds with children subject to a high prob-


21/30

17

ability of stunting (IFPRI 2000). The Red deProteccin Social in Nicaragua is most effectiveat targeting the most vulnerable, drawing 55

percent of its beneficiaries from the bottom in-come quintile. Progresa draws 40 percent and

PRAF Idraws 43 percent.

Progresa uses a two stage targeting mechanism.In the first stage, localities are selected throughgeographic targeting based on a marginality in-dex. The method is very successful in selectinglocalities with the highest concentration of poorfamiliesbetter even than a theoretically supe-rior method that selects localities using an index

based on families consumption expenditures.There is almost no undercoverage, but leakage isquite high. In a second targeting stage, familiesare selected by household income and other so-cioeconomic characteristics within the selected

localities, using discriminant analysis. At thisstage, targeting works better at identifying ex-tremely poor households but is more likely tofail to identify households that are moderately

poor. Considering the targeting costs, the two-stage procedure is viewed as more efficient forreducing the incidence and depth of poverty thansimple geographic targeting.

PRAF II rejected means test targeting. Benefitleakage of 40 percent and undercoverage of 15

percent were not thought to justify the high ad-

ministrative costs of household targeting.

TheRed de Proteccin Socialused geographictargeting based on a marginality index. Withinselected localities (comarcas), virtually allhouseholds (97.5 percent) are eligible to receivethe transfer except households that report own-ing a vehicle or a farm of more than 14 hectares.

Evaluating Impact

Results are available only for Progresa. Most

results reported here reflect impact two yearsinto program implementation (a few cover onlyeight months) using intermediate indicators,such as enrollment rates or visits to health cen-ters, rather than final indicators, such as years ofschooling or child mortality rates (IFPRI 2000).

In two years, the share of households living inpoverty (headcount) fell by 8 percent in Pro-gresa localities. Measures of poverty depth andseverity improved even more, implying that ex-treme poverty was greatly reduced. The povertygap dropped by 30 percent and the severity of

poverty index by 45 percent. These are the direct

impacts of Progresa and do not include in-creased future earnings of beneficiary families.

Progresaincreased enrollment ratesfor grades 3to 6 by 0.7 to 1.1 percentage points for boys and1.0 to 1.5 percentage points for girls, dependingon methodology used. For grades 7 to 9, enroll-ment increased by 3.5 to 5.8 percentage pointsfor boys and 7.2 to 9.3 percentage points forgirls. These figures indicate that Progresareached about a third of children not previouslyenrolled. Average years of schooling increased

by 0.66 year, which will translate into an esti-mated 8 percent increase in permanent futureearnings for these children. School attendancestill falls during the harvest season, particularlyfor boys and for children whose work force par-ticipation cannot easily be substituted for by in-creasing the labor supply of adults in the house-hold.

Preventive health care visit rates grew faster inProgresa localities than in control villages, witha significant increase in nutrition monitoring

visits, immunization rates, and pre-natal visits(Gertler 2000). The number of pre-natal visits inthe first three months of pregnancy increased by8 percent.

Two years into the program, 0-5 year old chil-dren in Progresa localities experienced 12 per-cent fewer incidences of disease than children incontrol villages. Among adults, Progresa bene-ficiaries had 19 percent fewer days of illnessthan nonbeneficiaries. As a result, the number ofin-patient hospitalizations was lower in Pro-

gresa localities.

Receiving the nutrition supplement regularlywas estimated to boost the annual mean growthof children ages 12 to 36 months by 16 percent(or 1 centimeter) and to increase childrensheight by about 1.2 percent. This is a potentially


22/30

18

important effect, which could increase lifetimeearnings by between 1.4 and 2.9 percent.

Eight months into the program,Progresa fami-lies had increased their consumption of milk andmilk derivatives by 33 percent; bread, 32 per-cent; meat, 24 percent, and fruits and vegetables,

19 percent relative to consumption in the controlgroup. After two years, average food consump-tion and caloric intake increased by about 10

percent in Progresa families compared with thecontrol group.

Eight months into the program,Progresa fami-lies had increased their spending relative to thecontrol group by 58 percent on childrens cloth-ing and 39 percent on shoes. Spending on chil-drens clothing and shoes also rose by a statisti-cally significant amount as a share of total fam-

ily expenditures. Consumption of tobacco andalcohol did not increase relative to the controlgroup.

Progresareduced boys remuneratedlabor forceparticipationby 25 percent in the age group 12-13 years. No statistically significant effects arerecorded for other age groups. Progresa doesnot appear to create negative incentives for adultlabor supply. Data show no reduction in laborforce participation rates for men or women.

Progresa increases womens decisionmakingrole in the family, particularly on decisions af-fecting children. Women report that they aremore likely to speak to other mothers abouthouse-related issues, are more comfortablespeaking out in groups, have become more edu-cated through health and nutrition training, andhave more control over household expenditures.

Progresa beneficiaries have organized them-selves to carry out new activities. For example,groups of women rented a vehicle to facilitate

collecting their transfers. A new range of small

productive activities has also emerged in Pro-gresalocalities.Affecting Institutional Structures and Social

Sector Policies

THDPs are expected to influence institutionalstructures and social sector policies and expen-

ditures. Adoption of an integrated approach in-creases coordination among line ministries(education, health, and other social sectors) andthe program executing agency. This coordina-tion is beneficial to social policy formulationand consistency. If a THDP substitutes for pro-grams shown by analysis to be ineffective, im-

plementation of a THDP can help to develop aunified framework for targeted interventions.

In many countries, particularly in Latin Americaand the Caribbean, the geographic allocation of

supposedly targeted social sector programs doesnot match the regional distribution of poverty.THDPs, with their robust targeting, facilitate themore efficient allocation of targeted public ex-

penditures across regions (urban and rural) andmunicipalities. Consider Mexico. Between 1994and 2000, the regional distribution of povertychanged little: the rural poor made up 60 percentof the total (Levy 2000). In 1994, before Pro-

gresastarted, only 39 percent of food subsidieswere targeted, and 31.4 percent of the subsidieswent to rural areas. With Progresa the situation

changed dramatically. Now more than 95 per-cent of food subsidies are targeted, and 76.4 per-cent of food-related transfers go to rural areas.The total amount of resources disbursed re-mained practically unchanged.

The integrated approach of THDPs stimulatescoordination among international lenders anddonorsmultilateral, bilateral, and NGOs. Thismay result in substantial efficiency gains, par-ticularly in countries and sectors where govern-ment and donor interventions overlap on both

the demand and supply sides.


23/30

19

Country Conditions

Country conditions should be analyzed to de-

termine whether a THDP could adequately re-spond to the particular situation. Country-specific conditions include the nature and distri-

bution of poverty and the nations institutionalcapacity and local organization.

NATURE AND DISTRIBUTION

OF POVERTY

Pockets of Poverty and Other Appropriate

Conditions

Given the targeted nature of THDPs, the pres-ence of well-confined pockets of poverty willhelp implementation. Other appropriate condi-tions include the existence of low-income fami-lies with low education levels, high labor earn-ings to total income ratio, little or no asset own-ership, high fertility rates, high incidence ofmalnutrition, delayed school enrollment, lowschool attendance, and high dropout rates. Inthese families, children often contribute to thefamilys monetary and non-monetary income.The presence of segments of the populationwhose demand for education, health, and nutri-tion is constrained by lack of resources is alsoan appropriate condition for the implementationof targeted human development programs.

TARGETING

Widespread Poverty

Widespread poverty within a country should notpreclude a THDP option. Critics argue that whenpoverty is widespread, there are no obvious cri-teria for selecting program beneficiaries. By the

same argument any intervention (road or school

construction, water and sanitation programs) thatfails to reach all the poor in a country should bequestioned as well. Poverty reduction policiescan be successfully implemented even whencoverage is limited.

A legitimate concern is where to begin a pro-gram,given available resources. If the objectiveis to reduce the depth and severity of poverty,

programs should target the poorest of the poor.If the objective is to reduce disparities betweengenders or ethnical groups, programs shouldtarget the disadvantaged group. A prerequisite isthe availability of good quality datacensusdata and national household expenditure andincome surveys. Developing targeting mecha-nisms that are technically robust and cost-effective becomes a priority.

Institutional Capacity

and Local Organization

THDP implementation requires a certain amountof institutional and administrative capacity andlocal organization. Institutional and administra-tive capacity is needed to design, implement,and maintain appropriate targeting, training,monitoring, and evaluation procedures.

4Local

organization, with the active involvement ofrepresentatives of beneficiary and non-

beneficiary families (school councils, localcommittees) is necessary to promote house-holds active participation and ownership of the

program. That kind of participation is essentialfor achieving permanent behavioral change,strengthening social control and ensuring trans-

parency and monitoring.

4Progresa in Mexico andPRAF-IIin Honduras havealready built this capacity. Other countries are build-

ing it, for example Nicaragua.


24/30

20

Bibliography

Targeted Human Development Programs

Adato, Michelle, David Coady and Marie Ruel. 2000. An Operations Evaluation ofProgresafromthe Perspective of Beneficiaries, Promotoras, School Directors, and Health Staff. Washington,D.C.: International Food Policy Research Institute.

Behrman, Jere and John Hoddinott. 2000.An Evaluation of the Impact ofPROGRESAon Pre-schoolChild Height. Washington, D.C.: International Food Policy Research Institute.

Behrman, Jere R., Piyali Sengupta and Petra Todd. 2000. The Impact ofPROGRESAon AchievementTest Scores in the First Year. Washington, D.C.: International Food Policy Research Institute.

Behrman, Jere R., Piyali Sengupta and Petra Todd. 2000. The Impact ofPROGRESA on SchoolingTransitions: A Markovian Approach. Washington, D.C.: International Food Policy ResearchInstitute.

Coady, David. 2000. The Application of Social Cost-Benefit Analysis to the Evaluation ofProgresa.Washington, D.C.: International Food Policy Research Institute.

Gertler, Paul J. 2000. Final Report: The Impact ofPROGRESA on Health. Washington, D.C.: Inter-national Food Policy Research Institute.

Glassman, Amanda. 2000. The New Social Protection Programs. Washington, D.C.: Inter-American Development Bank (RE3/SO3).

Inter-American Development Bank. 2000. Social Protection for Equity and Growth . Sustainable De-velopment Department, Poverty and Inequality Advisory Unit. Washington, D.C.: IDB.

International Food Policy Research Institute. 2000. Is Progress Working? Summary of the Results ofan Evaluation by IFPRI. Washington, D.C.: International Food Policy Research Institute.

International Food Policy Research Institute. 1998. Programa Nacional de Educacin, Salud y Ali-mentacin (PROGRESA): A Proposal for Evaluation. Washington, D.C.: International FoodPolicy Research Institute.

Levy, Santiago. 2000. Algunos aspectos del programa de educacin, salud y alimentacinPRO-GRESA. Presentation to the IDB Board of Executive Directors.Mexico, D.F.: Secretara de

Hacienda y Crdito Pblico.

Schultz, T. Paul. 1999. Preliminary Evidence of the Impact of Progresa on School Enrollments from1997 and 1998. Washington, D.C.: International Food Policy Research Institute.

Emmanuel Skoufias. 2000. IsProgresaWorking? Summary of the Results of an Evaluation by IF-PRI. Washington, D.C.: International Food Polciy Research Institute.


25/30

21

Sedlacek, Guilherme, Nadeem Ilahi and Emily-Gustafsson-Wright. 2000. Targeted ConditionalTransfers Programs in Latin America: An Early Survey. Washington, D.C.: The World Bank.

Sedlacek, Guilherme. 1999. An Assessment of theBolsa Escola Program in Brazil. Washington,D.C.: The World Bank.

Targeting

Judy L. Baker and Margaret E. Grosh. 1993. Measuring the Effects of Geographic Targeting on Pov-erty Reduction. Living Standards Measurement Study, Working Paper No. 99. Washington,D.C.: The World Bank.

Gomez de Len, Jos. 1998. Correlative Dimensions of Poverty in Mexico: Elements for TargetingSocial Programs. Paper presented at the First Annual Meeting of the LACEA/IDB/WorldBank Network on Inequality and Poverty. October. Buenos Aires, Argentina.

Legovini, Arianna. 1999. Targeting Methods for Social Programs.Poverty & Inequality TechnicalNote 1.Washington D.C.: Inter-American Development Bank.

Margaret E. Grosh and Judy L. Baker. 1995. Proxy Means Tests: Simulations and Speculation forSocial Programs. Living Standards Measurement Study, Working Paper No. 118. WashingtonD.C. The World Bank

Impact Evaluation

Behrman, Jere R. and Petra E. Todd. 1999. Randomness in the Experimental Samples ofPROGRESA(Education, Health and Nutrition Program). Research report prepared for IFPRI (mimeo).Philadelphia, PA: University of Pennsylvania.

Behrman, Jere R. and Petra E. Todd. 1999. An Evaluation of the Sample Sizes for Evaluation ofPROGRESA (Education, Health and Nutrition Program) of Mexico. Research report prepared

for IFPRI (mimeo). Philadelphia, PA: University of Pennsylvania.

Ezemenari, Kene, Andres Rudqvist and K. Subbarao. 1999. Impact Evaluation: A Note on Conceptsand Methods. Washington, D.C.: World Bank Institute.

Heckman, James J. and Jeffrey A. Smith. 1999. The Pre-Program Earnings Dip and the Determinantsof Participation in a Social Program: Implications for Simple Program Evaluation Strategies.

NBER Working PaperNo. W6983.

Heckman, James, Hidehiko Ichimura and Petra Todd. 1996. Sources of Selection Bias in EvaluationSocial Programs: An Interpretation of Conventional Measures and Evidence on the Effective-ness of Matching as a Program Evaluation Method.Proceedings National Academic Science,

Vol. 93, pp 13416-13420.

Heckman, James J., Jeffrey Smith and Nancy Clements. 1997. Making the Most Out of ProgrammeEvaluations and Social Experiments: Accounting for Heterogeneity in Programme Impacts.

Review of Economic Studies. Vol. 64 (October), pp. 487-535.


26/30

22

Heckman, James J., Hidehiko Ichimura and Petra Todd. 1997. Matching as an Econometric Evalua-tion Estimator.Review of Economic Studies, 65, 261-294.

Hicks, Norman. 1998. Poverty Impact of Projects: A Note on Methodology. Washington, D.C. TheWorld Bank.

. 1998. Measuring the Poverty Impact of Projects in Latin American Countries. Washington,

D.C.: The World Bank.

IFPRI/PROGRESA. 1999. An Evaluation of the Selection of Beneficiary Households in the Educa-tion, Health and Nutrition Program (PROGRESA) of Mexico. International Food Policy Re-search Institute, Washington, D.C.

Lipsey, Mark W. 2000. The Basics of Program Evaluation, Step by Step. Nashville, TN: VanderbiltUniversity.

Newman, John, Laura Rawlings and Paul Gertler. 1994. Using Randomized Control Designs inEvaluating Social Sector Programs in Developing Countries.The World Bank Research Ob-

server, vol. 9, no 2, pp 181-201. Washington, D.C.

Ravallion, Martin and Quentin Wodon. 1998. Evaluating a Targeted Social Program When Placementis Decentralized. Policy Research Working Paper 1945. Washington, D.C.: The World Bank.

Regalia, Ferdinando. 2000. Impact Evaluation Methods For Social Programs.Poverty & InequalityTechnical Note 2. Washington, D.C.: Inter-American Development Bank.

Skoufias, Emmanuel. On the Design and Implementation of Impact Evaluation Projects. IFPRIs ex-periences fromPROGRESA. Washington, D.C.: International Food Policy Research Institute.

Todd, Petra, Jere R. Behrman, and YingmeiCheng. 2000. Evaluating Preschool Programs whenLength of Exposure to the Program Varies: A Nonparametric Approach. Philadelphia, PA:

University of Pennsylvania, mimeo.

Other

Alderman, Harold, Jere R. Behrman, Victor Lavy, and Rekha Menon. 2001 (forthcoming). Child Nu-trition, Child Health, and School Enrollment: A Longitudinal Analysis.Journal of Human Re-

sources.

Behrman, Jere R. 1996. The Impact of Health and Nutrition on Education. World Bank Research Ob-server 11:23-37.

Bhalotra, Sonia. 2000. Is Child Work Necessary? Department of Economic. Cambridge: University of

Cambridge.

Glewwe, Paul, Hanan Jacoby, and Elizabeth King. 1998. Early Childhood Nutrition and AcademicAchievement: A Longitudinal Analysis. World Bank, Development Research Group, Wash-ington, D.C.


27/30

23

Glewwe, Paul, Hanan Jacoby 1995. An Economic Analysis of Delayed Primary School Enrollmentand Child Malnutrition in Low Income Countries.Review of Economics and Statistics 77(1):156-69.

Glewwe, Paul and Elizabeth King. 2000. The Impact of Early Childhood Nutrition Status on Cogni-tive Achievement: Does the Timing of Malnutrition Matter? World Bank, Development Re-search Group, Washington, D. C.

Handa, Sudhanshu 1999. Raising Primary School Enrollment in Developing Countries: The relativeimportance of Supply and Demand. International Food Policy Research Institute, Discussion

paper N. 76. Washington, D.C.

Jacoby, Hanan and Emmanuel Skoufias. 1997. Risk, Financial Markets and Human Capital in a De-veloping Country.Review of Economic Studies64, 311-335.

Legovini, Arianna and Nora Lustig. 2000. Low Investment in Human Capital and the Cycle of Pov-erty.Politica Internazionale. April.

Pollitt, E., K. S. Gorman, P. L. Engle, R. Martorell and J. Rivera. 1993. Early Supplementary Feeding

and Cognition.Monograph of the Society for Research in Child Development, 235 (58):7.

Ray, Ranjan 2000. Child Labor, Child Schooling and Their Interaction with Adult Labor: EmpiricalEvidence for Peru and Pakistan. World Bank Economic Review Vol. 13 N. 3


28/30

24

Annex Logical Framework for a Targeted Human DevelopmentProgram

Narrative Summary Key Performance Indica-tors5

Monitoring and Supervision Critical Assumptions andRisks

Overarching Objective

Reduce Poverty Incidence, depth and severity

of consumption poverty

Survey data

Poverty assessment

THDP Development Objec-

tives

Increase the human capital andearnings capacity of poorfamilies.

Education levelIlliteracy ratesLife expectancy

Survey data

(THDP Development Obj.s to

Overarching Objective)

Macroeconomic growth andstabilityMedium-term fiscal commit-

ment

Project Outputs (Outputs to Development Ob-

jectives)1. Increase in average

schooling and school per-

formance

1.1 Net enrollment rates1.2 Attendance rates

1.3 Average years ofschooling

1.4 Rate of functional liter-acy

1.5 Childrens average timeallocated to work leisureand study

1.6 Repetition and dropoutrates

1.7 Standardized test scores

1.1 Childrens school atten-dance record by teachers.

1.2 Attendance certificates1.3 External supervisors

random checks records

1.4 Test scores results1.5 Executing agencys bene-

ficiary database1.6 Survey data1.7 Beneficiary database1.8 Impact evaluation teams

reports

Supply side interventions:building of schools, ensuring

quality of education services,ensuring teachers presence inthe classroom

Coordination between lineministries

2. Improvement in health and

nutritional status

2.1 Maternal and infantmortality rates

2.2 Morbidity and incidenceof child malnutrition

2.3 Number of medical pre-natal visit in the firstthree months of preg-nancy

2.4 Number of birth assistedby qualified personnel

2.5 Number of control visitsfor growth and develop-

ment for infant and chil-dren

2.6 Vaccination coverage2.7 Incidence of malnutrition

2.1 Providers records2.2 Medical records2.3 Beneficiary database2.4 Survey data2.5 Anthropomorphic data2.6 Impact evaluation teams

reports

Supply side interventions:

availability of health posts,presence of medical staff,availability of drugs and vac-cines, quality of service

5Baseline and targeted values should be shown, with the latter divided into values expected at mid-term, end of project and

full impact.


29/30

25

Narrative Summary Key Performance Indicators Monitoring and Supervision Critical Assumptions and

Risks

3.1 Improve consumptionlevels and patterns

3.1 Share of food expendi-ture on total familys ex-

penditures

3.2 Changes in the consump-tion of milk, meat, bread,

fruit, vegetables and ba-sic staples

3.3 Number of productswhich enter familys con-sumption basket

3.4 The share of cloth, shoeand school util expendi-tures for kids relative tototal family expenditures

3.5 Familys consumption oftobacco and alcohol

3.6 Changes in the familypattern of expenditures

and asset accumulation3.7 Private transfers received

3.1 Survey data3.2 Impact evaluation teams

reports

4.1 Ensure targeting efficiency 4.1 Undercoverage and leak-age rates

4.2 Cost-benefit of differenttypes of targeting

4.1 Survey data4.2 Impact evaluation teams

reports

Project Components

1. Education component

Input:

$ __ million 1.1 Banking check issuancerecords

1.2 Executing units recordsand audits

(Components to Outputs)

Efficient targeting

Coordination between minis-triesTransparency

2. Health and Nutrition Com-

ponent

$ __ million 2.1 Beneficiary database

2.1 Providers contracts anddisbursements

2.2 Executing units recordsand audits

3. Impact Evaluation Compo-nent

$ __ million 3.1 Bidding documentation3.2 Impact evaluation contract


30/30

Contact Information

Br azil (Bol sa Escola)

Cristovam Buarque, Misso Criana, Brazil (email:[email protected])Alvaro Cubillo, IDB

Guilherme Sedlacheck, World Bank

Colombia (Apoyo F amili ar)

Pablo Adam, IDBMauricio Crdenas, former Director of the National Planning Department, ColombiaJesus Duarte, IDB

Amanda Glassman, IDBMaria Teresa LaFourie, Director, Project Executing Unit, Department of the Presidency, ColombiaManuel Salazar, Project Director, National Department of Planning, ColombiaMiguel Szkely, IDB

Honduras (Programa de Asignaciones Famil iar es (PRAF ))Carola Alvarez, IDBRafael Flores, IFPRI

Mexico (Progresa)Jere Berhman, University of Pennsylvania

Daniel Hernandez, former Director, Progresa, MexicoSantiago Levy, MexicoImmanuel Skoufias, IFPRI

Petra Todd, University of Pennsylvania

Ni caragua (Red de Proteccin Social )Carlos Lacayo, Technical Director, Emergency Social Investment Fund (FISE), e-mail:[email protected] Monin, IDBFerdinando Regalia, IDBMaria Eugenia Zavala, IDB

Documents

Targeted Human Development Programs_ Investing in the Next Generation